Citation
Data Collection Needs for Work Zone Incidents

Material Information

Title:
Data Collection Needs for Work Zone Incidents
Creator:
CARRICK, GRADY THOMAS
Copyright Date:
2008

Subjects

Subjects / Keywords:
Data collection ( jstor )
Databases ( jstor )
Focus groups ( jstor )
Law enforcement ( jstor )
Motor vehicle traffic ( jstor )
Police ( jstor )
Research methods ( jstor )
Roads ( jstor )
Traffic control ( jstor )
Transportation ( jstor )
Flagler County ( local )

Record Information

Source Institution:
University of Florida
Holding Location:
University of Florida
Rights Management:
Copyright Grady Thomas Carrick. Permission granted to University of Florida to digitize and display this item for non-profit research and educational purposes. Any reuse of this item in excess of fair use or other copyright exemptions requires permission of the copyright holder.
Embargo Date:
8/31/2006
Resource Identifier:
649810204 ( OCLC )

Downloads

This item has the following downloads:

carrick_g ( .pdf )

carrick_g_Page_093.txt

carrick_g_Page_095.txt

carrick_g_Page_038.txt

carrick_g_Page_046.txt

carrick_g_Page_080.txt

carrick_g_Page_028.txt

carrick_g_Page_042.txt

carrick_g_Page_040.txt

carrick_g_Page_088.txt

carrick_g_Page_045.txt

carrick_g_Page_096.txt

carrick_g_Page_025.txt

carrick_g_Page_082.txt

carrick_g_Page_090.txt

carrick_g_Page_035.txt

carrick_g_Page_065.txt

carrick_g_Page_084.txt

carrick_g_Page_054.txt

carrick_g_Page_059.txt

carrick_g_Page_017.txt

carrick_g_Page_066.txt

carrick_g_Page_049.txt

carrick_g_Page_024.txt

carrick_g_Page_051.txt

carrick_g_Page_052.txt

carrick_g_Page_037.txt

carrick_g_Page_013.txt

carrick_g_Page_101.txt

carrick_g_Page_018.txt

carrick_g_Page_073.txt

carrick_g_Page_014.txt

carrick_g_pdf.txt

carrick_g_Page_015.txt

carrick_g_Page_043.txt

carrick_g_Page_070.txt

carrick_g_Page_079.txt

carrick_g_Page_057.txt

carrick_g_Page_064.txt

carrick_g_Page_030.txt

carrick_g_Page_100.txt

carrick_g_Page_019.txt

carrick_g_Page_008.txt

carrick_g_Page_007.txt

carrick_g_Page_108.txt

carrick_g_Page_069.txt

carrick_g_Page_011.txt

carrick_g_Page_060.txt

carrick_g_Page_063.txt

carrick_g_Page_098.txt

carrick_g_Page_010.txt

carrick_g_Page_012.txt

carrick_g_Page_056.txt

carrick_g_Page_103.txt

carrick_g_Page_074.txt

carrick_g_Page_083.txt

carrick_g_Page_005.txt

carrick_g_Page_020.txt

carrick_g_Page_021.txt

carrick_g_Page_036.txt

carrick_g_Page_023.txt

carrick_g_Page_032.txt

carrick_g_Page_078.txt

carrick_g_Page_026.txt

carrick_g_Page_003.txt

carrick_g_Page_039.txt

carrick_g_Page_102.txt

carrick_g_Page_097.txt

carrick_g_Page_105.txt

carrick_g_Page_071.txt

carrick_g_Page_077.txt

carrick_g_Page_061.txt

carrick_g_Page_048.txt

carrick_g_Page_092.txt

carrick_g_Page_022.txt

carrick_g_Page_062.txt

carrick_g_Page_086.txt

carrick_g_Page_107.txt

carrick_g_Page_068.txt

carrick_g_Page_094.txt

carrick_g_Page_001.txt

carrick_g_Page_055.txt

carrick_g_Page_004.txt

carrick_g_Page_081.txt

carrick_g_Page_104.txt

carrick_g_Page_089.txt

carrick_g_Page_075.txt

carrick_g_Page_002.txt

carrick_g_Page_033.txt

carrick_g_Page_091.txt

carrick_g_Page_041.txt

carrick_g_Page_029.txt

carrick_g_Page_031.txt

carrick_g_Page_034.txt

carrick_g_Page_006.txt

carrick_g_Page_099.txt

carrick_g_Page_044.txt

carrick_g_Page_050.txt

carrick_g_Page_053.txt

carrick_g_Page_087.txt

carrick_g_Page_076.txt

carrick_g_Page_058.txt

carrick_g_Page_047.txt

carrick_g_Page_106.txt

carrick_g_Page_009.txt

carrick_g_Page_027.txt

carrick_g_Page_067.txt

carrick_g_Page_072.txt

carrick_g_Page_085.txt

carrick_g_Page_016.txt


Full Text












DATA COLLECTION NEEDS FOR WORK ZONE INCIDENTS


By

GRADY THOMAS CARRICK













A THESIS PRESENTED TO THE GRADUATE SCHOOL
OF THE UNIVERSITY OF FLORIDA IN PARTIAL FULFILLMENT
OF THE REQUIREMENTS FOR THE DEGREE OF
MASTER OF SCIENCE

UNIVERSITY OF FLORIDA


2006

































Copyright 2006

by

Grady Carrick














DEDICATION

I thank my mother for giving me the opportunity and encouragement to study as a

youth, and my wife for providing me positive motivation. My friend and colleague,

Chris Knight, deserves my gratitude for providing the professional support that is needed

to balance work and study.















ACKNOWLEDGMENTS

I thank Dr. Scott Washburn for his support and encouragement in my graduate

studies and in this project. Appreciation is given to Florida Highway Patrol Major Steve

Williams for assisting me with the database development part of this project. Without his

expertise, the supplemental data collection system could not have been possible. Many

thanks to the Florida Highway Patrol and Colonel Christopher Knight for providing the

institutional access and support that made this project possible. Finally, appreciation is

given to the STC for funding this project.
















TABLE OF CONTENTS

page

A C K N O W L E D G M E N T S ................................................................................................. iv

LIST OF TABLES .......... .. ... ....... ...................... .. ...................... vii

LIST O F FIG U R E S ......................................................... ......... .. ............. viii

ABSTRACT .............. .......................................... ix

CHAPTER

1 IN TR OD U CTION ............................................... .. ......................... ..

B ack g rou n d ...................................... .............................. .... ......... ...... .
Problem Statem ent................................................... ............... ... .... ..............
Research Objective and Supporting Tasks ........................................................4
D ocum ent O organization ................................................................ ....................... 5

2 LITERATURE REVIEW ........................................................................6

Current Crash R reporting System s ........................................ .......................... 6
W ork Zone Data Collection-Related Studies .................................. ............... 13

3 RESEARCH APPROACH ............................................................ ............... 17

O v e rv iew .........................................................................................1 7
M methodology ...................................................................................... ........... ...............17
Traditional Crash Report Content Decisions ..................................................19
Data M ining Using Qualitative M ethods ................................. ................ 20
Focus Groups In Qualitative Research.....................................................21
G roup com position .............................................. ..... ....... ............... 2 1
A dm inistrative preparation...................................... ......................... 23
M oderator and content preparation ................................... .................25
Adm inistration/conduct of m meetings ................................. ...... ............ ...26
Post-Session Processing ............................... ..................27
Data Element Identification-Qualitative Analysis.......................................28
Data Element Definition-Interpretation of Results...................................30
Data Element Testing-Supplemental Collection System..............................31
Technical framework of collection system ...............................................33



v









Operational framework of collection system ............................................34
D ata Elem ent V alidation .............................................................................. 36

4 RE SU LTS AN D AN ALY SIS............................................................................... 37

Qualitative Analysis Data Element Identification............................................37
A Priori Categorization..................... ......... ............................ 39
E m ergent C ategorization .......................................................... ... .............40
F inal C ategorization ............................................ .. ........ .... ...........40
A greem ent M measure .............................................. .... .... .. ............ 43
W eighted M measure of Intensity ................................. ............... ..................46
C om posite R anking of Item s ................................................................... ... ..48
Interpreting Analysis Definition of Elements .............................. ................49
Creating Linkages ..................................... ................................ 50
Converting Codes to Data Elements ............ ............................................. 54
Elem ents as Interrogatives /Binary Values.................................. ............... 55
Supplemental Collection System Element Testing.....................................59
Technical Framework of Collection System ..................... .................. .......... 59
Field Implementation and Testing of Collection System.............. .....................66
R results of C collection ..................... .. .... ..................... .... .. ........... 67
V alidation of E lem ents .............................................................. .... .. .... ........ 67

5 CONCLUSIONS AND RECOMMENDATIONS............................................. 71

C o n c lu sio n s.................................................... .................. 7 1
R ecom m endations....... ............ .............................. .. .. .. ...... .. ............72
Future W ork Zone Applications....................................... ................................. 73
O their A pplications........... .......................................................... .... ..74

APPENDIX

A OVERVIEW OF QUALITATIVE RESEARCH METHODS........................75

B FOCUS GROUP ADMINISTRATIVE FORMS ............... .................. ............85

C FOCUS GROUP MODERATOR GUIDE ..............................................................90

L IST O F R E F E R E N C E S ........................................................................ .....................94

B IO G R A PH IC A L SK E TCH ..................................................................... ..................98
















LIST OF TABLES

Table p

1 MMUCC Work Zone Elements and Attributes...........................................8

2 State Report Form Work Zone Methodologies..................................................9

3 Focus G roup C om position ............................................... ............................ 24

4 Focus Group M oderator Questions ........................................ ....... ............... 26

5 Unique Speakers by Group and Code ........................................... ............... 42

6 Agreement by Group and Code (Binary) ...................................... ............... 44

7 W eighted M measure of Intensity ................................. ............................................ 47

8 Comparison of Intensity and Agreement Measures ..............................................48

9 C om posite C odes........... ..... ............................................................ ......... ....... 49

10 D ata L inkage .........................................................................55

11 Converting Elements to Supplemental Report Questions .............. ...................58
















LIST OF FIGURES

Figure pge

1 Florida Traffic Crash Report .............. ........... .................. 10

2 W isconsin M otor Vehicle Accident Report .......... ........... .. ....... .......... 10

3 Commonwealth of Pennsylvania Police Crash Reporting Form............................ 11

4 South Carolina Traffic Collision Report Form ............................ ..................... 11

5 D ata Elem ent D evelopm ent Process ............................................. .............. 18

6 Meeting Commander Recording Software Interface ............................................28

7 Long Table M ethod of Analysis................................ .......... ............. .. 39

8 D ata A analysis Process .................................... ............... .... ....... 57

9 Supplemental W ork Zone Database Structure ............................... ............... .61

10 Supplem ental D ata Collection W eb Site.............................................................. 64

11 W ork Zone D ata R results Screen ........................................ ......................... 65

12 FHP Mobile Computing Architecture..... ...................... ...............66















Abstract of Thesis Presented to the Graduate School
of the University of Florida in Partial Fulfillment of the
Requirements for the Degree of Master of Science

DATA COLLECTION NEEDS FOR WORK ZONE INCIDENTS

By

Grady Thomas Carrick

August 2006

Chair: Scott Washburn
Major Department: Civil and Coastal Engineering

Roadway construction has become a common fixture in our daily travels.

According to the Federal Highway Administration (FHWA), fatalities in highway work

zones were up nearly 50% between 1997 and 2003. In 2003 alone, there were 41,000

injuries and 1,028 fatalities in these locations. Increasingly, safety interests are searching

for characteristics associated with work zones that contribute to the dangers of such areas.

Like many aspects of traffic safety, a better understanding of the contributing factors in

crashes can potentially lead to improved countermeasures. Examining crash data is a

principal method by which engineers, police, and safety advocates attempt to determine

those factors, but such data are often incomplete. The prospect of improving the data set

requires examining the potential of a supplemental data collection system. Using

qualitative research, work zone stakeholders potentially provide a better understanding of

work zone incidents, rendering new data elements. Creating a web-based supplemental

collection system can assist police in gathering the data while completing the current









traffic crash report. Supplemental data elements and collection systems have the

potential to enrich the data set, and bolster the cause of safety.














CHAPTER 1
INTRODUCTION

Background

According to the Federal Highway Administration [1], there was a nearly 50%

increase in U.S. work zone fatalities between 1997 and 2003. In 2003 alone, there were

1,028 work zone fatalities, a figure that represents about 2.4% of all roadway fatalities for

the year. A larger view of the problem is evident in the estimated 41,000 people injured

in the more than 102,000 work zone crashes for that same year. Florida statistics mirror

this compelling national problem. In 2003, 104 fatalities and 3,607 injuries occurred in

3,509 crashes in Florida highway work zones [2].

The danger that work zones pose for construction personnel is readily apparent.

The less visible statistic about these locations is the peril for motorists. Nationally, 85 %

of fatal crash victims in work zones are drivers or occupants, and in Florida that

proportion increases to 90 %, as 9 in 10 are motorists or pedestrians [1, 2]. Regardless of

the reason for being in a highway work zone, it is a potentially dangerous environment

for everyone.

Like many states around the nation, Florida is constantly trying to stay ahead of

rapid population growth with new road projects. When coupled with routine

maintenance efforts to address aging road and bridge infrastructures, the work zone has

become a common fixture on our roadways. Nationally, an average of 23,745 miles of

federal aid roadway improvement projects were underway annually from 1997 to 2001

[3]. These improvements are in addition to innumerable work zones associated with









municipal, county and state road departments, utility construction, and public works

projects.

Increasingly, safety interests are searching for characteristics associated with

crashes that occur in work zones. Like many aspects of traffic safety, a better

understanding of the contributing factors in crashes can potentially lead to new and

improved countermeasures. Examining aggregate crash data is usually the means by

which engineers, police, and safety advocates attempt to determine those contributing

factors. Individual police crash reports potentially focus attention on specific time,

location, and causation factors.

Problem Statement

Complete and accurate data is essential to the conduct of meaningful research.

Traffic safety research often relies in some part on data from police traffic crash reports

and aggregate crash statistics. In every state in the nation, crash report data elements

have evolved to capture relevant information about location, vehicles, persons,

conditions, and causation. The data derived from these reports are often the foundation

of safety-related research.

The basic crash report is designed to document facts for various governmental

purposes as well as satisfy insurance industry needs. With such general purpose, the

reports sometimes lack the detail necessary to be of value in examining unique crash

situations, like the work zone crash. While aggregate crash report data are suitable for

general frequencies analysis and cross tabulation, the lack of coding for roadway

construction variables makes work zone analysis difficult.

The Florida Traffic Crash Report Long Form (HSMV-90003) is completed anytime

there is a crash resulting in property damage in excess of $500 [4]. An abbreviated









report, the Law Enforcement Short Form Report (HSMV-90006), may be completed in

lieu of the previous for any crash that does not result in a vehicle being removed by

wrecker, an injury, or a crime (such as DUI or leaving the scene). The use of separate

report formats based on such criteria is commonly referred to as accident reporting

thresholds. The forms are very similar, and most of the coding for crash variables is, in

fact, identical. The main difference in the forms is that the diagram, narrative, and much

of the coding is optional for the officer with the use of the Short Form Report. Only

Long Form reports are used in tabulation of statewide crash statistics.

While the hand written crash report remains a staple, many agencies are migrating

to automated methods of reporting crashes. These systems usually consist of officers

using laptop computers in the field to input crash data electronically. Such systems have

proven to reduce errors and greatly improve the timeliness of data. Their continued

evolution will solve some data issues, but the completeness of data continues to be an

obstacle in the research of work zone incidents. These gaps in data are more of a

function of the format of the reports than anything else.

For most purposes, the current reports are well-designed and adequately capture the

most relevant data. The reports do, however, lack detail where highway work zones are

concerned. In Florida, linking crash reports to work zones is accomplished through a

code for "Work Area" (1-none, 2-nearby, 3-entered). Less prominent codes contained in

other report areas, may also indicate work zone involvement:

* "Road Conditions at Time of Crash" 04-Road Under Repair/Construction
* "First/Subsequent Harmful Event(s)" 24-Collision i/th Construction Barricade
Sign
* "Pedestrian Action" 07-Working in Road
* "Traffic Control" O0-Officer/Guard/Fl t.r/,'i ",in









When one can discern that a crash occurred in a work zone, there are generally

insufficient data to describe the nature of the work, the maintenance of traffic (MOT)

control present, or the conditions created by the road work. Information that may prove

valuable to improving work zone safety is typically not captured.

Difficulties associated with studying work zone crashes often stem from

incomplete data. The logical remedy is to improve the data set by adding data elements

to the police crash report form. Changing statewide crash report forms is not a task to be

undertaken lightly, however, since it involves months of planning, meetings and

significant printing costs. Adding data elements for these purposes should be

accomplished with a supplemental reporting methodology, to eliminate the impact on

status quo reporting procedures. While collecting more data in work zones will require

additional officer time and effort, such concerns are diminished by the relative

infrequency of crashes in these locations. Given the uniqueness of work zones and their

temporary nature, it is essential to capture as much detail as possible about incidents in

these areas in order to effect positive safety improvements.

Research Objective and Supporting Tasks

The fundamental question posed by this study is, "What can we document at the

scene of a work zone crash that will enhance our understanding of those incidents?" The

objective of this research effort is to answer that question through a qualitative research

approach.

Since police routinely respond to traffic crashes for reporting purposes, they are

logical candidates to assist in any enhanced data collection effort. Creating an instrument

that police use to better document work zone crashes is seen as a means whereby the data

sufficiency can be improved. For purposes of this study, such an instrument should be









supplemental to the statutorily required crash form, so it can be used selectively and cost

efficiently.

Identifying the data needed is as important as the collection and tabulation process

itself. Determining data elements to be contained in the new instrument to be used by

police is made possible through the use of focus groups conducted with various work

zone stakeholders. By using individuals who are knowledgeable in the field, across

several disciplines, there is a greater likelihood that the additional data collected will have

the desired utility. Such qualitative research methods will illuminate new data elements.

Through the design of a supplemental data collection instrument, new data

elements can be evaluated by select Florida Highway Patrol (FHP) personnel in

conjunction with actual work zone crash investigations. After newly created data

elements are field tested, an analysis will be conducted to determine relationships with

state of the practice work zone data elements.

Document Organization

Having identified work zone data sufficiency as a problem, the research objective

and supporting tasks described above serve to direct the overall research effort

envisioned. An effective research approach further involves systematically looking at the

research others have done, undertaking original study, analyzing the results of that study,

and ultimately drawing conclusions. This archetype forms the basis for chapters

contained in this document. Chapter 2 provides an overview of previous related research.

Chapter 3 describes the research approach and methodology used to satisfy the objectives

of this project. Chapter 4 presents analysis of the data collected in conjunction with this

project, and Chapter 5 provides conclusions and recommendations.














CHAPTER 2
LITERATURE REVIEW

Current Crash Reporting Systems

Police traffic crash reporting likely began in March of 1896, when an early

automobile struck a bicyclist in New York, resulting in the driver of the auto spending a

night in jail [5]. As the automobile gained in popularity in the early 20th century, police

standardized the way motor vehicle mishaps were reported, and ultimately states

standardized the forms for their respective jurisdictions. With the evolution of the

accident report at the state level, gathering statistics about crashes became possible. State

reporting and statistical practices have progressed to relatively efficient systems in most

states.

Because there is currently no mandatory national standard for police crash report

forms, state crash reports assume a wide range of designs that can be handwritten,

elaborately coded, bubble coded, or even computer generated. While they all certainly

have a different appearance, the reports typically capture similar information about the

location, persons, vehicles, and environment of traffic collisions. The information is

coded with varying degrees of detail, for inclusion in state collision databases.

Conclusions about vehicle movements and crash causation factors are typically the

product of police interpretation of these elements. Almost all formats allow for a

narrative and pictorial representation of the crash events.

While state formats differ, there are common data elements. It is through this

commonality that national statistical efforts have traditionally been undertaken. Even









with common data elements, getting data to integrate has often been a daunting task, due

mainly to a lack of uniformity in data elements, structure and definition. To improve data

portability, there remains a need for improved standardization.

In 1998, a joint effort was initiated to bring about greater uniformity in reporting.

The Model Minimum Uniform Crash Criteria Guideline (MMUCC) is a joint effort

between the National Highway Traffic Safety Administration (NHTSA), the Federal

Highway Administration (FHWA), the Federal Motor Carrier Safety Administration

(FMCSA), and the Governors Highway Safety Association (GHSA). "MMUCC

represents a voluntary and collaborative effort to generate uniform crash data that are

accurate, reliable and credible for data-driven highway safety decisions within a state,

between states and at the national level" [6:iii].

The MMUCC Guideline, 2nd Edition (2003) contains 111 data elements, of which

77 are collected by law enforcement at the scene of a crash [6]. The remaining elements

are derived from the data collected, for example, the number of vehicles involved,

number of people injured, etc. The MMUCC establishes a "minimum" set of data

elements, but does not dictate design of the actual report form. With standardized data

elements, the integration of data across multiple databases is made possible. Therein lays

the promise of the MMUCC, integrating state data for national statistical purposes.

It has only been in the last decade that crash reports have sought to capture data that

is unique to highway work zones. In 1992, only 27% of state crash reports contained an

explicit field for work zone presence [7], while today 67% contain such a section.

Historical codes for "roadway under construction" have increasingly given way to a more

specific representation of the presence of work zones.










Section C19 of the MMUCC establishes guidelines for work zone-related data

elements. The rationale for inclusion of data elements in crash reporting center on the

need to "assess the impact on traffic safety of various types of on-highway work activity,

to evaluate Traffic Control Plans used at work zones, and to make adjustments to the

Traffic Control Plans for the safety of workers and the traveling public" [6]. The Guide

also adeptly notes that the temporary nature of work-zones requires documentation of

their presence. Table 1 lists the work zone elements and attributes recommended by the

MMUCC.

Table 1. MMUCC Work Zone Elements and Attributes
Data Element Attribute

Was the crash in or near a Yes
construction, maintenance or No
utility work zone? Unknown

Before the First Work Zone Warning Sign
Advance Warning Area
Location of the crash Transition Area
Activity Area
Termination Area

Lane Closure
Lane Shift/Crossover
Type of Work Work on Shoulder or Median
Intermittent or Moving Work
Other

Yes
Workers present? No
Unknown


Because the MMUCC is only a guide, a review of state reporting forms is

necessary, in order to determine the scope of work zone data collected through police

reports. Specifically for this research effort, the police report forms used in all fifty states

and the District of Columbia were reviewed in detail for elements associated with work










zones. While a complete representation of the content in those reports is represented in

tabular form in the appendix of this report, Table 2 summarizes how work zone

information is depicted on the crash reports used in those jurisdictions. Analysis of those

forms is further described in the following paragraphs.

Table 2. State Report Form Work Zone Methodologies
Embedded Data Elements
Work Zone Specific Section Ebde Dale
Road Traffic Control Event Ped
Basic Detailed Const. Collision Working
(Y/Nor Data None Repainst. Flagger Sign or with in/on
(Y/N or Repair/Const.
Code) Captured Barricade Barricade Road
States 23 11 17 24 33 9 21 39
% 45 22 33 47 65 18 41 76


Work zone data are represented in US crash reports in two ways; the presence of

work zone specific sections on the reports, and the use of embedded coding for crash

attributes that may indicate the presence of a work zone. Work zone specific sections on

traffic crash reports are fairly new features, being added in the latest iteration of the forms

in most cases. For example, the state of Florida added the section in its January 2002

revision of the Florida Traffic Crash Report [8]. Of the 51 crash report forms reviewed,

34 (67%) had a separate or explicit section for work zone data. On a continuum, the

detail of these explicit work zone sections ranges from a simple 'yes/no' check box to

multiple data elements describing the work zone location. To further delineate the use of

explicit work zone sections, reports that use a special section can be can be grouped on

either end of this continuum, based on the level of detail, as either basic or detailed.

Basic work zone reporting sections generally have separate data elements) for

capturing the presence of a work zone, but they do not expound upon the circumstances

present. Of the 51 reports reviewed, 23 forms (45%) use methods of yes/no or simple








codes to indicate the presence of a work zone. Figure 1 and Figure 2 are examples from

Florida and Wisconsin respectively.

WORK ARFA
01 None
02 Nearby -
03 Entered



Figure 1. Florida Traffic Crash Report

Hit & Run CYt (N
Government Property J N)
Fire (Narrative) Y (NC
Photos Taken (Narrative) ( CN)
Trailer or Towed (Narrative) (Yi "j )
Truck or Bus (Last Page) c(Y ( CN
Load Spillage (Y) N)
Construction Zone CD-) (iC
Names Exchanged K CN

Figure 2. Wisconsin Motor Vehicle Accident Report

Some states capture greater detail about the presence of road work. In the case of

11 states, or 22% of those reviewed, the data capture went beyond simply noting the

presence of road work. The guidance of the MMUCC is evident in this detail, noting

where the crash occurred in relation to the work zone, the type of work being performed,

and the presence of workers. Only five states (Iowa, Nebraska, Ohio, Pennsylvania, and

South Carolina) follow the entire MMUCC guideline for elements and attributes relevant

to work zones. Pennsylvania uniquely captures the presence of law enforcement at the

work zone crash location and the presence of special speed limits. Figure 3 and Figure 4

are examples from Pennsylvania and South Carolina respectively.










Work Zone Tve Where in Work Zone? Work Zone Seed Lae closure?
OCs Br Advisovr Limit Secial Work Zone L
(Long Ton) Wa mlnI S- Law Enforcement Characteristics I I 1ed
S( _) Mvance warmmgA ~ea Officer Present (Mark all that
S i Maintenanc apply. I not f W ork o Shuker a
N S'rTn Transaibn Area Woyres involved or [ j
(39 U unknown, leave Intr"fft N
'-2 blank) Mong Wod&P
= T .... Area M Unknown
Othe Ot)No Un) know k Flagger Conlroa?
) Unknown
List all Waming Sins in Narrative om


Figure 3. Commonwealth of Pennsylvania Police Crash Reporting Form

1 Before 1" Sign 3 Transition Area 5 Termination 1 Yes 2 No Work Zone:
2 Advanced Warning Area 4 Activity Area Area --------------Work Zone Location
1 Shoulder/Median Work 3 Intermittent/Moving Work ---- --------Work Zone Type
2 Lane Shift/Crossover 4 Lane Closure 8 Other 9 Unk 1 Yes 2 No Workers Present:


Figure 4. South Carolina Traffic Collision Report Form

A total of 17 states (33%) do not have a designated section for capturing work zone

information on crash reports. Researchers are required to find other report data to link

these incidents with work zone activity. This is typically accomplished through

"embedded" attributes.

Embedded report attributes or codes can be mined to indicate the presence of work

zone conditions, albeit indirectly. These attributes can be used in conjunction with or

independent of a specific work zone report section. They are typically found in report

categories describing the roadway conditions, traffic control present, harmful event, or

pedestrian action. All 51 report formats from the states plus the District of Columbia

include some form of these embedded attributes. When cross tabulated with other report

variables, greater insight into the work zone crash is possible.

"Roadway conditions" is a data element contained in almost all police crash

reports. Defects in the roadway, obstructions, loose surface material, holes, and standing

water are all types of attributes that may be present. For purposes of this analysis,









specific attributes such as "road under repair" or "road under construction" can be

associated with a work zone. Of the 51 reports reviewed, 24 (47%) contained such an

attribute.

Traffic control can take many forms, but generally it involves the signs, signals,

and other controls present at the crash location. With respect to work zone analysis, the

presence of a flag person coded in this section of a report is a strong indicator that a

highway work zone may have been implicated. Coding for this type of traffic control

was present in 33 reports, representing 65% of the total. Additionally, 9 reports (18%)

specifically noted the presence of a construction sign or barricade as a traffic control

device.

Harmful events are captured to describe the damage or injury producing crash.

While first harmful events are generally captured, subsequent harmful events are

beneficial to illuminate the entire series of events in a crash. Practically anything that a

motor vehicle strikes, including the manner in which it strikes another vehicle, is

considered a harmful event. When a vehicle collides with a construction barricade,

barrel, or piece of work equipment, it may be an indication that the crash occurred in a

highway work zone. Collision with one of these objects is seen in 21 reports (41%) as a

"harmful event."

Like vehicles, pedestrians are represented in crash reporting as "traffic units." The

recognizable role of pedestrian construction workers readily comes to mind when one

considers highway work zones. For reporting purposes, the action of pedestrians is

classified in a section that describes what the pedestrian was doing at the time of the

collision. Typically this section is coded as crossing the street, walking along the road,









standing in the road, or similar descriptions of people that one might encounter near

roadways. Attributes that specifically denote the pedestrian as "working on/in the road"

are a good indicator of work zone involvement. All 51 report formats reviewed except 12

(76%) contain such a description of pedestrian action or movement.

Embedded attributes describing collision events, roadway conditions, traffic

control, and pedestrian actions were traditionally the indicators that were used to identify

the presence of a work zone. Statisticians were historically required to link these

elements to other data to determine if a work zone was implicated in the crash. Their

continued inclusion, coupled with specific work zone sections, can provide information

about work zones and the circumstances of crashes that occur within them.

Unfortunately, too little information about work zone crashes is still the norm in state and

national crash reporting systems.

Work Zone Data Collection-Related Studies

Several studies have examined alternatives to basic police reporting for collection

of data in highway work zone crashes. Reviewing these alternative methods is of value,

since this project seeks to go beyond the traditional crash report approach as well.

Wang, et al. conducted research using data from the Highway Safety Information

System (HSIS) to explore the issue of work zone crashes [9]. The conclusions of their

study indicate that the absence of a universal definition of "work zone" is problematic.

Additionally, they found that police reporting systems require modification to include

additional data elements that better describe work zone attributes. Data fields

recommended by their report are designed to answer the following questions:

* Did the crash occur in or near a construction, maintenance, or utility work area?
* Was there work activity at the time of the crash?









* Where did the crash occur in relation to the work?
* What was the type of work performed?
* Did the work area have an influence on or contribute to the crash?

Khattak and Targa investigated the role of large trucks in work zone traffic crashes,

using HSIS data and North Carolina Police Crash Reports [10]. Examining the narrative

and diagram in each crash report, researchers sought to enhance the data set by creating

unique data elements. They noted that this method of data acquisition was very laborious

and only possible given the limited scope of subject incidents, (i.e., large truck crashes in

work zones). Ha and Nemeth mined police reports in Ohio for unique work zone data

elements and also found the process to be difficult [11].

Garber and Zhao examined crash characteristics in work zones by examining police

crash reports believed to be work zone related from 1996 through 1999 [12]. The scope

of their study was 1,484 reports after nearly 500 reports were discounted because of

inconsistencies in reporting. To facilitate more detailed analysis, they recommended that

the Virginia Police Accident Report be modified to capture additional information

relevant to location, work activities, traffic control, speed limit, and presence of workers.

Raub et al. sought to determine causal factors in work zone crashes by enhancing

police reporting [13]. Making a case that the data derived from police reports was largely

insufficient with respect to work zones, researchers developed a separate data collection

instrument for the purpose of determining the contributing factors associated with 103

crashes in Illinois between 1998 and 1999. In these cases, police completed a

supplemental form to assist researchers in their analysis. The supplemental form did not

use a methodology for developing data elements, and was basically designed to be a

stand-alone data system, not linked to the larger crash reporting database.









Schrock et al. concluded research in 2004 that analyzed 77 fatal work zone crashes

in Texas [14]. Faced with issues about sufficiency of crash data, trained researchers

responded to the scene of fatal work zone crashes in an attempt to document factors at the

scene. Reviewers attempted to respond as soon after the crash as possible, but their

response was sometimes in the days that followed. Environmental factors generally

duplicated attributes already captured in reporting with the exception of identifying the

location of the crash within the work zone, and the nature of work being performed. The

Georgia Department of Transportation (DOT) has developed a practice whereby they

attempt to respond to all fatal work zone crashes as well [15].

The Florida DOT, like some other states, uses an "Engineers Maintenance of

Traffic (MOT) Evaluation at Accident Site" report to better understand the factors

associated with work zone crashes. After work zone crashes, FDOT engineers are tasked

with completing a report to capture information about the work zone. The report has

historically not consistently been completed by FDOT personnel, and Spainhour and

Mtenga sought to revise and automate the form [16]. While the electronic entry format

was superior to the paper form it replaced, the fact that it did not increase use by FDOT

engineers was seen as an indication that the data set continues to be notably incomplete.

Graham and Migletz noted similar problems with underreporting in their review of

project manager-based reporting systems used in Iowa and North Carolina [17].

Thielman examined the potential of expert systems in the collection of traffic crash

data, although not specifically work zone data [18]. The foundation of the expert system

was data elements derived from experts in the field. Panelists included traffic officers,

crash investigation trainers, safety analysts, reconstructionists, vehicle safety engineers,









and highway engineers. After determining data to be collected, officers equipped with

pen-based computers used the expert system to focus greater detail on seatbelt use,

vehicle damage, and roadside barriers. The expert system required an average of two

minutes of officer time to collect the data, and field tests of the system confirmed that

officers would be receptive to additional collection responsibilities. While it did not

specifically focus on work zone areas, the concept of enhanced data collection and the

use of an expert panel were viewed as successful.

In some cases, video images of vehicles entering work zones are used to measure

the effectiveness of merge operations [13]. Video can also be used to mitigate congestion

in conjunction with ITS applications [19, 20]. The potential for such technology in crash

data collection has not been empirically explored, but the wider use of these systems may

hold that potential. While the prospect of capturing actual crashes with these systems

would likely only be possible in very isolated instances, they may be well suited for

establishing the environment present at the time of a crash.

While there have been a number of studies relating to work zone crashes and the

sufficiency of data within these unique locations, substantial issues remain. Some studies

have focused loosely on crash attributes in work zones, but none broach the topic from

the perspective of perfecting the process by which data elements are identified in a

qualitative way. Similarly, despite efforts to bolster collection systems, potential exists

for improvement as well. No study approached the topic of supplementing work zone

data as a two-fold proposition requiring qualitative data identification and collection

system development.














CHAPTER 3
RESEARCH APPROACH

Overview

The prospect of improving the data set for work zone incidents requires defining

what information to collect and how to better accomplish that collection. With these

objectives in mind, the concentration of this project is to identify, test, and validate new

crash report data elements that specifically relate to work zone incidents. Subordinate to

that objective, a strategy for implementing the data elements is also tested, with the

creation of a web-based collection system for officer use. Through identifying new data

elements and actually collecting said data, the body of knowledge concerning work zone

incidents will be improved.

Methodology

A systematic approach to improving the data set requires identification of the

specific data needed. While some previous studies have sought to embellish the work

zone crash data set through supplemental reporting, none have qualitatively approached

the process of determining the data to be collected. Ruab et al. [13] used police in Illinois

to complete a supplemental report form, but the content of that form was not the product

of a qualitative effort. Many of the data elements duplicated normal crash report data

elements and some were quite subjective, asking for officer opinion. The supplemental

form was not designed to be linked with crash records, and was generally a stand-alone

product. Enhancing the data set requires a clear understanding of data needs of the crash










data consumers, a proposition that mandates their involvement in the development

process.

Development of work zone-specific crash data elements is a process that can be

described as several distinct steps. Mining potential data elements from discussions with

stakeholders provides the foundation of knowledge necessary for analysis. The content

of those meetings can yield specific issues/items when vetted using qualitative text

analysis methods. Once potential data elements are identified, they are contextually

linked to the original group discussion to develop the issue more precisely. Pilot testing

the product under real world conditions determines if the data element is mechanically

sound. The final step in the process involves validating the data elements against state of

the practice standards. The figure below is a graphic representation of the process

described herein.

k N K I k


Data Mining
(Focus Groups)


4


New
Data
Element


/ / V V r

Figure 5. Data Element Development Process

The development of a collection system is both a part of the data element

development process as well as an integral part of the overall data supplementation

proposition. The use of a supplemental collection system is described in greater detail as

a component of the data element testing section, and later in Chapter 4.









Traditional Crash Report Content Decisions

Because traffic crash reports serve a wide variety of purposes, their development

typically finds origins in public administration and public policy arenas. This is

supported by staff at the Florida Department of Highway Safety and Motor Vehicles

(DHSMV), who describe the process as highly collaborative and involving of

stakeholders. Input, clarification, and construction are products of meeting with a small

number of interested individuals. Inclusion in the group ranges from mere stakeholders

to subject matter experts. The leader of the group is generally a public administrator,

representing a state agency that is charged with the responsibility for the forms by statute

or administrative rule. The group could be referred to as an "expert panel", and their

meetings and activities are generally conducted informally.

Intuitively, the expert panel may be viewed as a practical way to approach the task

of determining the content of traffic crash report forms, but determining consensus of the

panel is often elusive. Some research suggests that the views of outliers are often

discounted to further the objective of consensus [21]. In addition, even the most

representative and well-intentioned panel may fall victim to common pitfalls associated

with individual and group dynamics. Closely related concepts such as groupthink, social

consciousness, and Abilene paradox must be considered in group settings. For additional

information on these issues, the readers is referred to the works of Elizabeth Scott, Emile

Durkhiem, and Jerry Harvey [21,22,23].

While a detailed analysis of how individual and/or group dynamics may affect

small group meetings is beyond the scope of this effort, it is important to note that

committees or groups assembled for almost any purpose may be susceptible to these and









other errors. When an expert panel is convened to examine crash report form changes,

the informal exchange may produce unexpected and unreliable results.

Data Mining Using Qualitative Methods

Qualitative research is, "Research involving detailed, verbal descriptions of

characteristics, cases, and settings. Qualitative research typically uses observation,

interviewing, and document review to collect data" [24:1]. This form of research is

rooted in the social sciences and is an excellent vehicle for examining things that may not

be measurable quantitatively. Qualitative research can be accomplished through surveys,

questionnaires, personal interviews, researcher observation, or similar methods. The

research seeks to learn more about things in their natural environment through people's

attitudes, perceptions, recollections, and feelings.

Getting people together for the purpose of determining direction for a project need

not fall victim to problems associated with group dynamics. One way to potentially

produce more reliable results is the use of qualitative research. For our purposes,

enhancing the content of crash reports is best undertaken as a function similar in

approach. Using qualitative research, investigating the topic of crash incident data

represents a form of collaboration. Similar to currently used methods, stakeholders such

as law enforcement, engineers, private contractors, drivers, and safety advocates form the

basis for input. But by employing a qualitative research methodology, potential errors

related to group dynamics can be minimized and a more reliable product is possible. A

more dependable consensus among stakeholders is also possible with qualitative

methods, since they employ an approach that is grounded in social science. A more

detailed examination of qualitative techniques is included as Appendix A, Overview of

Qualitative Methods.









Focus Groups In Qualitative Research

Choosing a qualitative research method for purposes of identifying potential work

zone data elements requires weighing available methods such as Delphi, survey, and

focus groups. While all three methods employ the use of stakeholders, focus groups

holds the greatest promise for success, given their interactive nature and structural

similarity to the traditional crash report development approach.

Focus groups have been used as a qualitative research tool for several decades, and

are readily associated with market and product research. Recently, the technique has

gained favor as an academic research and public policy tool. A focus group is described

as an assembly of people for the purpose of discussing a topic. The discussion is led by a

prepared moderator who guides the group, producing a collection of opinions,

perceptions, and experiences of the participants. Multiple focus group sessions broaden

input and improve the reliability of data collection.

As with any research tool, the use of focus groups requires following acceptable

guidelines for their conduct. Some of these guidelines are contained in Appendix A -

Overview of Qualitative Methods. The implementation of focus groups is best described

as functions of group composition, administrative preparation, moderator and content

preparation, administration, and post-session processing.

Group composition

Identifying focus group participants is an important factor in the development of

the research method. Having group members with similar backgrounds describes a

homogeneous group composition, and most researchers agree that this allows for

common threads of discussion [25]. Designing groups that represent work zone









stakeholders thus requires segregating FDOT traffic engineers, private contractors,

police, and average citizens into separate focus groups.

Traffic engineers in both the public and private sectors are those primarily

responsible for the design, setup, and maintenance of work zone traffic plans. Because of

their significant role and expertise, they are necessary participants. Segregating public

sector engineers (FDOT) from contractor personnel is beneficial because of sometimes

divergent priorities. Further segmenting FDOT personnel into field and headquarters

elements ensures all aspects of public-sector engineering are represented. The rationale

for separation within FDOT is that the people at the headquarters level are those mostly

responsible for developing and/or changing policy with regard to standards indices and

traffic control plans, while at the district level, they were mostly responsible for

implementation. A focus group with FDOT field personnel involves construction

personnel, who are in work zones on a daily basis. The FDOT headquarters focus group

included engineering representatives from the safety office, roadway design office, and

construction office. FDOT and industry representatives make up three separate focus

group sessions.

Law enforcement plays a significant role in work zones, and they represent

valuable stakeholders in gaining additional insight. Police are charged with enforcing a

myriad of laws within work zones, in order to promote the safety envisioned by

engineers. Additionally, they are responsible for investigating incidents at these

locations, and therefore document the conditions present and circumstances surrounding

incidents. Sworn law enforcement officers make up a separate focus group.









While they may lack traffic safety expertise, the layperson brings a unique

perspective to discussions about work zones. Such a perspective may be beneficial when

compared to the views of traffic safety practitioners. Since work zones are ostensibly

designed with the average motorist in mind, it is of value to gain their insight into these

locations. While they may not be direct consumers of incident data, one can readily

understand their value as stakeholders and benefactors of changes in the driving

environment. A collection of licensed drivers, representing the public at large, make up

the final focus group.

Administrative preparation

Conducting group meetings with individuals requires significant planning with

respect to scheduling activities. With the exception of the public meeting, all of the focus

groups were conducted without compensation to participants. Accommodating

participant work schedules and enlisting volunteers can be challenging. For public-sector

organizations such as law enforcement, and the FDOT, participants are supported by their

respective organizations. The session with law enforcement and two sessions with FDOT

personnel were all well attended, as expected. Conducting the meetings at their

respective work locations proved to be a wise accommodation.

The industry focus group was much more challenging from the perspective that

release of employees from work is more difficult for private business. Additionally,

assembling participants from diverse employers and work locations requires effort on the

part of the participant to travel and meet at a specified location. While 7 participants

were scheduled, work obligations precluded 3 from attending at the last minute.

Conducting the industry session with less than 6 participants was less than ideal, but the

level of discussion by participants made data collection acceptable.









The citizen focus group sought to gain input from the average driver. This group

was conducted at the Perceptive Market Research (PMR) facilities in Gainesville.

Participants were selected to be demographically representative of the community and

compensated by the market research firm.

As part of the research strategy, a composite group representing stakeholders was

deemed beneficial to review the homogeneous group findings. Using a regularly

scheduled meeting of the Alachua County Community Traffic Safety Team to facilitate

the composite stakeholder meeting was an excellent way to minimize impact on

participants. All meetings were scheduled at convenient locations, with suitable facilities

for privacy and comfort. The table below describes the setting, attendance, and duration

of all focus groups.

Table 3. Focus Group Composition
No Approx.
Group Date Location Part s Duration
Participants .
(hrs.)
Law Enforcement 11/08/05 Jacksonville, FHP 6 1.50
Industry 11/08/05 Jacksonville, FHP 4 1.75
FDOT Lake City 12/02/05 Lake City, DOT 8 1.50
Citizen 03/08/06 PMR, Gainesville 10 1.75
FDOT Tallahassee 03/27/06 Tallahassee, DOT 7 1.75
CTST* 04/20/06 Gainesville Technology 20 N/A
Enterprise Center
*Composite group not a focus group session.

In addition to scheduling, administrative preparation involves the use of a Group

Sign-In Form, Participant Instructions, and an Informed Consent Form. The Participant

Instructions provides each focus group participant with a one-page explanation of the

objectives, format, and basic guidelines for the conduct of the forum. These forms are all

included in Appendix B.









Moderator and content preparation

Content planning involves establishing clear objectives on the part of the research

team. The focus group is a moderator-led discussion, so the moderator must be well

equipped to perform in a way that promotes group interaction and adequately explores

the subject at hand. Dr. Scott Washburn is an experienced focus group moderator, and

the moderator for all focus group sessions. The role of the moderator is well established

with the use of a Moderator Guide. The Moderator Guide is a document that dictates the

steps followed by the moderator to ensure a systematic approach and efficient use of

time. The guide is a step-by-step script of sorts, that lists events, basic instructions, and

target time limits. The guide includes participant sign in, welcome and introductions,

background of the study, explanation of the format and scope of the meeting, and the

questions to be used to promote discussion. The guide also provides some reminders to

the moderator on methods for handling shy or reluctant participants, as well as those

participants who may be monopolizing discussion. The questions used in the focus

groups for are, by design, left quite general to promote maximum opportunity for

exploration of the subject of work zones. Four basic questions form the basis for the

work zone focus groups, however the moderator has a number of follow-up questions

available for use if necessary. Questions are not all inclusive, since the guided discussion

is designed to illuminate the subject. Latitude is afforded the moderator to explore areas

that may be beneficial to the objective of the research. The following table is

representative of the questions available to the moderator for the work zone focus group

sessions. The Moderator Guide is included as Appendix C.










Table 4. Focus Group Moderator Questions
1. Why do work zone crashes occur?
Physical features of the roadway
Issues with MOT and traffic control devices
Driver behaviors
Vehicle / Worker characteristics (trucks, exposed workers, etc)
2. What are positive things that are being done to help the situation?
Advance Warning
Speed Limits
Better MOT
Separation Barricades, barrels, walls, and other devices
Enforcement
Public Information and Education
3. What are things that still need to change?
Driver behaviors
Physical design of work zone
Traffic Control
Enforcement Issues
4. What can we learn from Incidents that occur in work zones?
Role of congestion
Secondary collisions
Location of incidents
Type of collision (rear end, sideswipe, run off road, with barricades/equipment, etc.)



Administration/conduct of meetings

Administration of the focus groups brings together the planning associated with

participants and the moderator(s). Site setup requires the focus group team to arrive at

the location early to prepare seating, visual aids, recording equipment, and refreshments.

Participants are greeted and offered refreshments, followed by the moderator welcoming









everyone and beginning the session. A PowerPoint presentation accompanies the

moderator's program, primarily highlighting main questions and providing for supporting

visual aids if necessary.

Audio recording of focus groups is essential to later analysis of content. Recording

is made possible by a laptop computer equipped with software and CM3 boundary

microphones strategically placed near participants. A backup cassette tape recorder with

one CM3 boundary microphone can be used as a backup device. Since it is important to

link participants with their respective comments, specialized recording software is

necessary to allow the research team to annotate speakers while recording. The Meeting

Commander software product is a commercial package [26] that allows the research team

to graphically depict speakers with on-screen icons, the same way they are seated in the

focus group setting. Once recording begins, the recording assistant simply clicks on

speaker icons each time there is a change in speakers. Speaker changes are captured in

the audio file, so that later playback depicts speaker changes. Elapsed time and speaker

change time stamps are included in the digital audio recording. Figure 6 represents an

illustration of the Meeting Commander screen, with speaker seating on the left and a

chronological representation of the recording, complete with speaker changes, on the

right.

Post-Session Processing

Qualitative analysis of the focus groups requires that content be reduced to a text

form. Post-session processing of the focus group audio recordings requires listening to

the audio to review speaker comments and subsequently transcribing the audio into word-

for-word text. The Meeting Commander software produces a time stamped chronological

listing of speakers as an exportable text file. Playback of the audio using the software or











Windows Media Player allows the person performing transcription to monitor speaker

changes and type text with the appropriate speakers. A USB foot-pedal facilitates

playback of the audio and eases transcription. The transcript is ultimately reviewed and

compared to the actual audio recording for final quality assurance.


; File Edit Speaker Help

/ MeetingMode r Deeign Mode Ev4 I nL .
Time I Event
I 0038 44 7 Attachment[ exe
S 004731 Rodney
I 0047351 Stop
S 0047351 Record


S 00 4 045 Stop
00 49 045 Record
0112074 Clair
"_0 0112083 Rodney

Corrmmenced recording at 3 24 19

PM StM uday Apr 2 2005


oIallilal




Figure 6. Meeting Commander Recording Software Interface

Data Element Identification-Qualitative Analysis

After data is collected through the use of focus groups, the data must be analyzed to


provide usable results. Text analysis has been widely used for mining news outlets and

more recently, for mining operations involving the internet, blogs, and email. Such


analysis generally focuses on obtaining frequencies for words or phrases to attach some

statistical significance to their occurrence.

Analyzing text obtained in conjunction with qualitative research is often described


as "content analysis" or "qualitative text analysis". The systematic and replicable

technique for compressing words or phrases into content categories, given established

rules, is a general description of the process [27]. The analysis of open-ended text









responses for qualitative research involves more than just determining word count or

frequency. The process employs coding techniques that allow the researcher to reduce

text which is produced by open-ended questions, in a categorical way. "A category is a

group of words with similar meaning or connotations." [28:37]

After scanning the text, numerous codes are created, without a great deal of

categorization. When screened in a more focused way, some codes are eliminated,

grouped into larger themes, and/or subdivided [29]. The process of analyzing text with

this coding enables the researcher to organize and digest the content of focus group data.

Focus group expert Richard Krueger describes the traditional process of focus

group content analysis using colored paper, colored markers, and cut and paste

techniques [30]. Using these items, the researcher identifies and rearranges text obtained

from research to create descriptive summaries. Such techniques may appear rudimentary,

but they are still used for qualitative analysis of focus group data. For our purposes,

similar methods are used.

Each focus group session is tape-recorded for later review and analysis. When

audio recordings from focus group sessions are transcribed, a wealth of text data becomes

available. These data, however, are unstructured and in a raw form, rather unusable for

research purposes. Similar to other forms of data obtained in qualitative research, the text

must be processed or analyzed in ways that make the data meaningful to researchers. The

fundamental question of this analysis centers on finding meaning in volumes of text.

That question is answered in the process of qualitative analysis.

The process of qualitative text analysis can be summarized as three basic steps; 1)

the reduction of the original database, 2) the construction of linkages, and 3) the









comparison of findings [31]. Reducing the text is the process of coding, where segments

of the data are given representations that are more easily manipulated and categorized.

Constructing linkages is the attempt to form coded units, based on subject meaning.

Finally, those subject meanings are compared to infer invariants [31]. Through this

process one can move the data from mere words to meaning.

Focus group data can also be analyzed by computer. Many commercially available

text analysis programs focus on word counting and frequency distributions. This form of

analysis is most useful in marketing fields, and for media applications. Qualitative

analysis software is a rapidly evolving area of research in the social sciences. Programs

that have the ability to perform coding and other analysis functions are eagerly being

embraced. Programs like SPSS Text Analyst, AQUAD, NUD*IST, ATLAS/ti, and

HyperRESEARCH are all packages that have include capabilities useful for the social

scientist and others who may need to qualitatively analyze data. It is important, however,

to remember that interpreting the meaning of text is something that the researcher cannot

delegate to a computer. Therefore, the role of computers in qualitative analysis should be

viewed as supportive of the researcher's duty to understand the text [32].

Analysis of the data derived from focus group sessions with work zone

stakeholders renders insight into the issues surrounding those locations and the types of

factors that are involved in work zone incidents. Potential data elements can be identified

from this form of qualitative research.

Data Element Definition-Interpretation of Results

Analytical techniques transform raw data in the form of focus group text into more

meaningful representations of the data. Analysis segregates data into categorical codes.

Codes are measured using their intensity and frequency among various homogeneous









groups. The product of the analysis is an ordered listings of codes. Linking that data

with content from the focus groups is necessary to form meaning in the coded data. The

product of linking codes with content reveals potential crash report data elements.

Since a composite group of stakeholders is used to determine if the product of the

qualitative research effort is representatives of stakeholders, analysis is essentially

verified. The composite group can determine if the product is defective in any way, and

also insure that it is fully exhaustive of the subject. No qualitative analysis of the

composite focus group is necessary, since their involvement is merely a review of

conclusions.

Data Element Testing-Supplemental Collection System

The newly created data elements must be presented to individuals responsible for

data collection in a way that maximizes the chances for success. Where Raub et al. [13]

created a paper supplemental report form for police use, creating an instrument for

collection now requires a format of electronic entry for officers, since their current crash

reports are now in electronic form in many cases (via a laptop computer in the police

vehicle). An ancillary benefit to this format is the elimination of secondary data entry,

from hand written crash reports to an electronic database. This speeds entry of data,

reduces the potential for data entry error, and makes data management easier.

Data entry for the new work zone elements can be accomplished in three basic

ways. The current officer reporting software can be reprogrammed to include added data

fields, a web-based application can be created, or a stand alone database can be

implemented.

The software currently used by FHP troopers to complete the Florida Traffic Crash

Report is licensed by the agency from a private vendor. Making modifications to the









software for purposes of this research would be cost prohibitive, involving programming

changes to an end user software product. Such changes would result in version changes

to the product, and require upgrade of all agency computers. This may be desirable for

permanent changes to fields of the crash report, but are not practical given the limited use

and distribution of data elements for this project. Because of these reasons, the current

crash report application will not be altered for testing purposes.

A web-based application offers utility and flexibility that are desirable for simple

data collection. Given the infrequency of work zone incidents, and the limited data fields

envisioned, such a format would be practical. One inherent disadvantage to web-based

collection is the requirement for officers to leave the report application, log into a web

site, and enter data. A second disadvantage is the requirement to post-process the data,

merging Florida Traffic Crash Report data with the newly created web-based data.

Neither of these disadvantages is viewed as significant, however. A common data field

ensures that the crash report and supplemental work zone databases can be merged. This

form of data collection will be used by the officers for testing, because it most closely

resembles current reporting technology.

A stand alone database that is not accessible by the end user is easily created with

readily available application software. This method, however, is the least desirable

alternative, particularly because of the need for post processing officer-collected data.

Officers would essentially complete a paper report that would be coded by clerical

personnel into the computer database. Such a system may be needed for wider

application of the newly created work zone elements, given that many agencies/officers









do not use vehicle-based computers. For purposes of this effort, a paper format will be

created, but not used, since all testing officers are issued laptop computers.

Technical framework of collection system

The traditional hand-written police traffic crash report would logically lend itself to

a supplemental report that is also hand-written. When specialized crash reporting

software is used by police, the traffic crash report process is automated and the need for

subsequent key punching or manual entry into a database is eliminated.

Both the hand-written report and the report created with reporting software pose

limitations when one considers modifying or supplementing those systems. Hand-written

reports require modifications to printing, officer training, and data entry systems.

Changing the many custom software packages used by police would involve significant

investment in additional programming. While each may have advantages, neither is a

pragmatic solution in an effort to add supplemental data. Both have institutional and

financial issues that would prove prohibitive.

A desirable solution for supplementing traffic crash reporting for work zone-

specific data would be cost effective, not affect current reporting systems, and ultimately

be palatable for the institutions that are charged with managing crash data. By creating a

stand alone web-based collection system, expensive programmatic software changes are

avoided, yet officers are able to take advantage of available mobile computing

technology. Officers who complete regular traffic crash reports in work zones are asked

to additionally access a web site to provide unique supplemental data. In this effort, the

Florida Highway Patrol supports a supplemental work zone data collection web site and

officers will use the system.









Since the Florida Highway Patrol supports a computer network that encompasses

all patrol vehicles, the framework for officer computing is in place. Each patrol car has a

laptop computer with continual access to the network and internet. The agency supports

an internet web site for public use, and an intranet web site exclusively for agency

personnel. Several online databases are used by the agency, accessible by personnel who

are authorized by means of a password-protected architecture.

For security reasons, Microsoft SQL 2000 [33] is well suited to create the database

and ASP.Net [34] for the HTML user interface. These applications provide an increased

level of security, since database operations can pose potential access to dangerous code or

controls. Florida Highway Patrol servers are excellent host systems for supplemental

collection applications, since they are already in place and they currently provide a level

of security that is desirable.

Microsoft SQL 2000 supports eXtensible Markup Language (XML) schema that

are becoming the standard for crash report data. This will ensure compatibility and

transportability of data collected in conjunction with this effort.

Since each Florida Traffic Crash Report Form has a unique number assigned by the

Department of Highway Safety and Motor Vehicles, such a number becomes an excellent

candidate for use as a primary key field. This field will link the traffic crash report with

the supplemental work zone database. As a backup, additional fields may be designated

as potential linkages between the databases.

Operational framework of collection system

After the data collection instrument is finalized, select Florida Highway Patrol

(FHP) troopers serve as "beta-testers" for a short evaluation period. The process of field

testing the elements, attributes, and collection instrument requires direction, training, and









verification. FHP Troop "G" encompasses nine (9) counties in northeast Florida. The

troop is staffed by approximately 140 sworn personnel. Troopers are introduced to the

study by a supervisory memorandum that provides an overview and instructions. This

direction to participate in the research effort will establish organizational commitment

and increase officer buy-in. Instructions are provided with this directive, and a time

period for testing is established.

Since all personnel involved with testing are familiar with incident (crash)

reporting, each has general knowledge about work zones, and basic computer skills are

present among users, it is anticipated that the learning curve for these testers will be

minimal. Training will be accomplished by written instructions, with an opportunity for

supervisory follow-up.

Given there are currently several large work zone projects underway in the study

area (Troop "G"), the supplemental data elements and collection system will receive a

reasonable amount of testing. There is no statistical requirement for sample size, since

such analysis is beyond the scope of this project.

Because the supplemental system is not mandated by law or administrative rule, the

system used by officers is essentially voluntarily. Organizational commitment from the

Florida Highway Patrol replaces a legal mandate for officers to use the supplemental

work zone data collection system. According to information provided by the DHSMV,

the FHP works about 65% of all work zone crashes in Florida and 95% of all work zone

crashes on interstate highways. They are excellent candidates for this project because of

their significant role in reporting work zone incidents. Colonel Christopher A. Knight's









directive serves to require personnel assigned to the pilot test area to use the supplemental

system in all cases where they complete a traffic crash report in a work zone.

Data Element Validation

Section 19 of the 2003 Edition of the MMUCC describes data elements and

attributes that are unique to work zones. These elements closely resemble the data

recommendations of the 1996 study by Wang et al. [9]. These elements are described in

Table 1 of Chapter 2, and they represent the current state of the practice for police

reporting of work zone elements, albeit the MMUCC guidance for work zones is not

widely embraced by the states.

Validating new data elements for work zone incidents is the final step in the

development process, and this involves comparing the newly created work zone data

elements and attributes with those that constitute the state of the practice. Such a

comparison can provide an indication of whether current practices require modification.

In addition, field testing data elements with officers in actual work zone crash

investigations provides a pragmatic validation.














CHAPTER 4
RESULTS AND ANALYSIS

Qualitative Analysis Data Element Identification

Because qualitative research assumes many forms and purposes, the method of

analyzing data is equally diverse. While quantitative measures are typical in scientific

and engineering disciplines, the social sciences and the analysis of qualitative data is not

necessarily rooted in numbers. Some researchers gravitate to the use of numbers while

others steer away from their use. "Those who can answer their research questions

without counting codes should feel well justified in doing so no appeals to imagined

problems with statistical independence or random sampling are necessary." [35:62] "In

quantitative analysis it is sometimes easy to get caught up in the logistics of data

collection and in the statistical analysis of data, thereby losing sight of theory for a short

time. This is less likely in qualitative research, where data collection, analysis, and

theory are more intimately intertwined" [36:370].

In the first step in qualitative analysis, text and audio is reduced to data segments.

A code is used to represent those segments [31]. Nearly all qualitative research is

analyzed using some form of coding. Coding is the process of transforming text data into

a form that is more standardized [36]. The most common form of coding for focus

groups concentrates on manifest content [35]. Manifest content is that which is on the

surface, seeking the occurrence of a term or concept [36]. Latent coding uses the

underlying meaning of what is said to standardize content [36]. Coding is accomplished

by analyzing transcript text and identifying the terms, concepts, and content that work









together to convey a similar meaning, that is, they fit similar categories. Categories can

be related to sub-categories, and each is considered a 'code'.

While mere occurrence of words or concepts may be analytically useful, it is also

beneficial to determine what groups felt were important issues. Morgan describes factors

that indicate the emphasis given to a topic by a group as the product of how many groups

mentioned it, how many in each group mentioned it, and how much energy they

associated with the topic [35]. The energy and enthusiasm that individuals display for a

topic appears somewhat subjective, and therefore this measure will not be attempted in

this effort.

Qualitative analysis for this inquiry involves steps that potentially produce a list or

set of work zone discussion issues. From those issues, the most important factors can be

determined using qualitative measures, notably measures of agreement and intensity.

Agreement describes the number of focus groups where a given code was discussed, and

intensity describes the number of mentions of the code within and across groups. A third

potential measure identifies cases where all participants mention a code, but it is more

appropriate for market research thus it is not included in this effort [35].

Automated methods of analysis using specialized computer software is gaining

popularity; however, more traditional manual approaches like the "long table" method are

still effective, if not as glamorous. As a process, analysis involves reducing focus group

audio recordings to word-for-word text, copied on uniquely colored paper for each

different focus group. The text is read to find common themes in content and notations

are made in the margin of the transcript. Then text is physically cut out of the document

using scissors and is grouped and later sub-grouped based on topic or code. The process











continues until an exhaustive analysis of the text renders many assembled bits of what


speakers said into logical groups. These groups form the categorical basis of the codes


which drive the process of transforming text into meaningful representations of what the


people said in the focus groups. The coded text is measured using the methods


previously described and some conclusions about importance and relevance can be made.


Focus group text Focus group text Code
Code1
Focus group text
Focus group text
cn Focus group tet
.nrlls nmlln Focus group text
Ft Focus group text
Focus group text t
Focus group text t Focus group text
Focus n t Focus group text
Focu
Focu Focus group text Focusup texFus up text Focus group text
Focut Focus group text
Focus Focus group text
Focut Focus group text Focus group text
ext Code 2
ext
ext Focus group text

Focus group text
Focus group text ext Focus group text
Focus group text ext
Focus group text
Focus group text Focus group text/ Focus group text
Focus group text
Focus group text Focus group text
Focus group text Focus group text
Focus group text
Focus group text Focus group text




Figure 7. Long Table Method of Analysis

A Priori Categorization

Before formal text analysis begins, a number of work zone crash factors can be


considered relevant, without biasing results. As is the case with any traffic safety


analysis, causation factors associated with drivers and the roadway environment readily


come to mind as potentially important. Because there are a myriad of traffic control


devices used in work zones like signs, cones, barrels, message boards, and barricades, a


category for traffic control devices is likely important as well. Driver behaviors, traffic


control devices, and construction/roadway conditions form a broad basis for categorizing









work zone crash factors. Creating a category for 'other' factors provides an opportunity

for coding to reach beyond the boundaries of these broad coding categories, and ensures a

comprehensive qualitative analysis is possible. In reviewing the audio of each focus

group, these general categories represent most discussion content. While these categories

only form an initial framework for analysis, they prove useful as a starting point for the

process of coding. From these a priori codes, deeper coding is possible using more

emergent techniques.

Emergent Categorization

A more detailed analysis is illuminating of the subject of work zone crash factors,

based on the discussions of the various focus groups. Using the traditional "long table"

method of focus group analysis, extracted text pieces can be grouped according to

content, and categories emerge from the process. As the text is analyzed and re-analyzed,

more and more groupings become possible. This process is aptly named emergent

categorization, because with each review of the text, more categories emerge.

In the first emergent categorization effort, 44 categorical codes were created to

represent the discussion of participants in the five focus groups. Another group of text

items remained, comprised mostly of single speaker issues, and topical items that were

isolated within the context of the larger work zone subject. Similar to the larger category

of 'other', a code for other was used to capture this orphaned subset of codes.

Final Categorization

Emergent categorization provides a level of detail in coding that is suitable for

purposes of content analysis. Because the long table method physically segregates text, it

does not always make it easy to identify strings of text that may contain multiple codes

however. For this reason, a complete version of each focus group text is reviewed after









long table, using the codes derived from the emergent process. Babbie points out that,

"You can always code and recode and even recode again if you want, making certain that

the coding is consistent" [36:324]. Salient differences in either code or content were

resolved by either expanding the code set or collapsing categories. This additional

review of the transcripts can demonstrate a need for more codes, representing more

specific content. The final process of coding rendered 69 unique categorical codes.

Since each focus group is homogeneous in composition, a group representative of a

cross-section of traffic safety stakeholders is needed to verify the codes. The Alachua

County, Florida Community Traffic Safety Team (CTST) provides an excellent

composite group of stakeholders to review codes. Their monthly meeting on April 20,

2006 provided a setting for presenting the findings of the researchers and soliciting

feedback. The group was comprised of individuals from law enforcement (6), FDOT (4),

local public works departments (3), private utility companies (2), emergency medical

services (2), and independent persons (3). The CTST group found no disagreement with

the codes derived from the homogeneous focus groups.

A spreadsheet listing 69 unique categorical codes was created to begin the process

of cataloging the responses of the focus groups. Columns representing the five focus

groups were aligned adjacent to each code, to correlate the code with the focus group.

Each focus group's content was analyzed to identify unique speakers that contributed to

discussion relating to the subject contained in the unique code categories. The numerical

speaker assignment for each focus group participant was noted for every participant who

entered into discussion relating to the code. The table below reflects several of the codes,

and the respective speaker assignments. For example, the code for "Movement of












Construction Equipment" was discussed by speakers 3, 6, 2, and 5 in the Police focus


group, and by speaker numbers listed in the other groups similarly.


Table 5. Unique Speakers by Group and Code
Discussion Issue Police Citizen Industry FDOT LC FDOT Tal
Movement of Construction Equipment 3,6,2,5 8,5, 5,1, 9,4, 1,2,4,5,3,6
Daytime vs. Nighttime Road Work 2,6, 7,8,9,2,10,11 1,5, 9,10, 2,6,5,1,4,
Worker Present 5,6, 2,6,8,10,11,5 5,1,2 4,9, 4,6,
Narrow Lanes 6,5 2, 5,1, 2, 1,5,
Queuing and backups 2,5,6,3, 10,4,8,7, 1,5, 10,1,6,4, 2,4,3,1,
Driver Speed 2,6,5 8,10,2,4,11 1,5,2,3 9,4,2, 1,7,
Driver Distraction (General) 6,1,5 7,8,10,4, 1,5,3 10,9,6,1,4,2 3,
Law Enforcement Static vs Ticketing 4,5,3,6 4,2,10,8,3,9,11 5,1,2 9,3,6,2,4, 2,7,4,
Law Enforcement Visibility 3,6,2,5 5,2,8,4,3 1, 10,6,4 2,4,6,
Enhanced Fines 2,5, 8,9,11,4, 5,1,2 9,4,
Advanced Warning 1,6,5, 3,11, 5,1, 10,9,4, 4,5,3,
Artificial Work Lighting at Night 6, 1,2 4, 5,
Lane Shifts 3,2 2,8,5,
TCD Lighting / Night Visibility 6, 5,2, 1, 1, 1,
Law Enforcement MOT Training 5,6,1, 5,1, 6,2, 2,
Advanced Notice of Work Zone 2,6,5 4,2,10,2,7,8,11 2,5,1, 3,10,4,9,5,
ITS and Variable Message Signs 2,6 8,10,4,6,7, 5, 10,4,5,6, 2,
TCD Maintenance 2,6, 5,7,8,2, 1,5, 1,2,10,9,
Work Zone Project Physical Length 7,8,3, 5,1 6,9, 3,6,
Work Zone Project Time to Complete 6, 2,9,10,7,8,4,3,5 5, 6,4 1,6,
No Shoulder/ Drop off 6, 6,2, 5,2,1,
Visual obstruction created by barrels or other TCD 6,3, 4,2, 1,2,
Work Zone Contributing 6,2,3 7, 7,4,2,
Lane Closures / Merge 1,5 7,10,4,5,8, 5,2 10,9, 2,5,1,6,
Driver Training and Education 2,5,4,3,6, 5,2, 9,6, 2,
Impaired Drivers 4,8,10,2, 9, 2,3,
Location of Crash within the Work Zone 6, 2,1,5 4, 2,1,
Photographs as part of reporting 1, 9,4,6, 4,1,3,5,
Police Reporting Narrative & Diagram 2,3,6,1, 1,5, 2,3,1,5
Flagperson / Worker Action 6,2, 10, 4,
Speed Trailers 2,6,3, 5,3,2 4,9,
TCD Clarity of message 6, 8,11,10,2, 2, 4,
Color / Reflectivity of Items in Work Zone 10, 1, 2,4,
Temporary Rumble Strips 4,9, 2,
Driver Licensing 4,5,2,6,7,8,10, 5,1, 4,6,
Side Street Control during Lane Closures 7,5, 6,4,9,
Liability 1,3 2,1,5 9,
Flagger / Worker Training 6,2 4,9,1,
Law 5,
Changing MOT plans 1,5,2 4,9,1, 1,6,
Highway Advisory Radio 2,8, 6,4,9
Commercial Motor Vehicles 6











Table 5. Continued.
Discussion Issue Police Citizen Industry FDOT LC FDOT Tal
FDOT MOT report 6 9,4,1, 2,1,
Harmful Even sequencing in reports 3,
TDC Speed Signs 3,6,5 3,5,8
Type of channeling device 2,
Length of Tapers 1 10, 5,2, 1,2
Worker Fatigue 9,6
Hydroplane, Standing Water 5,6
Type of Work Being Done 2
Law Enforcement Positioning with TCD (rolling) 6,10,9,2
Law Enforcement not part of MOT planning process 2
Alternate Routes & Detours 2,4,3,6 1
Commonplace of Road Work 2,5 7 5
Productivity or money outweighs safety 3 4,9
Human Toll of Work Zone Crashes 5,8 5
Business Access during Construction 1
Driver Training for Those Ticketed in Work Zones 2,1,5 6
Recurring crash sites 6 2
End Work Zone Signs Needed 2,4,10
Temporary Striping 3,6 9,4
Temporary Curbing 2 2
Presence of Temporary TCD* 2
Type of TCD present / used 5,1 2
Weather Conditions 6
Adding WZ data or fields on reports 2,5,6,3
Driver behavior (general) 6,5, 3,10 1
Police Rolling Roadblock 5,1 2
MOT Devices Hit 2



Agreement Measure

Whether each group's discussion contained a given code was noted as a measure in


analysis. Migrating from unique speakers in each group, based on subject codes, to a


measure of agreement is a relatively simple task. As a more general representation of the


unique speaker assignment, if any code had one or more speakers for the focus group, it


was considered included in the agreement measure. To simplify the task of tracking


these measures, a binary code is created with 1 representing agreement and 0 representing


cases where there was no discussion within the focus group.










When binary codes for individual focus groups are totaled, the level of agreement

among groups can be determined. The highest level of agreement would be represented

by a numerical measure of 5, representing that all 5 focus groups contained some

discussion of the particular code. Conversely, a measure of 1 would indicate that only

one of the 5 groups discussed a code. The measure of agreement being totaled for all 69

codes, a ranked listing is possible, listing those with the most agreement (5) to those with

the least agreement (1). The table below depicts all codes and their respective binary

measures of agreement.

Table 6. Agreement by Group and Code (Binary)
Discussion Issue Police Citizen Industry FDOT LC FDOT Tal Total
Movement of Construction Equipment 1 1 1 1 1 5
Daytime vs Nighttime Road Work 1 1 1 1 1 5
Worker Present 1 1 1 1 1 5
Narrow Lanes 1 1 1 1 1 5
Queuing and backups 1 1 1 1 1 5
Driver Speed 1 1 1 1 1 5
Driver Distraction (General) 1 1 1 1 1 5
Law Enforcement Static vs Ticketing 1 1 1 1 1 5
Law Enforcement Visibility 1 1 1 1 1 5
Advanced Warning 1 1 1 1 1 5
TCD Lighting / Night Visibility 1 1 1 1 1 5
ITS and Variable Message Signs 1 1 1 1 1 5
Work Zone Project Time to Complete 1 1 1 1 1 5
Lane Closures / Merge 1 1 1 1 1 5
Enhanced Fines 1 1 1 1 0 4
Artificial Work Lighting at Night 1 0 1 1 1 4
Law Enforcement MOT Training 1 0 1 1 1 4
Advanced Notice of Work Zone 1 1 1 1 0 4
TCD Maintenance 1 1 1 1 0 4
Work Zone Project Physical Length 0 1 1 1 1 4
Driver Training and Education 0 1 1 1 1 4
Location of Crash within the Work Zone 1 0 1 1 1 4
TCD Clarity of message 1 1 1 1 0 4
Length of Tapers 1 1 1 0 1 4
No Shoulder/ Drop off 1 0 0 1 1 3
Visual obstruction created by barrels or other TCD 1 1 1 0 0 3
Work Zone Contributing 1 1 0 0 1 3
Impaired Drivers 0 1 0 1 1 3
Photographs as part of reporting 0 0 1 1 1 3
Police Reporting Narrative & Diagram 1 0 1 0 1 3










Table 6. Continued.
Discussion Issue Police Citizen Industry FDOT LC FDOT Tal Total
Flagperson / Worker Action 1 1 0 1 0 3
Speed Trailers 1 0 1 1 0 3
Color / Reflectivity of Items in Work Zone 0 1 1 1 0 3
Driver Licensing 0 1 1 1 0 3
Liability 1 0 1 1 0 3
Changing MOT plans 0 0 1 1 1 3
FDOT MOT report 1 0 0 1 1 3
Commonplace of Road Work 1 1 1 0 0 3
Driver behavior (general) 1 1 1 0 0 3
Lane Shifts 1 1 0 0 0 2
Temporary Rumble Strips 0 0 0 1 1 2
Side Street Control during Lane Closures 0 1 0 1 0 2
Flagger / Worker Training 1 0 0 1 0 2
Highway Advisory Radio 0 1 0 1 0 2
TDC Speed Signs 1 1 0 0 0 2
Alternate Routes & Detours 0 1 0 0 1 2
Productivity or money outweighs safety 1 0 0 1 0 2
Human Toll of Work Zone Crashes 0 1 1 0 0 2
Driver Training for Those Ticketed in Work Zones 0 0 1 1 0 2
Recurring crash sites 1 0 0 0 1 2
Temporary Striping 1 0 0 1 0 2
Temporary Curbing 0 0 1 1 0 2
Type of TCD present / used 0 0 1 0 1 2
Police Rolling Roadblock 0 0 1 0 1 2
Law 0 0 1 0 0 1
Commercial Motor Vehicles 1 0 0 0 0 1
Harmful Even sequencing in reports 0 0 0 0 1 1
Type of channeling device 0 0 0 0 1 1
Worker Fatigue 0 0 0 1 0 1
Hydroplane, Standing Water 1 0 0 0 0 1
Type of Work Being Done 0 0 0 0 1 1
Law Enforcement Positioning with TCD (rolling) 0 0 0 1 0 1
Law Enforcement not part of MOT planning 0 0 0 0 1 1
process
Business Access during Construction 0 0 1 0 0 1
End Work Zone Signs Needed 0 1 0 0 0 1
Presence of Temporary TCD 0 0 0 0 1 1
Weather Conditions 1 0 0 0 0 1
Adding WZ data or fields on reports 1 0 0 0 0 1
MOT Devices Hit 0 0 0 0 1 1



A total of 14 codes represent agreement among all 5 groups. Another 10 codes

indicate agreement among 4 groups, and 15 codes each for agreement among 3, 2, and 1


groups.









Weighted Measure of Intensity

All mentions of a code within a focus group provide a potentially valuable measure

of intensity. Because each focus group varies in size, it was important to weight the

measure of intensity, based on the size of the group. Five individual speakers discussing

a code in a group of six has greater intensity than a similar number discussing an issue in

a group often. When homogeneous focus group intensity is weighted, measured, and

summed, an intensity score results. Equation 1 gives the formula for calculating this

intensity value.


I=- (1)



where
I = Weighted Measure of Intensity
S,= Unique speakers discussing code in focus group i
n, = Number of participants in focus group i

For example, "Movement of Construction Equipment" is a discussion topic in all

five focus groups, with 4, 2, 2, 2, and 6 speakers contributing to the discussion in each

group respectively. Since the total number of participants in each focus group is 6, 10, 4,

8, and 7 respectively, the weighted intensity formula would be applied as:

(4 x 6) + (2 x10) + (2 x 4) + (2 x 8) + (6 x 7)
6+10+4+8+7
110
35
=3.1

All 69 codes can be calculated for weighted intensity and their values ranked

accordingly. While the highest possible intensity measure is 7.6, the highest recorded








47



measure was 4.2, and the lowest 0.2. The table below depicts code weighting using the


formula above.


Table 7. Weighted Measure of Intensit
Discussion Issue I
Law Enforcement Static vs. Ticketing 4.2
Queuing and backups 3.8
Daytime vs. Nighttime Road Work 3.5
Driver Speed 3.5
Law Enforcement Visibility 3.5
Advanced Notice of Work Zone 3.4
Driver Distraction (General) 3.3
Lane Closures / Merge 3.3
Movement of Construction Equipment 3.1
Worker Present 3.0
ITS and Variable Message Signs 3.0
Advanced Warning 2.6
TCD Maintenance 2.6
Work Zone Project Time to Complete 2.6
Driver Licensing 2.4
Enhanced Fines 2.3
Driver Training and Education 2.3
Work Zone Project Physical Length 1.9
Impaired Drivers 1.8
Police Reporting Narrative & Diagram 1.7
TCD Clarity of message 1.7
Photographs as part of reporting 1.6
Narrow Lanes 1.5
Law Enforcement MOT Training 1.4
Work Zone Contributing 1.4
Changing MOT plans 1.4
TDC Speed Signs 1.4
TCD Lighting / Night Visibility 1.3
Speed Trailers 1.3
Side Street Control during Lane Closures 1.3
Highway Advisory Radio 1.3
FDOT MOT report 1.3
Alternate Routes & Detours 1.3
Lane Shifts 1.2
No Shoulder / Drop off 1.2
Visual obstruction created by barrels or
other TCD 1.1
Location of Crash within the Work Zone 1.1
Length of Tapers 1.1
Flagger / Worker Training 1.0
Driver behavior (general) 1.0
Flagperson / Worker Action 0.9
Color / Reflectivity of Items in Work Zone 0.9


Liability 0.9
Law Enforcement Positioning with TCD
(rolling) 0.9
End Work Zone Signs Needed 0.9
Artificial Work Lighting at Night 0.8
Temporary Striping 0.8
Temporary Rumble Strips 0.7
Commonplace of Road Work 0.7
Human Toll of Work Zone Crashes 0.7
Adding WZ data or fields on reports 0.7
Productivity or money outweighs safety 0.6
Driver Training for Those Ticketed in
Work Zones 0.6
Worker Fatigue 0.5
Recurring crash sites 0.4
Type of TCD present / used 0.4
Police Rolling Roadblock 0.4
Hydroplane, Standing Water 0.3
Temporary Curbing 0.3
Commercial Motor Vehicles 0.2
Harmful Even sequencing in reports 0.2
Type of channeling device 0.2
Type of Work Being Done 0.2
Law Enforcement not part of MOT
planning process 0.2
Presence of Temporary TCD 0.2
Weather Conditions 0.2
MOT Devices Hit 0.2
Law 0.1
Business Access during Construction 0.1











Composite Ranking of Items

Recall that according to Morgan, the way to determine the emphasis groups give to

a code is measuring how many persons in each group mention a code (intensity) and how

many different groups mention a code (agreement). He refers to this a "group-to-group

validation" [35]. To make the results of agreement and intensity rankings more

meaningful, the measures can be combined to produce a composite list of codes. The

composite list seeks to determine those items that have an agreement measure of 4 or 5

and an intensity in the 80th percentile. The 80th percentile was chosen, since it represents

a measure analogous to that of 4 out of 5 groups. A total of 15 code items met the criteria

of the composite ranking. Table 8 is a side-by-side comparison of the codes with an 80th

percentile intensity (15 codes) and those codes (24) with agreement measures of 5 or 4.

Both columns in the table are sorted alphabetically for ease of comparison.

Table 8. Comparison of Intensity and Agreement Measures
Agreement Weighted Intensity I
Advanced Notice of Work Zone Advanced Notice of Work Zone 3.4
Advanced Warning Advanced Warning 2.6
Artificial Work Lighting at Night Daytime vs Nighttime Road Work 3.5
Daytime vs Nighttime Road Work Driver Distraction (General) 3.3
Driver Distraction (General) Driver Speed 3.5
Driver Speed ITS and Variable Message Signs 3
Driver Training and Education Lane Closures / Merge 3.3
Enhanced Fines Law Enforcement Static vs Ticketing 4.2
ITS and Variable Message Signs Law Enforcement Visibility 3.5
Lane Closures / Merge Movement of Construction Equipment 3.1
Law Enforcement MOT Training Queuing and backups 3.8
Law Enforcement Static vs Ticketing TCD Maintenance 2.6
Law Enforcement Visibility Work Zone Project Time to Complete 2.6
Length of Tapers Worker Present 3
Location of Crash within the WZ
Movement of Const. Veh/Equipment
Narrow Lanes
Queueing and backups
TCD Clarity of message
TCD Lighting / Night Visibility











Table 8 Continued
Agreement Weighted Intensity I
TCD Maintenance
Work Zone Project Physical Length
Work Zone Project Time to Complete
Worker Present Highlighted = 4 of 5 groups


When the top 20% of the table for weighted intensity and all agreement measures

of 4 or 5 are integrated, table 8 is the resulting list of 14 codes. No ranking of these items

is necessary, so they are listed alphabetically for ease of reading between tables.

Table 9. Composite Codes
Composite Codes
Advanced Notice of Work Zone
Advanced Warning
Daytime vs Nighttime Road Work
Driver Distraction (General)
Driver Speed
ITS and Variable Message Signs
Lane Closures / Merge
Law Enforcement Static vs Ticketing
Law Enforcement Visibility
Movement of Construction Equipment
Queuing and backups
TCD Maintenance
Work Zone Project Time to Complete
Worker Present



Interpreting Analysis Definition of Elements

While reducing text and audio to codes is the important first step in qualitative

analysis, reconstructing content to derive some meaning is equally important. In

accomplishing this task, the researcher seeks to create linkages in sometimes subjective

meaning. Each focus group participant and each group discussion is unique. The task for

the researcher is to find common themes in words to 'discover' relationships. It is from

those relationships and meanings that we can look beyond mere codes and understand the

basis of the content. This process will ultimately allow codes to be transformed into

potential crash report data elements.









The following section examines the composite list of codes by presenting

descriptions of agreement and synopsizing focus group discussion.

Creating Linkages

Advanced Notice of Work Zone-Four of five groups discussed the need for public

notification of work zone activities, in advance of such work. The use of print and

electronic media, and the deployment of variable message boards before construction

begins were cited as important ways to inform drivers before changes in the driving

environment occur. A representative statement from the citizen focus group gives an

example of the discussion, "I think the community would be better off if there was a

mailing letting you know what exactly this project is."

Advanced Warning Signage-Advanced warning signage was noted as import in

all five focus groups. Warning drivers of potential changes in the driving environment

was cited as important because of factors related to driver expectancy. An example of the

discussion is evident in the police focus group where the following was said, "A lot of

times the construction company will go in and make changes in the pattern or flow of

traffic and they'll make these changes and there isn't often adequate signs for it."

Daytime vs. Nighttime Road Work -Every group discussed differences between

construction activities occurring during the nighttime versus the daytime. While there

was no consensus regarding which time of day was perceived to be safer, the effects on

both traffic flow and safety were noted and discussed. A representative comment from

one of the FDOT focus group noted, "There are also times where we require operations to

be done at night and you can't do them in the daytime because of the traffic volume

impact."









Driver Distraction-Similar to driver speed, driver distraction was noted in every

group as a significant reason that work zone collisions occur. Distractions inside and

outside the vehicle are viewed as important factors for drivers in work zones. An officer

from the law enforcement group pointed out, "More and more distractions take place

inside the vehicle, then they're really not paying attention to what's going on outside the

vehicle."

Driver Speed-Without fail, every group noted driver speed as one of the very first

discussion issues that contribute to work zone crashes. The failure of drivers to comply

with normal or reduced speed limits in work zones was particularly relevant in focus

group discussions. The FDOT focus group conducted in Tallahassee noted, "We try to

slow them down sometimes 10-20 mph below the speed limit and they continue to travel

10 mph faster than the speed limit." The police confirm the role of speed with comments

such as, "I got to concentrating at a place that is 45 and they are not usually running 45."

The contractor group noted succinctly, "They don't slow down."

ITS and Variable Message Signs-All focus groups discussed the use of variable

message signs as advanced warning devices and additionally their role as advance notice.

A representative statement from the FDOT focus group in Lake City was, "Variable

message boards, keep them up to date... change your message on it a lot. It keeps them

looking at it, helps out."

Lane Chi'm,,t'\ .erges-All focus groups discussed lane closures and merge

operations as factors in work zone crashes; however, the nature of the discussion was

varied. The citizen group discussed driver behavior in merge situations, while other

groups discussed the warning ahead of a merge, and the place within the merge where









crashes occur. A characteristic comment from the citizen group notes, "Sometimes its

really unclear if one lane has been closed." Another citizen participant added, "I notice

that's very irritating you see a sign that says 'right lane closed ahead,' and everybody

jumps in that right lane and they just keep going and going."

Law Enforcement Static vs. Ticketing-All groups discussed the importance of

active law enforcement in work zones, to modify driver behavior. The use of static patrol

cars, parked in or near work zones, was noted as not being as effective as the visible car

and officer engaged in enforcement action. The citizen focus group noted, "I think that

there should be an officer or two officers actually walking the construction site in

between the workers." Engineers from the FDOT focus group in Lake City made

statements like, "We don't want them (police) sitting on the side of the road, but want

them writing tickets."

Law Enforcement Visibility-Similar to extensive discussion by all groups about

law enforcement action in the work zone, it was certain among participant groups that

law enforcement visibility has an impact on driver behavior. Visible presence of

enforcement vehicles and officers is seen as relevant to compliance with speed limits in

work zones. A representative comment from the citizen focus group notes, "I like the

fact that they have a highway patrol car there. That always gets my attention. The

flashing lights." A participant from the contractor group adds, "One of the biggest

deterrents for us that I've seen is when you do have an officer out there and the blue

lights are on."

Movement of Construction Vehicles-All groups discussed the movement and

actions of construction vehicles as potentially contributing, directly or indirectly, to work









zone collisions. Vehicles entering or crossing traffic were noted, as well as slow moving

vehicles which enter or leave work zone areas. A comment from the Lake City FDOT

focus group is representative of the discussion, "I didn't have such a problem with pickup

trucks and vehicles that generally can accelerate and get across the road real fast. But

when you start putting a 20 ton dump truck trying to run across the road, I don't like

that."

Queuing andBackups-All groups mentioned rear-end-type collisions as being

particularly problematic in the work zone setting. Reductions in speed limits, physical

changes in the driving environment, and driver actions were all noted as contributing to

queues and backups. A representative comment from the police group was, "Traffic

stops and he rear ends it."

Traffic Control Device Maintenance-Four of five groups cited the maintenance of

traffic control devices as an important factor in evaluating crashes in work zones.

Misaligned cones and barrels, along with lighting on devices were specifically mentioned

as relevant. The contractor group noted, "I know for some of our job sites, we've got an

assigned MOT crew where it's sometimes two guys and they ride the site all day pretty

much setting up cones, taking them down, and maintaining what we've got that's out

there."

Work Zone Project-Time to Complete-The long duration of construction activity,

often months or years, was noted by all groups as a factor in crashes. Driver

complacency with warnings and environment changes was believed to occur when

projects take so long to complete. A representative comment from the citizen group

summarizes the issue, "Boy highway construction takes along time."









Workers Present-Every group noted differences in the dangers of a work zone

when workers and construction activity is present. All groups indicated that it was a

relevant factor in crashes whether or not workers are present. One comment from the

citizen group relates, "So they don't slow down figuring, well I can wait because if there

are no workers there then you are not endangering them."

The descriptive summary of how each code was discussed among focus groups is

quite revealing. When accompanied by just a few sample quotes from the focus groups,

the codes become more real and illustrate the process of linking codes back to actual

content.

Converting Codes to Data Elements

Focus group discussion items make up the codes used in qualitative analysis.

Codes that were determined to be of sufficient importance were subsequently identified

as potential crash data elements, through a process of linking them with focus group

content. The next step in the data element development process is to determine which

codes have potential value for inclusion in an effort to improve crash data. Moving

potential codes to data elements requires making a determination of whether the data

sought is available from other resources.

According to the MMUCC, it is desirable to create linkage to other sources of data

whenever possible to reduce the burden of data collection at the scene [6]. The term

linkage here is dissimilar from that which was used in qualitative analysis to clarify

codes. In this case, linkage refers to managing data sets so that they can be combined or

merged. For example, a traffic crash report would not capture some information about

injury mechanics, because that information could alternatively be obtained by linking

with EMS, hospital, or insurance records.










Department of Transportation project files would likely be a source for information

on advance notice and public information efforts, as well as the spatial and temporal

length ofprojects. The speed of the vehicle and the determination of whether speed was

a contributing cause are current data elements on the Florida Crash Report, as is driver

distraction. Law enforcement ticketing activities may be obtained from traffic citation

records or other data stored within enforcement agencies.

The role of daytime versus nighttime work is potentially a derived data element, if

current reporting fields capturing lighting conditions are married with a new element

capturing workers being present.

Table 10. Data Linkage
Composite Codes Linkage Potential
Advanced Notice of Work Zone FDOT project records
Advanced Warning
Daytime vs Nighttime Road Work New Element + Current
Driver Distraction (General) Current Report Field
Driver Speed Current Report Field
ITS and Variable Message Signs
Lane Closures / Merge
Law Enforcement Static vs Ticketing Citation Data
Law Enforcement Visibility
Movement of Construction Equipment
Queuing and backups
TCD Maintenance
Work Zone Project Time to Complete FDOT project records
Worker Present


Elements as Interrogatives / Binary Values

After codes are filtered using analysis techniques, eight codes remain, that are

potential work zone crash data elements. The guidance of the MMUCC requires that data

elements be appropriate, that is, they must be needed for traffic safety purposes and not

be administrative in nature [6]. All of the remaining codes meet that requirement. While

the MMUCC seeks to minimize the total number of data elements in the interest of

officer time, the proposition of supplementing data collection does not conflict with that









objective. Since work zone crashes are fairly infrequent events, supplemental data

collection would not unduly burden individual officers. Supplemental elements for work

zones would not be used in cases where there was not a work zone involved. Null data

would not be captured.

Having identified topical work zone issues in the form of codes and subsequently

linking discussion to form context, it is then possible to form the basis of data collection,

the data element. Traditional crash reports make extensive use of terse categories that

represent data elements and pick lists of attributes. Attributes are typically given alpha or

numeric codes to make them suitable for data entry and analysis. Given the objective of

broadening data about work zones without burdening officers, the use of interrogative

statements can speed data collection. Transforming potential data elements into complete

questions allows for the use of simple "yes/no" responses from officers. This allows for a

more accurate description of work zone features that are sometimes not familiar to police

officers. The state of Wisconsin (Figure 2) makes use of "yes/no" input fields, and

Pennsylvania (Figure 3) employs interrogatives with check boxes to capture certain

information in their reporting format. The use of both the "yes/no" input method and the

interrogative as a data element are both proven techniques for data collection.

Creating questions from potential data elements is the product of again examining

the context of focus group discussion. Stakeholders discuss work zone issues in various

degrees of detail, and consequently provide insight into their data needs.


















Mth Percande Weatghted bntetntty
Advanced Notice of Work Zone 3-4
Advanced Warning 2_6
Daytime vs Nighttime Road Work 3.5
Driver Distraction (General) 33
Driver Speed 3.5
ITS and Varriable Message Signs 3
Lane Closures / Merge 3-3
Law Enforcement Static vs Ticketing 4.2
Law Enforcement Visibility 3.5
Movement of Construction Equipment 3.1
Queueing and backups 3 8
TCO Maintenance 2.6
Work Zone Project Time to Complete 2.6
Worker Present 3
----------------------


Data Analysis Process


coGoampe teCoOde
Advanced Notice of Work Zone
Advanced Waring
Daytime vs Nighttime Road Work
Driver Distraction (General)
Driver Speed
ITS and Varriable Message Signs
Lane Closures I Merge
Law Enforcement Static vs Ticketing
Law Enforcement Visibility
Movement of Construction Equipment
Queueing and backups
TCD Maintenance
Work Zone Project Time to Complete
Worker Present


UnlNage Potantsl
FDOT records

New Element + Current
current Report Field
current Report Field


Citation Data




FDOT records


Renam riag aoa

Advanced Waming



ITS and Varriable Message Signs
Lane Closures Merge

Law Enforcement Static vs Ticketing
Movement of Construction Equipment
Queueirn and backups
TCD Maintenance

Worker Present


IWorker Present
Figure 8. Data Analysis Process


Agreement Level or81
Advancd Nicce of Work Zone


Advanced Warning
ifArtW.alWoV LOling at Niht
Daytime vs Nighttime Road Work
Diver Distraction (General)
Driver Speed
Drver Trainieg and Educafiont
Enhanced Fines-
ITS and Varriable Message Signs
Lane Closures / Merge
Law urtpram .MOCT Twra
Law Enforcement Static vs Ticketing
Law Enforcement Visibility
Length:dfTapewre
Locatin of CTash within thfeWZ
Movement of Const. Veh/Equipment
Narrow Lanes
Queueing and backups
r Clarity ofmeasage
TCD Lighting I Night Visiblllty
TCDEMaintenance
ioiak Zone Prject Phyical e Leth
Work Zone Project Time to Complete


~


IL










One can determine the type of data they seek, and construct report questions that

officers can answer, thereby improving the data set. .It would be impossible to create a

data element and associated set of attributes for every conceivable work zone incident

scenario. Even the availability of multiple attributes cannot satisfy the myriad of

situations one may observe. The crash report diagram and narrative are excellent tools

for documenting things that are not suitable for coding. Officers often neglect to fully

utilize the crash report narrative and diagram to document work zone attributes and

observations. The supplemental system for collecting work zone incident data can also

serve as a pointer system, reminding the officer to collect additional information, and

informing analysts examining the data element that the master crash report form may

contain additional data. This bolsters the effectiveness of the supplemental system,

without duplicating the value of the crash report narrative and diagram.

Table 11. Converting Elements to Supplemental Report Questions
Potential Element (Code) Report Interrogative (Yes/No Format)
Queuing and backups Did a backup or queuing of traffic contribute to the
crash?
Law Enforcement Visibility Was an on or off-duty police officer working in the
construction zone nearby?
Lane Closures / Merge Did the crash occur within a lane closure or merge
section of roadway?
Movement of Construction Equipment Did the movement of construction trucks or equipment
contribute? If so, explain the type of equipment or
vehicle and nature of the movement in the crash report
narrative.
ITS and Variable Message Signs Was a variable message sign or arrow board used to
warn of construction ahead? If so, please include the
location in your crash report diagram.
Workers) Present Were workers present in the vicinity of the crash?
Advanced Warning Were advanced warning signs in place? Please be
sure to include the location of advanced warning signs
in the crash report diagram.
Traffic Control Device Maintenance Were temporary traffic control devices (signs,
barricades, cones, pavement markings, etc.) in good
condition and proper working order at the time of the
crash? If "No", please describe in the crash report
narrative.









By reminding the officer to include the type of construction vehicle or equipment

and the nature of the movement in the crash report narrative, the data set is accomplished.

When the officer adds the location and position of message boards and advance warning

devices to the report diagram, valuable information concerning those two elements is

documented.

Supplemental Collection System Element Testing

Developing new traffic crash report data elements, specifically for work zone

incidents, sets the stage for improving the crash data set. Merely identifying potential

data elements, however, does not fulfill the objective of this research effort. The second

component required for supplementing work zone data collection requires a mechanism

for actually collecting data at the officer level. Once such a mechanism is chosen, actual

testing by field officers, in real crash scenarios, can help transform conceptual

supplementing of work zone data into an actual proof of concept.

While creating a simple paper form may be the easiest way to capture additional

data, it is decidedly not the preferred method. Using a web-based system allows officers

to expedite data collection, and takes full advantage of technologies currently available to

officers. To implement such an electronic system, technical and operational frameworks

must be considered.

Technical Framework of Collection System

A web-based collection system requires, in its simplest form, a database structure

and a computer network structure. Database structure describes the data fields used,

while the network structure describes the user interface, hosting, and storage systems.

Using an eXtensible Markup Language (XML) data schema is a requirement for

this type of activity, since the overall objective is maximum transportability of the data.









Any number of commercial databases could be used to create the database structure, but

since ActiveX controls are a security concern, Microsoft SQL 2000 was chosen. This

powerful database application allows for the expedient development of databases using

tabular systems. Most importantly, enhanced security features ensure that host systems

will be protected from potential intrusion.

Before simply including work zone specific data elements, the prospect of linking

the supplemental database with a larger traffic crash record system is essential to

complete data. Primary key fields must be created to provide the link between

independent data sets and/or tables used by the database. The unique and sequential

"HSMV" number on each Florida Traffic Crash Report Form provides an excellent field

for linking data sets. The supplemental database will include an HSMV number that will

use the same numeric format as the original. The database will also include a feature that

returns an error if the user enters a duplicate number. Subordinate to the HSMV report

number key field, additional data fields can capture data that duplicates other Traffic

Crash Report Form data, for the purpose of providing additional linkages, and also to

make the supplemental database somewhat stand alone. These additional data fields are

date, county, officer ID, and location of the crash. The officer ID is automatically

captured from the officer login, via an authorized user table. An additional variable for

"Long Form" or "Short Form" differentiates the traffic crash report type, based on

reporting thresholds established by Florida Statute. This variable provides an additional

mechanism by which to sort, search, and link the supplemental database with the two

types of master crash report. This is beneficial, since "Short Form" reports are not

considered for Florida statistical reporting purposes. These reports are however,











automated by the FHP, and therefore potentially enhance the overall data set, when

combined with the supplemental work zone reporting data.

Having created a database structure to handle the basic data necessary for linking

the supplemental database, the remainder of the database structure involves the inclusion

of the twelve data elements that were created herein. The structure of these data elements

is simplified by the use of binary codes that represent agreement with the interrogative

that forms the basis for the element. For example, the question "Did a backup or queuing

of traffic contribute to the crash?" would be captured as a yes/no response, and stored in

the supplemental database as a value of FALSE for "no" and TRUE for "yes". The

figure below depicts the data structure used for the supplemental work zone data

collection database.



FLORIDA TRAFFIC CRASH REPORT DO NOT WRITE IN THIS PACE
LONG FORM
MALTa i0.OFHMHwIGHETY WMTOR & VBE* TRUnFC CMH
OCORIAM E no aOalaF T t CRSSSEE. nH W
A1TEOFCIt IHE OFLL~AI TIME OFFICER NOT1FEO TIM OTFRCi AfRRVED RgVEST AG~YHC RBI'IT NIEIfMeP (tiW~HRPORT UBE
I I A D. D D60 1H,60571037
COUNYr CfCD FEET U CT g/ & ON (OM at.i,^ --I CO RGY
tblWZdata
ATNooXm a iM a S u,, HSMVnum DM.. o vwa oM -A-
wzDate TI "
E ATT)EIHEtilT)OF w f EK FB(fat*BMEtI OFCF
AT L-1-I--R- ~wzCounty
-wzPlace
wzReporttype tbCounNames
OfficerlD-- CountylD
Queing CountyName
LE TrooplD
Iblwzlocation Merge
wzlD Equipment tblOfficerLoaon
wzD Equipment
wzlocation TCDmaint trpNam
trpName
VMS
Workers
Warning


*Florida Traffic Crash Report shown for illustration purposes only

Figure 9. Supplemental Work Zone Database Structure

Microsoft ASP.NET was selected to create a simple web-based design interface for

officers who would use the supplemental database system. The software offers excellent









tools for the user interface and exceptional support for desirable security features.

Officers accessing the supplemental reporting intranet web page are screened in, based on

their access to the wireless network and their logon password. Their ID number is

automatically captured and that information populates the screen and database. A date

picker tool is provided to assist officers with the date field. Table-driven drop down lists

are used to assist officers in selecting the county of crash and major work zone project

locations underway. Radio buttons are provided to toggle between the "Long Form" and

"Short Form" report types. The HSMV report number is a manual entry, with a feature

that prohibits duplicate numbers being used. Each of the seven work zone-specific

supplemental data elements is displayed as a question, listed on the screen for the user to

see. "Yes" and "No" radio buttons accompany each question. Edit rules require one of

the two buttons to be selected or an error message will appear. A "Yes" selection is

saved as a "True" value and a "No" choice is saved as a "False" value in the database. A

"submit" button on the bottom of the page allows the user to store the record and exit the

system. Minimum data requirements are date, county, location, report type, and HSMV

number, along with a choice of "Yes" or "No" for each question. If the minimum data

requirements are not met, the user cannot save the record and an error message prompts

them to reconsider the offending value. Figure 10 depicts a screen shot of the actual web

page.

The data collection web page is designed to be simple to use and self-explanatory.

As part of the web design, however, help is available for officer users to explain

individual data elements. Help buttons associated with each data element direct the user

to a text file where context sensitive help is provided. The scope of the help file is to









better explain the objective of the data element, and provide illustrations or other support

when appropriate.

A separate web page was developed for administrative review of the supplemental

database. Microsoft ASP.net provides a simple report that lists all records in tabular

form, with rows representing unique records and columns representing all of the database

fields. The data is readily imported into any database project or Microsoft Excel for

manipulation. Figure 11 is a screen shot of the ASP.Net report page.

For purposes of the supplemental work zone data collection system, the existing

architecture of the FHP system lends itself well to a web-based approach. Since mobile

computing, communications, and server systems are currently in place within the agency,

setting up a database for data collection is quite simple.

The Florida Highway Patrol computer network provides the backbone for the

supplemental data collection system. Every patrol vehicle in the agency is equipped with

a laptop computer connected to the network via a continuous cellular link. Officer laptop

computers are primarily used for applications associated with computer aided dispatch

and crash reporting software. Officers also use their computers for accessing local, state,

and national information systems, obtaining information on persons, vehicles, and

articles. The agency recently migrated to an "i-evidence" system that allows troopers to

catalog property and evidence they seize, prior to entering the items into storage rooms.

The job of the FHP trooper is highly automated and most personnel are very comfortable

with their use.
















1..lioo W n Re rin soft Iinternt IEx. It51i'


File Edit View Favorites Tools Help

Back hi D Search wFavorites C0 W Li .
Address 14 http: /www2.fhp.stateFl.us/troopglWZInputasp


v 1 Go Links
Ak Al


Date: __

HSMV:


County I-Select County- I Work Zone Location: [-Select Location-

Report Type: 0 Long 0 Short Employee ID#:


Please answer the following questions by clicking the appropriate "Yes" or "No" button


Yes 0 No 0 Did a backup or queuinlg of traffic contribute to the crash? B
Yes No 0 Was an on or off-duty police officer working in the construction zone nearby? B
Yes 0 No 0 Didthe crash occur within a lane closure or merge? U
Yes 0 No 0 Did the movement of construction trucks or equipment contribute? B
If "yes", explain the type of equipment or vehicle and nature of the
movement in the crash report narrative.
Yes 0 No 0 Was a variable message sign or arrow board used to warn of construction ahead? B
If "yes", please include the location or position in your crash report diagram.
Yes 0 No 0 Were workers present in the vicinity of the crash? B
Yes 0 No 0 Were advanced warning signs in place? B
If "yes", please be sure to include the location orposition
of advancedwarning signs in the crash report diagram.
Yes O No O Were temporarytraffic control devices (signs, barricades, cones, pavement markings, etc.)
in good condition and proper working order at the time of the crash? B
If "No", please describe in the crash report narrative. (ie., "cones in the roadway" or "signs faded")


SSubmit Report
111


] (4 digits)


Figure 10. Supplemental Data Collection Web Site


S )>1















File Edit View Favorites Tools Help
Back Search Favorites *
Address I| ] http :,www2.fhp.state.fl.usitroopg.records.aspoffset=-1 V Go Links





Work Zone Data Report

Records 16to 30 of 30

IHSMV IDate Coii-uty rLocation Report Type QuIein1g LEO Merge Equipment VMS Workers [Warninlg TCD Mailt Officer ID
76984282 1611 82006 iDuval 11-95 Trout River Bridge Long Yes |Yes Yes No [Yes Yes Yes No 1183
76976690 16i202006 iDuval 11-95 Trout River Bridge Long No No No INo .Yes Yes Yes No 602
76983770 162212006 Flagler 11-95 FlaglerWidening Long No FNo No No rNo Yes Yes No 1624
76975941 16f232006 Flagler 1-95 FlaglerWidening Long Yes No Yes No No Yes Yes No 1909
76975944 162512006 St.Johns IOther Location Long No No No I No NNIo o Yes INo 1909
76976692 16f2812006 Duval 11-95 Trout River Bridge Long No No No No Yes No IYes No 602 J
76986067 61/282006 Flagler 11-95 FlaglerWidening Long No N No No No No Yes No 1302
76986265 162912006 Duval 11-95 1-10 Interchange Long No No lYes No |Yes Yes No No 1522
76983771 613012006 Flagler 1-95 FlaglerWidening Long IYes No No No |No Yes No No 624
76983772 7f1f2006 Flagler 11-95 FlaglerWidening Long Yes FNo No No |No Yes IYes No 624
76984042 1712006 IPutnam I0ther Location Long No No INo INo INo No Yes Yes 1438
76985819 17f22006 Putnam IOther Location Long No No No INo |Yes No Yes No 2155
76983773 7/412006 Flagler 11-95 FlaglerWidening Long No No No INo -No Yes Yes No 624
7698404317f5f2006 IPutnam IOther Location Long No NoNo No No Yes No Yes Yes 1438
76986069 1762006 Ist. Johns Other Location Long No No No o NNo o Yes Yes Yes 1302

Show All Records

First Previous

View/Edit Work Zone Locations I Work Zone Data Input


Figure 11. Work Zone Data Results Screen









Laptop computers in each patrol vehicle communicate to a central State of Florida

shared resource switch via Global System for Mobile Communications (GSM)

technology. From the switch, information is routed to various destinations, based on type

of traffic. For the purposes of a supplemental work zone database, that information

would pass to an FHP proxy server, where it would further be routed to a web server. A

firewall would enhance protection as the data moves to a database server, and ultimately

to the storage system. The figure below depicts the basic architecture of the FHP system.

State of Proxy Web Database
Florida Z

Laptop computer hared
Co Tower urce Server ever Firewall Server
Figure 12. FHP Mobile Computing Architecture

Field Implementation and Testing of Collection System

Knowing the information to collect at the scene of a work zone traffic collision is

the central purpose of this report. Creating a mechanism for actually collecting the data

was a subordinate, but integral component of the effort. For the concept to be proven,

however, a demonstration project is required. The Florida Highway Patrol worked

closely with the researchers on this project and volunteered to participate in a small pilot

of the supplemental data collection system. Troop "G" covers nine northeast Florida

counties (Nassau, Duval, Baker, Clay, St. Johns, Bradford, Union, Putnam, and Flagler)

and is based in Jacksonville, Florida. Since several major road construction projects are

underway in the troop, they were selected to test the supplemental system. The troop is

large enough to obtain a reasonable usage, but far short of a statewide rollout of the

product.









Florida Highway Patrol Director Colonel Christopher A. Knight issued a

memorandum to all 147 sworn personnel assigned to Troop "G". The memorandum

provided information about the supplemental work zone system and directed them to use

the system anytime they conducted a traffic crash investigation in a work zone. In

addition, the logon screen for individual officer laptop computers displayed a reminder

message to troopers. Within the first 5 days of implementation, several records were

recorded, indicating that troopers were actively using the system. Rollout coincided with

other laptop training given to all personnel during the first week of June 2006, so all

personnel were able to obtain assistance on the use of the system if required. Because of

the familiarity of troopers with both crash reporting and the use of computers, as well as

the simplicity of the web-based collection system, there were very few user issues.

Results of Collection

The early days and weeks of data collection reinforced the acceptance of officers to

the concept of supplemental reporting using a web-based approach. The collection

mechanism proves simple in design and application. The collection system and pilot

project continues through the end of 2006.

Records from the supplemental database were successfully linked to their

counterpart records in the traffic crash reporting system. This proves portability of the

data and reinforces the objective that the work zone crash data set be improved and

expanded.

Validation of Elements

The Model Minimum Uniform Crash Criteria, MMUCC, provides general

guidelines for work zone data elements. One or more of these data elements are

generally used by the states that gather data about work zone crashes. A priori, one









would look to these data elements as the state of the practice, against which potential data

elements would be measured.

Two of the four work zone data elements contained in the MMUCC are consistent

with the results of this research effort. Essentially, through qualitative research, these

common data elements are mutually validated.

"Was the crash in or near a construction, maintenance or utility work zone?" This

MMUCC data element is already present in the Florida Traffic Crash Report Form. The

Florida form uses "none, nearby, or entered" for attributes while the MMUCC offers,

"yes, no, unknown". Subtle differences in attributes can be potentially complicating,

creating more justification for binary values, yes/no, although this can also lead to more

ambiguity. Because this particular data element is already included in the Florida format,

it will not be duplicated as part of a supplemental effort.

"Location of the crash" is an MMUCC data element that seeks to pinpoint the

location of the crash within the limits of the work zone. The attributes associated with

this element are, "before the first warning sign, advanced warning area, transition area,

work area, termination area". These variables do not represent descriptions that officers

would readily understand, and the values, as depicted in the MMUCC, were not

supported by qualitative research. This effort identified "merge areas" and "lane

closures" as relevant locations within a work zone. The location of the crash within the

work zone is simplified by determining if it occurred within a "merge" or "lane closure".

The value in pinpointing the location more specifically was not reinforced by research.

"Workers present?" is another MMUCC data element that directly corresponds to

the findings of this report. The attributes for this variable are, "yes, no, or unknown".









Qualitative research produced a high emphasis on the presence of workers, thus the data

element is valid.

"Type of Work" is an MMUCC data element that seeks to identify the nature of the

road work in some general terms. The attributes for this data element are, "lane closure,

lane shift/crossover, work on shoulder or median, intermittent or moving work, and

other". Similar to the MMUCC guidance on location, this particular data element may be

difficult for the officer to identify, given the variety and complexity of work zone

projects. The utility of the data element did not screen into a supplemental collection

system through qualitative research. The "type of work" data element will be discarded

for purposes of this effort.

In addition to the four data elements found in the MMUCC, other data elements

were derived from the qualitative research process. These data elements should be

included in any supplemental data collection effort. The influence of queuing or backup,

presence of police in the work zone, nighttime work activity, the presence of advanced

warning devices, the use of message/arrow boards, and the movement of construction

equipment were determined to be additional variables of value.

Comparing the state of the practice data elements of the MMUCC with the data

elements produced through qualitative research, we find that the latter are more

encompassing and more representative of stakeholder needs. Two of four MMUCC

elements are directly supported, one is supported with modification, and one is not

supported as relevant. A total of fourteen data elements were produced through

qualitative research, representing six new data elements that should be considered for






70


inclusion in any collection methodology that seeks to fully explain the work zone

incident.














CHAPTER 5
CONCLUSIONS AND RECOMMENDATIONS

Conclusions

With continued demands to improve the roadway transportation infrastructure, it is

certain that work zones will continue to be a prominent part of the driving environment.

Since a primary objective of roadway design, maintenance, and improvement is safety, it

is only logical that the activity undertaken in that improvement, work zones, be made as

safe as possible. Understanding roadway incidents in work zones continues to be a way

in which trends can be identified and countermeasures developed. The difficulty is

translating what occurs in actual traffic crashes into data that can be later analyzed.

Stakeholders have a great deal to contribute when crash reporting systems are

modified. This research effort has shown that qualitative research methods are a

desirable way to capture what stakeholders have to say. Qualitative analysis has been

demonstrated to be a viable way in which we can better understand that contribution.

Qualitative research produced a total of 14 potential data elements for use in

conjunction with supplementing the work zone data collection effort. Of those 14

elements, 8 actually screen in a new collection methodology, given the remainder are

capable of being obtained through linkages with other data sets. Advanced warning, law

enforcement visibility, lane closures and merges, backup and/or queuing traffic, variable

message signs, moving construction vehicles, traffic control device maintenance, and the

presence of workers are the elements produced by analysis.









Qualitative analysis provides a basis for making conclusions about stakeholder

input, where mere meetings with interested parties cannot. While there are many ways to

approach analysis, this effort determined which items were important by measuring

agreement between groups and the number of persons within groups who discussed those

items. Measuring agreement and intensity provides "group-to-group validation" and

increases the chances that the correct items are seen as those most significant.

Qualitative results are desirable, given they are more defensible, and ostensibly more

valid.

The second component of enhancing the work zone incident data set involves

exploring alternative collection mechanisms. Chapter 4 details the considerations for a

supplemental collection system. A desirable solution for supplementing traffic crash

reporting would be cost effective, not affect current reporting systems, and ultimately be

palatable for the institutions that are charged with managing crash data. By creating a

stand alone web-based collection system, expensive programmatic software and/or form

changes are avoided, and new data can easily merge with existing data by observing

XML schema. Since current reporting systems were not affected by this project,

institutional support was not an issue. It was concluded that this was the best approach

for this project.

Recommendations

A tradeoff exists when using law enforcement to supplement work zone incident

data. They do not necessarily possess the detailed knowledge of work zone traffic

control design, nor the FDOT standards indexes that dictate their setup. They do

however respond to almost all incidents regardless of time or location, providing

maximum access to scenes. While reporting by engineers potentially resolves issues of









technical knowledge of work zones, their access is limited by their lack of 24-hour

availability. The short-lived nature of the temporary traffic control is precisely the

variable for which their expertise is needed. Either methodology has potential, but their

respective shortcomings should be resolved for best results. The need to balance access

and expertise must continue to be evaluated in future efforts.

Similar to the role that both law enforcement and engineers potentially assume in

incident reporting, this research illuminated a greater need for law enforcement

participation in the work zone planning process. The design and implementation of

temporary traffic control in work zones can benefit from law enforcement input, to

promote safety for motorists and officers who travel in those areas.

Through this research effort, the notion of photographing work zones in

conjunction with incident investigation received much discussion among stakeholders.

Although the topic of photography did not screen in as a potential data element, it is

certain that such a use of technology may bridge the gap between access and expertise.

Cost-effective digital photographic equipment and readily available storage and retrieval

systems make routine photography of work zone crashes a tangible proposition. Such

applications should be considered for future studies.

Future Work Zone Applications

This research has identified ways to supplement the work zone incident data set.

While this effort originally envisioned a way to apply supplemental collection to all

crashes that occur in work zones, the pilot test conducted by the Florida Highway Patrol

revealed that a narrower objective would be more appropriate. Rather than require

supplemental work zone data collection in all work zone crashes throughout Florida, such

a system may be better used in a project-specific way. The data derived from a specific









work zone location, say a major interchange project, would likely be more meaningful

than an attempt to capture additional data on all work zones. When married with project-

specific information, the supplemental crash report data would have the potential to be

much more illuminating. In addition, a project-specific approach has the potential to

provide real time data, expediting countermeasures when necessary.

Other Applications

Unique crash scenarios are often difficult to study because they occur infrequently,

because current reporting systems do not adequately address data collection needs, or

some combination of both. Like the case of work zones, current traffic crash report

forms do not contain the level of detail needed for analysis in such cases. Crashes

involving school buses or school zones, fatal crashes, crashes involving emergency

vehicles or motorcycles, or those occurring on bridges are all examples of unique crashes

where supplemental data collection may be helpful. A method for identifying data

elements using qualitative research and collecting data using web-based systems may be

useful.














APPENDIX A
OVERVIEW OF QUALITATIVE RESEARCH METHODS

Qualitative research is, "Research involving detailed, verbal descriptions of

characteristics, cases, and settings. Qualitative research typically uses observation,

interviewing, and document review to collect data."[24] This form of research is rooted

in the social sciences and is an excellent vehicle for examining things that may not be

measurable quantitatively. Qualitative research can be accomplished through surveys,

questionnaires, personal interviews, researcher observation, or similar methods. The

research seeks to learn more about things in their natural environment through people's

attitudes, perceptions, recollections, and feelings.

Getting people together for the purpose of determining direction for a project need

not fall victim to problems associated with group dynamics. One way to potentially

produce more reliable results is the use of qualitative research. For our purposes,

enhancing the content of crash reports is best undertaken as a function similar in

approach. Using qualitative research, investigating the topic of crash incident data

represents a form of collaboration. Similar to currently used methods, stakeholders such

as law enforcement, engineers, construction industry representatives, drivers, and safety

advocates form the basis for input. But by employing a qualitative research

methodology, issues related to group dynamics can be minimized and a more reliable

product is possible. A more dependable consensus among stakeholders is also possible

with qualitative methods, since they employ an approach that is grounded in social

science.









To set the stage for qualitative research as a methodology herein, examining several

qualitative methods will be beneficial. Delphi technique, survey research, and focus

groups are qualitative research methods that may be useful. Survey research streamlines

the collection and analysis of input from larger audiences, through a systematic method

of data collection. When analyzed qualitatively, the content of multiple focus group

sessions with stakeholders may lend insight into potential data elements.

Delphi Technique

The Delphi technique uses iterative ranking methods to determine levels of

importance. Like other forms of qualitative research, the Delphi technique is a way to

obtain information and judgments from participants to facilitate problem-solving,

planning, and/or decision-making [37]. The technique was developed by the RAND

Corporation in the 1960s as a forecasting technique. The US government subsequently

improved upon the model and promoted its use as a group decision-making tool.

The Delphi technique can be used in a group setting, or with the proliferation of

communications technologies like fax and email, conducted independent of actually

assembling groups of people together. Not physically assembling participates creates

logistical advantages that are often appealing to participants and researchers alike. From

the perspective of group dynamics, the Delphi technique may hold advantages since the

technique sidesteps many of the issues that accompany groups of people working

together. In either case, the technique is similarly implemented. Group sizes in the

Delphi technique range from several people to several hundred.

The Delphi technique employs an iterative process that encourages people to offer

their view of the relative importance of an idea, concept, or topic. Participants are









typically knowledgeable concerning the area of study, much the same as the case of an

expert panel or focus group. In the case of non-assembled participants, they may respond

anonymously to a coordinator, who asks questions and then simply assembles the

responses for redistribution for additional input. Panelists make individual estimates that

are summarized and circulated among participants, and each can alter his or her opinion.

The process is repeated until a consensus is reached.

To initiate the technique, a coordinator prepares a simple open-ended question and

asks participants to offer brief ideas. These ideas usually take the form of words and

phrases, and not fully developed concepts. From the responses, the coordinator

assembles a second questionnaire, requesting participants offer commentary on all of the

responses. Participants list strengths and weaknesses of those responses and resubmit

them to the coordinator. The coordinator once again reassembles the responses, creating

a 3rd questionnaire, asking again for input, including new ideas. This process can

continue to the point that the coordinator feels that no new thoughts are being introduced.

After the iterative brainstorming process, the coordinator is charged with resolving

the results. Resolution can manifest itself through the emergence of a consensus, at

which point the process if complete or it can move to the conduct of a formal evaluation

by the coordinator [37]. If a formal resolution is used, the coordinator will ask

participants to utilize a scale to rank ideas on a continuum from zero to seven (or other

number), with zero being the least weighted and the upper limit, the most effective in

dealing with the issue. Responses are tabulated to create a rank-order listing of the ideas.

Another way to implement the formal resolution is the use of Nominal Group

Technique for participant "voting" for ideas [37] In this case, the members are asked to









identify the top five ideas and assign numeric points to the most promising, on a

descending scale to the least promising of the five. The "votes" are tallied by the

coordinator who again creates a rank-order listing of ideas, based on the number of votes

each received.

Depending upon the subject studied, the process of ranking that is described above

may also have to be iterated, to form a statistically suitable result. Showing panelists the

results of first iteration ranking may allow for clarification, discussion, or re-evaluation

by the panelists. This can be helpful when responses do not reveal clear conclusions. The

final ranking process should produce a consensus, and a subordinate list of next best

group choices.

Variations of the Delphi technique alter the number of participants, the number of

iterations, the number of graphic scoring points, anonymity of the participants, and the

definition of statistical consensus. Implementation of the technique requires the

researcher have a clear understanding of these factors before entering into the process.

It should be noted that the Delphi technique mentioned herein is a tool for

qualitative research. If properly implemented, Delphi can produce valid results, suitable

for research, planning, or decision-making, particularly in a monovariable study. Like

any form of research, it must be used with competency, credibility and integrity. The

results of the Delphi method are statistically arrived at through the process of iterative

ranking, however one must be reminded that the basis of those statistics are only as good

as the opinions of the participants[36]. Like any form of qualitative research, participant

selection is critical.









Recently the name Delphi has been attached to efforts where groups are essentially

led by skilled "manipulators" who seek to move consensus rather than uncover it. These

sometimes antagonistic coordinators are neither interested in research nor truth, and their

misuse of the technique is unfortunate. Given there is some negative sentiment

concerning this misuse of the technique, a qualifying statement is in order to separate

appropriate and inappropriate uses of the Delphi technique. The Delphi method

described herein is not related to some of these recent abuses that have been improperly

associated with the properly used research tool.

Survey Research

Perhaps the most recognizable form of qualitative research is survey research. The

term "survey" generally describes a method whereby information is gathered from a

number of people who ostensibly represent a larger population. Within statistical

parameters decided by the researcher, the responses of those who are surveyed will be an

indication of how the larger population would answer the same set of questions. Reliable

research data can be achieved with survey research, without the expense of asking

everyone in the population the same question.

Survey research takes many forms including self-administered questionnaires

(mail, online), personal interviews, and telephone interviews. It may be as innocuous as

entering your zip code upon entering a web page, or as involved as the complete multi-

page form of the US Census. Surveys form the basis for everything from consumer tastes

to major public policy decisions.

As a form of social/qualitative research, surveys provide a way to collect data when

observation is not possible. While they often seek opinion, a well designed survey









instrument can mine factual information from individuals as well. The design of surveys

and questionnaires is rooted in social science, therefore they offer validity and allow for

statistical processing of respondent data. While the technical design of questions and the

form of the instrument are beyond the scope of this composition, it is sufficient to say

that, as in other forms of qualitative research, it must be designed with care.

The premise of a survey is to sample a subset of a larger population. It is essential

that the selected sample be representative of the population being studied. Surveys can

quantify variables to make statistical conclusions about cause and effect. They can also

be used as a gauge of public opinion, as is the case in the plethora of polls conducted by

academic, media, and marketing entities.

Survey research generally requires adherence to rules governing sample size and

composition. If the researcher is not strictly bound by such sampling conditions, a

general survey can provide some information or insight into an issue, although the results

would be more an estimation than a provable fact. Potential respondents might be asked

to participate in an online survey about traffic safety issues, similar to a recent

undertaking of the Florida Department of Transportation. In complying with Federal

requirements for states to have a Strategic Highway Safety Plan, the FDOT created an

online survey for the purpose of gauging the sentiment of stakeholders. By selecting

items from a list, respondents are able to offer their opinion about traffic safety priorities.

A sample of the survey and introductory instructions can be viewed at

www.dot.state.fl.us/safety/ [38]. Traffic safety stakeholders would potentially makeup

the population for potential respondents in an effort to identify crash reporting variables

or content changes in reports.















Fie Edit View Favortes Tool Help

a Iek P earh F aorte o- L it


Addre7 Flhrpasurve01.uamysTaomosurvertwab.e l S
^^^B u IBF Sjn -MTT- B i .T ~E u -


Transportation safety Concerns
1. In your own experience, what do you think are the most important transportation related safety problems facing Florida today? Please
check up to five (5) choices.

SAggressive or reckless dnving
Distractions or inattention to driving (including cell phone use, eatng/dnnking)
SAnimal Conflicts
O Impaired driving (drugs or alcohol)
l Older dnvers and pedestrians
SUnskilled or unlicensed drivers
D Young drivers
0 Speeding
SIgnonng red lights or stop signs
E Driver fatigue
ENon-use or improper use of safety belts/child safety seats
L Conflicts between bicycles and vehicles
SConflicts between pedestnans and vehicles
SSharing the road with trucks or buses
E Crashes involving buses or trains
LThrough trucks on local streets
SSchool-related (e.g., walking, student drop off/pick up school buses)
SExcessive traffic congestion
E Trucks carrying hazardous cargo
L Presence of sport-utiity vehicles
SNight-time dnving (e.g., darkness, headlght glare)
D Bridges narrow bndges, metal surfaces, etc
SPotholes or poor pavement
LPoor Intersection geometry (e.g left-hand turns)
O Narrow, broken, or missing sidewalks
OPoorly designed roads and ramps (narrow lanes, short ramps, sharp turns steep grade, etc.)
EWork zone safety
O Personal safety concerns at bus stops, train stations
LBad weather (rain, snow, icesun glare)
7Inadequate visiblty due to signs, utility poles, trees, etc.
ORa]i crossings
L Poorly designed parking lots or private property
O Traffic circles
O Too many driveways
IToll booths
L Poor signage/lane markings



Figure 1 Florida DOT Online Safety Survey



While surveys offer advantages in terms of time and money, they are not capable of



probing an issue any deeper than the original question. The designer of the survey must



have a clear concept of the questions, before they are asked, because there is no



opportunity to rephrase or clarify with the respondent. For example, a questionnaire is



generally reduced to the least common denominator for all of the potential respondents,



often neglecting specifics that may be more appropriate for some than others [36].



Unlike the open-ended questions used in the focus group and Delphi method, the survey



is sometimes limited by the form of the instrument. This structure limits the exploratory


M


Is


1 Flvid OnineS~wy:TraspotatonSafty icrsof Ineret xplrerlm]Tl









value of the survey, when compared to the other qualitative methods, making them less

effective at exploration.

Analysis of surveys involves the use of descriptive and inferential statistics.

Descriptive statistics allow the researcher to present quantitative information in a way

that is more easily understood. When describing single variables or associations between

two or more variables, descriptive methods like association, regression analysis, or other

multivariate techniques are used. Inferential statistics are more common in the social

sciences, and they make estimates of larger populations from samples. Test for statistical

significance and estimations of relationships between variables is generally the basis for

such methods [24].

Using surveys to explore potential crash report data elements may be difficult to

implement, given the immense volume and diversity of the topic. They may be

worthwhile for focusing on specific elements and attributes associated with specific

scenarios. Examining how to better capture data about unique crashes, like those

involving work zones for example, may be sufficiently limited to lend themselves to the

use of surveys.

Focus Groups

Focus groups are a relatively new qualitative research technique, and they are used

to obtain information from people in a group setting. Focus groups are defined as, "A

group of individuals selected and assembled by researchers to discuss and comment on,

from personal experience, the topic that is the subject of the research." [37] They have

been widely used for marketing purposes, in an attempt to measure consumer opinion or

sentiment. In the focus group, a moderator promotes discussion among a relatively small









group of individuals. The group discusses the issues) presented by the moderator in

such a way that their opinions, attitudes, and observations are brought to the surface.

Group forces and dynamics are seen as advantageous parts of the process with

participants discussing issues with each other, rather than simply dialoging the

moderator. Many researchers believe this process produces richer and more detailed data

than possible with other research methods [36].

The moderator of the focus group creates a comfortable environment for the

participants and seeks to stimulate their thoughts and discussion by asking a series of

open-ended questions. The role of the moderator is an important one, for he/she must

introduce the topic, use probing questions when necessary, maintain order, and finally

summarize the meeting.

Focus group meetings usually last about one to two hours, and are generally held at

locations suitable for privacy and comfort. The recommended number of participants for

focus groups is somewhat subject to debate; 6-10 [39], 5-6 [40], 6-8 [41], and up to

fifteen [42]. Since some assert that a group of 6-12 is appropriate if the group is

homogeneous [30].

The number of separate focus group sessions to be conducted is another salient

issue that the researcher must decide. If only one focus group is used, it runs the risk of

observing the dynamics of a group and little else [43]. Conducting more than one session

is desirable, particularly if distinct subgroups are present. Additional sessions serve to

increase the available data, and insure that individual group dynamics do not skew

results. In some cases, multiple sessions may be conducted with the same group of

people, particularly when temporal trends may be an issue.






84


Focus groups are a valuable form of social research. As a qualitative form of

research, they provide the inputs that are necessary for identifying work zone issues. In

some ways, these forums emulate the collaborative process that has been described in the

development of crash report forms. Such similarity affirms the appropriateness of the

method for data collection. Focus groups were selected as the preferred method of

qualitative research and data collection for this effort.















APPENDIX B
FOCUS GROUP ADMINISTRATIVE FORMS








AAE< UNIVERSITY OF
M FLORIDA

Transportation Research Center


Work Zone Data Collection Focus Group Meeting


DATE:
TIME:


MODERATOR:
ASSISTANT:


LOCATION:


PARTICIPANTS' LIST (please print your name)

1)

2)

3)

4)

5)

6)

7)

8)

9)

10)

11)









# UNIVERSITY OF
.... FLORIDA

Transportation Research Center

Data Collection Requirements for Work Zone Incidents
Focus Group Interview


Roadway construction has become a common fixture in our daily travels. These
locations can present unique challenges to traffic safety interests, as well as the motoring
public. Like many aspects of traffic safety, a better understanding of the contributing
factors in crashes can potentially lead to new insight and improved countermeasures.

The University of Florida Transportation Research Center, under a grant from the
Southeastern Transportation Center (STC), is conducting group interviews called "focus
groups" with FDOT engineers, private contractors, law enforcement, and drivers to
obtain a better understanding work zones. By examining the unique perspective and
expertise of each group, we hope to develop a better understanding of work zone
dynamics and develop the tools for more effective analysis of crashes in these areas.

Objectives of this focus group exercise:
1. Identify factors that might contribute to incidents in work zones.
2. Determine attributes associated with those factors.

Format of this focus group session:
1. The background, objectives, and benefits of this focus group interview will be
explained by the moderator.
2. The moderator will describe the format of the focus group session, and the points to
keep in mind.
3. An open-ended question will be presented by the moderator.
4. For the given question, participants discuss the issue and provide their perceptions
or opinions relevant to the issue. Additional questions may also be asked by the
moderator to fully explore the issue.
5. A total of approximately 4 to 6 questions will be presented.

Points to keep in mind:
When possible, discuss the issue in non-technical terms. Minimize the use of
acronyms and jargon that may be common within your field of expertise.
The terms incident, accident, crash, and collision will all be considered equivalent
in our discussion of work zones.
For efficient use of the time period allotted for the interview, the moderator may
need to interrupt and/or redirect the discussion, to insure that all questions are
covered.










Informed Consent Form for Focus Group Interview
Protocol Title: Data Collection Requirements for Work Zone Incidents


Purpose of the research study:
The purpose of this study is to identify data elements that can potentially be collected at the scene
of work zone crashes, to better understand the factors surrounding work zone incidents.


What you will be asked to do in this study:
After everyone in the room introduces themselves, a moderator will lead a group discussion about
highway work zones, and incidents that occur in those locations. You will be asked to offer
opinions about construction zones, and their impact on the driving environment.


Time Required:
Up to 2 hours


Risks and Benefits:
There are no risks involved in this study. While we do not anticipate that you will benefit directly
from participating in this study, it may lead to improved data collection at the scene of work zone
incidents. This additional data may contribute to better statistical analysis of work zone incidents,
and potentially a better understanding of the factors surrounding these incidents.

Compensation:
There is no monetary compensation for participation in this study.


Confidentiality:
We will record the names of those participating in this study, however, your name will not be
associated with your individual comments in the focus group interview. Your name will not be
used in any published reports and your identity will be kept confidential to the extent provided by
law.


Voluntary participation:
Your participation in this study is completely voluntary. There is no penalty for not participating.

Right to withdraw from the study:
You have the right to withdraw from the study at any time without consequence.

Whom to contact if you have questions about the study:
Scott S. Washburn, Ph.D, P.E.
Civil and Coastal Engineering, 365 Weil Hall, P.O. Box 116580, Phone: (352) 392-9537 x1453.



Whom to contact about your rights as a research participant in this study:
University of Florida Institutional Review Board (UFIRB).
UFIRB Office, P.O. Box 112250, University of Florida, Gainesville, FL 32611-2250,







89


Phone: (352) 392-0433.

Agreement:
I have read the procedure described above. I voluntarily agree to take part in the procedure and I
have received a copy of this description.


Participant:


Interviewer:


Date


Date















APPENDIX C
FOCUS GROUP MODERATOR GUIDE




Full Text

PAGE 1

DATA COLLECTION NEEDS FOR WORK ZONE INCIDENTS By GRADY THOMAS CARRICK A THESIS PRESENTED TO THE GRADUATE SCHOOL OF THE UNIVERSITY OF FLOR IDA IN PARTIAL FULFILLMENT OF THE REQUIREMENTS FOR THE DEGREE OF MASTER OF SCIENCE UNIVERSITY OF FLORIDA 2006

PAGE 2

Copyright 2006 by Grady Carrick

PAGE 3

DEDICATION I thank my mother for giving me the opportunity and encouragement to study as a youth, and my wife for providing me positiv e motivation. My friend and colleague, Chris Knight, deserves my gratitude for providi ng the professional support that is needed to balance work and study.

PAGE 4

iv ACKNOWLEDGMENTS I thank Dr. Scott Washburn for his support and encouragement in my graduate studies and in this project. Appreciation is given to Florida Highw ay Patrol Major Steve Williams for assisting me with the database deve lopment part of this project. Without his expertise, the supplemental data collection system could not have been possible. Many thanks to the Florida Highwa y Patrol and Colonel Christ opher Knight for providing the institutional access and support that made this project possible. Finally, appreciation is given to the STC for funding this project.

PAGE 5

v TABLE OF CONTENTS page ACKNOWLEDGMENTS.................................................................................................iv LIST OF TABLES............................................................................................................vii LIST OF FIGURES.........................................................................................................viii ABSTRACT....................................................................................................................... ix CHAPTER 1 INTRODUCTION........................................................................................................1 Background...................................................................................................................1 Problem Statement........................................................................................................2 Research Objective and Supporting Tasks...................................................................4 Document Organization................................................................................................5 2 LITERATURE REVIEW.............................................................................................6 Current Crash Reporting Systems................................................................................6 Work Zone Data Collection-Related Studies.............................................................13 3 RESEARCH APPROACH.........................................................................................17 Overview.....................................................................................................................17 Methodology...............................................................................................................17 Traditional Crash Report Content Decisions.......................................................19 Data Mining Using Qualitative Methods............................................................20 Focus Groups In Qualitative Research................................................................21 Group composition.......................................................................................21 Administrative preparation...........................................................................23 Moderator and content preparation..............................................................25 Administration/conduct of meetings............................................................26 Post-Session Processing...............................................................................27 Data Element IdentificationQualitative Analysis............................................28 Data Element DefinitionInterpretation of Results...........................................30 Data Element TestingSuppl emental Collection System..................................31 Technical framework of collection system..................................................33

PAGE 6

vi Operational framework of collection system...............................................34 Data Element Validation.....................................................................................36 4 RESULTS AND ANALYSIS.....................................................................................37 Qualitative Analysis Data Element Identification...................................................37 A Priori Categorization........................................................................................39 Emergent Categorization.....................................................................................40 Final Categorization............................................................................................40 Agreement Measure.............................................................................................43 Weighted Measure of Intensity...........................................................................46 Composite Ranking of Items...............................................................................48 Interpreting Analysis Definition of Elements..........................................................49 Creating Linkages................................................................................................50 Converting Codes to Data Elements....................................................................54 Elements as Interrogatives / Binary Values.........................................................55 Supplemental Collection Syst em Elemen t Testing................................................59 Technical Framework of Collection System.......................................................59 Field Implementation and Tes ting of Collection System....................................66 Results of Collection...........................................................................................67 Validation of Elements...............................................................................................67 5 CONCLUSIONS AND RECOMMENDATIONS.....................................................71 Conclusions.................................................................................................................71 Recommendations.......................................................................................................72 Future Work Zone Applications..........................................................................73 Other Applications...............................................................................................74 APPENDIX A OVERVIEW OF QUALITATIV E RESEARCH METHODS...................................75 B FOCUS GROUP ADMINISTRATIVE FORMS.......................................................85 C FOCUS GROUP MODERATOR GUIDE.................................................................90 LIST OF REFERENCES...................................................................................................94 BIOGRAPHICAL SKETCH.............................................................................................98

PAGE 7

vii LIST OF TABLES Table page 1 MMUCC Work Zone Elements and Attributes..........................................................8 2 State Report Form Wo rk Zone Methodologies..........................................................9 3 Focus Group Composition.......................................................................................24 4 Focus Group Moderator Questions..........................................................................26 5 Unique Speakers by Group and Code......................................................................42 6 Agreement by Group and Code (Binary).................................................................44 7 Weighted Measure of Intensity................................................................................47 8 Comparison of Intensity and Agreement Measures.................................................48 9 Composite Codes......................................................................................................49 10 Data Linkage............................................................................................................55 11 Converting Elements to Supplemental Report Questions........................................58

PAGE 8

viii LIST OF FIGURES Figure page 1 Florida Traffic Crash Report....................................................................................10 2 Wisconsin Motor Vehicle Accident Report.............................................................10 3 Commonwealth of Pennsylvania Police Crash Reporting Form..............................11 4 South Carolina Traffic Collision Report Form........................................................11 5 Data Element Development Process........................................................................18 6 Meeting Commander Recording Software Interface...............................................28 7 Long Table Method of Analysis...............................................................................39 8 Data Analysis Process..............................................................................................57 9 Supplemental Work Zone Database Structure.........................................................61 10 Supplemental Data Collection Web Site..................................................................64 11 Work Zone Data Results Screen..............................................................................65 12 FHP Mobile Computing Architecture......................................................................66

PAGE 9

ix Abstract of Thesis Presen ted to the Graduate School of the University of Florida in Partial Fulfillment of the Requirements for the Degree of Master of Science DATA COLLECTION NEEDS FOR WORK ZONE INCIDENTS By Grady Thomas Carrick August 2006 Chair: Scott Washburn Major Department: Civil and Coastal Engineering Roadway construction has become a comm on fixture in our daily travels. According to the Federal Highway Administ ration (FHWA), fatalitie s in highway work zones were up nearly 50% between 1997 and 2003. In 2003 alone, there were 41,000 injuries and 1,028 fatalities in these locations. Increasingly, safety in terests are searching for characteristics associated with work zones th at contribute to the da ngers of such areas. Like many aspects of traffic safety, a bette r understanding of the contributing factors in crashes can potentially lead to improved c ountermeasures. Examining crash data is a principal method by which engineers, police, and safety advocates attempt to determine those factors, but such data are often incomp lete. The prospect of improving the data set requires examining the potential of a suppl emental data collection system. Using qualitative research, work zone stakeholders potentially provide a be tter understanding of work zone incidents, rendering new data elements. Creating a web-based supplemental collection system can assist police in gath ering the data while completing the current

PAGE 10

x traffic crash report. Supplemental data elements and collection systems have the potential to enrich the data set, and bolster the cause of safety.

PAGE 11

1 CHAPTER 1 INTRODUCTION Background According to the Federal Highway Admini stration [1], there was a nearly 50% increase in U.S. work zone fatalities between 1997 and 2003. In 2003 alone, there were 1,028 work zone fatalities, a figure that repres ents about 2.4% of all roadway fatalities for the year. A larger view of the problem is evident in the estimated 41,000 people injured in the more than 102,000 work zone crashes for th at same year. Florida statistics mirror this compelling national problem. In 2003, 104 fatalities and 3,607 injuries occurred in 3,509 crashes in Florida highway work zones [2]. The danger that work zones pose for cons truction personnel is readily apparent. The less visible statistic about these locations is the peril for motori sts. Nationally, 85 % of fatal crash victims in work zones are dr ivers or occupants, and in Florida that proportion increases to 90 %, as 9 in 10 are moto rists or pedestrians [1, 2]. Regardless of the reason for being in a highway work zone it is a potentially dangerous environment for everyone. Like many states around the nation, Florida is constantly trying to stay ahead of rapid population growth with new road projects. When coupled with routine maintenance efforts to address aging road and bridge infrastructures, the work zone has become a common fixture on our roadways. Nationally, an average of 23,745 miles of federal aid roadway improvement project s were underway annually from 1997 to 2001 [3]. These improvements are in addition to innumerable work z ones associated with

PAGE 12

2 municipal, county and state road department s, utility construction, and public works projects. Increasingly, safety interests are search ing for characteristics associated with crashes that occur in work zones. Like many aspects of traffic safety, a better understanding of the contribu ting factors in crashes can potentially lead to new and improved countermeasures. Examining aggreg ate crash data is usually the means by which engineers, police, and safety advocat es attempt to determine those contributing factors. Individual police cr ash reports potentially focus attention on specific time, location, and causation factors. Problem Statement Complete and accurate data is essential to the conduct of meaningful research. Traffic safety research often relies in some part on data from poli ce traffic crash reports and aggregate crash statistics. In every st ate in the nation, crash report data elements have evolved to capture relevant info rmation about location, vehicles, persons, conditions, and causation. The data derived from these reports are often the foundation of safety-related research. The basic crash report is designed to document facts for various governmental purposes as well as satisfy insurance industr y needs. With such general purpose, the reports sometimes lack the detail necessary to be of value in examining unique crash situations, like the work zone crash. While aggregate crash report data are suitable for general frequencies analysis and cross tabulation, the lack of coding for roadway construction variables makes wo rk zone analysis difficult. The Florida Traffic Crash Report Long Fo rm (HSMV-90003) is completed anytime there is a crash resulting in property dama ge in excess of $500 [4]. An abbreviated

PAGE 13

3 report, the Law Enforcement Short Form Report (HSMV-90006), may be completed in lieu of the previous for any crash that doe s not result in a vehi cle being removed by wrecker, an injury, or a crime (such as DUI or leaving the scene). The use of separate report formats based on such criteria is commonly referred to as accident reporting thresholds. The forms are very similar, and mo st of the coding for crash variables is, in fact, identical. The main difference in the forms is that the diagram, narrative, and much of the coding is optional for the officer with the use of the Short Form Report. Only Long Form reports are used in tabulati on of statewide crash statistics. While the hand written crash report remain s a staple, many agencies are migrating to automated methods of reporting crashes. These systems usually consist of officers using laptop computers in the field to input crash data electronically. Such systems have proven to reduce errors and greatly improve the timeliness of data. Their continued evolution will solve some data issues, but the completeness of data continues to be an obstacle in the research of wo rk zone incidents. These gaps in data are more of a function of the format of the reports than anything else. For most purposes, the current reports are well-designed and adequately capture the most relevant data. The reports do, however, lack detail where highway work zones are concerned. In Florida, linking crash repor ts to work zones is accomplished through a code for Work Area (1-none, 2-nearby, 3-ente red). Less prominent codes contained in other report areas, may also indi cate work zone involvement: Road Conditions at Time of Crash 04-Road Under Repair/Construction First/Subsequent Harmful Event(s) 24-Collision with Construction Barricade Sign Pedestrian Action 07-Working in Road Traffic Control 10-Officer/Guard/Flagperson

PAGE 14

4 When one can discern that a crash occurre d in a work zone, there are generally insufficient data to describe the nature of the work, the maintenance of traffic (MOT) control present, or the conditi ons created by the road work. Information that may prove valuable to improving work zone safety is typically not captured. Difficulties associated with studying work zone crashes often stem from incomplete data. The logical remedy is to im prove the data set by adding data elements to the police crash report form. Changing statew ide crash report forms is not a task to be undertaken lightly, however, since it involves months of planning, meetings and significant printing costs. Adding data el ements for these purposes should be accomplished with a supplemental reporting methodology, to eliminate the impact on status quo reporting procedures. While collec ting more data in work zones will require additional officer time and effort, such concerns are diminished by the relative infrequency of crashes in these locations. Given the uniqueness of work zones and their temporary nature, it is essential to capture as much detail as possi ble about incidents in these areas in order to effect positive safety improvements. Research Objective and Supporting Tasks The fundamental question posed by this study is, What can we document at the scene of a work zone crash that will enhan ce our understanding of t hose incidents? The objective of this research effort is to answ er that question through a qualitative research approach. Since police routinely respond to traffic crashes for reporting purposes, they are logical candidates to assist in any enhanced data collection e ffort. Creating an instrument that police use to better document work zone cr ashes is seen as a means whereby the data sufficiency can be improved. For purposes of this study, such an instrument should be

PAGE 15

5 supplemental to the statutorily required crash form, so it can be used selectively and cost efficiently. Identifying the data needed is as import ant as the collection and tabulation process itself. Determining data elements to be c ontained in the new instrument to be used by police is made possible through the use of focus groups conducted with various work zone stakeholders. By using individual s who are knowledgeable in the field, across several disciplines, there is a greater likelihood that the additi onal data collected will have the desired utility. Such qualitative resear ch methods will illuminate new data elements. Through the design of a supplemental da ta collection instrument, new data elements can be evaluated by select Flor ida Highway Patrol (FHP) personnel in conjunction with actual work zone crash i nvestigations. After newly created data elements are field tested, an analysis will be conducted to determine relationships with state of the practice work zone data elements. Document Organization Having identified work zone data sufficie ncy as a problem, the research objective and supporting tasks described above serve to direct the overall research effort envisioned. An effective research approach further involves system atically looking at the research others have done, undertaking origin al study, analyzing the results of that study, and ultimately drawing conclusions. This archetype forms the basis for chapters contained in this document. Chapter 2 provide s an overview of previous related research. Chapter 3 describes the resear ch approach and methodology used to satisfy the objectives of this project. Chapter 4 presents analysis of the data collected in conjunction with this project, and Chapter 5 provides conclusions and recommendations.

PAGE 16

6 CHAPTER 2 LITERATURE REVIEW Current Crash Reporting Systems Police traffic crash reporting likely be gan in March of 1896, when an early automobile struck a bicyclist in New York, re sulting in the driver of the auto spending a night in jail [5]. As the automobile gained in popularity in the early 20th century, police standardized the way motor vehicle mishaps were reported, and ultimately states standardized the forms for their respective jurisdictions. With the evolution of the accident report at the state level, gathering st atistics about crashes be came possible. State reporting and statistical practices have progressed to relatively efficient systems in most states. Because there is currently no mandatory national standard for police crash report forms, state crash reports assume a wide range of designs that can be handwritten, elaborately coded, bubble coded, or even comput er generated. While they all certainly have a different appearance, the reports typi cally capture similar information about the location, persons, vehicles, and environment of traffic collisions. The information is coded with varying degrees of detail, for inclusion in state collision databases. Conclusions about vehicle movements and crash causation factors are typically the product of police interpretati on of these elements. Almost all formats allow for a narrative and pictorial represen tation of the crash events. While state formats differ, there are comm on data elements. It is through this commonality that national statis tical efforts have traditionally been undertaken. Even

PAGE 17

7 with common data elements, getting data to integrate has often been a daunting task, due mainly to a lack of uniformity in data elemen ts, structure and definition. To improve data portability, there remains a need for improved standardization. In 1998, a joint effort was in itiated to bring a bout greater uniformity in reporting. The Model Minimum Uniform Crash Criteria Guideline (MMUCC) is a joint effort between the National Highway Traffic Safe ty Administration (NHTSA), the Federal Highway Administration (FHWA), the Federa l Motor Carrier Safety Administration (FMCSA), and the Governors Highway Safety Association (GHSA). MMUCC represents a voluntary and colla borative effort to generate uniform crash data that are accurate, reliable and credible for data-drive n highway safety decisions within a state, between states and at the national level [6:iii]. The MMUCC Guideline, 2nd Edition (2003) contains 111 data elements, of which 77 are collected by law enforcement at the scen e of a crash [6]. The remaining elements are derived from the data collected, for example, the number of vehicles involved, number of people injured, etc. The MMU CC establishes a minimum set of data elements, but does not dictate design of the act ual report form. With standardized data elements, the integration of data across multiple databases is made possible. Therein lays the promise of the MMUCC, integrating stat e data for national st atistical purposes. It has only been in the last decade that cras h reports have sought to capture data that is unique to highway work zones. In 1992, onl y 27% of state crash reports contained an explicit field for work zone presence [7], while today 67% contai n such a section. Historical codes for roadway under construction have incr easingly given way to a more specific representation of the presence of work zones.

PAGE 18

8 Section C19 of the MMUCC establishes guidelin es for work zone-related data elements. The rationale for inclusion of da ta elements in crash reporting center on the need to assess the impact on traffic safety of various types of on-highway work activity, to evaluate Traffic Control Plans used at wo rk zones, and to make adjustments to the Traffic Control Plans for the safety of workers and the traveling public [6]. The Guide also adeptly notes that the temporary natu re of work-zones requires documentation of their presence. Table 1 lists the work zone elements and attributes recommended by the MMUCC. Table 1. MMUCC Work Zone Elements and Attributes Data Element Attribute Yes No Was the crash in or near a construction, maintenance or utility work zone? Unknown Before the First Work Zone Warning Sign Advance Warning Area Transition Area Activity Area Location of the crash Termination Area Lane Closure Lane Shift/Crossover Work on Shoulder or Median Intermittent or Moving Work Type of Work Other Yes No Workers present? Unknown Because the MMUCC is only a guide, a re view of state reporting forms is necessary, in order to determine the scope of work zone data collected through police reports. Specifically for this research effort, the police report forms used in all fifty states and the District of Columbia were reviewed in detail for elements associated with work

PAGE 19

9 zones. While a complete representation of th e content in those repor ts is represented in tabular form in the appendix of this re port, Table 2 summarizes how work zone information is depicted on the crash reports us ed in those jurisdictions. Analysis of those forms is further described in the following paragraphs. Table 2. State Report Form Work Zone Methodologies Embedded Data Elements Work Zone Specific Section Road Traffic Control Event Ped Basic Data (Y/N or Code) Detailed Data Captured None Under Repair/Const. Flagger Const. Sign or Barricade Collision with Barricade Working in/on Road States 23 11 17 24 33 9 21 39 % 45 22 33 47 65 18 41 76 Work zone data are represented in US cr ash reports in two wa ys; the presence of work zone specific sections on the reports and the use of embedded coding for crash attributes that may indicate the presence of a work zone. Work zone specific sections on traffic crash reports are fairly new features, being added in th e latest iteration of the forms in most cases. For example, the state of Florida added the section in its January 2002 revision of the Florida Traffi c Crash Report [8]. Of the 51 crash report forms reviewed, 34 (67%) had a separate or explicit section for work zone data. On a continuum, the detail of these explicit work zone sections ranges from a simple yes/no check box to multiple data elements describing the work zone location. To further delineate the use of explicit work zone sections, reports that us e a special section can be can be grouped on either end of this continuum, based on the le vel of detail, as either basic or detailed. Basic work zone reporting sections genera lly have separate data element(s) for capturing the presence of a work zone, but they do not expound upon the circumstances present. Of the 51 reports reviewed, 23 fo rms (45%) use methods of yes/no or simple

PAGE 20

10 codes to indicate the presence of a work zone Figure 1 and Figure 2 are examples from Florida and Wisconsin respectively. Figure 1. Florida Traffic Crash Report Figure 2. Wisconsin Motor Vehicle Accident Report Some states capture greater detail about th e presence of road work. In the case of 11 states, or 22% of those reviewed, the data capture went beyond simply noting the presence of road work. The guidance of th e MMUCC is evident in this detail, noting where the crash occurred in relation to the wo rk zone, the type of work being performed, and the presence of workers. Only five st ates (Iowa, Nebraska, Oh io, Pennsylvania, and South Carolina) follow the entire MMUCC guide line for elements and attributes relevant to work zones. Pennsylvania uniquely capture s the presence of law enforcement at the work zone crash location and the presence of special speed limits. Figure 3 and Figure 4 are examples from Pennsylvania and South Carolina respectively.

PAGE 21

11 Figure 3. Commonwealth of Penns ylvania Police Crash Reporting Form Figure 4. South Carolina Traffic Collision Report Form A total of 17 states (33%) do not have a de signated section for capturing work zone information on crash reports. Researchers ar e required to find othe r report data to link these incidents with work zone activity. This is typically accomplished through embedded attributes. Embedded report attributes or codes can be mined to indicate the presence of work zone conditions, albeit indirectl y. These attributes can be us ed in conjunction with or independent of a specific work zone report se ction. They are typically found in report categories describing the roadway conditions, tr affic control present, harmful event, or pedestrian action. All 51 report formats from the states plus the Di strict of Columbia include some form of these em bedded attributes. When cro ss tabulated with other report variables, greater insight into th e work zone crash is possible. Roadway conditions is a data element contained in almost all police crash reports. Defects in the roadway, obstructions loose surface material, holes, and standing water are all types of attribut es that may be present. Fo r purposes of this analysis,

PAGE 22

12 specific attributes such as road under re pair or road under construction can be associated with a work zone. Of the 51 re ports reviewed, 24 (47%) contained such an attribute. Traffic control can take ma ny forms, but generally it i nvolves the signs, signals, and other controls present at the crash location. With respect to work zone analysis, the presence of a flag person coded in this se ction of a report is a strong indicator that a highway work zone may have been implicated. Coding for this type of traffic control was present in 33 reports, representing 65% of the total. Additionally, 9 reports (18%) specifically noted the presence of a construc tion sign or barricade as a traffic control device. Harmful events are captured to describe the damage or injury producing crash. While first harmful events are generally captured, subsequent harmful events are beneficial to illuminate the entire series of events in a crash. Practically anything that a motor vehicle strikes, including the manner in which it strikes another vehicle, is considered a harmful event. When a vehi cle collides with a construction barricade, barrel, or piece of work equipment, it may be an indication that the crash occurred in a highway work zone. Collision with one of th ese objects is seen in 21 reports (41%) as a harmful event. Like vehicles, pedestrians are represented in crash reporting as traffic units. The recognizable role of pedestri an construction workers readily comes to mind when one considers highway work zones. For reporti ng purposes, the action of pedestrians is classified in a section that describes what the pedestrian was doing at the time of the collision. Typically this sect ion is coded as crossing the street, walking along the road,

PAGE 23

13 standing in the road, or similar descriptions of people that one might encounter near roadways. Attributes that sp ecifically denote the pedestrian as working on/in the road are a good indicator of work zone involvement. All 51 report formats reviewed except 12 (76%) contain such a description of pedestrian action or movement. Embedded attributes describing collisi on events, roadway conditions, traffic control, and pedestrian actions were traditionally the indicators that were used to identify the presence of a work zone. Statisticians were historically required to link these elements to other data to determine if a wo rk zone was implicated in the crash. Their continued inclusion, coupled with specific wo rk zone sections, can provide information about work zones and the circumstances of crashes that occur within them. Unfortunately, too little information about work zone crashes is still the norm in state and national crash reporting systems. Work Zone Data Collection-Related Studies Several studies have examined alternatives to basic police reporting for collection of data in highway work zone crashes. Revi ewing these alternative methods is of value, since this project seeks to go beyond the tradi tional crash report approach as well. Wang, et al. conducted research using data from the Highway Safety Information System (HSIS) to explore the issue of work zone crashes [9]. The conclusions of their study indicate that the absence of a universal definition of work zone is problematic. Additionally, they found that police reporting systems require modification to include additional data elements that better descri be work zone attributes. Data fields recommended by their report are designe d to answer the following questions: Did the crash occur in or near a construc tion, maintenance, or utility work area? Was there work activity at the time of the crash?

PAGE 24

14 Where did the crash occur in relation to the work? What was the type of work performed? Did the work area have an influenc e on or contribute to the crash? Khattak and Targa investigated the role of large trucks in work zone traffic crashes, using HSIS data and North Carolina Police Cr ash Reports [10]. Examining the narrative and diagram in each crash report, researchers sought to enhance the data set by creating unique data elements. They noted that this method of data acquisition was very laborious and only possible given the limited scope of subj ect incidents, (i.e., large truck crashes in work zones). Ha and Nemeth mined police re ports in Ohio for unique work zone data elements and also found the pr ocess to be difficult [11]. Garber and Zhao examined crash characte ristics in work zones by examining police crash reports believed to be work zone related from 1996 through 1999 [12]. The scope of their study was 1,484 reports after nearly 500 reports were discounted because of inconsistencies in reporting. To facilitate more detailed an alysis, they recommended that the Virginia Police Accident Report be modified to capture additional information relevant to location, work activit ies, traffic control, speed lim it, and presence of workers. Raub et al. sought to determine causal fact ors in work zone crashes by enhancing police reporting [13]. Making a ca se that the data derived fr om police reports was largely insufficient with respect to work zones, rese archers developed a separate data collection instrument for the purpose of determining th e contributing factors associated with 103 crashes in Illinois between 1998 and 1999. In these cases, police completed a supplemental form to assist researchers in th eir analysis. The suppl emental form did not use a methodology for developing data elements and was basically designed to be a stand-alone data system, not linked to the larger crash reporting database.

PAGE 25

15 Schrock et al. concluded rese arch in 2004 that analyzed 77 fatal work zone crashes in Texas [14]. Faced with issues about sufficiency of crash data, trained researchers responded to the scene of fatal work zone cras hes in an attempt to document factors at the scene. Reviewers attempted to respond as soon after the crash as possible, but their response was sometimes in the days that followed. Environmental factors generally duplicated attributes already captured in reporting with th e exception of identifying the location of the crash within the work zone, a nd the nature of work being performed. The Georgia Department of Tran sportation (DOT) has develope d a practice whereby they attempt to respond to all fatal wo rk zone crashes as well [15]. The Florida DOT, like some other states uses an Engineers Maintenance of Traffic (MOT) Evaluation at Accident Site report to better un derstand the factors associated with work zone crashes. After wo rk zone crashes, FDOT engineers are tasked with completing a report to capture informa tion about the work zone. The report has historically not consisten tly been completed by FDOT personnel, and Spainhour and Mtenga sought to revise and automate the fo rm [16]. While the electronic entry format was superior to the paper form it replaced, th e fact that it did not increase use by FDOT engineers was seen as an indica tion that the data set continue s to be notably incomplete. Graham and Migletz noted similar problems with underreporting in their review of project manager-based reporting systems us ed in Iowa and North Carolina [17]. Thielman examined the potential of expert systems in the collection of traffic crash data, although not specifically work zone data [18]. The fo undation of the expert system was data elements derived from experts in th e field. Panelists incl uded traffic officers, crash investigation trainers, safety analysts, reconstructionists, vehi cle safety engineers,

PAGE 26

16 and highway engineers. After determining da ta to be collected, officers equipped with pen-based computers used the expert system to focus greater detail on seatbelt use, vehicle damage, and roadside barriers. The expert system required an average of two minutes of officer time to collect the data, a nd field tests of the system confirmed that officers would be receptive to additional coll ection responsibilities. While it did not specifically focus on work zone areas, the c oncept of enhanced da ta collection and the use of an expert panel were viewed as successful. In some cases, video images of vehicles entering work zones are used to measure the effectiveness of merge opera tions [13]. Video can also be used to mitigate congestion in conjunction with ITS applic ations [19, 20]. The potential for such technology in crash data collection has not been empirically expl ored, but the wider use of these systems may hold that potential. While the prospect of capturing actual crashes with these systems would likely only be possible in very isolated instances, th ey may be well suited for establishing the environment pres ent at the time of a crash. While there have been a number of studies relating to work zo ne crashes and the sufficiency of data within these unique locati ons, substantial issues remain. Some studies have focused loosely on crash attributes in work zones, but none broach the topic from the perspective of perfecting the process by which data elements are identified in a qualitative way. Similarly, desp ite efforts to bolster collecti on systems, potential exists for improvement as well. No study approach ed the topic of supplementing work zone data as a two-fold propositi on requiring qualitati ve data identification and collection system development.

PAGE 27

17 CHAPTER 3 RESEARCH APPROACH Overview The prospect of improving the data set fo r work zone incidents requires defining what information to collect and how to better accomplish that collection. With these objectives in mind, the c oncentration of this project is to identify, test, and validate new crash report data elements that specifically rela te to work zone incide nts. Subordinate to that objective, a strategy for implementing th e data elements is also tested, with the creation of a web-based collec tion system for officer use. Through identifying new data elements and actually collecting said data, the body of knowledge concerning work zone incidents will be improved. Methodology A systematic approach to improving the data set requires identification of the specific data needed. While some previous studies have sought to embellish the work zone crash data set through supplemental re porting, none have qualitatively approached the process of determining the data to be collect ed. Ruab et al. [13] used police in Illinois to complete a supplemental report form, but th e content of that form was not the product of a qualitative effort. Many of the data elements duplicated normal crash report data elements and some were quite subjective, asking for officer opinion. The supplemental form was not designed to be linked with cr ash records, and was generally a stand-alone product. Enhancing the data se t requires a clear u nderstanding of data needs of the crash

PAGE 28

18 data consumers, a proposition that mandate s their involvement in the development process. Development of work zone-specific crash da ta elements is a process that can be described as several distinct steps. Mining pot ential data elements from discussions with stakeholders provides the foundation of knowle dge necessary for analysis. The content of those meetings can yield specific issu es/items when vetted using qualitative text analysis methods. Once potential data elemen ts are identified, th ey are contextually linked to the original group discussion to de velop the issue more precisely. Pilot testing the product under real world cond itions determines if the data element is mechanically sound. The final step in the process involves va lidating the data elements against state of the practice standards. The figure below is a graphic representation of the process described herein. Figure 5. Data Element Development Process The development of a collection system is both a part of the data element development process as well as an integral part of the overall data supplementation proposition. The use of a supplem ental collection system is desc ribed in greater detail as a component of the data element tes ting section, and later in Chapter 4.

PAGE 29

19 Traditional Crash Report Content Decisions Because traffic crash reports serve a wide variety of purposes, their development typically finds origins in public administra tion and public policy arenas. This is supported by staff at the Florida Department of Highway Safety and Motor Vehicles (DHSMV), who describe th e process as highly collabo rative and involving of stakeholders. Input, clarification, and constr uction are products of meeting with a small number of interested individua ls. Inclusion in the group rang es from mere stakeholders to subject matter experts. The leader of the group is generally a public administrator, representing a state agency th at is charged with the respons ibility for the forms by statute or administrative rule. The group could be re ferred to as an expert panel, and their meetings and activities are ge nerally conducted informally. Intuitively, the expert panel may be viewed as a practical way to approach the task of determining the content of traffic crash re port forms, but determining consensus of the panel is often elusive. Some research s uggests that the views of outliers are often discounted to further the objective of cons ensus [21]. In addition, even the most representative and well-inten tioned panel may fall victim to common pitfalls associated with individual and group dynamics. Closely related concepts such as groupthink, social consciousness, and Abilene paradox must be cons idered in group settings. For additional information on these issues, the readers is refe rred to the works of Elizabeth Scott, Emile Durkhiem, and Jerry Harvey [21,22,23]. While a detailed analysis of how individual and/or group dynamics may affect small group meetings is beyond the scope of th is effort, it is impor tant to note that committees or groups assembled for almost a ny purpose may be susceptible to these and

PAGE 30

20 other errors. When an expert panel is conve ned to examine crash report form changes, the informal exchange may produce une xpected and unreliable results. Data Mining Using Qualitative Methods Qualitative research is, Research involving detailed, verbal descriptions of characteristics, cases, and settings. Qua litative research typica lly uses observation, interviewing, and document review to collect data [24:1]. This form of research is rooted in the social sciences and is an excel lent vehicle for examining things that may not be measurable quantitatively. Qualitative re search can be accomplished through surveys, questionnaires, personal interv iews, researcher observation, or similar methods. The research seeks to learn more about things in their natural environment through peoples attitudes, perceptions, re collections, and feelings. Getting people together for the purpose of determining direction for a project need not fall victim to problems associated with group dynamics. One way to potentially produce more reliable results is the use of qualitative research. For our purposes, enhancing the content of crash reports is best undertaken as a function similar in approach. Using qualitative re search, investigating the topi c of crash incident data represents a form of collaborat ion. Similar to currently used methods, stakeholders such as law enforcement, engineers, private contractors, drivers, and safety advocates form the basis for input. But by employing a qualita tive research methodology, potential errors related to group dynamics can be minimized a nd a more reliable product is possible. A more dependable consensus among stakehol ders is also possi ble with qualitative methods, since they employ an approach that is grounded in social science. A more detailed examination of qualitative techniques is included as Appendix A, Overview of Qualitative Methods

PAGE 31

21 Focus Groups In Qualitative Research Choosing a qualitative research method fo r purposes of identifying potential work zone data elements requires weighing avai lable methods such as Delphi, survey, and focus groups. While all three methods employ the use of stakeholders, focus groups holds the greatest promise for success, given their interactive natu re and structural similarity to the traditional cras h report development approach. Focus groups have been used as a qualitativ e research tool for several decades, and are readily associated with market and pr oduct research. Recently, the technique has gained favor as an academic research and pub lic policy tool. A focus group is described as an assembly of people for the purpose of di scussing a topic. The discussion is led by a prepared moderator who guides the gr oup, producing a collection of opinions, perceptions, and experiences of the particip ants. Multiple focus group sessions broaden input and improve the reliab ility of data collection. As with any research tool, the use of focus groups requires following acceptable guidelines for their conduct. Some of th ese guidelines are contained in Appendix A Overview of Qualitative Methods The implementation of focus groups is best described as functions of group composition, administ rative preparation, moderator and content preparation, administration, and post-session processing. Group composition Identifying focus group participants is an important factor in the development of the research method. Having group member s with similar backgrounds describes a homogeneous group composition, and most rese archers agree that this allows for common threads of discussion [25]. Desi gning groups that represent work zone

PAGE 32

22 stakeholders thus requires segregating FDOT traffic engineers, private contractors, police, and average citizens into separate focus groups. Traffic engineers in both the public a nd private sectors ar e those primarily responsible for the design, setup, and maintenanc e of work zone traffic plans. Because of their significant role and expertise, they ar e necessary participants Segregating public sector engineers (FDOT) from contractor personnel is bene ficial because of sometimes divergent priorities. Further segmenting FDOT personnel into fi eld and headquarters elements ensures all aspects of public-sector engineering ar e represented. The rationale for separation within FDOT is that the people at the headqua rters level are those mostly responsible for developing and/or changing pol icy with regard to standards indices and traffic control plans, while at the district level, they were mostly responsible for implementation. A focus group with FDOT field personnel involves construction personnel, who are in work zones on a daily basis. The FDOT headquarters focus group included engineering representa tives from the safety office, roadway design office, and construction office. FDOT and industry repr esentatives make up three separate focus group sessions. Law enforcement plays a significant role in work zones, and they represent valuable stakeholders in gaining additional in sight. Police are charged with enforcing a myriad of laws within work zones, in order to promote the safety envisioned by engineers. Additionally, they are responsible for invest igating incidents at these locations, and therefore document the condi tions present and circumstances surrounding incidents. Sworn law enforcement o fficers make up a separate focus group.

PAGE 33

23 While they may lack traffic safety e xpertise, the layperson brings a unique perspective to discussions about work zones. Such a perspective may be beneficial when compared to the views of traffic safety practitioners. Since work zones are ostensibly designed with the average motorist in mind, it is of value to gain their insight into these locations. While they may not be direct c onsumers of incident data, one can readily understand their value as stakeholders and benefactors of changes in the driving environment. A collection of licensed driver s, representing the public at large, make up the final focus group. Administrative preparation Conducting group meetings with individuals requires significant planning with respect to scheduling activities. With the ex ception of the public meeting, all of the focus groups were conducted without compensati on to participants. Accommodating participant work schedules a nd enlisting volunteers can be ch allenging. For public-sector organizations such as law enforcement, and the FDOT, participants are supported by their respective organizations. The session with la w enforcement and two sessions with FDOT personnel were all well attended, as exp ected. Conducting the meetings at their respective work locations proved to be a wise accommodation. The industry focus group was much more challenging from the perspective that release of employees from work is more difficult for private business. Additionally, assembling participants from diverse employers and work locations requires effort on the part of the participant to travel and meet at a specified location. While 7 participants were scheduled, work obligations preclude d 3 from attending at the last minute. Conducting the industry session with less than 6 participants was less than ideal, but the level of discussion by participants made data collection acceptable.

PAGE 34

24 The citizen focus group sought to gain input from the average driver. This group was conducted at the Perceptive Market Rese arch (PMR) facilities in Gainesville. Participants were selected to be demogra phically representative of the community and compensated by the market research firm. As part of the research strategy, a composite group re presenting stakeholders was deemed beneficial to review the homoge neous group findings. Using a regularly scheduled meeting of the Alachua County Comm unity Traffic Safety Team to facilitate the composite stakeholder meeting was an excellent way to minimize impact on participants. All meetings were scheduled at convenient locations, with suitable facilities for privacy and comfort. The table below de scribes the setting, attendance, and duration of all focus groups. Table 3. Focus Group Composition Group Date Location No. Participants Approx. Duration (hrs.) Law Enforcement 11/08/05Jacksonville, FHP 6 1.50 Industry 11/08/05Jacksonville, FHP 4 1.75 FDOT Lake City 12/02/05Lake City, DOT 8 1.50 Citizen 03/08/06PMR, Gainesville 10 1.75 FDOT Tallahassee 03/27/06Tallahassee, DOT 7 1.75 CTST* 04/20/06Gainesville Technology Enterprise Center 20 N/A *Composite group not a focus group session. In addition to scheduling, administrative preparation involves the use of a Group Sign-In Form, Participant Inst ructions, and an Informed Co nsent Form. The Participant Instructions provides each focus group partic ipant with a one-page explanation of the objectives, format, and basic guidelines for th e conduct of the forum. These forms are all included in Appendix B.

PAGE 35

25 Moderator and content preparation Content planning involves establishing clear objectives on the part of the research team. The focus group is a moderator-led di scussion, so the moderator must be well equipped to perform in a way that promotes group interaction and adequately explores the subject at hand. Dr. Scott Washburn is an experienced focus group moderator, and the moderator for all focus group sessions. Th e role of the moderato r is well established with the use of a Moderator Gu ide. The Moderator Guide is a document that dictates the steps followed by the moderator to ensure a systematic approach and efficient use of time. The guide is a step-by-step script of so rts, that lists events, basic instructions, and target time limits. The guide includes pa rticipant sign in, welcome and introductions, background of the study, explanation of the format and scope of the meeting, and the questions to be used to promote discussion. The guide also provides some reminders to the moderator on methods for handling shy or reluctant participants, as well as those participants who may be monopolizing discus sion. The questions used in the focus groups for are, by design, left quite ge neral to promote maximum opportunity for exploration of the subject of work zones. Four basic questions form the basis for the work zone focus groups, however the moderator has a number of follow-up questions available for use if necessary. Questions are not all inclusive, since the guided discussion is designed to illuminate the subject. Latitude is afforded the moderator to explore areas that may be beneficial to the objective of the research. The following table is representative of the questions available to the moderator for the work zone focus group sessions. The Moderator Guide is included as Appendix C.

PAGE 36

26 Table 4. Focus Group Moderator Questions 1. Why do work zone crashes occur? Physical features of the roadway Issues with MOT and traffic control devices Driver behaviors Vehicle / Worker characteristics (trucks, exposed workers, etc) 2. What are positive things that are be ing done to help the situation? Advance Warning Speed Limits Better MOT Separation Barricades, barre ls, walls, and other devices Enforcement Public Information and Education 3. What are things that still need to change? Driver behaviors Physical design of work zone Traffic Control Enforcement Issues 4. What can we learn from Incidents that occur in work zones? Role of congestion Secondary collisions Location of incidents Type of collision (rear end, sideswipe, run off road, with barricades/equipment, etc.) Administration/conduct of meetings Administration of the focus groups brings together the planning associated with participants and the moderator(s ). Site setup requires the focus group team to arrive at the location early to prepare se ating, visual aids, recording eq uipment, and refreshments. Participants are greeted and offered refres hments, followed by the moderator welcoming

PAGE 37

27 everyone and beginning the session. A Po werPoint presentation accompanies the moderators program, primarily highlighting main questions and providing for supporting visual aids if necessary. Audio recording of focus groups is essential to later analys is of content. Recording is made possible by a laptop computer equipped with software and CM3 boundary microphones strategically placed near participan ts. A backup cassette tape recorder with one CM3 boundary microphone can be used as a backup device. Since it is important to link participants with their respective comments, specialized recording software is necessary to allow the resear ch team to annotate speakers while recording. The Meeting Commander software product is a commercial package [26] that allows the research team to graphically depict speakers with on-screen icons, the same way they are seated in the focus group setting. Once recording begins, the recording assistant simply clicks on speaker icons each time there is a change in speakers. Speaker changes are captured in the audio file, so that later playback depict s speaker changes. Elapsed time and speaker change time stamps are included in the digi tal audio recording. Fi gure 6 represents an illustration of the Meeting Commander screen, with speaker seating on the left and a chronological representation of the recording, complete w ith speaker changes, on the right. Post-Session Processing Qualitative analysis of the focus groups requi res that content be reduced to a text form. Post-session processing of the focus group audio recordings requires listening to the audio to review speaker co mments and subsequently tran scribing the audio into wordfor-word text. The Meeting Commander soft ware produces a time stamped chronological listing of speakers as an exportabl e text file. Playback of the audio using the software or

PAGE 38

28 Windows Media Player allows the person pe rforming transcription to monitor speaker changes and type text with the appropriat e speakers. A USB foot-pedal facilitates playback of the audio and eases transcription. The transcript is ultimately reviewed and compared to the actual audio reco rding for final quality assurance. Figure 6. Meeting Commander Recording Software Interface Data Element IdentificationQualitative Analysis After data is collected thr ough the use of focus groups, the data must be analyzed to provide usable results. Text analysis has been widely used for mining news outlets and more recently, for mining operations involving the internet, blogs, and email. Such analysis generally focuses on obtaining frequenc ies for words or phrases to attach some statistical significance to their occurrence. Analyzing text obtained in conjunction with qualitative research is often described as content analysis or qualitative text analysis. The systematic and replicable technique for compressing words or phrases in to content categories, given established rules, is a general description of the pro cess [27]. The analysis of open-ended text

PAGE 39

29 responses for qualitative research involves mo re than just determining word count or frequency. The process empl oys coding techniques that al low the researcher to reduce text which is produced by open-ended questions in a categorical way. A category is a group of words with similar mean ing or connotations. [28:37] After scanning the text, numerous codes are created, without a great deal of categorization. When screened in a more focused way, some codes are eliminated, grouped into larger themes, and/or subdivided [29]. The process of analyzing text with this coding enables the researcher to organize and digest the conten t of focus group data. Focus group expert Richard Krueger desc ribes the traditiona l process of focus group content analysis using colored pape r, colored markers, and cut and paste techniques [30]. Using these items, the rese archer identifies and rearranges text obtained from research to create descriptive summaries. Such techniques may appear rudimentary, but they are still used for qualitative anal ysis of focus group data. For our purposes, similar methods are used. Each focus group session is tape-recorded for later review and analysis. When audio recordings from focus gr oup sessions are transcribed, a w ealth of text data becomes available. These data, however, are unstruc tured and in a raw form, rather unusable for research purposes. Similar to other forms of da ta obtained in qualitati ve research, the text must be processed or analyzed in ways that make the data meaningful to researchers. The fundamental question of this analysis centers on finding mean ing in volumes of text. That question is answered in th e process of qualitative analysis. The process of qualitative text analysis can be summarized as three basic steps; 1) the reduction of the original database, 2) the constructi on of linkages, and 3) the

PAGE 40

30 comparison of findings [31]. Reducing the text is the process of coding, where segments of the data are given representations that are more easily manipulated and categorized. Constructing linkages is the attempt to fo rm coded units, based on subject meaning. Finally, those subject meanings are compared to infer invariants [31]. Through this process one can move the data from mere words to meaning. Focus group data can also be analyzed by computer. Many commercially available text analysis programs focus on word counting and frequency distributi ons. This form of analysis is most useful in marketing fields, and for media applications. Qualitative analysis software is a rapidly evolving area of research in the social sciences. Programs that have the ability to perform coding a nd other analysis functi ons are eagerly being embraced. Programs like SPSS Text Analys t, AQUAD, NUD*IST, ATLAS/ti, and HyperRESEARCH are all packages that have include capabilities useful for the social scientist and others who may need to qualitatively analyze data It is important, however, to remember that interpreting the meaning of te xt is something that the researcher cannot delegate to a computer. Therefore, the role of computers in qualitat ive analysis should be viewed as supportive of th e researchers duty to u nderstand the text [32]. Analysis of the data derived from focus group sessions with work zone stakeholders renders insight into the issues surrounding those locati ons and the types of factors that are involved in work zone incident s. Potential data elements can be identified from this form of qualitative research. Data Element DefinitionInterpretation of Results Analytical techniques transfor m raw data in the form of focus group text into more meaningful representations of the data. Analysis segregates data into categorical codes. Codes are measured using their intensity and frequency among various homogeneous

PAGE 41

31 groups. The product of the analysis is an or dered listings of codes. Linking that data with content from the focus groups is necessary to form meaning in the coded data. The product of linking codes with content reve als potential crash report data elements. Since a composite group of stakeholders is used to determine if the product of the qualitative research effort is representative s of stakeholders, anal ysis is essentially verified. The composite group can determine if the product is defective in any way, and also insure that it is fully exhaustive of the subject. No qualitative analysis of the composite focus group is necessary, since th eir involvement is merely a review of conclusions. Data Element TestingSupplemental Collection System The newly created data elements must be presented to individuals responsible for data collection in a way that maximizes the ch ances for success. Where Raub et al. [13] created a paper supplemental report form fo r police use, creating an instrument for collection now requires a format of electronic entry for officers, since their current crash reports are now in electronic form in many cas es (via a laptop computer in the police vehicle). An ancillary benefit to this format is the elimination of secondary data entry, from hand written crash reports to an electr onic database. This speeds entry of data, reduces the potential for data entry erro r, and makes data management easier. Data entry for the new work zone elements can be accomplished in three basic ways. The current officer reporting software can be reprogrammed to include added data fields, a web-based application can be cr eated, or a stand alone database can be implemented. The software currently used by FHP troopers to complete the Florida Traffic Crash Report is licensed by the agency from a pr ivate vendor. Making modifications to the

PAGE 42

32 software for purposes of this research woul d be cost prohibitive, involving programming changes to an end user software product. Such changes would resu lt in version changes to the product, and require upgrade of all agen cy computers. This may be desirable for permanent changes to fields of the crash repor t, but are not practical given the limited use and distribution of data elements for this pr oject. Because of these reasons, the current crash report application will not be altered for testing purposes. A web-based application offers utility and flexibility that are desirable for simple data collection. Given the infrequency of work zone incidents, and the limited data fields envisioned, such a format w ould be practical. One inhere nt disadvantage to web-based collection is the requirement for officers to leave the report applic ation, log into a web site, and enter data. A second disadvantage is the requirement to post-process the data, merging Florida Traffic Crash Report data wi th the newly created web-based data. Neither of these disadvantage s is viewed as significant, however. A common data field ensures that the crash report and supplemental work zone databases can be merged. This form of data collection will be used by the officers for testing, because it most closely resembles current re porting technology. A stand alone database that is not accessi ble by the end user is easily created with readily available application software. Th is method, however, is the least desirable alternative, particularly becau se of the need for post processing officer-collected data. Officers would essentially complete a paper report that would be coded by clerical personnel into the computer database. Su ch a system may be needed for wider application of the newly created work zone elements, given that many agencies/officers

PAGE 43

33 do not use vehicle-based computers. For purposes of this effort, a paper format will be created, but not used, since all testing officers are issued laptop computers. Technical framework of collection system The traditional hand-written police traffic cr ash report would logically lend itself to a supplemental report that is also hand-w ritten. When specia lized crash reporting software is used by police, the traffic crash report process is automated and the need for subsequent key punching or manual entr y into a database is eliminated. Both the hand-written report and the re port created with reporting software pose limitations when one considers modifying or supplementing those systems. Hand-written reports require modifications to printing, o fficer training, and data entry systems. Changing the many custom software packages used by police would involve significant investment in additional programming. While each may have advantages, neither is a pragmatic solution in an effort to add suppl emental data. Both have institutional and financial issues that would prove prohibitive. A desirable solution for supplementing tr affic crash reporting for work zonespecific data would be cost effective, not aff ect current reporting systems, and ultimately be palatable for the institutions that are ch arged with managing crash data. By creating a stand alone web-based collec tion system, expensive programmatic software changes are avoided, yet officers are able to take advantage of av ailable mobile computing technology. Officers who complete regular traffic crash report s in work zones are asked to additionally access a web site to provide uniq ue supplemental data. In this effort, the Florida Highway Patrol supports a supplemental work zone da ta collection web site and officers will use the system.

PAGE 44

34 Since the Florida Highway Patrol supports a computer network that encompasses all patrol vehicles, the framework for officer co mputing is in place. Each patrol car has a laptop computer with continual access to the network and internet. The agency supports an internet web site for public use, and an intranet web site exclusively for agency personnel. Several online databases are us ed by the agency, accessible by personnel who are authorized by means of a passw ord-protected architecture. For security reasons, Microsoft SQL 2000 [33] is well suited to create the database and ASP.Net [34] for the HTML user interface. These applications provide an increased level of security, since database operations can pose potential access to dangerous code or controls. Florida Highway Patrol servers are excellent host systems for supplemental collection applications, since th ey are already in place and th ey currently provide a level of security that is desirable. Microsoft SQL 2000 supports eXtensible Markup Language (XML) schema that are becoming the standard for crash report da ta. This will ensure compatibility and transportability of data collected in conjunction with this effort. Since each Florida Traffic Crash Report Form has a unique number assigned by the Department of Highway Safety and Motor Vehi cles, such a number becomes an excellent candidate for use as a primary key field. This field will link the traffic crash report with the supplemental work zone database. As a backup, additional fields may be designated as potential linkages between the databases. Operational framework of collection system After the data collection instrument is finalized, select Flor ida Highway Patrol (FHP) troopers serve as beta-testers for a short evaluation period. The process of field testing the elements, attributes and collection instrument re quires direction, training, and

PAGE 45

35 verification. FHP Troop G encompasses nine (9) counties in northeast Florida. The troop is staffed by approximately 140 sworn pe rsonnel. Troopers ar e introduced to the study by a supervisory memorandum that provide s an overview and instructions. This direction to participate in the research e ffort will establish organizational commitment and increase officer buy-in. Instructions are provided with this directive, and a time period for testing is established. Since all personnel involved with testi ng are familiar with incident (crash) reporting, each has general knowledge about work zones, and basic computer skills are present among users, it is anticipated that the learning curve for these testers will be minimal. Training will be accomplished by written instructions, with an opportunity for supervisory follow-up. Given there are curren tly several large work zone projects un derway in the study area (Troop G), the supplemental data elem ents and collection system will receive a reasonable amount of testing. There is no stat istical requirement for sample size, since such analysis is beyond the scope of this project. Because the supplemental system is not ma ndated by law or administrative rule, the system used by officers is essentially volunt arily. Organizational commitment from the Florida Highway Patrol repla ces a legal mandate for officers to use the supplemental work zone data collection system. Acco rding to information provided by the DHSMV, the FHP works about 65% of all work zone crashes in Florida and 95% of all work zone crashes on interstate highways. They are exce llent candidates for this project because of their significant role in reporting work zone incidents. Colonel Christopher A. Knights

PAGE 46

36 directive serves to require pe rsonnel assigned to the pilot test area to use the supplemental system in all cases where they complete a traffic crash report in a work zone. Data Element Validation Section 19 of the 2003 Edition of the MMUCC describes data elements and attributes that are unique to work zones. These elements closely resemble the data recommendations of the 1996 study by Wang et al. [9]. These elements are described in Table 1 of Chapter 2, and they represent th e current state of the practice for police reporting of work zone elements, albeit th e MMUCC guidance for work zones is not widely embraced by the states. Validating new data elements for work z one incidents is the final step in the development process, and this involves co mparing the newly created work zone data elements and attributes with those that c onstitute the state of the practice. Such a comparison can provide an indication of whet her current practices require modification. In addition, field testing data elements w ith officers in actual work zone crash investigations provides a pragmatic validation.

PAGE 47

37 CHAPTER 4 RESULTS AND ANALYSIS Qualitative Analysis Data Element Identification Because qualitative research assumes many forms and purposes, the method of analyzing data is equally diverse. While qua ntitative measures are typical in scientific and engineering disciplines, the social sciences and the analysis of qualitative data is not necessarily rooted in numbers. Some resear chers gravitate to the use of numbers while others steer away from their use. Thos e who can answer their research questions without counting codes should f eel well justified in doing so no appeals to imagined problems with statistic al independence or random sampling are necessary. [35:62] In quantitative analysis it is sometimes easy to get caught up in th e logistics of data collection and in the statistical analysis of da ta, thereby losing sight of theory for a short time. This is less likely in qualitative researc h, where data collection, analysis, and theory are more intimately intertwined [36:370]. In the first step in qualitative analysis, text and audio is reduced to data segments. A code is used to represent those segments [31]. Nearly all qual itative research is analyzed using some form of coding. Coding is the process of transforming text data into a form that is more standardized [36]. The most common form of coding for focus groups concentrates on manifest content [35]. Manifest content is that which is on the surface, seeking the occurrence of a term or concept [36]. Latent coding uses the underlying meaning of what is sa id to standardize content [36] Coding is accomplished by analyzing transcript text and identifying the te rms, concepts, and content that work

PAGE 48

38 together to convey a similar meaning, that is, they fit similar categories. Categories can be related to sub-categories, a nd each is considered a code. While mere occurrence of words or concepts may be analytically useful, it is also beneficial to determine what groups felt were important issues. Morgan describes factors that indicate the emphasis given to a topi c by a group as the product of how many groups mentioned it, how many in each group men tioned it, and how much energy they associated with the topic [35] The energy and enthusiasm that individuals display for a topic appears somewhat subjective, and therefore this measure will not be attempted in this effort. Qualitative analysis for this inquiry involve s steps that potentially produce a list or set of work zone discussion issues. From t hose issues, the most important factors can be determined using qualitative measures, notably measures of agreement and intensity. Agreement describes the number of focus gr oups where a given code was discussed, and intensity describes the number of mentions of the code with in and across groups. A third potential measure identifies cases where all participants mention a code, but it is more appropriate for market research thus it is not included in this effort [35]. Automated methods of analysis using speci alized computer software is gaining popularity; however, more traditional manual a pproaches like the long table method are still effective, if not as glamorous. As a process, analysis involves reducing focus group audio recordings to word-for-word text, copied on uniquely colored paper for each different focus group. The text is read to find common themes in content and notations are made in the margin of the transcript. Then text is physically cut out of the document using scissors and is grouped and later subgrouped based on topic or code. The process

PAGE 49

39 continues until an exhaustive analysis of the text renders many assembled bits of what speakers said into logical groups. These groups form the categorical basis of the codes which drive the process of transforming text in to meaningful represen tations of what the people said in the focus groups. The coded text is measured using the methods previously described and some conclusions about importance and relevance can be made. Figure 7. Long Table Method of Analysis A Priori Categorization Before formal text analysis begins, a nu mber of work zone crash factors can be considered relevant, without biasing results. As is the case with any traffic safety analysis, causation factors associated with drivers and the roadwa y environment readily come to mind as potentially important. Becau se there are a myriad of traffic control devices used in work zones like signs, cones, barrels, message boards, and barricades, a category for traffic control devices is likely im portant as well. Driv er behaviors, traffic control devices, and construction/roadway condi tions form a broad basis for categorizing

PAGE 50

40 work zone crash factors. Creating a category for other factors provides an opportunity for coding to reach beyond the boundaries of these broad coding categories, and ensures a comprehensive qualitative analysis is possibl e. In reviewing the audio of each focus group, these general categories represent most di scussion content. While these categories only form an initial framework for analysis, th ey prove useful as a starting point for the process of coding. From these a priori code s, deeper coding is possible using more emergent techniques. Emergent Categorization A more detailed analysis is illuminating of the subject of work zone crash factors, based on the discussions of the various focu s groups. Using the traditional long table method of focus group analysis, extracted text pieces can be grouped according to content, and categories emerge from the process. As the text is analyzed and re-analyzed, more and more groupings become possible. This process is aptly named emergent categorization, because with each review of the text, more categories emerge. In the first emergent categorization effo rt, 44 categorical codes were created to represent the discussion of par ticipants in the five focus groups. Another group of text items remained, comprised mostly of single sp eaker issues, and topical items that were isolated within the context of the larger work zone subject. Similar to the larger category of other, a code for other was used to capture this orphane d subset of codes. Final Categorization Emergent categorization provide s a level of detail in c oding that is suitable for purposes of content analysis. Because the l ong table method physically segregates text, it does not always make it easy to identify stri ngs of text that may contain multiple codes however. For this reason, a complete versi on of each focus group text is reviewed after

PAGE 51

41 long table, using the codes de rived from the emergent proces s. Babbie points out that, You can always code and recode and even r ecode again if you want, making certain that the coding is consistent [36:324]. Salient di fferences in either code or content were resolved by either expanding the code set or collapsing categories. This additional review of the transcripts can demonstrate a need for more codes, representing more specific content. The final process of c oding rendered 69 unique categorical codes. Since each focus group is homogeneous in composition, a group representative of a cross-section of traffic safety stakeholders is needed to verify the codes. The Alachua County, Florida Community Traffic Safety Team (CTST) provides an excellent composite group of stakeholders to review codes. Their monthly meeting on April 20, 2006 provided a setting for presenting the fi ndings of the resear chers and soliciting feedback. The group was comprised of individua ls from law enforcement (6), FDOT (4), local public works departments (3), private utility companies (2), emergency medical services (2), and independent persons (3). The CTST group found no disagreement with the codes derived from the homogeneous focus groups. A spreadsheet listing 69 unique categorical codes was created to begin the process of cataloging the responses of the focus gr oups. Columns representing the five focus groups were aligned adjacent to each code, to correlate the code with the focus group. Each focus groups content was analyzed to identify unique speakers that contributed to discussion relating to the subjec t contained in the unique code categories. The numerical speaker assignment for each focus group partic ipant was noted for every participant who entered into discussion relating to the code. The table below reflects several of the codes, and the respective speaker assignments. Fo r example, the code for Movement of

PAGE 52

42 Construction Equipment was discussed by speakers 3, 6, 2, and 5 in the Police focus group, and by speaker numbers listed in the other groups similarly. Table 5. Unique Speakers by Group and Code Discussion Issue Police Citizen Industry FDOT LC FDOT Tal Movement of Construction Equipment 3,6,2,5 8,5, 5,1, 9,4, 1,2,4,5,3,6 Daytime vs. Nighttime Road Work 2,6, 7,8,9,2,10,11 1,5, 9,10, 2,6,5,1,4, Worker Present 5,6, 2,6,8,10,11,5 5,1,2 4,9, 4,6, Narrow Lanes 6,5 2, 5,1, 2, 1,5, Queuing and backups 2,5,6,3, 10,4,8,7, 1,5, 10,1,6,4, 2,4,3,1, Driver Speed 2,6,5 8,10,2,4,11 1,5,2,3 9,4,2, 1,7, Driver Distraction (General) 6,1,5 7,8,10,4, 1,5,3 10,9,6,1,4,2 3, Law Enforcement Static vs Ticketing 4,5,3,6 4,2,10,8,3,9,11 5,1,2 9,3,6,2,4, 2,7,4, Law Enforcement Visibility 3,6,2,5 5,2,8,4,3 1, 10,6,4 2,4,6, Enhanced Fines 2,5, 8,9,11,4, 5,1,2 9,4, Advanced Warning 1,6,5, 3,11, 5,1, 10,9,4, 4,5,3, Artificial Work Lighting at Night 6, 1,2 4, 5, Lane Shifts 3,2 2,8,5, TCD Lighting / Night Visibility 6, 5,2, 1, 1, 1, Law Enforcement MOT Training 5,6,1, 5,1, 6,2, 2, Advanced Notice of Work Zone 2,6,5 4,2,10,2,7,8,11 2,5,1, 3,10,4,9,5, ITS and Variable Message Signs 2,6 8,10,4,6,7, 5, 10,4,5,6, 2, TCD Maintenance 2,6, 5,7,8,2, 1,5, 1,2,10,9, Work Zone Project Physical Length 7,8,3, 5,1 6,9, 3,6, Work Zone Project Time to Complete 6, 2,9,10,7,8,4,3,5 5, 6,4 1,6, No Shoulder / Drop off 6, 6,2, 5,2,1, Visual obstruction created by barrels or other TCD 6,3, 4,2, 1,2, Work Zone Contributing 6,2,3 7, 7,4,2, Lane Closures / Merge 1,5 7,10,4,5,8, 5,2 10,9, 2,5,1,6, Driver Training and Education 2,5,4,3,6, 5,2, 9,6, 2, Impaired Drivers 4,8,10,2, 9, 2,3, Location of Crash within the Work Zone 6, 2,1,5 4, 2,1, Photographs as part of reporting 1, 9,4,6, 4,1,3,5, Police Reporting Narrative & Diagram 2,3,6,1, 1,5, 2,3,1,5 Flagperson / Worker Action 6,2, 10, 4, Speed Trailers 2,6,3, 5,3,2 4,9, TCD Clarity of message 6, 8,11,10,2, 2, 4, Color / Reflectivity of Items in Work Zone 10, 1, 2,4, Temporary Rumble Strips 4,9, 2, Driver Licensing 4,5,2,6,7,8,10, 5,1, 4,6, Side Street Control during Lane Closures 7,5, 6,4,9, Liability 1,3 2,1,5 9, Flagger / Worker Training 6,2 4,9,1, Law 5, Changing MOT plans 1,5,2 4,9,1, 1,6, Highway Advisory Radio 2,8, 6,4,9 Commercial Motor Vehicles 6

PAGE 53

43 Table 5. Continued. Discussion Issue Police Citizen Industry FDOT LC FDOT Tal FDOT MOT report 6 9,4,1, 2,1, Harmful Even sequencing in reports 3, TDC Speed Signs 3,6,5 3,5,8 Type of channeling device 2, Length of Tapers 1 10, 5,2, 1,2 Worker Fatigue 9,6 Hydroplane, Standing Water 5,6 Type of Work Being Done 2 Law Enforcement Positioning with TCD (rolling) 6,10,9,2 Law Enforcement not part of MOT planning process 2 Alternate Routes & Detours 2,4,3,6 1 Commonplace of Road Work 2,5 7 5 Productivity or money outweighs safety 3 4,9 Human Toll of Work Zone Crashes 5,8 5 Business Access during Construction 1 Driver Training for Those Ticketed in Work Zones 2,1,5 6 Recurring crash sites 6 2 End Work Zone Signs Needed 2,4,10 Temporary Striping 3,6 9,4 Temporary Curbing 2 2 Presence of Temporary TCD* 2 Type of TCD present / used 5,1 2 Weather Conditions 6 Adding WZ data or fields on reports 2,5,6,3 Driver behavior (general) 6,5, 3,10 1 Police Rolling Roadblock 5,1 2 MOT Devices Hit 2 Agreement Measure Whether each groups discussion contained a given code was noted as a measure in analysis. Migrating from unique speakers in each group, based on subject codes, to a measure of agreement is a relatively simple ta sk. As a more general representation of the unique speaker assignment, if any code had one or more speakers for the focus group, it was considered included in the agreement measure. To simplify the task of tracking these measures, a binary code is created w ith 1 representing agreement and 0 representing cases where there was no disc ussion within the focus group.

PAGE 54

44 When binary codes for individual focus gr oups are totaled, the level of agreement among groups can be determined. The highest level of agreement would be represented by a numerical measure of 5, representing that all 5 focus groups contained some discussion of the particular code. Conversel y, a measure of 1 woul d indicate that only one of the 5 groups discussed a code. The m easure of agreement being totaled for all 69 codes, a ranked listing is possibl e, listing those with the most agreement (5) to those with the least agreement (1). The table below de picts all codes and their respective binary measures of agreement. Table 6. Agreement by Group and Code (Binary) Discussion Issue Police Citizen Industry FDOT LC FDOT Tal Total Movement of Construction Equipment 1111 15 Daytime vs Nighttime Road Work 1111 15 Worker Present 1111 15 Narrow Lanes 1111 15 Queuing and backups 1111 15 Driver Speed 1111 15 Driver Distraction (General) 1111 15 Law Enforcement Static vs Ticketing 1111 15 Law Enforcement Visibility 1111 15 Advanced Warning 1111 15 TCD Lighting / Night Visibility 1111 15 ITS and Variable Message Signs 1111 15 Work Zone Project Time to Complete 1111 15 Lane Closures / Merge 1111 15 Enhanced Fines 1111 04 Artificial Work Lighting at Night 1011 14 Law Enforcement MOT Training 1011 14 Advanced Notice of Work Zone 1111 04 TCD Maintenance 1111 04 Work Zone Project Physical Length 0111 14 Driver Training and Education 0111 14 Location of Crash within the Work Zone 1011 14 TCD Clarity of message 1111 04 Length of Tapers 1110 14 No Shoulder / Drop off 1001 13 Visual obstruction created by barrels or other TCD 1110 03 Work Zone Contributing 1100 13 Impaired Drivers 0101 13 Photographs as part of reporting 0011 13 Police Reporting Narrative & Diagram 1010 13

PAGE 55

45 Table 6. Continued. Discussion Issue Police Citizen Industry FDOT LC FDOT Tal Total Flagperson / Worker Action 1101 03 Speed Trailers 1011 03 Color / Reflectivity of Items in Work Zone 0111 03 Driver Licensing 0111 03 Liability 1011 03 Changing MOT plans 0011 13 FDOT MOT report 1001 13 Commonplace of Road Work 1110 03 Driver behavior (general) 1110 03 Lane Shifts 1100 02 Temporary Rumble Strips 0001 12 Side Street Control during Lane Closures 0101 02 Flagger / Worker Training 1001 02 Highway Advisory Radio 0101 02 TDC Speed Signs 1100 02 Alternate Routes & Detours 0100 12 Productivity or money outweighs safety 1001 02 Human Toll of Work Zone Crashes 0110 02 Driver Training for Those Ticketed in Work Zones 0011 02 Recurring crash sites 1000 12 Temporary Striping 1001 02 Temporary Curbing 0011 02 Type of TCD present / used 0010 12 Police Rolling Roadblock 0010 12 Law 0010 01 Commercial Motor Vehicles 1000 01 Harmful Even sequencing in reports 0000 11 Type of channeling device 0000 11 Worker Fatigue 0001 01 Hydroplane, Standing Water 1000 01 Type of Work Being Done 0000 11 Law Enforcement Positioning with TCD (rolling) 0001 01 Law Enforcement not part of MOT planning process 0000 11 Business Access during Construction 0010 01 End Work Zone Signs Needed 0100 01 Presence of Temporary TCD 0000 11 Weather Conditions 1000 01 Adding WZ data or fields on reports 1000 01 MOT Devices Hit 0000 11 A total of 14 codes represent agreem ent among all 5 groups. Another 10 codes indicate agreement among 4 groups, and 15 codes each for agreement among 3, 2, and 1 groups.

PAGE 56

46 Weighted Measure of Intensity All mentions of a code within a focus group provide a potentially valuable measure of intensity. Because each focus group varies in size, it was important to weight the measure of intensity, based on the size of th e group. Five individual speakers discussing a code in a group of six has greater intensity than a similar number discussing an issue in a group of ten. When homogeneous focus gr oup intensity is weighted, measured, and summed, an intensity score results. Equation 1 gives the formula for calculating this intensity value. i i i i in n S I (1) where I = Weighted Measure of Intensity Si = Unique speakers discussing code in focus group i ni = Number of participants in focus group i For example, Movement of Construction E quipment is a discussion topic in all five focus groups, with 4, 2, 2, 2, and 6 speak ers contributing to the discussion in each group respectively. Since the total number of participants in each focus group is 6, 10, 4, 8, and 7 respectively, the weighted inte nsity formula would be applied as: 1 3 35 110 7 8 4 10 6 7) (6 8) (2 4) (2 10) (2 6) (4 I All 69 codes can be calculated for weight ed intensity and their values ranked accordingly. While the highest possible inte nsity measure is 7.6, th e highest recorded

PAGE 57

47 measure was 4.2, and the lowest 0.2. The tabl e below depicts code weighting using the formula above. Table 7. Weighted Measure of Intensity Discussion Issue I Law Enforcement Static vs. Ticketing 4.2 Queuing and backups 3.8 Daytime vs. Nighttime Road Work 3.5 Driver Speed 3.5 Law Enforcement Visibility 3.5 Advanced Notice of Work Zone 3.4 Driver Distraction (General) 3.3 Lane Closures / Merge 3.3 Movement of Construction Equipment 3.1 Worker Present 3.0 ITS and Variable Message Signs 3.0 Advanced Warning 2.6 TCD Maintenance 2.6 Work Zone Project Time to Complete 2.6 Driver Licensing 2.4 Enhanced Fines 2.3 Driver Training and Education 2.3 Work Zone Project Physical Length 1.9 Impaired Drivers 1.8 Police Reporting Narrative & Diagram 1.7 TCD Clarity of message 1.7 Photographs as part of reporting 1.6 Narrow Lanes 1.5 Law Enforcement MOT Training 1.4 Work Zone Contributing 1.4 Changing MOT plans 1.4 TDC Speed Signs 1.4 TCD Lighting / Night Visibility 1.3 Speed Trailers 1.3 Side Street Control during Lane Closures 1.3 Highway Advisory Radio 1.3 FDOT MOT report 1.3 Alternate Routes & Detours 1.3 Lane Shifts 1.2 No Shoulder / Drop off 1.2 Visual obstruction created by barrels or other TCD 1.1 Location of Crash within the Work Zone 1.1 Length of Tapers 1.1 Flagger / Worker Training 1.0 Driver behavior (general) 1.0 Flagperson / Worker Action 0.9 Color / Reflectivity of Items in Work Zone 0.9 Liability 0.9 Law Enforcement Positioning with TCD (rolling) 0.9 End Work Zone Signs Needed 0.9 Artificial Work Lighting at Night 0.8 Temporary Striping 0.8 Temporary Rumble Strips 0.7 Commonplace of Road Work 0.7 Human Toll of Work Zone Crashes 0.7 Adding WZ data or fields on reports 0.7 Productivity or money outweighs safety 0.6 Driver Training for Those Ticketed in Work Zones 0.6 Worker Fatigue 0.5 Recurring crash sites 0.4 Type of TCD present / used 0.4 Police Rolling Roadblock 0.4 Hydroplane, Standing Water 0.3 Temporary Curbing 0.3 Commercial Motor Vehicles 0.2 Harmful Even sequencing in reports 0.2 Type of channeling device 0.2 Type of Work Being Done 0.2 Law Enforcement not part of MOT planning process 0.2 Presence of Temporary TCD 0.2 Weather Conditions 0.2 MOT Devices Hit 0.2 Law 0.1 Business Access during Construction 0.1

PAGE 58

48 Composite Ranking of Items Recall that according to Morgan, the way to determine the emphasis groups give to a code is measuring how many persons in each group mention a code (intensity) and how many different groups mention a code (agreemen t). He refers to this a group-to-group validation [35]. To make the results of agreement and intensity rankings more meaningful, the measures can be combined to produce a composite list of codes. The composite list seeks to determine those items that have an agreement measure of 4 or 5 and an intensity in the 80th percentile. The 80th percentile was chosen, since it represents a measure analogous to that of 4 out of 5 groups. A total of 15 code items met the criteria of the composite ranking. Table 8 is a side -by-side comparison of the codes with an 80th percentile intensity (15 codes) and those codes (24) with agre ement measures of 5 or 4. Both columns in the table are sorted alphabetically for ease of comparison. Table 8. Comparison of Intensity and Agreement Measures Agreement Weighted Intensity I Advanced Notice of Work Zone Ad vanced Notice of Work Zone 3.4 Advanced Warning Advanced Warning 2.6 Artificial Work Lighting at Night Daytime vs Nighttime Road Work 3.5 Daytime vs Nighttime Road Work Driver Distraction (General) 3.3 Driver Distraction (General) Driver Speed 3.5 Driver Speed ITS and Variable Message Signs 3 Driver Training and Education Lane Closures / Merge 3.3 Enhanced Fines Law Enforcement Static vs Ticketing 4.2 ITS and Variable Message Signs La w Enforcement Visibility 3.5 Lane Closures / Merge Movement of Construction Equipment 3.1 Law Enforcement MOT Traini ng Queuing and backups 3.8 Law Enforcement Static vs Ticketing TCD Maintenance 2.6 Law Enforcement Visibility Work Z one Project Time to Complete 2.6 Length of Tapers Worker Present 3 Location of Crash within the WZ Movement of Const. Veh/Equipment Narrow Lanes Queueing and backups TCD Clarity of message TCD Lighting / Night Visibility

PAGE 59

49 Table 8 Continued Agreement Weighted Intensity I TCD Maintenance Work Zone Project Physical Length Work Zone Project Time to Complete Worker Present Highlighted = 4 of 5 groups When the top 20% of the table for weight ed intensity and all agreement measures of 4 or 5 are integrated, table 8 is the resulting list of 14 code s. No ranking of these items is necessary, so they are listed alphabetica lly for ease of reading between tables. Table 9. Composite Codes Composite Codes Advanced Notice of Work Zone Advanced Warning Daytime vs Nighttime Road Work Driver Distraction (General) Driver Speed ITS and Variable Message Signs Lane Closures / Merge Law Enforcement Static vs Ticketing Law Enforcement Visibility Movement of Construction Equipment Queuing and backups TCD Maintenance Work Zone Project Time to Complete Worker Present Interpreting Analysis Definition of Elements While reducing text and audio to codes is the important first step in qualitative analysis, reconstructing content to derive some meaning is equally important. In accomplishing this task, the researcher seeks to create linkages in sometimes subjective meaning. Each focus group participant and each group discussion is un ique. The task for the researcher is to find common themes in wo rds to discover relati onships. It is from those relationships and meanings that we can look beyond mere codes and understand the basis of the content. This process will ul timately allow codes to be transformed into potential crash report data elements.

PAGE 60

50 The following section examines the composite list of codes by presenting descriptions of agreement and synopsizing focus group discussion. Creating Linkages Advanced Notice of Work Zone Four of five groups discussed the need for public notification of work zone activ ities, in advance of such work. The use of print and electronic media, and the deployment of va riable message boards before construction begins were cited as important ways to in form drivers before changes in the driving environment occur. A representative stat ement from the citizen focus group gives an example of the discussion, I think the co mmunity would be better off if there was a mailing letting you know what exactly this project is. Advanced Warning Signage Advanced warning signage was noted as import in all five focus groups. Warning drivers of potential changes in the driving environment was cited as important because of factors related to driver ex pectancy. An example of the discussion is evident in the police focus gr oup where the following was said, A lot of times the construction company will go in and make changes in the pattern or flow of traffic and they'll make these changes and there isn't often adequate signs for it. Daytime vs. Nighttime Road Work Every group discusse d differences between construction activities occurring during the nighttime versus the daytime. While there was no consensus regarding which time of day wa s perceived to be safer, the effects on both traffic flow and safety were noted a nd discussed. A representative comment from one of the FDOT focus group noted, There are also times where we require operations to be done at night and you cant do them in the daytime because of the traffic volume impact.

PAGE 61

51 Driver Distraction Similar to driver speed, driver distraction was noted in every group as a significant reason that work zone co llisions occur. Distractions inside and outside the vehicle are viewed as important fact ors for drivers in work zones. An officer from the law enforcement group pointed out, More and more distractions take place inside the vehicle, then they're really not paying attention to what s going on outside the vehicle. Driver Speed Without fail, every group noted driver speed as one of the very first discussion issues that contribute to work zone crashes. The failure of drivers to comply with normal or reduced speed limits in work zones was particularly relevant in focus group discussions. The FDOT focus group cond ucted in Tallahassee noted, We try to slow them down sometimes 10-20 mph below the speed limit and they continue to travel 10 mph faster than the speed limit. The poli ce confirm the role of speed with comments such as, I got to concentrating at a place that is 45 and they are not usually running 45. The contractor group noted succi nctly, They dont slow down. ITS and Variable Message Signs All focus groups discussed the use of variable message signs as advanced warning devices and additionally their role as advance notice. A representative statement from the FDOT focus group in Lake City was, Variable message boards, keep them up to datechange your message on it a lot. It keeps them looking at it, helps out. Lane Closures/Merges All focus groups discussed lane closures and merge operations as factors in work zone crashes; however, the nature of the discussion was varied. The citizen group di scussed driver behavior in merge situations, while other groups discussed the warning ahead of a merg e, and the place within the merge where

PAGE 62

52 crashes occur. A characteri stic comment from the citizen group notes, Sometimes its really unclear if one lane has been closed. Another citizen participant added, I notice thats very irritating you see a sign that says right lane closed ahead, and everybody jumps in that right lane and they just keep going and going. Law Enforcement Static vs. Ticketing All groups discussed the importance of active law enforcement in work zones, to modify driver behavior. The use of static patrol cars, parked in or near work zones, was noted as not being as effective as the visible car and officer engaged in enforcement action. The citizen focus group noted, I think that there should be an officer or two officers actually walking the construction site in between the workers. Engineers from th e FDOT focus group in Lake City made statements like, We dont want them (police) sitting on the side of the road, but want them writing tickets. Law Enforcement Visibility Similar to extensive disc ussion by all groups about law enforcement action in the work zone, it was certain among participant groups that law enforcement visibility has an impact on driver behavior. Visible presence of enforcement vehicles and officers is seen as relevant to compliance with speed limits in work zones. A representative comment from the citizen focus group notes, I like the fact that they have a highwa y patrol car there. That al ways gets my attention. The flashing lights. A participant from the contractor group adds, One of the biggest deterrents for us that Ive seen is when you do have an officer out there and the blue lights are on. Movement of Construction Vehicles All groups discussed the movement and actions of construction vehicles as potentially contributing, directly or indirectly, to work

PAGE 63

53 zone collisions. Vehicles entering or crossi ng traffic were noted, as well as slow moving vehicles which enter or leave work zone ar eas. A comment from the Lake City FDOT focus group is representative of the discussion, I didnt have such a problem with pickup trucks and vehicles that generally can acceler ate and get across the road real fast. But when you start putting a 20 ton dump truck tryi ng to run across the road, I dont like that. Queuing and Backups All groups mentioned rear-endtype collisions as being particularly problematic in the work zone se tting. Reductions in speed limits, physical changes in the driving environment, and driver actions were all note d as contributing to queues and backups. A representative co mment from the police group was, Traffic stops and he rear ends it. Traffic Control Device Maintenance Four of five groups cited the maintenance of traffic control devices as an important fact or in evaluating crashes in work zones. Misaligned cones and barrels, along with ligh ting on devices were sp ecifically mentioned as relevant. The contractor group noted, I know for some of our job sites, weve got an assigned MOT crew where its sometimes two guys and they ride the site all day pretty much setting up cones, taking them down, and maintaining what weve got thats out there. Work Zone ProjectTime to Complete The long duration of construction activity, often months or years, was noted by all gr oups as a factor in crashes. Driver complacency with warnings and environment changes was believed to occur when projects take so long to complete. A re presentative comment from the citizen group summarizes the issue, Boy highway construction takes along time.

PAGE 64

54 Workers Present Every group noted differences in the dangers of a work zone when workers and construction activity is pres ent. All groups indi cated that it was a relevant factor in crashes whether or not workers are present. One comment from the citizen group relates, So they dont slow dow n figuring, well I can wait because if there are no workers there then you are not endangering them. The descriptive summary of how each c ode was discussed among focus groups is quite revealing. When accompanied by just a few sample quotes from the focus groups, the codes become more real and illustrate the process of linking codes back to actual content. Converting Codes to Data Elements Focus group discussion items make up the codes used in qualitative analysis. Codes that were determined to be of suffi cient importance were subsequently identified as potential crash data elements, through a process of linking them with focus group content. The next step in the data elemen t development process is to determine which codes have potential value for inclusion in an effort to improve crash data. Moving potential codes to data elements requires making a determination of whether the data sought is available from other resources. According to the MMUCC, it is desirable to create linkage to ot her sources of data whenever possible to reduce the burden of da ta collection at the scene [6]. The term linkage here is dissimilar from that which wa s used in qualitative analysis to clarify codes. In this case, linkage refers to managi ng data sets so that they can be combined or merged. For example, a traffic crash repor t would not capture some information about injury mechanics, because that informati on could alternatively be obtained by linking with EMS, hospital, or insurance records.

PAGE 65

55 Department of Transportation project file s would likely be a source for information on advance notice and public information efforts, as well as the spatial and temporal length of projects The speed of the vehicle and the determination of whether speed was a contributing cause are current data elements on the Florida Crash Report, as is driver distraction Law enforcement ticketing activities ma y be obtained from traffic citation records or other data stored within enforcement agencies. The role of daytime versus nighttime work is potentially a de rived data element, if current reporting fields capturing lighting conditions are married with a new element capturing workers being present. Table 10. Data Linkage Composite Codes Linkage Potential Advanced Notice of Work Zone FDOT project records Advanced Warning Daytime vs Nighttime Road Work New Element + Current Driver Distraction (General) Current Report Field Driver Speed Current Report Field ITS and Variable Message Signs Lane Closures / Merge Law Enforcement Static vs Ticketing Citation Data Law Enforcement Visibility Movement of Construction Equipment Queuing and backups TCD Maintenance Work Zone Project Time to Complete FDOT project records Worker Present Elements as Interrogatives / Binary Values After codes are filtered using analysis techniques, eight codes remain, that are potential work zone crash data elements. The guidance of the MMUCC requires that data elements be appropriate, that is, they must be needed for traffic safety purposes and not be administrative in nature [6]. All of the remaining codes meet that requirement. While the MMUCC seeks to minimize the total number of data elements in the interest of officer time, the proposition of supplementing data collection does not conflict with that

PAGE 66

56 objective. Since work zone crashes are fa irly infrequent events, supplemental data collection would not unduly burde n individual officers. Supplemental elements for work zones would not be used in cases where ther e was not a work zone involved. Null data would not be captured. Having identified topical work zone issues in the form of codes and subsequently linking discussion to form context, it is then possible to form the basis of data collection, the data element. Traditional crash reports make extensive use of terse categories that represent data elements and pick lists of attributes. Attribut es are typically given alpha or numeric codes to make them suitable for data entry and analysis. Given the objective of broadening data about work zones without bur dening officers, the use of interrogative statements can speed data collection. Transf orming potential data elements into complete questions allows for the use of simple yes/no responses from officers. This allows for a more accurate description of work zone featur es that are sometimes not familiar to police officers. The state of Wisconsin (Figure 2) makes use of yes/no input fields, and Pennsylvania (Figure 3) employs interrogativ es with check boxes to capture certain information in their reporting format. The use of both the yes/no input method and the interrogative as a data element are bot h proven techniques for data collection. Creating questions from potential data el ements is the product of again examining the context of focus group discussion. Stakehol ders discuss work zone issues in various degrees of detail, and consequently pr ovide insight into their data needs.

PAGE 67

57 Figure 8. Data Analysis Process

PAGE 68

58 One can determine the type of data they seek, and construct report questions that officers can answer, thereby improving the data set. .It would be impossible to create a data element and associated set of attributes for every conceivable work zone incident scenario. Even the availability of multiple attributes cannot satisfy the myriad of situations one may observe. The crash repor t diagram and narrative are excellent tools for documenting things that are not suitable for coding. Officers of ten neglect to fully utilize the crash report narra tive and diagram to document work zone attributes and observations. The supplemental system for colle cting work zone incident data can also serve as a pointer system, reminding the o fficer to collect additional information, and informing analysts examining the data element that the master crash report form may contain additional data. This bolsters the effectiveness of the supplemental system, without duplicating the value of the crash report narrative and diagram. Table 11. Converting Elements to Supplemental Report Questions Potential Element (Code) Report Interrogative (Yes/No Format) Queuing and backups Did a backup or queuing of traffic contribute to the crash? Law Enforcement Visibility Was an on or off-duty police officer working in the construction zone nearby? Lane Closures / Merge Did the crash occur within a lane closure or merge section of roadway? Movement of Construction Equipment Did the movement of construction trucks or equipment contribute? If so, explain the type of equipment or vehicle and nature of the movement in the crash report narrative. ITS and Variable Message Signs Was a variable message sign or arrow board used to warn of construction ahead? If so, please include the location in your crash report diagram. Worker(s) Present Were workers present in the vicinity of the crash? Advanced Warning Were advanced warning signs in place? Please be sure to include the location of advanced warning signs in the crash report diagram. Traffic Control Device Maintenance Were temporary traffic control devices (signs, barricades, cones, pavement markings, etc.) in good condition and proper working order at the time of the crash? If No, please describe in the crash report narrative.

PAGE 69

59 By reminding the officer to include the t ype of construction ve hicle or equipment and the nature of the movement in the crash re port narrative, the data set is accomplished. When the officer adds the location and position of message boards and advance warning devices to the report diagram, valuable info rmation concerning those two elements is documented. Supplemental Collection System Element Testing Developing new traffic crash report data elements, specifically for work zone incidents, sets the stage for improving the cr ash data set. Merely identifying potential data elements, however, does not fulfill the object ive of this research effort. The second component required for supplementing work z one data collection requires a mechanism for actually collecting data at the officer level. Once such a mechanism is chosen, actual testing by field officers, in real crash scenarios, can help transform conceptual supplementing of work zone data into an actual proof of concept. While creating a simple paper form may be the easiest way to capture additional data, it is decidedly not the preferred method. Using a web-ba sed system allows officers to expedite data collection, and takes full adva ntage of technologies currently available to officers. To implement such an electronic system, technical and operational frameworks must be considered. Technical Framework of Collection System A web-based collection system requires, in its simplest form, a database structure and a computer network structure. Database structure describes the data fields used, while the network structure describes the us er interface, hosting, and storage systems. Using an eXtensible Markup Language (XML ) data schema is a requirement for this type of activity, since the overall objective is maximum transportability of the data.

PAGE 70

60 Any number of commercial data bases could be used to create the database structure, but since ActiveX controls are a security c oncern, Microsoft SQL 2000 was chosen. This powerful database application allows for th e expedient development of databases using tabular systems. Most importantly, enhanced security features ensure that host systems will be protected from potential intrusion. Before simply including work zone specifi c data elements, the prospect of linking the supplemental database with a larger traffic crash reco rd system is essential to complete data. Primary key fields must be created to provide the link between independent data sets and/or tables used by the database. The unique and sequential HSMV number on each Florida Traffic Crash Report Form provides an excellent field for linking data sets. The supplemental data base will include an HSMV number that will use the same numeric format as the original. The database will also include a feature that returns an error if the user enters a duplicate number. S ubordinate to the HSMV report number key field, additional data fields can capture data that dup licates other Traffic Crash Report Form data, for the purpose of providing additional linka ges, and also to make the supplemental database somewhat stan d alone. These additional data fields are date, county, officer ID, and location of the crash. The officer ID is automatically captured from the officer login, via an authorized user table. An additional variable for Long Form or Short Form differentiates the traffic crash report type, based on reporting thresholds established by Florida Statut e. This variable provides an additional mechanism by which to sort, search, and li nk the supplemental database with the two types of master crash report This is beneficial, sin ce Short Form reports are not considered for Florida statistical reporti ng purposes. These reports are however,

PAGE 71

61 automated by the FHP, and therefore potential ly enhance the overall data set, when combined with the supplemental work zone reporting data. Having created a database structure to ha ndle the basic data necessary for linking the supplemental database, the remainder of th e database structure involves the inclusion of the twelve data elements that were created herein. The structure of these data elements is simplified by the use of binary codes that represent agreement with the interrogative that forms the basis for the element. For example, the question Did a backup or queuing of traffic contribute to the cr ash? would be captured as a yes/no response, and stored in the supplemental database as a value of FA LSE for no and TRUE for yes. The figure below depicts the data structure us ed for the supplemental work zone data collection database. *Florida Traffic Crash Report shown for illustration purposes only Figure 9. Supplemental Work Zone Database Structure Microsoft ASP.NET was selected to create a simple web-based design interface for officers who would use the supplemental database system. The software offers excellent

PAGE 72

62 tools for the user in terface and exceptional support for desirable security features. Officers accessing the supplemental reporting intr anet web page are screened in, based on their access to the wire less network and their logon pass word. Their ID number is automatically captured and that information populates the screen and database. A date picker tool is provided to assi st officers with the date fiel d. Table-driven drop down lists are used to assist officers in selecting the county of crash and major work zone project locations underway. Radio buttons are prov ided to toggle between the Long Form and Short Form report types. The HSMV report number is a manual en try, with a feature that prohibits duplicate numbers being used. Each of the seven work zone-specific supplemental data elements is displayed as a question, listed on the screen for the user to see. Yes and No radio buttons accompa ny each question. Edit rules require one of the two buttons to be selected or an erro r message will appear. A Yes selection is saved as a True value and a No choice is saved as a False value in the database. A submit button on the bottom of the page allows the user to store the record and exit the system. Minimum data requirements are da te, county, location, re port type, and HSMV number, along with a choice of Yes or N o for each question. If the minimum data requirements are not met, the user cannot save the record and an error message prompts them to reconsider the offending value. Figure 10 depicts a screen s hot of the actual web page. The data collection web page is designed to be simple to use a nd self-explanatory. As part of the web design, however, help is available for offi cer users to explain individual data elements. Help buttons associ ated with each data element direct the user to a text file where context sensitive help is provided. The scope of the help file is to

PAGE 73

63 better explain the objective of the data elemen t, and provide illustra tions or other support when appropriate. A separate web page was developed for ad ministrative review of the supplemental database. Microsoft ASP.net provides a simple report that lists all records in tabular form, with rows representing unique records an d columns representing all of the database fields. The data is readily imported into any database project or Microsoft Excel for manipulation. Figure 11 is a screen shot of the ASP.Net report page. For purposes of the supplemental work z one data collection system, the existing architecture of the FHP system lends itself well to a web-based approach. Since mobile computing, communications, and server systems are currently in plac e within the agency, setting up a database for data collection is quite simple. The Florida Highway Patrol computer network provides the backbone for the supplemental data collection system. Every patr ol vehicle in the agency is equipped with a laptop computer connected to the network via a continuous cellular link. Officer laptop computers are primarily used for applications associated with computer aided dispatch and crash reporting software. Officers also us e their computers for accessing local, state, and national information systems, obtaini ng information on persons, vehicles, and articles. The agency recently migrated to an i-evidence system that allows troopers to catalog property and evidence they seize, prior to entering the items into storage rooms. The job of the FHP trooper is highly automate d and most personnel ar e very comfortable with their use.

PAGE 74

64 Figure 10. Supplemental Data Collection Web Site

PAGE 75

65 Figure 11. Work Zone Data Results Screen

PAGE 76

66 Laptop computers in each patrol vehicle co mmunicate to a central State of Florida shared resource switch via Global System for Mobile Communications (GSM) technology. From the switch, information is rout ed to various destinat ions, based on type of traffic. For the purposes of a supplemen tal work zone database, that information would pass to an FHP proxy server, where it woul d further be routed to a web server. A firewall would enhance protection as the data moves to a data base server, and ultimately to the storage system. The figure below depi cts the basic architecture of the FHP system. Figure 12. FHP Mobile Co mputing Architecture Field Implementation and Te sting of Collection System Knowing the information to collect at the s cene of a work zone traffic collision is the central purpose of this report. Creating a mechanism for actually collecting the data was a subordinate, but integral component of the effort. For the concept to be proven, however, a demonstration project is require d. The Florida Highway Patrol worked closely with the researchers on this project and volunteered to participate in a small pilot of the supplemental data co llection system. Troop G c overs nine northeast Florida counties (Nassau, Duval, Bake r, Clay, St. Johns, Bradfor d, Union, Putnam, and Flagler) and is based in Jacksonville, Florida. Sin ce several major road construction projects are underway in the troop, they were selected to test the supplemental system. The troop is large enough to obtain a reasonable usage, but far short of a stat ewide rollout of the product.

PAGE 77

67 Florida Highway Patrol Di rector Colonel Christophe r A. Knight issued a memorandum to all 147 sworn personnel as signed to Troop G. The memorandum provided information about the supplemental wo rk zone system and directed them to use the system anytime they conducted a traffic crash investigation in a work zone. In addition, the logon screen for individual o fficer laptop computers displayed a reminder message to troopers. Within the first 5 da ys of implementation, several records were recorded, indicating that troopers were activel y using the system. Rollout coincided with other laptop training given to all personnel during the first week of June 2006, so all personnel were able to obtain assistance on the use of the system if required. Because of the familiarity of troopers with both crash re porting and the use of computers, as well as the simplicity of the web-based collection sy stem, there were very few user issues. Results of Collection The early days and weeks of data collecti on reinforced the acceptance of officers to the concept of supplemental reporting using a web-based approach. The collection mechanism proves simple in design and appl ication. The collection system and pilot project continues through the end of 2006. Records from the supplemental database were successfully linked to their counterpart records in the tra ffic crash reporting system. This proves portability of the data and reinforces the objective that the work zone crash data set be improved and expanded. Validation of Elements The Model Minimum Uniform Crash Criteria, MMUCC, provides general guidelines for work zone data elements. One or more of these data elements are generally used by the states th at gather data about work z one crashes. A priori, one

PAGE 78

68 would look to these data elements as the state of the practice, agains t which potential data elements would be measured. Two of the four work zone data elements contained in the MMUCC are consistent with the results of this research effort. Essentially, through qualita tive research, these common data elements are mutually validated. Was the crash in or near a construction, ma intenance or utility work zone? This MMUCC data element is already present in th e Florida Traffic Crash Report Form. The Florida form uses none, nearby, or entere d for attributes while the MMUCC offers, yes, no, unknown. Subtle differences in at tributes can be poten tially complicating, creating more justification for binary values, yes/no, although this can also lead to more ambiguity. Because this particular data elem ent is already included in the Florida format, it will not be duplicated as part of a supplemental effort. Location of the crash is an MMUCC data element that seeks to pinpoint the location of the crash within the limits of the work zone. The attributes associated with this element are, before the first warning sign, advanced warning area, transition area, work area, termination area. These variable s do not represent descriptions that officers would readily understand, and the values, as depicted in the MMUCC, were not supported by qualitative research. This e ffort identified merge areas and lane closures as relevant locations within a work zone. The location of the crash within the work zone is simplified by determining if it o ccurred within a merge or lane closure. The value in pinpointing the location more sp ecifically was not reinforced by research. Workers present? is another MMUCC data element that directly corresponds to the findings of this report. The attributes for this variable are, yes, no, or unknown.

PAGE 79

69 Qualitative research produced a high emphasis on the presence of workers, thus the data element is valid. Type of Work is an MMUCC data element that seeks to identify the nature of the road work in some general terms. The attribut es for this data element are, lane closure, lane shift/crossover, work on shoulder or median, intermittent or moving work, and other. Similar to the MMUCC guidance on loca tion, this particular data element may be difficult for the officer to identify, given the variety and complexity of work zone projects. The utility of the data element did not screen into a supplemental collection system through qualitative research. The type of work data element will be discarded for purposes of this effort. In addition to the four data elements found in the MMUCC, other data elements were derived from the qualitative research process. These data elements should be included in any supplemental data collection e ffort. The influence of queuing or backup, presence of police in the work zone, nightti me work activity, the pr esence of advanced warning devices, the use of message/arrow boa rds, and the movement of construction equipment were determined to be additional variables of value. Comparing the state of the practice data elements of the MMUCC with the data elements produced through qualitative resear ch, we find that the latter are more encompassing and more representative of stakeholder needs. Two of four MMUCC elements are directly supported, one is supported with modification, and one is not supported as relevant. A total of fourte en data elements were produced through qualitative research, representing six new data elements that should be considered for

PAGE 80

70 inclusion in any collection methodology that seeks to fully explain the work zone incident.

PAGE 81

71 CHAPTER 5 CONCLUSIONS AND RECOMMENDATIONS Conclusions With continued demands to improve the ro adway transportation infrastructure, it is certain that work zones will continue to be a prominent part of the driving environment. Since a primary objective of roadway design, ma intenance, and improvement is safety, it is only logical that the activity undertaken in that improvement, work zones, be made as safe as possible. Understanding roadway inci dents in work zones continues to be a way in which trends can be identified and count ermeasures developed. The difficulty is translating what occurs in actual traffic cras hes into data that can be later analyzed. Stakeholders have a great deal to cont ribute when crash reporting systems are modified. This research effort has show n that qualitative research methods are a desirable way to capture what stakeholders ha ve to say. Qualitative analysis has been demonstrated to be a viable way in which we can better understand th at contribution. Qualitative research produced a total of 14 potential data elements for use in conjunction with supplementing the work zone data collection e ffort. Of those 14 elements, 8 actually screen in a new co llection methodology, given the remainder are capable of being obtained through linkages with other data sets. Advanced warning, law enforcement visibility, lane closures and me rges, backup and/or queuing traffic, variable message signs, moving construction vehicles, tr affic control device maintenance, and the presence of workers are the elements produced by analysis.

PAGE 82

72 Qualitative analysis provides a basis fo r making conclusions about stakeholder input, where mere meetings with interested parties cannot. While there are many ways to approach analysis, this effort determined which items were important by measuring agreement between groups and the number of persons within groups who discussed those items. Measuring agreement and intensity provides group-to-group validation and increases the chances that th e correct items are seen as those most significant. Qualitative results are desirable, given they are more defensible, and ostensibly more valid. The second component of enhancing the wo rk zone incident data set involves exploring alternative collection mechanisms. Chapter 4 details the considerations for a supplemental collection system. A desirabl e solution for supplementing traffic crash reporting would be cost effective, not affect current reporting systems, and ultimately be palatable for the institutions that are char ged with managing crash data. By creating a stand alone web-based collecti on system, expensive programmatic software and/or form changes are avoided, and new data can eas ily merge with existing data by observing XML schema. Since current reporting system s were not affected by this project, institutional support was not an issue. It was concluded that this was the best approach for this project. Recommendations A tradeoff exists when using law enforcement to supplement work zone incident data. They do not necessarily possess the detailed knowledge of work zone traffic control design, nor the FDOT standards inde xes that dictate their setup. They do however respond to almost all incidents re gardless of time or location, providing maximum access to scenes. While reporting by engineers potentially resolves issues of

PAGE 83

73 technical knowledge of work zones, thei r access is limited by their lack of 24-hour availability. The short-lived nature of th e temporary traffic control is precisely the variable for which their expe rtise is needed. Either me thodology has potential, but their respective shortcomings should be resolved for best results. The n eed to balance access and expertise must continue to be evaluated in future efforts. Similar to the role that both law enforcement and engineers potentially assume in incident reporting, this research illumi nated a greater need for law enforcement participation in the work zone planning pr ocess. The design and implementation of temporary traffic control in work zones can benefit from law enforcement input, to promote safety for motorists and offi cers who travel in those areas. Through this research effort, the notion of photographing work zones in conjunction with incident inve stigation received much di scussion among stakeholders. Although the topic of photography did not screen in as a pote ntial data element, it is certain that such a use of technology may bri dge the gap between access and expertise. Cost-effective digital photographic equipment and readily available storage and retrieval systems make routine photography of work zone crashes a tangible proposition. Such applications should be consid ered for future studies. Future Work Zone Applications This research has identified ways to supplem ent the work zone incident data set. While this effort originally envisioned a way to apply supplemental collection to all crashes that occur in work zones, the pilo t test conducted by the Fl orida Highway Patrol revealed that a narrower objective would be more appropriate. Rather than require supplemental work zone data collection in all work zone crashes throughout Florida, such a system may be better used in a project-sp ecific way. The data derived from a specific

PAGE 84

74 work zone location, say a major interchange project, would likely be more meaningful than an attempt to capture a dditional data on all work zones. When married with projectspecific information, the supplemental crash repo rt data would have the potential to be much more illuminating. In addition, a projec t-specific approach has the potential to provide real time data, expediting countermeasures when necessary. Other Applications Unique crash scenarios are often difficult to study because they occur infrequently, because current reporting systems do not ade quately address data collection needs, or some combination of both. Like the case of work zones, current traffic crash report forms do not contain the level of detail need ed for analysis in such cases. Crashes involving school buses or school zones, fatal crashes, crashes involving emergency vehicles or motorcycles, or those occurring on bridges are all examples of unique crashes where supplemental data collection may be helpful. A method for identifying data elements using qualitative research and colle cting data using web-based systems may be useful.

PAGE 85

75 APPENDIX A OVERVIEW OF QUALITATIVE RESEARCH METHODS Qualitative research is, Research involving detailed, verbal descriptions of characteristics, cases, and settings. Qua litative research typica lly uses observation, interviewing, and document review to collect data.[24] This form of research is rooted in the social sciences and is an excellent vehicle for examining things that may not be measurable quantitatively. Qualitative re search can be accomplished through surveys, questionnaires, personal interv iews, researcher observation, or similar methods. The research seeks to learn more about things in their natural environment through peoples attitudes, perceptions, re collections, and feelings. Getting people together for the purpose of determining direction for a project need not fall victim to problems associated with group dynamics. One way to potentially produce more reliable results is the use of qualitative research. For our purposes, enhancing the content of crash reports is best undertaken as a function similar in approach. Using qualitative re search, investigating the topi c of crash incident data represents a form of collaborat ion. Similar to currently used methods, stakeholders such as law enforcement, engineers, construction industry representatives, drivers, and safety advocates form the basis for input. But by employing a qualitative research methodology, issues related to group dynamics can be minimized and a more reliable product is possible. A more dependable c onsensus among stakeholde rs is also possible with qualitative methods, since they employ an approach that is grounded in social science.

PAGE 86

76 To set the stage for qualitative research as a methodology herein, examining several qualitative methods will be beneficial. De lphi technique, survey research, and focus groups are qualitative research methods that ma y be useful. Survey research streamlines the collection and analysis of input from la rger audiences, through a systematic method of data collection. When analyzed qualit atively, the content of multiple focus group sessions with stakeholders may lend insi ght into potential data elements. Delphi Technique The Delphi technique uses iterative ra nking methods to determine levels of importance. Like other forms of qualitative research, the Delphi technique is a way to obtain information and judgments from par ticipants to facilitate problem-solving, planning, and/or decision-mak ing [37]. The technique was developed by the RAND Corporation in the 1960s as a forecasting tec hnique. The US government subsequently improved upon the model and promoted its use as a group decision-making tool. The Delphi technique can be used in a gr oup setting, or with the proliferation of communications technologies like fax and email, conducted independent of actually assembling groups of people together. Not physically assembling participates creates logistical advantages that are often appealing to participants and researchers alike. From the perspective of group dynamics, the Delphi technique may hold a dvantages since the technique sidesteps many of the issues that accompany groups of people working together. In either case, the technique is similarly implemented. Group sizes in the Delphi technique range from se veral people to several hundred. The Delphi technique employs an iterative process that encourages people to offer their view of the relative im portance of an idea, concept, or topic. Participants are

PAGE 87

77 typically knowledgeable concer ning the area of study, much the same as the case of an expert panel or focus group. In the case of non-assembled participants, they may respond anonymously to a coordinator, who asks questions and then simply assembles the responses for redistribution for additional input. Panelists make individual estimates that are summarized and circulated among particip ants, and each can alter his or her opinion. The process is repeated until a consensus is reached. To initiate the technique, a coordinator prepares a simple open-ended question and asks participants to offer brief ideas. Th ese ideas usually take the form of words and phrases, and not fully developed concepts. From the responses, the coordinator assembles a second questionnaire, requesting pa rticipants offer commentary on all of the responses. Participants list strengths and weaknesses of those responses and resubmit them to the coordinator. The coordinator once again reassembles the responses, creating a 3rd questionnaire, asking again for input, including new ideas. This process can continue to the point that the coordinator feels that no new thought s are being introduced. After the iterative brainstorming process, the coordinator is charged with resolving the results. Resolution can manifest itself through the emergence of a consensus, at which point the process if complete or it can move to the conduct of a formal evaluation by the coordinator [37]. If a formal re solution is used, the coordinator will ask participants to utiliz e a scale to rank ideas on a continuum from zero to seven (or other number), with zero being the least weighted and the upper limit, the most effective in dealing with the issue. Respons es are tabulated to create a rank-order listing of the ideas. Another way to implement the formal re solution is the use of Nominal Group Technique for participant voting for ideas [37] In this case, the members are asked to

PAGE 88

78 identify the top five ideas and assign nu meric points to the most promising, on a descending scale to the least promising of the five. The votes are tallied by the coordinator who again creates a rank-order listing of ideas, based on the number of votes each received. Depending upon the subject studied, the proce ss of ranking that is described above may also have to be iterated, to form a stat istically suitable result. Showing panelists the results of first iteration ranking may allow fo r clarification, discu ssion, or re-evaluation by the panelists. This can be helpful when responses do not reveal clear conclusions. The final ranking process should pr oduce a consensus, and a subordinate list of next best group choices. Variations of the Delphi t echnique alter the number of pa rticipants, the number of iterations, the number of graphi c scoring points, anonymity of the participants, and the definition of statistical consensus. Impl ementation of the technique requires the researcher have a clear unders tanding of these factors before entering into the process. It should be noted that th e Delphi technique mentioned herein is a tool for qualitative research. If properly implemente d, Delphi can produce valid results, suitable for research, planning, or decision-making, pa rticularly in a monova riable study. Like any form of research, it must be used with competency, credibility and integrity. The results of the Delphi method are statisticall y arrived at through th e process of iterative ranking, however one must be reminded that th e basis of those sta tistics are only as good as the opinions of the participants[36]. Like any form of qualitative research, participant selection is critical.

PAGE 89

79 Recently the name Delphi has been attached to efforts where groups are essentially led by skilled manipulators who seek to move consensus rather than uncover it. These sometimes antagonistic coordinato rs are neither interested in research nor truth, and their misuse of the technique is unfortunate. Given there is some negative sentiment concerning this misuse of the technique, a qua lifying statement is in order to separate appropriate and inappropriate uses of th e Delphi technique. The Delphi method described herein is not relate d to some of these recent abus es that have been improperly associated with the properly used research tool. Survey Research Perhaps the most recognizable form of qualitative research is survey research. The term survey generally describes a method whereby information is gathered from a number of people who ostensibly represen t a larger popul ation. Within statistical parameters decided by the researcher, the res ponses of those who are surveyed will be an indication of how the larger population would an swer the same set of questions. Reliable research data can be achieved with survey research, without the expense of asking everyone in the population the same question. Survey research takes many forms incl uding self-administered questionnaires (mail, online), personal interviews, and telephon e interviews. It may be as innocuous as entering your zip code upon entering a web page or as involved as the complete multipage form of the US Census. Surveys form the basis for everything from consumer tastes to major public policy decisions. As a form of social/qualitative research, surveys provide a way to collect data when observation is not possible. While they often seek opinion, a well designed survey

PAGE 90

80 instrument can mine factual information from individuals as well. The design of surveys and questionnaires is rooted in social scienc e, therefore they offer validity and allow for statistical processing of respondent data. Wh ile the technical design of questions and the form of the instrument are beyond the scope of this composition, it is sufficient to say that, as in other forms of qualitative research, it must be designed with care. The premise of a survey is to sample a subs et of a larger populati on. It is essential that the selected sample be representative of the population being st udied. Surveys can quantify variables to make statistical conclusi ons about cause and effect. They can also be used as a gauge of public opinion, as is the case in the plethor a of polls conducted by academic, media, and marketing entities. Survey research generally requires adherence to rule s governing sample size and composition. If the researcher is not strictly bound by such sampling conditions, a general survey can provide some information or insight into an issue, although the results would be more an estimation than a provable fact. Potential respondents might be asked to participate in an online survey about traffic safety issues, similar to a recent undertaking of the Florida Depa rtment of Transportation. In complying with Federal requirements for states to have a Strategic Highway Safety Plan, the FDOT created an online survey for the purpose of gauging the se ntiment of stakeholders. By selecting items from a list, respondents are able to offer their opinion about traffi c safety priorities. A sample of the survey and introduct ory instructions can be viewed at www.dot.state.fl.us/safety/ [38]. Traffic safety stakeh olders would potentially makeup the population for potential respondents in an e ffort to identify crash reporting variables or content changes in reports.

PAGE 91

81 Figure 1 Florida DOT Online Safety Survey While surveys offer advantages in terms of time and money, they are not capable of probing an issue any deeper than the original question. The designer of the survey must have a clear concept of the questions, be fore they are asked, because there is no opportunity to rephrase or clarif y with the respondent. For example, a questionnaire is generally reduced to the least common denomin ator for all of the potential respondents, often neglecting specifics that may be more appropriate for some than others [36]. Unlike the open-ended questions used in the focus group and Delphi method, the survey is sometimes limited by the form of the instru ment. This structure limits the exploratory

PAGE 92

82 value of the survey, when compared to th e other qualitative methods, making them less effective at exploration. Analysis of surveys involves the use of descriptive and inferential statistics. Descriptive statistics allow the researcher to present quantitative information in a way that is more easily understood. When describi ng single variables or associations between two or more variables, descriptive methods li ke association, regressi on analysis, or other multivariate techniques are used. Inferentia l statistics are more common in the social sciences, and they make estimates of larger popul ations from samples. Test for statistical significance and estimations of relationships be tween variables is generally the basis for such methods [24]. Using surveys to explore potential crash re port data elements may be difficult to implement, given the immense volume and di versity of the topic. They may be worthwhile for focusing on specific elements and attributes associated with specific scenarios. Examining how to better capt ure data about unique crashes, like those involving work zones for example, may be suffi ciently limited to lend themselves to the use of surveys. Focus Groups Focus groups are a relatively new qualitative research technique, and they are used to obtain information from people in a group setting. Focus groups are defined as, A group of individuals selected and assemble d by researchers to discuss and comment on, from personal experience, the topic that is the subject of the research. [37] They have been widely used for marketing purposes, in an attempt to measure consumer opinion or sentiment. In the focus group, a moderator promotes discussion am ong a relatively small

PAGE 93

83 group of individuals. The group discusses th e issue(s) presented by the moderator in such a way that their opinions, attitudes, and observations are brought to the surface. Group forces and dynamics are seen as a dvantageous parts of the process with participants discussing issues with each other, rather than simply dialoging the moderator. Many researchers believe this pro cess produces richer and more detailed data than possible with other research methods [36]. The moderator of the focus group create s a comfortable environment for the participants and seeks to s timulate their thoughts and discus sion by asking a series of open-ended questions. The role of the modera tor is an important one, for he/she must introduce the topic, use probing questions when necessary, maintain order, and finally summarize the meeting. Focus group meetings usually last about one to two hours, and are generally held at locations suitable for privacy and comfort. The recommended number of participants for focus groups is somewhat subject to debate ; 6-10 [39], 5-6 [40] 6-8 [41], and up to fifteen [42]. Since some assert that a gr oup of 6-12 is appropriate if the group is homogeneous [30]. The number of separate focus group sessi ons to be conducted is another salient issue that the researcher must decide. If only one focus group is used, it runs the risk of observing the dynamics of a group and little else [43]. Conducting more than one session is desirable, particularly if distinct subgroups are present. Additional sessions serve to increase the available data, and insure th at individual group dy namics do not skew results. In some cases, multiple sessions may be conducted with the same group of people, particularly when temporal trends may be an issue.

PAGE 94

84 Focus groups are a valuable form of social research. As a qualitative form of research, they provide the input s that are necessary for identifying work zone issues. In some ways, these forums emulate the collaborat ive process that has been described in the development of crash report forms. Such si milarity affirms the appropriateness of the method for data collection. Focus groups we re selected as the preferred method of qualitative research and data collection for this effort.

PAGE 95

APPENDIX B FOCUS GROUP ADMINISTRATIVE FORMS

PAGE 96

86 Transportation Research Center Work Zone Data Collection Focus Group Meeting DATE: __________________ MODERATOR: ________________ TIME: __________________ A SSISTANT: ___________________ LOCATION:___________________ _____________________________ PARTICIPANTS LIST (please print your name) 1) ______________________________ _____________________________ 2) ______________________________ _____________________________ 3) ______________________________ _____________________________ 4) ______________________________ _____________________________ 5) ______________________________ _____________________________ 6) ______________________________ _____________________________ 7) ______________________________ _____________________________ 8) ______________________________ _____________________________ 9) ______________________________ _____________________________ 10)__________________ ____________________ _____________________ 11)__________________ ____________________ _____________________

PAGE 97

87 Transportation Research Center Data Collection Requirements for Work Zone Incidents Focus Group Interview Roadway construction has become a common fixture in our daily travels. These locations can present unique challenges to tra ffic safety interests, as well as the motoring public. Like many aspects of traffic safe ty, a better understanding of the contributing factors in crashes can potentially lead to new insight and improved countermeasures. The University of Florida Transportation Research Center, under a grant from the Southeastern Transportation Center (STC), is conducting group interviews called focus groups with FDOT engineers, private contra ctors, law enforcement, and drivers to obtain a better understanding work zones. By examining the unique perspective and expertise of each group, we hope to deve lop a better understandi ng of work zone dynamics and develop the tools for more effec tive analysis of cras hes in these areas. Objectives of this focus group exercise: 1. Identify factors that might contribut e to incidents in work zones. 2. Determine attributes associated with those factors. Format of this focus group session: 1. The background, objectives, and benefits of this focus group interview will be explained by the moderator. 2. The moderator will describe the format of the focus group session, and the points to keep in mind. 3. An open-ended question will be presented by the moderator 4. For the given question, participants discu ss the issue and provide their perceptions or opinions relevant to the issue. Additional questions may also be asked by the moderator to fully explore the issue. 5. A total of approximately 4 to 6 questions will be presented. Points to keep in mind: When possible, discuss the issue in non-t echnical terms. Minimize the use of acronyms and jargon that may be common w ithin your field of expertise. The terms incident, accident, crash, and coll ision will all be considered equivalent in our discussion of work zones. For efficient use of the time period allotted for the interview, the moderator may need to interrupt and/or redirect the di scussion, to insure th at all questions are covered.

PAGE 98

88 Informed Consent Form fo r Focus Group Interview Protocol Title: Data Collection Requirements for Work Zone Incidents Purpose of the research study: The purpose of this study is to identify data elemen ts that can potentially be collected at the scene of work zone crashes, to better understand the factors surrounding work zone incidents. What you will be asked to do in this study: After everyone in the room introduces themselves, a moderator will lead a group discussion about highway work zones, and incidents that occur in those locations. You will be asked to offer opinions about construction zones, and their impact on the driving environment. Time Required: Up to 2 hours Risks and Benefits: There are no risks involved in this study. While we do not anticipate that you will benefit directly from participating in this study, it may lead to improved data collection at the scene of work zone incidents. This additional data may contribute to be tter statistical analysis of work zone incidents, and potentially a better understanding of the factors surrounding these incidents. Compensation: There is no monetary compensation fo r participation in this study. Confidentiality: We will record the names of those participating in this study, however, your name will not be associated with your individual comments in the focus group interview. Your name will not be used in any published reports and your identity will be kept confidential to the extent provided by law. Voluntary participation: Your participation in this study is completely vol untary. There is no penalty for not participating. Right to withdraw from the study: You have the right to withdraw from the study at any time without consequence. Whom to contact if you have questions about the study: Scott S. Washburn, Ph.D, P.E. Civil and Coastal Engineering, 365 Weil Hall, P.O. Box 116580, Phone: (352) 392-9537 x1453. Whom to contact about your rights as a research participant in this study: University of Florida Institutional Review Board (UFIRB). UFIRB Office, P.O. Box 112250, Univers ity of Florida, Gainesville, FL 32611-2250,

PAGE 99

89 Phone: (352) 392-0433. Agreement: I have read the procedure described above. I volunta rily agree to take part in the procedure and I have received a copy of this description. Participant: __________________________________________ Date ____________ Interviewer:__________________________________________ Date ____________

PAGE 100

APPENDIX C FOCUS GROUP MODERATOR GUIDE

PAGE 101

91 Transportation Research Center Focus Group Moderators Guide (Data Collection Requirements for Work Zone Incidents) This document is designed to assist the moderato r in the conduct of focus groups relating to the titled research. The guide covers both administra tive and subject matter issues, and is presented in a chronological form for ease of use. Suggested time notations are made in the interest of maximizing the time available with the participants. 1) Sign-in Form and Informed Consent Form Distribution (Duration = 5 minutes) Each participant is asked to provide his/her name on the Sign-in Form and fill out the Informed Consent Form before taking part in the focus group meeting. The Informed Consent Form is required by the University of Florida Institutional Review Board (UFIRB) to ensure that the participants were made of aware of the risks and benefits of participating in this study by the researchers, and that they vol untarily agreed to participate in it 2) Welcome and Introductions (Duration = 5 minutes) The moderator introduces themselves and any assist ant(s) that may be present. Appreciation is offered to participants for their time and participation. Each participant is asked to introduce themselves. 3) Overview of Study Background, Objectives, and Benefits (Duration = 5 minutes) The background, objectives, and potential benefits of this study will be briefly described by the moderator. A separate hand-out can be pr ovided to participants that highlights these comments. 4) Explanation of Format and Scope of the Focus Group Session (Duration = 5 minutes) The moderator will briefly explain that the session is designed to solicit the ideas and opinions of the participants. The format of the forum will be reviewed, and the moderator will establish some simple ground rules. The moderator will inform participants that the session is being recorded for later analysis, but also assure them that their comments will be kept confidential.

PAGE 102

92 5) Focus Group Questions During the course of each focus group session, several open-ended or subject-specific questions will be presented by the moderator to the participants. The participants will then discuss each topic amongst themselves and with the moderator. Each question should be written on a white board, or presented on an electronic slide, or the like, for all participants to easily see. The ( tentative ) questions are listed below in chronological order with the approximate time assigned for discussion of each subject within a two-hour focus group meeting. Factors and issues that the researc hers thought are worth discussing are presented under each question for the moderator to inquire about the importance of the factors, in case they are not brought up by participants. Additi onal questions may also be asked about why the discussed factors are important or the par ticipants experience related to the factors. 1. Why do work zone crashes occur? (expected time = 30 minutes) Physical features of the roadway Issues with MOT and traffic control devices Driver behaviors Vehicle / Worker characteristics (trucks, exposed workers, etc) 2. What are positive things that are being done to help the situation? (expected time = 20 minutes) Advance Warning Speed Limits Better MOT Separation Barricades, barre ls, walls, and other devices Enforcement Public Information and Education 3. What are things that still need to change? (expe cted time = 20 minutes) Driver behaviors Physical design of work zone Traffic Control Enforcement Issues 4. What can we learn from Incidents that occur in work zones? (expected time =

PAGE 103

93 20 minutes) Role of congestion Secondary collisions Location of incidents Type of collision (rear end, sideswipe, run off the roa d, with barricades/equipment, etc) 6) Concluding remarks and wrap up (Duration = 5 minutes) The moderator brings about a conclusion to the discussion of the subject and thanks the participants for their contribution. Volunteers are sought for a follow-up group meeting that will be conducted with a com posite group at a later date. Strategies for a moderator to solicit input from shy or quiet participants The moderator should maximize eye contact with shy or quiet participants to encourage them to speak in the discussion. The moderator can also call on a quiet participant by name. Jordan, you havent had a chance to say anything about this. What do you think? or Jordan, I dont want to miss what you have to say. Would you like to add something? Strategies for a moderator to int errupt remarks when it is necessary A moderator is usually not recommended to interr upt participants remarks during the meeting. There may be cases where the moderator needs inte rvene (i.e., to redirect discussion when it gets off the right track, to give others the opportuni ty to speak in the discussion, etc.). When it happens, the moderator may find a chance to interrupt with some comments observing the breathing pattern of the speaker. It should be remembered that a person can not speak while inhaling. If a participant says too much, the modera tor might say Thank you, Jordan, thats been helpful. Lets hear from others in the group. E xcuse me, Jordan, but what we want to focus on now is may be said when a participant is getting off the track. A big board showing each question will be prepared to get the participan ts on the right track as well. Asking favor to participants about potential need of interruptio n and clear explanation of the scope of the discussion will be helpful.

PAGE 104

94 LIST OF REFERENCES 1. Work Zone Safety Facts and Statistics. US Department of Transportation, Federal Highway Administration. http://safety.fhwa.dot .gov/wz/wz_facts.htm Accessed August 3, 2005. 2. Work Zone Safety: Its Everyones Job. Florida Department of Transportation. http://www.itseveryonesjob.com/ Accessed April 5, 2005. 3. Highway Statistics 2001 Publication No. FHWA-PL-02-020. Federal Highway Administration, US Departme nt of Transportation, 2001. 4. State of Florida. Florida Statutes, Ch apter 316.066, Written Reports of Crashes. State of Florida, Tallahassee, Florida. 2005. 5. Bellis, Mary. Inventors: The Du ryea Brothers. About.com. http://inventors.about.com/library/inventors/blDuryea.htm Accessed July 14, 2005. 6. NHTSA (National Highway Traffic Safe ty Administration). Guideline for Minimum Uniform Crash Criteria (MMUCC), 2nd Edition. NHTSA, Washington, D.C. 2003. 7. Ullman, Gerald L. and Scriba, Tracy A. Revising the Influence of Crash Report Forms on Work Zone Crash Data. In Transportation Research Record: Journal of the Transportation Research Board No. 1897, TRB, National Research Council, Washington, D.C., 2004, pp. 180-182. 8. Instructions for Completing the Florida Uniform Traffic Crash Forms Florida Department of Highway Safety and Motor Vehicles, Tallahassee, Florida. January 2002. 9. Wang, Jun, Warrren E. Huges, Forrest M. Council, and Jeffery F. Paniati. Investigation of Highway Work Zone Crashes: What We Know and What We Dont Know. In Transportation Research Record: Journal of the Transportation Research Board No. 1529, TRB, National Research Council. Washington, D.C., 1996, pp. 54-62. 10. Khattak, A. and Targa, F. Injury Severity and Total Harm in Truck-Involved Work Zone Crashes. Presented at 83rd Annual Meeting of the Tran sportation Research Board, Washington, D.C., 2004.

PAGE 105

95 11. Ha, T-J, and Nemeth, Z.A. Detailed Study of Accident Experience in Construction and Maintenance Zones. In Transportation Research Record: Journal of the Transportation Research Board No. 1509, TRB, National Research Council, Washington, D.C., 1995, pp. 38-45. 12. Garber, N.J. and Zhao, Ming Zhao. Distribu tion and characteristics of crashes at different work zone locations in Virginia. In Transportation Research Record: Journal of the Transportation Research Board No. 1794, TRB, National Research Council, Washington, DC., 2002, pp. 19-25. 13. Raub, Richard A., Sawaya, Omar B., Schofer, Joseph L., Ziliaskopoulos, Athanasios. Enhanced Crash Reporting to Explore Workzone Crash Patterns Northwester University Center fo r Public Safety, Evanston, IL, 2001. 14. Schrock, Steven D., Ullman, Gerald L.; Cothron, Scott A., Draus, Edgar, Voigt, Anthony An Analysis of Fatal Work Zone Crashes in Texas Publication FHWA/TX-05/0-0428-1, Texas Transportation Institute, 2004. 15. National Work Zone Safety Information Clearinghouse: Work Zone Safety Research Database. Texas Transportation Institute. http://wzsafety.tamu.edu/searches/practices.stm Accessed July 21, 2005. 16. Spanhour, Lisa and Mtenga, Primus. Analysis of Work Zone MOT Data Collection and Usage Procedures FM No. 4066331B201. Florida Department of Transportation, 2002. 17. Migletz, J. and Graham, J.L. Coll ection of Work Zone Accident Data (abridgement). In Transportation Research Record: Journal of the Transportation Research Board No. 933, TRB, National Research Council, Washington, D.C., 1983, pp. 15-18. 18. Theilman, Carol Y. Expert Systems for Crash Data Collection Publication No. FHWA-RD-99-052. FHWA, US Depa rtment of Transportation, 1998. 19. Keenan, Carol. Work Zones that Work. Public Roads Nov-Dec 2004, pp. 22-29. 20. Scriba, Tracy. Intelligent Transportation Systems in Work Zones A Cross Cutting Study. http://www.benefitcost.its.dot.gov /ITS/benecost.nsf/ID/A93D71768BA029E28525 6E4E006A6864 Accessed July 29, 2005. 21. Scott, Elizabeth A.and Black, Nick. When Does Consensus Exist in Expert Panels? Journal of Public Health Vol. 13, 2003, pp. 35-39. 22. Durkheim, Emile. The Rules of Sociological Method The Free Press, New York, 1982. 23. Harvey, Jerry B. The Abilene Paradox Lexington Books, Landham, MD, 1996.

PAGE 106

96 24. Bureau of Justice Assistance, Center for Program Evaluation Glossary. US Dept. of Justice. http://www.ojp.usdoj.gov/BJA/evalua tion/glossary/glossary_q.htm Accessed September 15, 2005. 25. Vaughn, S., Schumm, J.S., & Sinagub, J. Focus Group Interviews in Education and Psychology Sage, Thousand Oaks, 1996 26. Meeting Commander Home Page. Digital Dictation & Transcription Service. http://www.meetingcommander.com/ Accessed June 23, 2006. 27. Berelson, B. Content Analysis in Communication Research Free Press, Glencoe, IL, 1952. 28. Weber, R.P. Basic Content Analysis, 2nd ed Sage, Newbury Park, CA, 1990. 29. Analyzing, Interpreting, and Reporting Fo cus Group Data For Publication. UT Austin. University of Texas at Austi n. Division of Instru ctional Innovation and Assessment, Austin. http://www.utexas.edu/academic/diia/assessment/iar/how_to/interpreting_data/focu s_groups/publication.php Accessed September 15, 2005. 30. Krueger, R.A. Moderating Focus Groups: Focus Group Kit 4 Sage, Thousand Oaks, CA, 1998. 31. Huber, Gunter L. How To Conduct a Qualitative Analysis. Aquad Six. www.aquad.de/eng/m-chap05.pdf September 18, 2005. 32. Cattarall, M. and Maclaran, P. Focus Group Data and Qualitative Analysis Programs: Coding the Moving Picture as Well as the Snapshots. Sociological Research Online Vol. 2, no. 1, http://www.socresonline.org.uk/2/1/6.html Accessed September 18, 2005. 33. Microsoft SQL 2005 Product Informati on Page. Microsoft Corporation. http://www.microsoft.com/sql/prodinfo/overview/default.mspx Accessed June 23, 2006. 34. Microsoft ASP.Net Product Information Page. Microsoft Corporation. http://msdn.microsoft.com/asp.net/ Accessed June 23, 2006. 35. Morgan, D.L. Focus Groups as Qualitative Research Sage, London, 1988. 36. Babbie, Earl. The Practice of Social Research, 10th Edition ThomasonWadsworth, 2004. 37. Dunham, Randall, B. The Delphi Technique University of Wisconsin School of Business. http://instruction.bus.wisc.e du/obdemo/readings/delphi.htm Accessed April 5, 2006.

PAGE 107

97 38. Florida Online Survey: Transportation Safety. Florida Department of Transportation.. http://survey01.camsys.com/survey/wsb.dll/webmaster/fdottranssafety3.htm Accessed March 21, 2006. 39. Powell, R.A. and Single, H.M. Focus Groups, International Journal of Quality in Health Care Vol., 8 1996, pp. 499-504. 40. Asbury, J.E. Overview of Focus Group Research, Qualitative Health Research Vol. 5, no. 4, 1996, pp. 414-420. 41. MacIntosh, J. Focus groups in distance nursing education, Journal of Advanced Nursing Vol. 18, 1981. 42. Barbour, R.S., & Kitzinger, J. Developing Focus Group Research: Politics, Theory and Practice Sage, London, 1999. 43. Goss, J.D., Leinbach, T.R. Focus Groups as Alternative Research Practice, Area 28 Vol. 2, 1996, pp. 115-23.

PAGE 108

98 BIOGRAPHICAL SKETCH Grady Carrick is a candidate for the Master of Science Degree in civil engineering, transportation. He currently holds both a b achelors and masters degree in criminal justice, as well as a Master of Public Ad ministration (MPA) degree all obtained from Florida International University (FIU) in Miam i, Florida. As a student who is also a working professional, he has been employe d with the Florida Highway Patrol for 24 years, and currently holds an upper-level ma nagement position with the agency. He is among the most experienced managers in the agency, holding a command position for the past 13 years. He currently manages a ni ne county troop which covers northeast Florida, and he is responsi ble for oversight of approximately 200 sworn and civilian personnel. He is a graduate of the pres tigious FBI National Academy and the Florida Criminal Justice Executive Institute.