|UFDC Home||myUFDC Home | Help|
This item has the following downloads:
IDENTIFICATION OF PREFERRED PERFORMANCE MEASURES FOR THE
ASSESSMENT OF TRUCK LEVEL OF SERVICE
A DISSERTATION PRESENTED TO THE GRADUATE SCHOOL
OF THE UNIVERSITY OF FLORIDA IN PARTIAL FULFILLMENT
OF THE REQUIREMENTS FOR THE DEGREE OF
DOCTOR OF PHILOSOPHY
UNIVERSITY OF FLORIDA
O 2007 Byungkon Ko
To the graduate students of the University of Florida
It may not have been possible for me to complete my Ph.D. study without the advice or
assistance of those who have supported me throughout my life in Gainesville. They are my
family members, friends, and University of Florida faculties and administrative staffs. Although
I cannot adequately describe how blessed I am to have been with them, I would like to show my
gratitude to them with this opportunity.
I am especially grateful to my parents, wife, and four siblings. Their consistent care and
support have been a true driving factor for me to succeed in preparing for my future as well as
earning the degree.
I would like to express my appreciation to Dr. Scott S. Washburn who has served as my
academic advisor for my Ph.D. study. He supported me financially and his exceptional
knowledge and wisdom has provoked academic curiosity that encouraged me to enhance my
research ability onto a higher level.
I also would like to thank my Ph.D. supervisory committee members for their time and
effort to provide me with invaluable advice for my research. Dr. Lily Elefteriadou has guided
me into the right direction academically and also provided many graduate students with
convenient study environment as the director of Transportation Research Center. Dr. Yafeng
Yin lent me his acute insight on mathematical expressions and recommended better ways to
organize the dissertation. Dr. Siva Srinivasan helped me to improve statistical data analysis and
modeling part of my work. Dr. Joseph Geunes in the Industrial Engineering department served
as an external committee member. His consistent concerns and advice about my work has
become a motivation to improve overall performance of my study.
My friends in Korea or Gainesville have always cheered me up for my successful career at
the University of Florida. I will never forget the time and memory I have shared with them.
I certainly appreciate the assistance of University of Florida faculties and staffs. They
have always tried to answer my academic or administrative questions in spite of their tight time
TABLE OF CONTENTS
ACKNOWLEDGMENTS .............. ...............4.....
LIST OF TABLES ............ ...... .__ ...............10...
LIST OF FIGURES .............. ...............13....
LIST OF TERMS ............_ ..... ..__ ...............14...
AB S TRAC T ............._. .......... ..............._ 17...
1 INTRODUCTION ................. ...............20.......... ......
1.1 Background ........._..... ...._... ...............20....
1.2 Problem Statement ........._..... ...._... ...............21....
1.3 Study Obj ectives and Tasks ........._..... ...._... ...............23..
1.4 Organization of the Dissertation ........._..... ...._... ...............25...
2 LITERATURE REVIEW ........._.._.. ...._... ...............26....
2.1 Truck Level of Service .............. ...............26....
2.2 User-Perceived Level of Service ............... ...............30....
2.3 Trucking Community Focus Groups and Surveys............... ..... .............3
2.4 Florida Trucking Community Focus Group Participant Recruitment Sources ........._.....52
2.5 Trucking Community Survey Methods .............. ...............54....
3 RESEARCH APPROACH .........._.... ...............59._.._. ......
3.1 Focus Group Sessions ........._...... ...............61._.__. ....
3.1.1 Participant Recruitment ................. ...............62........... ..
3.1.2 Participant Selection ................. ...............63...............
3.1.3 Focus Group Questionnaires .............. ...............66....
3.1.4 Conducting Interviews. ................ ................ .................. ....__..68
3.1.5 Summary of Discussions .......................... .................. ................69
3.2 Survey Studies .............. ...............70....
3.2. 1 Question Types ............... ... ...............71.......... ...
220.127.116.11 Interval-rating questions............... ...............7
3.2. 1.2 Ratio-scale questions ................. ...............72...............
18.104.22.168 Forced-ranking questions .............. ...............73....
3.2. 1.4 Discrete-choice questions ................. ...............74...............
22.214.171.124 Fixed-sum questions............... ...............7
3.2.2 Survey Development .............. ...............76....
3.2.3 Data Collection............... ...............8
3.2.4 Data Reduction ................ ...............88................
3.2.5 Data Analysis................. ...............8
126.96.36.199 Descriptive statistics............... ...............8
188.8.131.52 Exploratory factor analysis............... ...............90
184.108.40.206 Multiple comparison test ................. ...............97..............
220.127.116.11 Non-parametric test ................. ...............99........... ...
18.104.22.168 Chi-squared test............... ...............101.
3.3 Truck LOS Measurement .............. ...............103....
3.3.1 Truck LOS Service Measures............... ... .............10
22.214.171.124 Single performance measure approach .............. ..... ............... 10
126.96.36.199 Multiple variable approach............... ...............10
3.3.2 Truck LOS Estimation Model .............. ...............105....
188.8.131.52 Data collection............... ..............10
184.108.40.206 Statistical modeling ................ ......... ....___ ...........10
4 FOCUS GROUP STUDY RESULTS ........._._. ......... ...............115.
4. 1 Perceptions of Truck Drivers ........._..... ........._.__ ...............115..
4. 1.1 Truck Route and Departure Time Selection ................. ............................116
4. 1.2 Factors Affecting Truck Trip Quality............... .................118
4. 1.2. 1 Factors affecting truck trip quality on freeways............_. .. ........._._ ...118
4. 1.2.2 Factors affecting truck trip quality on urban arterials .............. .... ........._..122
4. 1.2.3 Factors affecting truck trip quality on two-lane highways......................... 126
4. 1.2.4 Factors affecting truck trip quality on hub facilities .............. ................. 128
4. 1.3 Improvement Priority of Transportation Services ................. .......................128
4. 1.4 Truck Delivery Schedule Reliability ................. ...............129........... ..
4.2 Perceptions of Truck Company Managers .............. .....................130
4.2. 1 Truck Route and Departure Time Selection ................. ............................131
4.2.2 Factors Affecting Truck Trip Quality............... ...............133
220.127.116.11 Factors affecting truck trip quality on freeways............... .................3
18.104.22.168 Factors affecting truck trip quality on urban arterials .............. .... ........._..136
22.214.171.124 Factors affecting truck trip quality on two-lane highways.........................140
126.96.36.199 Factors affecting truck trip quality on hub facilities ................. ...............141
4.2.3 Truck Delivery Schedule Reliability .............. ... .. .......... ....... ....... ........14
4.3 Perceptional Difference between Truck Drivers and Truck Company Managers..........1 43
5 SURVEY DATA ANALYSIS RESULTS .............. ...............145....
5.1 Backgrounds of the Participants ................ ...............145....._... ..
5.1.1 Truck Driver Participants ................. .......... ...............145...
5.1.2 Truck Company Manager Participants ................. ........... ............... ....14
5.2 Perceptions on the Relative Importance of Each Factor on Freeways .................. .........147
5.2. 1 Relative Importance of Each Factor ............... ............ ..............__ ....14
5.2.2 Applicability of Single Hypothetical Performance Measure to Estimate Truck
Trip Q uality............... ..... .. .... .... ... ......... 5
5.3 Perceptions on the Relative Importance of Each Factor on Urban Arterials. ........._.......155
5.3.1 Relative Importance of Each Factor ........._._.._......_.. ...............155.
5.3.2 Applicability of Single Hypothetical Performance Measure to Estimate Truck
Trip Q uality............... ..... .. .... .. ..... ......... 5
5.4 Perceptions on the Relative Importance of Each Factor on Two-Lane Highways......... 162
5.4. 1 Relative Importance of Each Factor ................. .... ... ... ...... ... .............. .... 16
5.4.2 Applicability of Single Hypothetical Performance Measure to Estimate Truck
T rip Q uality................ .. ....... .. ......... .. .. ......... 6
5.5 Relative Importance of Each Category of Factors to Quality of a Truck Trip ...............168
5.5.1 Relative Importance of Each Category of Factors for Freeways ................... ......169
5.5.2 Relative Importance of Each Category of Factors for Urban Arterials ................169
5.5.3 Relative Importance of Each Category of Factors for Two-lane Highways ........170
5.6 Comparisons of the Importance of Each Factor Category on Various Roadway
Facilities ........... __...... .. ..... ... .._ ... .. ...........17
5.7 Perceptions on the Improvement Priority of Various Roadway Types ........................171
5.8 Relationships between Truck Drivers' Backgrounds and Their Perceptions on the
Applicability of Each Hypothetical LOS Performance Measure .............. .. .................172
5.8.1 Truck Drivers' Backgrounds that Explain Their Perceptions on the
Importance of Each Hypothetical LOS Performance Measure on Freeways ............173
5.8.2 Truck Drivers' Backgrounds that Explain Their Perceptions on the
Importance of Each Hypothetical LOS Performance Measure on Urban Arterials...175
5.8.3 Truck Drivers' Backgrounds that Explain Their Perceptions on the
Importance of Each Hypothetical LOS Performance Measure on Two-Lane
H ighw ays ................ .... .. .. .. ................17
5.9 Perceptions on Truck Driving Environment by Time of Day ................. ........._......180
5.9. 1 Current and Preferred Truck Driving Time of Day ................. .... ....._._.........181
5.9.2 Relationships between Truck Drivers' Backgrounds and Their Perceptions on
Preferred Truck Driving Time of Day ................. ........ ....._. ...........8
5.10 Other Transportation Service Issues for Truck Drivers ........__. ........ _.._.............186
5.10.1 Freeway Truck Operations .............. ...............186....
5. 10.2 Urb an Arteri al Truck Operati ons ....._.__._ ........___ ...............187
5.10.3 Two-Lane Highway Truck Operations............... ..............18
6 CONCLUSIONS AND RECOMMENDATIONS .............. ...............239....
6.1 Conclusions.................... .................23
6. 1.1 Quality of a Truck Trip on Freeways .............. ...............240....
6. 1.2 Quality of a Truck Trip on Arterials............... ...............24
6. 1.3 Quality of a Truck Trip on Two-Lane Highways............... .. ................ 24
6. 1.4 Improvement Priority of Various Transportation Facilities for Trucks ...............244
6. 1.5 Improvement Priority of the Factors on Each Transportation Facility for
Trucks .............. ... .. ......... .. .............24
6. 1.6 Preference on Truck Driving Time of Day ................. .. ........... ..... ............. ..245
6. 1.7 Relationships between Truck Drivers' Backgrounds and Their Perceptions on
Truck Trip Quality ................. ... .. ....... .. ..........4
6. 1.8 Overall Effectiveness of Research Approach ....._.__._ .........__ ................ 247
6.2 Recommendations............... ............. ..........4
6.2.1 Truck LOS Estimation Model Development............... ..............24
188.8.131.52 Truck LOS on freeways .............. ...............248....
6.2. 1.2 Truck LOS on arterials ........._._. ...._._ ...............251.
184.108.40.206 Truck LOS on two-lane highways ................ ......... ... .........5
6.2.2 Transportation Service Improvement for the Trucking Community ........._._.......252
6.2.3 Trucking Community Surveys .............. ...............253....
A COOPERATION REQUEST LETTER SENT TO FTA .............. ..... ............... 25
B FOCUS GROUP INSTRUCTION .............. ...............259....
C GUIDELINES FOR FOCUS GROUP PARTICIPANT SELECTION SENT TO FTA......261
D FOCUS GROUP MODERATOR' S GUIDE ...._. ......_._._ .......__. ...........6
E FOCUS GROUP PARTICIPANTS' BACKGROUND SURVEY RESULTS .........._.......268
F TRUCK DRIVER FOCUS GROUP BACKGROUND SURVEY FORM.............._._........272
G TRUCK COMPANY MANAGER FOCUS GROUP BACKGROUND SURVEY
FORM ........._._ ...... .. ...............275...
H TRUCK DRIVER SURVEY FORM............... ...............278.
I TRUCK COMPANY MANAGER SURVEY FORM .............. ...............285....
J IMPROVEMENT PRIORITY SCORE ...._. ......_._._ ......._. .............9
K PO STAGE-PAID TRUCK DRIVER SURVEY FORM ........._.._.. .....___ ..............294
L PO STAGE-PAID TRUCK COMPANY MANAGER SURVEY FORM................_..__........298
M POSTAGE-PAID MANAGER SURVEY COVER LETTER ....._____ ....... ......_........3 02
N SURVEY DATA FILTERING CRITERIA ................ ...............304........... ...
LIST OF REFERENCES ................. ...............307................
BIOGRAPHICAL SKETCH ................. ...............3.. 11.............
LIST OF TABLES
2-1 Comparison of Survey Methods .............. ...............58....
3-1 Survey Development............... ..............10
3-2 Postage-Paid Truck Driver Survey Response Rates ................. .......... ................1 09
3-3 Survey Participation of the Selected Carriers by Each Conference and Chapter ............ 110
3-4 Survey Collection by Each Conference and Chapter ................. ......... ................11 1
3-5 FTDC Truck Driver Survey Data Usability ................. ...............111........... ..
3-6 Postage-Paid Truck Driver Survey Data Usability ................. .............................111
3-7 Overall Truck Driver Survey Data Usability ................. ...............112........... ..
3-8 Overall Truck Company Manager Survey Data Usability ................. ............ .........112
3-9 Current HCM Service Measures used for LOS Determination .........._.... ........._......114
5-1 Truck Driver Survey Respondent Background Summary Statistics ............... .... ........._..188
5-2 Additional Truck Driver Survey Respondent Background Summary Statistics ..............189
5-3 Truck Company Manager Survey Respondent Background Summary Statistics ...........190
5-4 Additional Truck Company Manager Survey Respondent Background Summary
Statistics ................ ...............191................
5-5 Truck Drivers' Perceptions on Each Factor on Truck Travel Quality of Service on
Freew ay s .............. ...............192....
5-6 Managers' Perceptions on Relative Importance of Each Factor on Freeways ................193
5-7 Exploratory Factor Analysis Results .............. ...............194....
5-8 Importance of Each Factor on Truck Travel Quality of Service on Freeways ................195
5-9 Truck Drivers' Perceptions on Applicability of Single Performance Measure to
Determine Truck Travel Quality of Service on Freeways ....._____ ..... ....__ ..........196
5-10 Truck Company Managers' Perceptions of Relative Importance of Each Truck
Driving Condition on Freeways for trucking business .............. ...............196....
5-11 Games-Howell Post Hoc Test Results ...._.._.._ .... .._._. ...._.._ ...........9
5-12 Truck Drivers' Perceptions of Each Factor on Truck Travel Quality of Service on
Urban Arterial s............... ...............19
5-13 Managers' Perceptions on Relative Importance of Each Factor on Urban Arterials.......199
5-14 Exploratory Factor Analysis Results .............. ...............200....
5-15 Importance of Each Factor on Truck Travel Quality of Service on Urban Arterials ......201
5-16 Truck Drivers' Perceptions of Applicability of Single Performance Measure to
Determine Truck Travel Quality of Service on Urban Arterials .............. ...................202
5-17 Truck Company Managers' Perceptions of Relative Importance of Each Truck
Driving Condition on Urban Arterials for trucking business .............. ....................20
5-18 Games-Howell Post Hoc Test Results ...._.._.._ .... .._._. ...._.._ ...........0
5-19 Truck Drivers' Perceptions of Each Factor on Truck Travel Quality of Service on
Two-Lane Highways............... ...............20
5-20 Managers' Perceptions of Relative Importance of Each Factor on Two-Lane
Highways .............. ...............205....
5-21 Exploratory Factor Analysis Results .............. ...............206....
5-22 Importance of Each Factor on Truck Travel Quality of Service on Two-Lane
Highways .............. ...............207....
5-23 Truck Drivers' Perceptions of Applicability of Single Performance Measure to
Determine Truck Travel Quality of Service on Two-Lane Highways ............................208
5-24 Truck Company Managers' Perceptions of Relative Importance of Each Truck
Driving Condition on Two-Lane Highways for trucking business............... ................20
5-25 Games-Howell Post Hoc Test Results ...._.._.._ .... .._._. ...._.._ ...........0
5-26 Definitions of Independent Variables used in Statistical Tests ........._.._... ........_.._... ...217
5-27 Definitions of Independent Variables used in Statistical Tests ........._.._... ........_.._... ...218
5-28 Kruskal-Walli s and Mann-Whitney Test Stati stics ................. ................. ..........21 9
5-29 Kruskal-Walli s and Mann-Whitney Test Stati stics ................. ................. ..........220
5-30 Kruskal-Walli s and Mann-Whitney Test Stati stics ................. ................. ..........22 1
5-3 1 Kruskal-Walli s and Mann-Whitney Test Stati stics ................. ................. ..........222
5-32 Kruskal-Walli s and Mann-Whitney Test Stati stics ................. ................. ..........223
5-33 Kruskal-Walli s and Mann-Whitney Test Stati stics ................. ................. ..........224
5-34 Mann-Whitney Test Statistics............... ..............22
5-35 Kruskal-Walli s and Mann-Whitney Test Stati stics ................. ................. ..........226
5-36 Kruskal-Walli s and Mann-Whitney Test Stati stics ................. ................. ..........227
5-37 Mann-Whitney Test Statistics............... ..............22
5-38 Chi-Squared Test Statistics 1 .............. ...............232....
5-39 Chi-Squared Test Statistics 2 ................ ...............233........... ..
5-40 Chi-Squared Test Statistics 3 .............. ...............233....
5-41 Chi-Squared Test Statistics 4 ................ ...............234........... ..
5-42 Other Factors Affecting Truck Trip Quality on Freeways............... ...............23
5-43 Other Drivers' Behavior Affecting Truck Trip Quality on Freeways .............................236
5-44 Other Factors Affecting Truck Trip Quality on Urban Arterials............... ................3
5-45 Other Factors Affecting Truck Trip Quality on Two-Lane Highways ................... .........238
6-1 Truck Trip Quality of Service Determinants on Freeways ................. ......................254
6-2 Truck Trip Quality of Service Determinants on Arterials .............. ....................25
6-3 Truck Trip Quality of Service Determinants on Two-Lane Highways ...........................255
6-4 Top Six Factors in the Need for the Improvement on Each Roadway Type for Trucks .256
E-1 1st Truck Driver Focus Group Participants' Background Survey Results .......................268
E-2 1st Truck Company Manager Focus Group Participants' Background Survey Results...269
E-3 2nd Truck Driver Focus Group Participants' Background Survey Results ......................270
J-1 Improvement Priority Scores ................. ...............292...___ ....
LIST OF FIGURES
3-1 Research Approach ................ ...............108................
3-2 Truck Driving Competition Course ................. ......... ...............109 ....
3-3 Survey Table Setup at Truck Driving Competition ................ ............................110
3-4 Example of a Path Diagram for an EFA Model by Principle Component Extraction
M ethod ........... ..... .._ ...............113..
3-5 Normal Q-Q Plot of the Importance of 'Ease of Obtaining Useful Traveler
Information' on Freeways ........... ..... .._ ...............113..
5-1 Relative Importance of Each Factor Category for Freeways.............__ ..........___.....210
5-2 Relative Importance of Each Factor Category for Urban Arterials .............. .... ........._...21 1
5-3 Relative Importance of Each Factor Category for Two-lane Highways. ................... ......212
5-4 Relative Importance of Roadway Conditions on Different Roadway Types ................ ..213
5-5 Relative Importance of Traffic Conditions on Different Roadway Types ......................214
5-6 Relative Importance of Other Drivers' Behavior on Different Roadway Types .............215
5-7 Improvement Priority of Various Roadway Facilities for Truck Trip Quality ........._.....216
5-8 Truck Drivers' Current and Preferred Truck Driving Times of Day ............... .... ...........229
5-9 Truck Driving Time of Day Preference of Current Users ................. ............ .........230
5-10 Truck Company Managers' Preference on Truck Driving Times of Day ................... ....23 1
LIST OF TERMS
American Association of State Highway Officials (now AASHTO)
American Association of State Highway and Transportation Officials
Analysis of Variance
American Trucking Association
Bureau of Transportation Statistics
Computer-Aided Telephone Interview
Central Business District
Citizen Band Radio
Commercial Driver' sLicense
Center for Economic Development and Research
Changeable Message Sign
Commercial Motor Vehicle
Exploratory Factor Analysis
Quarterly Census of Employment and Wages
Florida Department of Highway Safety & Motor Vehicles
Florida Department of Transportation
Federal Highway Administration
Florida Intrastate Highway System
America' s traveler information phone number
Florida Trucking Association
Florida Truck Driving Championship
Global Positioning System
Highway Advisory Radio
Highway Capacity Manual
HIS Highway Information System
IRI International Roughness Index
KYTC Kentucky Transportation Cabinet
LOS Level of Service
LSD Fisher' s Least Significance Difference multiple comparison test
NATSO National Association of Truck Stop Operators
NHS National Highway System
NTA National Truckers Association
NYC New York City
ODOT Oregon Department of Transportation
OTA Ontario Trucking Association
P&D Pick-up and Delivery
PCE Passenger Car Equivalent
PSI Present Serviceability Index
PSR Present Serviceability Rating
PTBF Percent-Ti me-B ei ng-F followed
PTSF Percent-Time- Spent-F following
QOS Quality of Service
Q-Q plot Quantile-Quantile plot
RCI Roadway Characteristics Inventory
RMI Relative Maneuverability Index
RV Recreational Vehicle
SIS Strategic Intermodal System
SMC Safety Management Council
Student-Newman-Keuls multiple comparison test
Southwestern Pennsylvania Regional Planning Commission
Traveler Information System
Third Party Logistics
Transportation Technical Services, Inc
Text Unit Block
Urban Distribution Centre
University of Hawaii at Manoa
Variable Message Sign
a satellite radio service
Abstract of Dissertation Presented to the Graduate School of the University of Florida in Partial
Fulfillment of the Requirements for the Degree of Doctor of Philosophy
IDENTIFICATION OF PREFERRED PERFORMANCE MEASURES FOR THE
ASSESSMENT OF TRUCK LEVEL OF SERVICE
Chair: Scott S. Washburn
Major: Civil Engineering
Commercial trucks, the leading transportation mode for freight movement, are vital to our
economy and people's lives. The importance of this mode has become greater as the demand for
just-in-time delivery, lower inventory, electronic commerce, and Less-than-TruckLoad (LTL)
shipping has increased. This study was conducted under the Florida' s Strategic Intermodal
System (SIS) plan for the Florida Department of Transportation (FDOT) in an effort to better
understand the needs of the Florida trucking community by investigating their perceptions of
truck trip quality on various roadway facilities.
The current Highway Capacity Manual (HCM) provides analytical methods to evaluate
performance levels of transportation facilities and estimate Level Of Service (LOS) perceived by
the users. The HCM methodologies yield a single LOS value for all the users in a traffic stream.
However, due to the unique size and operating characteristics of trucks, it is possible that truck
mode users perceive LOS on various roadway facilities based on different criteria from those of
the other mode users. This study focused on identifying the determinants of LOS perceived by
truck mode users and measuring their relative importance based on which truck LOS estimation
methodologies should be developed.
Three focus group sessions (two with truck drivers and the other one with truck company
managers) were held to elicit the factors affecting truck LOS on various transportation facilities
and the relative importance of those factors were evaluated through a follow-on survey study
with a larger audience. A total of 459 truck drivers and 38 truck company managers responded
to the written surveys collected at Florida Truck Driving Championship (FTDC) event or the
postage-paid mail-back surveys distributed at four agricultural inspection stations. The survey
responses were analyzed with various statistical methods such as descriptive statistics,
Exploratory Factor Analysis (EFA), multiple comparisons of the means, non-parametric tests,
and chi-squared tests of independence.
The quality of a truck trip generally was found to depend on three issues: travel safety;
travel time; and physical and psychological driving comfort. Truck drivers showed more
concerns about the driving comfort, while truck company managers were more concerned with
travel time. The travel safety aspect of a truck trip was a major concern for both groups. 'Speed
Variance' and 'Pavement Quality' were the two most important determinants of truck trip quality
on freeways. Truck trip quality on arterials was a function of multiple factors including
'Pavement Quality', 'Turning Maneuvers', 'Speed Variance', and 'Traffic Density'. 'Percent-
Time-Being-Followed' (PTBF), 'Percent-Time-Spent-Following' (PTSF), 'Travel Lane and
Shoulder Widths and their Pavement Quality' were identified as truck LOS determinants on two-
lane highways. Among many factors contributing to the quality of a truck trip, 'Passenger Car
Drivers' Road Etiquette and Knowledge about Truck Driving Characteristics', 'Traffic
Congestion', and 'Frequency and Timing of Construction Activities' were perceived to be in
significant need of improvement regardless of the type of roadway facility. Truck travel
restrictions (e.g., truck lane restrictions) on freeways, inadequate curb radii and poor traffic
signal coordination on arterials, and inappropriate shoulder width and condition and poor night-
time lighting condition on two-lane highways were also considered to be greatly in need of
The results of this study provide the FDOT with guidelines and recommendations to
develop truck LOS estimation methodologies to effectively assess how well it is addressing the
needs of freight transportation on the state roadway system and offer transportation service
providers and other stakeholders valuable insights for the prioritization of transportation
improvement proj ects for commercial truck traffic.
Transportation services for freight are vital to our economy and people's daily lives.
Various kinds of freight are relocated daily in and out of Florida through several transportation
modes such as truck, train, ship, and plane. Among all the modes, trucks moved 76.3 percent of
freight value, 79.4 percent of tonnage, and 66.9 percent of ton-miles of the total freight
originating in Florida in 2002, according to the Commodity Flow Survey (CFS) (Bureau of
Transportation Statistics, 2004). This makes the truck mode the leading mode for freight
movement in the U.S. Truck traffic is also expected to grow significantly throughout the State of
Florida over the next couple of decades, along with increasing population, as noted in the Freight
Analysis Framework (FAF) (U. S. Department of Transportation and Federal Highway
Administration, 2002). In addition, the increased demands for just-in-time delivery, low
inventory, electronic commerce (e-commerce), Less-than-TruckLoad (LTL) shipping, and more
distributed manufacturing, make it necessary for transportation service providers and other
stakeholders to look for ways to better accommodate truck traffic on existing roadway systems.
Florida's Strategic Intermodal System (SIS) plan was established by the Florida
Department of Transportation (FDOT) in 2003 to support transportation facilities that are
necessary for Florida' s rapidly growing and ever changing economy (FDOT, 2003). The main
goal of the SIS program is to provide safe, efficient, and convenient transportation services for
all types of transportation users on the most critical transportation facilities in Florida. The SIS
facilities were selected based on national or industry standards for measures of transportation and
economic activity. The SIS includes three different types of facilities; hubs, corridors, and
intermodal connectors. Hubs are ports and terminals, corridors are highways, rail lines, or
waterways that connect maj or markets, and connectors are highways, rail lines, or waterways
that connect hubs and corridors. As one of its first research proj ects under the SIS program, this
study has been conducted for FDOT to better understand the needs of the Florida trucking
community by investigating its perceptions and opinions about transportation services for trucks
on various state roadway facilities.
1.2 Problem Statement
The Highway Capacity Manual (HCM) (Transportation Research Board, 2000) provides
analytical methods to estimate capacity and key performance measures for a wide variety of
roadway facilities. The HCM also provides methods for translating performance measure values
into a Level Of Service (LOS) value. Level Of Service (LOS) is a qualitative measure used to
describe general operating conditions within a traffic stream. The HCM uses a scale of 'A' to
'F' for LOS, with 'A' corresponding to excellent operating conditions and 'F' corresponding to
extremely poor operating conditions. These LOS values are typically based on performance
measures such as speed and travel time, freedom to maneuver, traffic interruptions, and comfort
and convenience, and are intended to reflect drivers' perception of those conditions. The
selected performance measures, upon which LOS is based, referred to in the HCM as service
measures, can vary from one roadway facility to another. The methods of the HCM have been
widely adopted by the transportation community as the primary means of measuring system
performance. They are often used as valuable tools for most transportation agencies to monitor
or improve the performance levels of existing transportation facilities, or plan for future
The current HCM utilizes engineering-based measures such as density, delay, and average
travel speed in evaluating the performance level of transportation facilities. These measures are
conceptually reasonable and easy to apply, but are not necessarily consistent with the measures
that the actual roadway users base their perception of the facility performance level on.
Furthermore, for the roadway facilities in the HCM (such as arterial, highways, and freeways),
the HCM methods result in a single LOS value for the entire traffic stream; that is, they do not
distinguish between different modes within the same traffic stream. The methods are generally
designed to establish LOS for the passenger vehicle mode. However, commercial trucks
(hereafter referred to as just 'trucks') are unique in their size and operating characteristics among
various vehicle types in traffic streams. Trucks require more space and time to maneuver due to
their difference in size, weight, off-tracking, acceleration, and braking. So they are often used as
design vehicles for roadway facilities, and sometimes subj ect to various restrictions such as lane,
route, speed, or time-of-day. In addition, most truck drivers operate their trucks primarily for
business purposes and spend a significant amount of time driving, while the trip purposes of
other vehicle users vary and they usually drive less frequently.
Due to these special characteristics of truck operations on transportation facilities, it is
possible that truck mode users perceive LOS on various roadway facilities based on different
criteria from those of the other mode users. However, the current HCM (Transportation
Research Board, 2000) only accounts for the mode of trucks in the traffic stream through the
heavy vehicle adjustment factor (f, ). This adjustment factor is based upon the percentage of
heavy vehicles (e.g., trucks, buses, or recreational vehicles) in the traffic stream and their
passenger car equivalents (PCEs). A PCE is the number of passenger cars expected to have
same effect on a traffic stream as a single heavy vehicle under specific roadway and traffic
conditions. The PCE is used to convert a traffic stream flow rate with some portion of heavy
vehicles into an equivalent one with passenger cars only. So, although the current HCM
procedures account for the effects of trucks in the traffic stream, the overall methodology does
not provide any exclusive LOS evaluation criteria or procedures to estimate how well a
transportation facility accommodates truck traffic in a traffic stream.
1.3 Study Objectives and Tasks
The primary obj ective of this study was to determine the roadway, traffic, and/or control
factors important for Florida' s trucking community and estimate the relative significance of each
factor on overall quality of truck operation on the SIS facilities. Based on information obtained
about the perceptions of truck mode users, this study provides the FDOT with guidelines and
recommendations to develop actual LOS estimation methodologies to effectively assess how
well it is addressing the needs of freight transportation on the state roadway system. These
efforts provide transportation service providers, researchers, and transportation agencies with
valuable insights as to what should be prioritized to improve truck operations on various
To achieve this obj ective, it was essential to obtain input from both truck drivers and truck
company managers, as these two groups represent the maj or stakeholders with regard to truck
operations. For this study, data collection was confined to focus group and written survey
methods. The maj or tasks performed to complete this study are described below and presented
in their chronological order:
Task 1: Literature review. A literature review was performed to facilitate and enhance
this study with regard to the following five areas: 1) trck LOS studies; 2) user-perception based
LOS studies; 3) trucking community focus group and/or survey studies, 4) Florida trucking
community focus group participant recruitment sources, and 5) trucking community survey
Task 2: Preparation for focus group sessions. In advance of the focus group meetings,
the methodology for conducting the focus group studies was determined. This includes
clarification of the purpose and scope of the interviews, participant selection guidelines,
participant recruitment plans, moderator selection, development of participant background
surveys, moderator's guide regarding focus group questionnaires and instructions, focus group
discussion recording, and methods of analysis.
Task 3: Focus group sessions of truck drivers and/or truck company managers.
Several focus group meetings were held with selected participants to explore the perceptions and
opinions of the trucking community on the factors important to truck operations on various
Task 4: Summary of focus group findings. The focus group discussions were
summarized to identify the factors affecting quality of a truck trip on each facility and by
different user groups (truck drivers and truck company managers). The summary results were
then used as inputs to a follow-on survey.
Task 5: Survey development. Survey studies were intended to reconfirm the focus group
Endings with a broader audience and to measure the relative importance of each factor
quantitatively. Two different survey forms were prepared; one for truck drivers and the other for
truck company managers. Both forms contain questions about participants' personal and
working backgrounds and their perceptions about the relative importance of each factor
identified in the previous focus group studies.
Task 6: Survey data collection. Truck driver surveys and truck company manager
surveys were distributed at multiple locations with the assistance of the FDOT and the Florida
Trucking Association (FTA).
Task 7: Analysis of survey data. Collected surveys were analyzed statistically to
determine the relative importance of each factor on truck trip quality and/or overall trucking
business. Relationships between the participants' backgrounds and their perceptions were also
Task 8: Study results and recommendations. The factors influencing truck trip quality
and their relative importance were determined based on the focus group and survey results.
Potential performance measures to estimate truck LOS were recommended by each
transportation facility type.
1.4 Organization of the Dissertation
The remainder of this dissertation is as follows. Chapter 2 provides a summary of relevant
literature. Chapter 3 describes the study methods and detailed steps taken to complete this study.
The study results are presented in Chapters 4 and 5. Chapter 4 discusses focus group study
findings and Chapter 5 discusses the results of statistical analyses on the survey data. In chapter
6, conclusions are presented and recommendations are made with regard to roadway facility
improvement priorities for truck LOS considerations. Potential truck LOS service measures are
This chapter describes previous research efforts by other researchers related to this study,
conceptually or methodologically. It provides descriptions of the research methodologies and
results of those studies, and a summary of the findings. First, the studies regarding truck LOS
development are presented. They contain some implications on potential truck LOS service
measures. Other research efforts to identify and test potential LOS service measures on a
specific roadway facility are described in the next section of this chapter. They provide some
insights about how to elicit the perception of travelers and convert the information to develop
user-perceived LOS. The final section reviews previous focus group and/or survey studies of the
trucking community. Potential sources of focus group recruitment and characteristics of various
survey methods were separately summarized to help develop the methodologies suitable for the
purpose and scope of this study.
2.1 Truck Level of Service
A study by Washburn (2002) explored and demonstrated the development of a potential
method to assess level of service specific to the truck mode. The current LOS performance
measure for a freeway segment (density) applies to the traffic stream as a whole, regardless of
vehicle types, although they typically have very different physical and operating characteristics
from passenger cars in the traffic stream. To assess the level of service of heavy trucks on a
freeway segment, a 'relative density' concept was proposed. Relative density for trucks could be
obtained by dividing density for the traffic stream by a Relative Maneuverability Index (RMI),
which is a function of the ratio of percentage of free-flow speed of trucks to percentage of free-
flow speed of passenger cars. Under low density, ideal conditions, percentage of free-flow speed
for both trucks and cars should be at or near 100%, yielding the same LOS for both modes (e.g.,
'A'). Likewise, under high density, the percentage of free-flow speed for both trucks and cars
would both be very low, producing the same LOS for both modes (e.g., 'F'). At densities in
between, the percentage of free-flow speed for trucks will probably be less that that for cars,
resulting in higher relative density values for trucks, which can be referenced to the current HCM
density thresholds for determining LOS. Other possible measures for determining LOS for
trucks have also been introduced in the document; acceleration noise, passing opportunities,
percent-time spent-following, and heavy vehicle factor.
A study by Hostovsky and Hall (2004) focused on the perceptions of tractor-trailer drivers
on the factors affecting quality of service on freeways. A focus group meeting was conducted
for this study with 7 Road Knight Team members at the annual convention of the Ontario
Trucking Association (OTA) in November, 2001. The participants were asked what makes for a
good or bad trip for themselves to drive trucks on a freeway. The discussion was then
transcribed to be analyzed through NUD.IST (Non-numerical Unstructured Data Indexing
Searching and Theorizing), an industry standard qualitative data analysis software program using
five criteria (intensity, relevance, frequency, universality, and emphasis). The OTA, Canadian
Trucking Association (CTA), and American Trucking Association (ATA) showed an agreement
with the study's conclusions. The study results were organized in this paper by the number of
Text Unit Blocks (TUB), which represent the number of times a certain theme was discussed by
the participants. With respect to freeway conditions (TUB = 34), the factors identified were road
surface (pavement condition, snow removal, road debris), lane restrictions, signage, and lane
width and markings.
As far as traffic conditions (TUB = 22) are concerned, congestion, steady flow, and
maneuverability were discussed. Attitudes toward other vehicles (TUB = 27) were another big
issue mentioned in the meeting. Courteous interplay by automobile drivers was found to be
important for them to have a good trip. In terms of safety (TUB = 3), road etiquette, weather,
and rubbernecking were considered to influence their trip quality. The participants also
mentioned the hazard of aggressive driving and road rage (TUB = 23) and regionally different
perceptions of trip quality (TUB = 23) with respect to such various factors as level of congestion,
courteous interplay by other drivers, driving skills of other drivers around them in snow
condition, and low speed limits. With all the findings mentioned above, the authors indicated
that what really matters to truck drivers in terms of traffic condition is traffic flow, not traffic
density, which is a service measure used in the current HCM. In this respect, steadiness of traffic
flow and an ability to drive at a comfortable operating range of highway speeds without much
braking or accelerating were considered to be important for truck drivers' perception of the
quality of service.
A study by Hall, et al. (2004) developed a method to evaluate the access routes for large
trucks between intermodal or other truck-traffic-generating sites and the National Highway
System (NHS) and used it to prioritize and program the truck access routes for improvement.
The study began with identifying clusters of truck-traffic generating facilities based on total
trucks per day, distance to NHS, and recommendations on the sites with truck access problems
by transportation planners from highway district and area development district offices. Then
telephone surveys were conducted with the operators/managers of the selected sites to identify
problem routes to be evaluated in this study. Each route was evaluated with respect to three
types of features: subjective, point, and continuous. Using the Kentucky Transportation Cabinet
(KYTC) statewide Highway Information System (HIS) database and some field observations, the
following characteristics of each route were collected: 1) point features: curve off-tracking,
maximum safe speed on horizontal curves, stopping sight distance, turning radii; 2) continuous
features: grade, lane width, shoulder, route LOS. The point and continuous elements on each
route were ranked as "preferred", "adequate", and "less than adequate" based on the
recommendations in AASHTO (American Association of State Highway and Transportation
Officials)' s "A Policy of Geometric Design of Highways and Streets" and "Roadway Design
Guide." The features subjectively evaluated by the researchers include clear zone, pavement
condition, accident history, parking, pedestrian traffic, land use conflicts, dust/noise issues, and
so on. The rankings of point and continuous elements were converted to a relative urgency
rating by assigning a relative weight with respect to truck volume and section length. Problem
truck-points and problem truck-miles were calculated based on these rankings to prioritize the
problem routes. They were adjusted by predominant subj ective features, where appropriate.
Finally, with all the evaluation data of each problem route, the researchers inspected the routes
and graded them on a subj ective scale of 1 to 10 (10 represents good access for trucks).
All the three studies presented in this section focused on developing methods to evaluate
operating conditions on a specific roadway type for truck traffic. Washburn (2002) devised the
RMI concept to evaluate truck LOS on freeways, based on a hypothetical reasoning that
maneuverability of trucks is inferior to that of passenger cars because of unique physical and
operating characteristics of trucks. A study by Hostovsky and Hall (2004) found that traffic flow
is more important than traffic density to truck drivers' perception of trip quality on freeways.
The two above mentioned studies imply that traffic density, the current HCM service measure
for freeway facility, may not reflect the perception of truck drivers adequately. Steady traffic
flow and maneuverability may be more important to truck drivers than traffic density,
considering the large size, heavy weight, low acceleration and deceleration capability of trucks.
Hall, et al. (2004) developed a methodology for evaluating large truck access routes between
NHS and truck-traffic generating facilities to prioritize and program the routes for improvement.
Overall quality of each route for truckk access was determined by many various factors. They
included geometric adequacy measures, pavement condition, clear zone, accident history, traffic
LOS, and other subj ective measures such as parking, pedestrian traffic, land use conflicts,
dust/noise issues, etc. As this study investigated so many different characteristics of each route,
it may not be appropriate to identify one or two performance measures to effectively estimate
quality of an access route for trucks.
2.2 User-Perceived Level of Service
A study by Hall, et al. (2001) conducted two focus groups (one with Hyve members, and the
other with seven members) to explore what freeway users perceive as important to level of
service on freeways. Focus group participants were chosen by a snowball sample selection
process to identify commuters going from the city of Toronto, Ontario, Canada to McMaster
University (the research center) in Hamilton, Ontario Canada. The participants were all
university faculty members in various departments, and were selected to ensure that all the
participants traveled the same stretch of freeway so that all knew about the situations that were
being discussed and all had had relatively similar experiences. Both groups were moderated by
the same member of the research team to ensure consistency in technique across groups. The
following general questions were asked in the discussions;
* Tell me about the usual freeway route you take when you are commuting.
* Have your perceptions of the drive changed over time?
* I want you to talk about both the ideal or best trip you've ever made and also the worst trip
The factors important to perceptions of trip quality were listed with respect to the number
of Text Units (TUs) in which each theme was mentioned in the two discussions, as follows;
* Travel time (TUs = 103 (6%))
* Density/maneuverability (TUs = 86 (5%))
* Road safety (TUs = 81 (5%))
* Commuter information and communication (TUs = 65 (4%))
* Civility (TUs = 38 (2%))
* Photo radar (TUs = 31 (2%))
* Weather (TUs = 23 (1%))
The participants indicated travel time as the first thing to describe the quality of particular
trips. Having time constraints on their arrival from a trip was thought to greatly increase the
stress involved in the trip. Time spent commuting was also considered to be lost time.
Maneuverability also came up in the discussions in light of travel time, accident avoidance,
adequate gaps between cars, weaving in and out of traffic, and changing lanes. Another
important issue for participants was safety. They were concerned about accidents not only in
terms of congestion but also because of the risk to their personal safety. Materials from trucks
and other debris on the road were concerns for many participants, and the effects of sport-utility
vehicles (SUVs) and minivans on other drivers' visibility were also mentioned in the safety
perspectives. Having adequate information about what was happening to traffic while they were
on the road was also important to the participants. This issue was discussed along with the
possibility for them to avoid traffic congestion by finding alternative routes. Some other factors
were mentioned occasionally in the focus groups as well; driver civility, the use of photo radar,
and weather and its effect on driving conditions. Passengers' perceptions also were different
from drivers' with respect to travel time, mainly due to their ability to undertake activities other
than driving. The participants indicated that travelers on the bus would have different
perspectives; they usually do not notice the traffic conditions when they are in the bus. The
authors also indicated that the participants do not think of freeways in terms of distinct segments
classified by the HCM (basic segments, ramp-freeway junctions, and weaving segments),
proposing further studies about LOS breakpoints.
The factors that affect traveler-perceived quality of service on rural freeways were
explored using in-field surveys of motorists traveling on rural freeways (Washburn, et al., 2004).
A total of 233 responses from a good mix of respondents were collected at several different
locations along I-75 and the Turnpike in Florida. The researchers decided to perform an 'in-
field' survey data collection effort, as opposed to a mail-back survey or something similar, as
this provided the advantage of obtaining input while the specific characteristics of the traveler' s
trip were still fresh in their mind. The surveys at the rest stops, which offered respondents a $2
food discount voucher, also showed a good overall productivity (about six respondents per hour).
The importance of 16 traffic and roadway factors was ranked by respondents on a seven-
point ordinal scale (1 being not at all important, and 7 being extremely important). The six top
ranked factors based on mean scores and/or percent time in top three are as follows:
* Ability to consistently maintain your desired travel speed (mean = 6.09, % time in top
three = 64.3%)
* Ability to change lanes and pass other vehicles easily (mean = 5.79, % time in top three =
* Smooth and quiet road surface condition (mean = 5.68, % time in top three = 20.3%)
* Ability to travel at a speed no less than the posted speed limit (mean = 5.58, % time in top
three = 33.0%)
* Other drivers' etiquette/courtesy (mean = 5.38, % time in top three = 22.1%)
* Infrequent construction zones (mean = 5.37, % time in top three = 23.4%)
A probit modeling technique was used to discover any possible relationships between
personal characteristics and current trip information of the respondents, and their opinions about
the top four ranked factors (from 1 to 4 above). For the first factor, the more highly educated an
individual, the more likely they were to rank this factor highly. The respondents that were
driving a tractor-trailer, or with higher estimated average speeds, were also more likely to rank
this factor highly. For the second factor, the respondents that had higher education or with
higher estimated average speeds, were more likely to rank this factor highly. Travelers that
indicated they were experiencing mostly very dense or dense traffic flow conditions on their trip
were less likely to rank this factor highly. For the third factor, travelers with higher income were
more likely to rank this factor highly. Travelers who indicated their trip purpose was 'other',
that is, neither business nor leisure, were less likely to rank this factor highly. Travelers that
were on a business trip and very familiar with the route they were traveling were also less likely
to rank this factor highly. For the fourth factor, older people were more likely to rank this factor
highly. On the other hand, the respondents that were strictly drivers on the trip or driving a large
auto (pickup truck, SUV, minivan, or full-size van) were less likely to rank this factor highly.
The authors indicated that in addition to density, there are some factors that are just as
important to travelers, such as speed variance and percent of free-flow speed. Some non-traffic
performance measures were also found to be important through the study, such as pavement
quality, and driver etiquette.
A web-based survey was conducted at the University of Hawaii at Manoa (UHM) in 2005
to see how road users evaluate signalized intersection LOS (Zhang and Prevedouros, 2005). E-
mail messages with group-specific survey hyperlinks were sent to selected samples asking them
to fill out the motorist survey on-line. Six groups comprise the sample; random UHM students,
faculty, and staff, Hawaii engineering and transportation professionals, drivers from other groups
in Hawaii, and drivers from the mainland U.S. The response rate was 12.2%, 16.0%, 18.9% for
UJHM students, faculty, and staff respectively, yielding 2,017 usable surveys. The survey
questions focused on factors important to drivers at signalized intersections, driver opinions on
protected left-tumn signals for various sizes of intersections, and trade-offs between perceived
safety risk and delay. Analysis of variance (ANOVA) was applied to dependent variables with
ratio- or interval-level data to assess if there were significant differences among the independent
groups. The influence of several independent variables such as gender, age, and driving
experience on the dependent variables were investigated simultaneously through ANOVA
(Analysis of Variance).
The ten factors important to drivers at signalized intersections were evaluated with a five-
point ordinal scale (1 being not important, and 5 being extremely important). The factors are
listed as follows in order of decreasing importance.
1. Traffic signal responsiveness
2 Ability to go through the intersection within one cycle of light changes
3 Availability of left tumn only lanes and protected left turn signal for vehicles turning left
4 Pavement markings for separating and guiding traffic
5 Availability of a protected left tumn signal for vehicles turning left
6 Availability of left turn only lanes for vehicles turning left
7 Pavement quality
8 Waiting time
9 Heavy vehicles such as trucks and buses that are waiting ahead
10 Availability of right-turn only lanes for vehicles turning right
The respondents were also asked to rate the difficulty in making a left turn without a
protected left tumn signal and their preference for a protected left turn signal, on a scale form 1
(not difficult or not preferred) to 5 (extremely difficult or extremely preferred). The difficulty
increases with intersection size, and female drivers perceived more difficulty than male drivers.
The preference also increases with intersection size, and female drivers prefer protected left tumn
signals more than male drivers. As high as 91% of the respondents stated that they much or
extremely prefer a protected left turn signal at an intersection where left turn vehicles have to
cross three lanes of opposing through traffic.
Safety was stated to be three to six times more important than delay, depending on the type
of conflict; drivers turning left are more concerned than drivers going through, and pedestrians
are more concerned than drivers turning left. Female drivers, non-risk-prone drivers, and
pedestrians valued safety more than delay to a significantly greater degree. Drivers and
pedestrians would be willing to wait a longer time at signalized intersections in exchange for
protected left turn signals. The average additional delay to exchange for protected left turn
signal is 20.6 sec, 25.4 sec, and 27.2 sec for drivers going through, drivers turning left, and
Overall findings suggest that the current measure, delay, should be supplemented by a
number of quantifiable attributes of signalized intersections for determining a LOS that
represents road user perceptions.
Some potential performance measures for LOS on rural freeways were evaluated with
microscopic traffic simulation (Kim, et al., 2003). Although density-based LOS is thought to be
well suited to the assessment of urban freeways, some have questioned whether density is also a
proper indicator of the quality of service on rural freeways because drivers may think more in
terms of psychological or emotional comfort for rural freeways, which generally serve long,
high-speed trips and rarely experience more than moderate congestion levels. Acceleration
noise, number and duration of cruise control applications, and percent time spent following
(PTSF) were examined as some possible means of determining Level of Service (LOS) of rural
freeway sections as they have at least an intuitive relationship to the concept of driver comfort.
The three measures were estimated for a hypothetical section of rural freeway. Acceleration
noise is a measure of the physical turbulence in the traffic stream. The instantaneous
acceleration of each vehicle at each second is computed from the speed differential with respect
to the previous second, and the standard deviation of acceleration for each vehicle is computed
over the 6,000 feet segment. A post-processing routine was written to analyze the simulation
results and infer the application and release of cruise control. Cruise control was applied to a
vehicle when it had been traveling at its desired speed for three or more seconds. Cruise control
was released when the vehicle began to decelerate for any reason. For PTSF, a vehicle is
considered to be following its leader if the relationship between its position and speed with
respect to the leader places it within the car-following influence zone. A nonlinear relationship
between acceleration noise and traffic volume shows that acceleration noise increases more
rapidly in the lower volume range and levels off as volume increases. It is also observed that the
nonlinear effect is most pronounced on single-lane freeways and diminishes as the number of
lanes increases. The nonlinearity for proportion of time with cruise control applied was only
discernible to any degree in the case of the single-lane freeway. The nonlinearity for number of
cruise control applications was too pronounced to be useful. The nonlinearity for PTSF was
more pronounced than that for acceleration noise, and there was no discernible difference
between two and three lane freeways. Based on the investigation of the nonlinear relationship
shapes between level of volumes and three candidate measures by number of lanes, the authors
concluded that a nonlinear relationship between acceleration noise and traffic volume on rural
freeways could be used directly as the basis for a new set of LOS criteria for rural freeways,
given further studies focusing on driver opinions, behavior, and field measurement to support
this finding. The other two measures also have conceptual appeal, but used individually would
not be suitable as the basis for determining LOS on a rural freeway.
The above described studies concerning the exploration of user-perceived level of service
performance measures identified some interesting findings. The study by Hall, et al. (2001)
found from two focus group meetings that total travel time is the most important LOS
determinant for travelers on a freeway. It was noted that passengers did not consider the travel
time as important as drivers did because they can do something else other than driving. It would
be worth investigating how freeway drivers' perceptions on importance of total travel time can
be influenced by their trip purposes. The study by Washburn, et al. (2004) found from surveys
of rural freeway travelers that "consistently maintaining desired travel speed" is more important
than "ability to change lanes and pass other vehicles easily" or "traveling at or not less than the
posted speed limit." This implied that the cruise control factor is more important than the
density factor or percent of free-flow speed factor. Other important factors included pavement
condition, other drivers' etiquette/courtesy, and infrequent construction zones. A web-based
survey study by Zhang and Prevedouros (2005) found that the current service measure of delay
for signalized intersections is less important than a number of other factors such as "traffic signal
responsiveness", "ability to go through the intersection within one cycle of light changes", etc. It
may be possible that drivers are more sensitive to stop and go conditions than actual delay
experienced at signalized intersections. Kim, et al. (2003) identified acceleration noise as a
potential service measure for rural freeways due to its desirable properties of relating to drivers'
sense of safety and comfort and its non-linear relationship with traffic volume. Many previous
studies of this type verified that some current service measures do not reflect the perception of
the users adequately.
2.3 Trucking Community Focus Groups and Surveys
With increasing importance on urban freight mobility, Morris, et al. (1998) conducted a
study on cost, time, and barriers related to moving freight into New York City's Central Business
District (CBD). The study consisted of 13 focus groups with different industry-sector senior
logistics executives and freight mobility interviews with logistics, transportation, or distribution
managers. The research team cooperated with the Center for Logistics and Transportation
Executive Committee in the development of a focus group moderator' s guide, recruitment of
focus group participants, pretest of the guide, and the guidance of the freight mobility interview.
Each focus group was scheduled at participants' convenience, allowing the use of
speakerphones, and included 2-4 participants to explore the issues in depth within a 2-hour time
frame. Participants also received the focus group guide and informational material before the
meetings so that they would have time to think about the issues in advance. At the meetings, a
flip chart listing seminal points for focus group questions and probes, and a large CBD area map
were displayed to reinforce attendees' attention to topics. Transportation barriers listed in the
order of greatest frequency of mention in all the meetings were; congestion, inadequate docking
space, curb space for commercial vehicles, security, and excessive ticketing of high-profile
companies. A freight mobility interview followed with logistics, distribution, or transportation
managers, who were recruited with the help of the previous focus group participants or from the
membership lists of Council for Logistics Management and the Center for Logistics and
Transportation. An on-site interview was planned initially, but the response was generally
negative. So phone interviews were requested to be scheduled after sending letters containing a
freight mobility interview, the study goals and purpose, and focus group findings. Follow-up
calls were made 2-4 weeks after the letters were sent. But due to the lower than expected
response rate, another attempt was made with follow-up calls made I week after sending letters,
resulting in 51 completed interviews. Data on costs, time, distance, product types, and major
barriers in the movement of freight into Manhattan's CBD by shippers and carriers were
collected. Maj or barriers to freight mobility identified through the interviews were; widespread
congestion, security, physical constraints, and institutional barriers.
In a study by Veras, et al. (2005), various types of efforts such as focus group studies, in-
depth interviews, and internet surveys were made to obtain the perceptions on challenges and the
potential of off-peak deliveries to congested areas. The targeted groups for this study included
private stake holders; shippers, receivers, third party logistics (3PLs), trucking companies, and
warehouses. Off-peak deliveries to the New York City (NYC) metropolitan region were
proposed to avoid traffic congestion and lack of parking spots hampering daytime commercial
vehicle deliveries. The strategy may reduce the price accompanied by congestion and illegal
parking, pollution, and frustration to the public, but it may pose additional costs to shippers,
receivers, and carriers mainly due to workers necessary for night shifting. It may also involve
regulatory and legal impediment. On January 20, 2004, two focus groups with truck dispatchers
were conducted in NYC as part of the Evaluation Study of the Port Authority of New York and
New Jersey's Value Pricing Initiative. The focus group findings indicated that both shipper costs
and receiver costs would increase due to the need of night time workers, regardless of feasible
toll discounts, or compensation due to traveling hours. In-depth interviews with 17 stakeholders
of various types (trucking companies, shippers, receivers, lobbyists, trucking-warehouse
combination companies, and shipping-trucking-warehouse combination companies) were
performed to explore the issues further. Companies with trucking operations prefer to make
deliveries during off-peak hours due to congestion and parking problems, but they are often
discouraged to do it because of the recruitment of night-shift workers, and security of drivers,
receivers, and products. The shippers were natural on the subject of off-peak deliveries. They
do not care when their products are delivered only if the products get to the destination on time.
The two receivers operating restaurants stated that off-peak deliveries would help reduce severe
parking problems, but fresh food is typically not delivered during off-peak hours. Thirty-three
internet surveys were gathered from private stakeholders (shippers, receivers, 3PLs, trucking
companies, warehouses) in several states. Seventy-Hyve percent of warehouses and 70% of
shippers, 3PLs, and trucking companies were performing off-peak deliveries. Many others at
least have considered using this alternative. None of the shippers was currently performing off-
peak deliveries, but maj ority of them indicated that they could do them if they are provided with
some incentives. Reasons given by stakeholders for performing off-peak deliveries included
faster deliveries, faster turn-around times, and lower costs. Reasons given for not performing
off-peak deliveries included businesses not being open at those times, customers not accepting
off-peak deliveries, and employee costs. With respect to incentives to implement off-peak
deliveries, most stakeholders were interested in all of the tax incentives and subsidies. A
significant number of trucking companies and 3PLs indicated that they would be responsive to a
request from many receivers to do off-peak deliveries. The author also brought up carrier-
centered policies, receivers as key stumbling blocks, Einancial incentives, and targeted maj or
Two studies focused on developing an effective methodology to survey the freight
community (i.e., shippers and carriers) in Oregon state by comparing various survey data
collection methods (Lawson and Riis, 2001 and Lawson, et al., 2002). The researchers did an
extensive literature review on the previous trucking community survey studies and experimented
with several data collection methods to Eind out the most effective method to survey shippers and
carriers. Traditional structured interviews conducted in person or by telephone have a high
response rate from purposeful sampling procedure. These surveys can focus on broad issues and
are useful in obtaining detailed information about the specific topic, but may not accurately
reflect the large population. Computer-Aided Telephone Interviews use a script of
predetermined questions with random sampling techniques. This method allows for a large
sample size, and is less costly, but, reduces the opportunity to fully explore a certain aspect of
any given topic. The response rate from the method is lower than traditional structured
interviews, but higher than mail out surveys. Response rates for previous written surveys ranged
from 8 to 24 percent, but can be improved with telephone follow up. Written surveys are the
least costly, good for broad sampling, but it typically produces low response rates and does not
ensure that the right person will complete the survey.
Nine types of survey methods were tested as a pilot study to search for the best one to use
for surveying the freight community.
* Type 1 mail out/mail back questionnaire with follow up reminders (ES202 database,
response rate of 15%)
* Type 2 mail out/mail back questionnaire and a map, with follow up reminders (ES202
database, response rate of 11%)
* Type 3 postcard invitation to participate, for positive respondents, mail out/mail back
questionnaire with follow up reminders (ES202 database, response rate of 6%)
* Type 4 postcard invitation to participate, for positive respondents, mail out/mail back
questionnaire and a map, with follow up reminders (ES202 database, response rate of 4%)
* Type 5 telephone invitation to participate, for positive respondents, mail out/mail back
questionnaire with follow up reminders (ES202 database, response rate of 19%)
* Type 6 telephone survey (ES202 database, response rate of 60%)
* Type 7 telephone survey (Oregon DOT Motor Carrier Transportation Division truck
registration database, response rate of 64%)
* Type 8 mail out/mail back questionnaire with telephone follow up reminders (Oregon
DOT Motor Carrier Transportation Division truck registration database, response rate of
* Type 9 mail out/mail back questionnaire with follow up reminders (Oregon DOT Driver
and Motor Vehicle Services Division Commercial Driver' s License (CDL) Database,
response rate of 12%)
In every case, the respondents were given information on the availability of the survey via
e-mail or the website, or both. However, little evidence was found that the option to use the web
site or to communicate via e-mail was of interest to the survey participants. In addition to the
infrastructure problems, various other issues such as regulations, taxation, and enforcement were
covered in the survey. The survey response rates from telephone surveys were highest among all
the methods. The use of postcard invitation before a mail out survey resulted in very low
response rates. Telephone invitation with a mail out survey yielded a 19% response rate, while a
mail out survey with follow up phone calls produced a 33% response rate. Freight firms said that
a written survey is the most preferred method, but proved to yield a lower response rate than
Prior to carrying out the Oregon State-wide Freight Shipper and Carrier Survey, 34
previous freight surveys already implemented were reviewed by Loudon (2000). Considerations
for conducting surveys of the freight movement community, and lessons learned from previous
freight research in Oregon are summarized. The options to be determined according to the
purpose of the surveys are:
* One company questionnaire versus separate company questionnaires for shippers and
* Survey transportation managers only versus survey managers and drivers.
* Direct driver contact versus distribution to drivers by managers.
* Focus on truck movements only versus focus on all modes of freight.
* Explore problems through open-ended questions versus structured list of possible
* Ask the respondent to rank problems versus list problems only.
* Ask about problem and practices for inbound and outbound freight movements or
* Use a written (self-completing) questionnaire versus interviewing.
* Use cold mailing of written questionnaire versus pre-arranged participation.
Lessons learned from previous freight research in Oregon are:
* Freight movement logistics are complex. So, special considerations for flexibility should
be made to design each issue (shipment size, timing of shipments, etc).
* Methods of shipping freight are changing rapidly. For example, more inventories are
maintained in trucks on the highway rather than in manufacturing plants, warehouses, or
stores. The survey questionnaires must allow sufficient flexibility to pick up on the current
* It is not easy to get participation from private businesses.
* Limit the number of issues covered in the survey so that sufficient depth of understanding
on those issues can be achieved.
* Survey the right person. A transportation manager will generally know the most about
decisions about how goods are shipped and received, while an accountant of the company
will be able to say something about impact of congestion on costs and profitability.
* Explore the reasons why transportation bottlenecks are a problem and how they affect the
* Nonrecurring congestion is a significant problem, and access to the maj or highways was as
important as level of service on the maj or highways.
The feasibility and an estimation of the potential for using Urban Distribution Centres
(UDCs) in the city of Dublin were studied using two surveys (Finnegan, et al., 2004 and
Finnegan, et al., 2005). UDCs can fulfill a number of functions including warehousing,
transshipment, consolidation of loads, efficient dispatch and collection of goods. Several
associations or groups including Dublin City Centre Business Association have participated
consultation meeting to indicate freight transportation issues in Dublin City Centre prior to the
surveys. The fist survey was deployed on a weeklong survey of deliveries to Trinity College
Dublin. The intercept survey method was used at the gate of the campus, resulting in an 82%
response rate (299 participated out of 365 requested). The second survey was distributed to
several trade association members, capturing 906 individual deliveries with a 10% response rate.
Many aspects of freight movement in the Dublin City Centre (time of delivery arrival and
departure, types of goods, how packaged, quantity of packages, types of delivery vehicles, who
supplied the goods, location of supplier, loading and unloading places) were yielded by the
surveys to discover the use and location of a UDC. Some conclusions were made at this point;
Food related deliveries are expected to be improved by the use of a UDC in Dublin, a UDC
could assists in the process of backhauling, the operation of a delivery platform in the city centre
was suggested. But, no definite recommendations for the use, location, and impact of UDC were
The Southwestern Pennsylvania Regional Planning Commission (SPRPC) conducted a
broad mail survey of freight transportation users and service providers about various issues in
freight transportation (SPRPC, 1996). The surveys were drafted under the guidance of the
Freight Forum and distributed to the 700 freight service providers from SPRPC's Freight
Transportation Database, and 800 area manufacturers listed in the Southwestern Pennsylvania
Regional Development Council's Computer Assisted Product Search database (CAPS). The
postage-paid surveys were sent to them by mail. The original response rate was 4%, but the final
response rate was 9% after follow up phone call efforts. With various types of freight service
providers, and manufacturers answering the survey, 22% of manufacturers indicated that they
utilize intermodal transportation, and 42% of service providers said that they do. Specific
impediments to the respondents with locations and comments were also identified with their
percent frequencies; traffic congestion (46%), rush hour deliveries (32%), roadway turning
radius (25.5%), turning at traffic lights (24%), poor bridge or tunnel clearance (18%), curfews on
high or wide roads (17%), merge lanes (16.5%), at-grade railroad crossings (14%), lack of
receiving areas at malls (13%), lack of trailer drop-off and pick-up (12%), poor truck access to
airport air cargo area (1 1%), lack of adequate warehousing (8%), delays or other problems at
customs (5%), poor truck access to river terminals (4.5%), poor truck access to intermodal
facilities (4.5%), lack of rail to highway access (3.5%), poor signage (3%), highway interference
with railroad (2%). Women's compensation and other labor costs, regulations, and taxation were
also introduced to be pressing issues affecting the freight transportation industry.
In 1998, Regan and Golob (1999) studied the perceptions of motor carriers about traffic
congestion, congestion-relief policies, usefulness of information technologies, and efficiency of
intermodal facilities in California through computer-aided telephone interviews (CATI). A total
of 5258 freight operators, including California-based for-hire trucking companies, private fleets,
and for-hire large national carriers were chosen from the databases maintained by Transportation
Technical Services, Inc, producing an overall response rate of 22.4% (1 177 responses). Most
responding operators believe that traffic congestion will get worse over the next five years. Over
half the respondents indicated significant or maj or problems of increased fuel and maintenance
costs due to stop and go traffic, high numbers of accidents and insurance costs, driver
frustrations and morale, and scheduling problems due to unreliable travel times. Similarly, more
than half indicated that stop and go driving, speeders and other traffic violators, and poor road
surface quality are important causes of loss of equipment, damaged goods, or even injury to
drivers. Almost 90% of the freight operators also replied that at least sometimes schedules are
missed, drivers are re-routed due to congestion, or customer time-windows force drivers to work
in congestion. Preferred congestion relief policies included adding more freeway lanes, truck
only lanes on freeways or arterials, better traffic signal coordination along the arterials, and so
forth. The surveyed operators perceive that dedicated highway advisory radio, traffic reports on
commercial radio stations, and face-to-face reports among drivers at truck stops and terminals
are somewhat useful by drivers on the road, while traffic reports on television and computer
traffic maps on the Internet are very useful to dispatchers. With respect to intermodal operations,
about 45% of the operators pointed out that operations of carriers are often or very often
impacted by congestion or other problems at maritime ports, while only 25% of them indicated
this for airports or rail terminals.
The American Trucking Association (ATA) surveyed 470 stratified, randomly selected
private and for-hire motor carriers based in the Baltimore region (ATA, 1997). It yielded a
response rate of 13.1% (62 returned surveys) after two times mailing and follow-up calls. The
ATA worked with the Baltimore Metropolitan Council and the Maryland Motor Transportation
Association for the survey questions to check out various needs of transportation users in
Baltimore region. The survey questions included company characteristics, major routes of
travel, impediments in freight flows, infrastructure improvements needed, downtown freight pick
up and delivery, time of day travel, freight origins and destinations, intermodal freight activity,
company future plans. The survey results describe facts and motor carriers' perceptions about
freight transportation in the Baltimore region. Especially, traffic congestion was thought to be
one major structural impediment to freight movement. Intersection design/function, ramp
design, tools, constructions projects were listed by some motor carriers as well.
In 2001, two surveys were conducted to obtain the opinions of the public and large truck
drivers on road safety issues (Center for Public Policy, 2001). A computer-assisted telephone
surveys produced 2415 samples of randomly selected adult residents from Virginia (n=602),
Maryland (n=600), North Carolina (n=610), and West Virginia (n=603) with a cooperation rate
of 52%. For truck drivers, an intercept survey was used at three truck stops (Lee Hi Truck Stop
(n=206), Truck Stops of America (n=3 18), and White' s Truck Stop (n=102) on interstate 81 and
95, producing 618 samples. The surveys discovered some overall aspects of truck drivers and
difference in perspectives on safety issues from truckers and public respondents. The main
findings from the surveys are as follows:
* More than 70% of truckers in the survey were company drivers opposed to owner-
* Most truck drivers (about 70%) get paid based on miles driven.
* More than 90% of truckers in the survey are driving tractor-semi trailers.
* More than 80% of truckers spend 3-7 days away from home every week.
* Most truckers take time for a sleep break at private truck stops or public rest areas rather
than motels or roadside.
* Both truckers and public respondents perceived the highways driven on most often by
them to be somewhat safe.
* Truck drivers and public respondents tend to attribute conflicts or crashes between cars and
trucks to each other.
* Both truckers and public respondents agreed that the driving habits of large bus drivers are
considered to be the most safe and least aggressive around cars.
* Both truckers and public respondents agreed that drivers of large trucks drive somewhat
aggressively around cars.
* Truck drivers get information of crashes between cars and trucks mostly from other truck
drivers or citizen band (CB) radio, while public respondents get it from television or
Truck drivers' and motorists' opinions of such restrictions at three sites (I-5 Southcenter
Hill, SR-520, and I-5 Southbound to Tacoma Mall) in the Puget Sound region of Washington
State were obtained through surveys in 1992 (Koehne, et al., 1997). The trucker survey was
performed at two truck stop locations (one day at each location), slightly more than four months
after the last restrictions were put into place on I-5 Southcenter Hill, yielding 129 completed
surveys. A mail-back survey was distributed to motorists who traveled each of the three
restricted sections of highway more than three months after the last restrictions. About 400
license plate numbers were collected from each site for this purpose. The survey produced 153
completed responses (response rate of about 16%). The maj or findings are as follows:
* A relatively high 3 1.4 percent of truckers indicated that they had disobeyed the lane
restrictions, while about 78 percent of motorists indicated that they have seen truckers
* Only about 32 percent of the truckers are in favor of keeping Puget Sound lane restrictions,
while 91 percent of the motorists are in favor of them. The negative view of most truckers
toward restrictions could also be simply because they believe they are not necessary since
trucks rarely use the leftmost lanes on ascending grades.
* About 65 percent of truckers and motorists indicated that it is not clear which vehicles or
which lanes are subj ect to the lane restrictions.
* Only about 30 percent of the truckers believe that lane restrictions improve freeway
operations, while 86 percent of the motorists do.
* Only about 3 1 percent of the truckers believe that lane restrictions improve safety, while
82 percent of the motorists do.
* About 66 percent of the truckers believe that lane restrictions should include buses, and 74
percent of the motorists do.
* Only about 21 percent of the truckers believe that lane restrictions should be expanded to
more freeway sections, while 83 percent of the motorists do.
Three logit models were also developed for exploring truckers' and motorists' favorability
and awareness of truck restrictions. There is a profile of a trucker who is least likely to favor
truck restrictions, one who admits to violating restrictions, frequently changes lanes to avoid
rough pavement, typically carries nonperishable cargo, is between 20 and 40 years old, and has
been licensed for many years. Motorists most likely to favor restrictions also fit a definite
profile; one who frequently changes lanes to avoid being followed by trucks, typically drives a
passenger car, is between 30 and 45 years old, and has been a long-time licensed driver. The
motorists most likely to be aware of Puget Sound truck lane restrictions are male passenger car
operators who have been licensed relatively few years. Efforts to improve motorist awareness
should focus on those motorists who do not fit this profile.
A study by Golob and Regan (2002) investigated the perceptions of truck company
managers on usefulness of different source of traffic information to trucking operations.
Managers from 1177 trucking companies including 34% private carriers and 66% for-hire
carriers were asked how useful they consider different sources of traffic information are to the
dispatchers and to their drivers. The surveys were distributed to 5258 companies containing 804
California-based for-hire companies, 2129 California-based private carriers, and for-hire large
national carriers based outside of California, overall response rate of 22.4%. The relationships
between 6 characteristics of the companies (load type, carrier type, primary service, location of
logistics manager, intermodal operations, and average length of load moves) and manager-
perceived usefulness of different source of traffic information to dispatchers or drivers, are
discovered through canonical correlation analysis. The respondents were asked to evaluate the
sources in one of the three categories (i.e., very useful, somewhat useful, and not useful).
With respect to the overall usefulness to dispatchers, "reports from the drivers on the road"
are judged to be most useful followed by "traffic reports on the radio." Least useful was "traffic
reports on television", followed by "internet traffic maps." "Reports from their own drivers" are
valued most highly by carriers with either rail or multiple intermodal operations, and "traffic
reports on commercial radio stations" are useful to general LTL carriers and operations based in
the Greater Los Angeles Area. "Traffic reports on television" are thought to be useful to movers,
and intermodal operations. "Internet traffic information" is judged to be useful to operators with
long moves, and "phone calls to Caltrans" are considered to be useful to operations based in
California, but outside of the two largest metropolitan areas.
Concerning overall usefulness to drivers, "changeable message signs (CMS)" was thought
to be most useful, followed by "CBR (Citizen Band Radio) or other radio reports from other
drivers." "Traffic reports on commercial radio station" were considered as useful as "face-to-
face reports among drivers", while "Dedicated highway advisory radio (HAR)" was rated to be
least useful for drivers. "Face-to-face reports among drivers at truck stops and terminals" are
useful with rail intermodal operations, and "CMS" was useful to common carriers and operations
from outside of CA carriers with long load moves. "CBR reports from other drivers" are deemed
to be useful exclusively to truckload carriers, and "HAR" is judged to be useful to common
carriers or carriers with long load moves.
As far as future traffic information sources are concerned, "HAR and CMS" are considered
to be most useful followed by "in-vehicle navigation system." "In-vehicle navigation",
"computer traffic map", "CMS", and "traffic information kiosk" are expected to be useful to
operations from outside of CA, and carriers with long moves, while "HAR" is thought to be
useful for carriers with both truckload and LTL in the future.
Crum, et al. (2001) developed a conceptual level of commercial motor vehicle (CMV)
driver fatigue model by reviews of the 55 literature and focus group meetings with industry
professionals. It was later used for developing a survey for truck drivers to explore driving
environment effects on driver fatigue and crash rates. The model categorized three factors
influencing driver fatigue and crashes; CMV driving environments, economic pressures, and
carrier support for driving safety. The CMV driving environments were subcategorized into
three issues as regularity of time, quality of rest, and trip control, under which a total of 25
individual measures fall, while fatigue and crash outcome measures included 15 items. The data
for analysis were collected at five truck stops each in different states, yielding 502 usable
surveys. A $10 cash inducement was offered to participants in 1999 with the assistance of the
National Association of Truck Stop Operators (NATSO). Twelve driving environment
indicators were found to be meaningfully related to 15 fatigue and crash outcome measures; two
regularity of time items, six measures of trip control, and four items for quality of rest. Factor
analysis identified three constructs underlying the 15 fatigue and crash measures; close call due
to fatigue, the perception of fatigue as a problem for self and other drivers, and crashes. From
the regression analysis, "long load time" and "start workweek tired" is found to be significant in
increasing frequency of close calls due to fatigue, while the frequency of the use of "6-hour time
zone" was negatively related to "close calls" unexpectedly. "Long load time" and "start
workweek tired" also were associated with more fatigue, while more "uninterrupted hours of
sleep", more use of "regular routes", and more times "driving the same hours" were associated
with less fatigue. More "average stops per day" and "start workweek tired" were found to
significantly increase the number of crashes.
At the front part of this study, it was decided to use focus group and survey methods in
obtaining the perceptions and opinions of the trucking community. Thus, some previous studies
relevant to this issue were reviewed in this section to search for an effective way to conduct
focus group and survey studies for the obj ective of this study. The authors of the studies provide
their experiences and recommendations about focus group preparation, survey data collection
and perceptions of the trucking community on various trucking-related issues as results. For the
sake of this study, potential Florida trucking community focus group and survey participant
recruitment sources, and advantages and disadvantages of various trucking community survey
methods are summarized in the following three sections, based on the findings from this section.
2.4 Florida Trucking Community Focus Group Participant Recruitment Sources
It is generally difficult to invite truck drivers to a meeting at one place and time. They
usually spend a significant portion of their time driving on the road, and their schedules are apt to
change for time-variant demand for deliveries. One survey on truck safety issues at three truck
stops (Center for Public Policy, 2001) showed that more than 80 percent of the total of 618
surveyed truck drivers spends 3 to 7 days away from home every week for deliveries. Given
these constraints, following potential recruitment sources were found for truck driver focus
* Florida Trucking Association (FTA) Road Team members
* Florida Truck Driving Championship (FTDC) participants
* Florida-based National Truckers Association (NTA) members
* Truck drivers from one or two trucking companies
Many previous trucking-related studies have benefited from the cooperation of national or
state trucking associations. The American Trucking Association (ATA) is a maj or national
association for United States trucking industry, which often provides assistance to the
researchers, even conducting research studies by itself (ATA, 1997). Most states have either the
state trucking association or motor vehicle carrier association, which is the state division of
ATA. The Florida Trucking Association (FTA) is in that category.
A focus group study was conducted in Toronto at the annual convention of the Ontario
Trucking Association (OTA) with its Road Knight Team members, which consist of 10
professional truck drivers (Hostovsky and Hall, 2004). FTA Road Team is the equivalent of the
OTA Road Knight Team. The FTA Road Team includes 8 professional drivers from 6 different
companies. They are highly informed professionals (each with more than 15 years of truck
driving experience) and care about their industry and profession enough to take time from their
daily jobs to speak at any public gatherings and give safety demonstration to the general public.
The Florida Truck Driving Championship (FTDC) is an annual competition of truck
drivers' driving skills and knowledge on how to operate trucks safely and efficiently. Each
participant competes for the championship in one of eight classes of trucks (i.e., single truck,
three-axle, four-axle, five-axle, tank truck, flat bed, sleeper berth, and twins). The champions of
this event represent Florida truck drivers in the annual National Truck Driving Championship
(NTDC). Florida-licensed truck drivers, who performed regular duties of a full-time professional
truck driver with no accident history for at least a year and no criminal record in the past 5 years,
are eligible to participate in the competition.
The National Truckers Association (NTA) is an organization designed to protect and
support the trucking business of independent (owner-operator) truck drivers. For the reason,
most NTA members are independent truck drivers.
One or two big-sized trucking companies may help get their drivers together for a focus
group meeting. The following potential sources of ticking company contacts were found:
* FTA membership directory
* Center for Economic Development Research (CEDR) Data Center (ES202)
* Florida Department of Highway Safety & Motor Vehicles (FDHSMV)
* Transportation Technical Services, Inc (TTS)
A total of 180 Florida-based trucking companies are listed in the FTA membership directory.
They are categorized by 5 chapters (geographical locations), or 6 conferences (carrier types).
One study (Lawson, et al., 2002) utilized the Oregon Employment Department database (ES202),
Oregon Department of Transportation (ODOT) Motor Carrier Transportation Division truck
registration database, and ODOT Driver and Motor Vehicle Services Division Commercial
Driver's License (CDL) database to survey freight shippers and carriers. Likewise, ES202
database from CEDR Data Center or FDHSMV database can be used to contact the trucking
companies in Florida. The information by TTS was used in the Golob and Regan (2002) study.
It includes national motor carrier directory, private fleet directory, and owner-operator directory
database. FDOT personnel also recommended the following companies as potential sources of
truck driver participant recruitment:
* Watkins Motor Lines, Inc (one of the nation' s largest LTL carriers)
* Landstar Systems, Inc (a big logistics and transportation provider)
* Roundtree Transport & Rigging, Inc
It seems to be also difficult to recruit manager-level participants for focus group meetings.
A study by Morris, et al. (1998) performed 13 industry sector focus groups, but with only 2-4
participants per group. Some of them also used a conference call. They seem to be tight in their
time schedules. The following potential recruitment sources were found for truck company
manager focus groups.
* FTA Leadership Conference participants
* FTA Safety Management Council (SMC) members
* Truck company managers from two or more trucking companies
In 2004, Veras, et al. (2005) conducted 2 focus groups with truck dispatchers as a part of
the Evaluation Study of the Port Authority of New York and New Jersey's Value Pricing
Initiative. FTA Leadership Conference is held annually to support and enhance trucking
business for Florida trucking community. Florida-based Truck company owners and managers
are the main attendees in this event. SMC members consist of professional safety managers from
FTA member companies. Their primary goal is to make trucking in Florida as safe as possible.
It is also possible to contact some of the trucking companies by using the same sources presented
earlier to recruit truck company managers.
2.5 Trucking Community Survey Methods
In-field survey methods have often been used by other researchers to collect truck driver
surveys. The perceptions of truck drivers on roadway safety issues (Center for Public Policy,
2001) were surveyed at three truck stops (Lee Hi Truck Stop, Truck Stops of America, and
White' s Truck Stop on interstate 81 and 95), producing a total of 618 surveys. A truck driver
survey by Koehne, et al. (1997) was performed at two truck stop locations (one day at each
location), yielding 129 completed surveys. Another study about truck driver fatigue and crash
rates (Crum, et al., 2001) conducted surveys at 5 truck stop locations: Maryland, Georgia,
California, Iowa, and Colorado. The National Association of Truck Stop Operators (NATSO)
assisted with the study, producing a total of 502 usable surveys (overall effective response rate
was 97.3% with $10 cash incentives). A week-long survey of deliveries to Trinity College
Dublin (TCD) was also conducted with an intercept survey method at the gate of the campus,
resulting in an 82% response rate (Finnegan et al., 2004). Surveying truck drivers at rest stops or
truck stops have following benefits:
* It usually shows good overall productivity with some types of incentives. The completed
surveys are collected promptly on site.
* It asks for the perceptions of truck drivers on truck operation issues while they are on a trip,
so their experiences are fresh.
* It enables face-to-face interactions between the participants and surveying staff. This helps
yield more completed surveys and reduce the risk of the participants' misunderstanding of
However, it is required for the surveying staff to spend a fair amount of time and effort in the
Considering the irregular working hours and j ob characteristics of the truck drivers, it
would be effective to distribute the surveys directly to them. It may be possible that written
surveys are filled out by truck drivers where they can take time to fill them out. At places where
truck drivers stay for a short time, written postage-paid surveys could be distributed to them so
that they can return them later by mail. Otherwise, written postage-paid surveys may be sent to a
number of ticking companies that can distribute the surveys to their drivers. The sources of
contacting trucking companies are listed in the previous section.
Different survey methods can be considered for surveying truck company managers.
Two studies by Lawson and Riis, and Lawson, et al. (2001 and 2002) compared several methods
to survey the freight community (i.e., shippers and carriers) by a literature review and a pilot
study. It was found that the most effective survey method in terms of response rates is a phone-
based survey although a written survey is preferred by the freight firms. In the pilot study, the
highest response rate (60-64%) was achieved by a telephone interview survey (with five
callbacks). The other study (Regan and Golob, 1999) obtained a response rate of 35% with a
phone survey. However, the response rates for previous written surveys of the trucking
community only ranged from 8-24% (Lawson and Riis, 2001).
General characteristics of the survey methods are described in Table 2-1. Mail-based
surveys are less costly and time-consuming while producing relatively low response rate. It
would be helpful to contact the potential respondents in person or by phone to improve the
response rate. In one study (SPRPC, 1996), follow-up phone calls resulted in an 5% increase of
the response rate (from 4% to 9 %). The other study (ATA, 1997) showed a response rate of
13.1% by mailing twice with follow-up phone calls. The surveys were personally distributed to
the potential respondents to improve the response rate in another study (Finnegan, et al., 2005).
This approach also enabled to obtain helpful feedback and comments about the survey issues
directly from them. The survey questions should be clear and simple to reduce the possibility
that the respondents misunderstand them. A postage-paid survey format is typically used to
reduce the respondents' efforts to return the surveys.
Phone-based surveys usually require more money and effort while yielding relatively
high response rate. The two studies (Lawson and Riis, 2001 and Lawson, et al., 2002) indicated
that a traditional structured phone survey is useful in obtaining detailed information about
specific issues, but is not suitable for a sample size large enough to represent the views of a big
population. The Computer-Aided Telephone Interview (CATI) allows for a large sample size
while reducing the opportunity to fully explore a certain aspect of any given topic. The CATI is
typically conducted by survey companies because it requires trained interviewers and interactive
CATI computer systems. The participants' responses are keyed directly into a computer and
administration of the interview is managed by a specifically designed software program. The
program does not accept invalid surveys. In a survey study of freight operators (Regan and
Golob, 1999 and Golob and Regan, 2002), the CATI was conducted by Strategic Consulting and
Research, an Irvine, California-based survey company, yielding a response rate of 35%.
The Web-based survey method is less costly and easy to administer, but the response rate
from this method may greatly depend on publicity and advertisement of the surveys. Some types
of contacts (e.g., phone, email, mail) with potential respondents are highly encouraged to
increase the response rate. Little is known about the use of this method for surveying the
trucking community. The Idaho Technology Transfer Center, in conjunction with the ATA, is
conducting a web-based study about the perceptions of truck drivers or fleet managers on the
benefits of anti-icing chemicals and reductions of potential vehicle corrosion (Alexander and
Regardless of the survey methodss, a clear indication of the contributions) of a survey
study to the trucking community or industry may encourage the potential respondents to
participate in the survey. The sponsorships of trcking-related associations or institutes (e.g.,
FTA, ATA, or FDOT, etc.) may help reinforce the importance of the survey study, resulting in
an improvement of the response rate as well.
Table 2-1. Comparison of Survey Methods
* Relatively high
* Chance to correct
* Chance to get more
* No respondents' efforts
required to retum
0 More costly and time-
0 Dependent on
o Not suitable for large
0 Can not be used for
0 May present lack of
* Less costly and time-
* Interviewer bias is not
* Uniform survey method
* Provide respondents
with enough time to
give thoughtful answers
* Suitable for obtaining
larger and more
* Relatively low response
* Potential long time
* Hard to ensure that the
right person will
complete the survey
misunderstanding of the
questionnaires by the
* Respondents' efforts
required to return
Lawson and Riis (2001)
Lawson, et al. (2002)
Finnegan, et al. (2005)
* Less costly, easy to
* Fast results
* Provide respondents
with enough time to
give thoughtful answers
* No respondents' efforts
required to return
* Response rates may
greatly depend on
of the surveys
* Hard to ensure that the
right person will
complete the survey
Alexander and Moore
Studies in which
Lawson and Riis (2001)
Lawson, et al. (2002)
Regan and Golob (1999)
Golob and Regan (2002)
This exploratory study was aimed at discovering the factors important to estimate LOS on
existing roadway systems, as perceived by truck mode users. This was intended to provide the
FDOT with the information about the determinants of truck LOS and the levels of their
significance with which it can develop methods to effectively evaluate LOS provided for trucks
on various roadway facilities. To accomplish this, it was necessary to obtain the perceptions and
opinions of the trucking community on what factors are important for the quality of a truck trip
on various transportation facilities and the relative significance of each of those factors. Focus
group and survey studies were conducted to satisfy this requirement.
Truck drivers and truck company managers are the two maj or stakeholder groups for the
LOS provided for trucks. Truck drivers deliver goods by driving trucks while truck company
managers operate the trucks and drivers to make a profit. Truck drivers are the most important
group concerning truck LOS, in that they are the ones who actually drive the trucks on the road,
so their performance and satisfaction is directly affected by the LOS provided by various
transportation facilities. Their performance levels also have major effects on trucking business
in such aspects as on-time performance, operating cost, etc. Truck company managers are
primarily concerned with those trucking business concerns, but in most cases they need to
manage their fleet on existing roadway facilities based on the perceptions and opinions of their
drivers due to the lack of regular truck driving experience. Thus, the focus was on the
perceptions of truck drivers, while the perceptions of truck company managers were also sought
to be compared with those of truck drivers.
Initially, very limited knowledge was present on the factors determining LOS perceived by
the trucking community. This required a small number of observations to be obtained through
some type of qualitative study to get enough information to develop a more formal survey, which
would then be used to obtain the perceptions of a larger sample to represent the trucking
community. Typically, there are two types of qualitative study methods; personal interviews and
focus group studies. Individual interviews can require a considerable amount of time and effort.
A respondent is also apt to be unwittingly influenced by an interviewer, or, often not able to
come up with his/her opinions about various aspects of the subj ects during the interview.
Homogeneous and information-rich participants in a focus group study can boost the diverse
conversation about the subjects with little guidance from a moderator. Thus, focus group studies
were performed for this study.
Focus group studies do not provide enough observations for quantitative analysis. Thus, a
follow-on survey study was conducted to confirm the focus group findings with a broader
audience and measure the relative importance of each factor quantitatively. The surveys focused
only on potentially relevant and important items.
Based on the focus group and survey studies, the guidelines for developing truck LOS
estimation methodologies were developed to provide the FDOT with potential service measures
for truck LOS and/or a list of factors that should be considered to develop truck LOS estimation
models on various roadway facilities. The overall description of the research approach of this
study is presented in Figure 3-1.
To develop actual truck LOS estimation models in future studies, experimental data should
be collected to quantitatively measure the correlations between the service measures and/or the
list of factors identified in this study and the perceptions of a representative sample of truck
drivers on trip quality. This often requires some experimental efforts with in-field driving, video
simulation, or driving simulator methods.
3.1 Focus Group Sessions
The primary obj ective of this focus group study was to identify the roadway, traffic, and
control factors that are important to the trucking community for truck trip quality on various
transportation facilities and to explore the perceptional differences between truck drivers and
truck company managers on the relative importance of each factor.
Ultimately, this study sought to inform transportation service providers of what should be
focused on to better accommodate truck traffic on existing roadway facilities. Thus, the factors
that cannot be controlled by the transportation service providers were of less interest in the
discussions, although the discussions were open to any factor important to the quality of a
complete truck trip from origin to destination. The focus was on the operational and design
policies and issues relative to truck operations on various roadway facilities (e.g., lane widths,
traveler information systems, etc.) as opposed to truck industry regulation issues such as number
of continuous hours of driving, maximum non-permit weight loads, etc. Weather-related factors
were also less of a concern.
Three focus group interviews were conducted to elicit the factors affecting LOS on various
transportation facilities perceived by truck mode users (truck drivers and truck company
managers). The participants for each focus group session were recruited by the cooperation of
the Florida Trucking Association (FTA). Several discussion topics were selected carefully for
the overall obj ective of this study, which is to Eind out what should be focused on to better
accommodate truck traffic on the existing roadway system. During each focus group session, the
topics were introduced to the participants with several open-ended questions so that they discuss
each topic amongst themselves with only a little guidance from the moderator. The focus group
discussions were transcribed and summarized to be used as inputs to a follow-on survey.
3.1.1 Participant Recruitment
Based on the review and considerations of focus group participant recruitment presented in
Chapter 2, it was decided that the most effective and efficient approach to recruiting candidates
for the focus group sessions would be with the cooperation of the FTA. With the FTA' s contacts
and presence in the industry, they would be much more capable of identifying willing
participants for this study. Thus, assistance from the FTA was solicited (Appendix A for a
cooperation request letter), and they were happy to assist. FTA staff were pleased to hear that
the FDOT was conducting a research proj ect targeted at the LOS needs of the trucking
For the focus group participant recruitment, several documents were provided to the FTA
to inform them of preferred participant characteristics, what questions would be asked
throughout the sessions, and how the sessions would be conducted. The documents included
focus group instructions (Appendix B), guidelines for participant selection (Appendix C), and a
focus group moderator's guide (Appendix D). For the truck drivers' focus group sessions, the
FTA recruited members from its Road Team. Two other truck driver candidates were also
recruited. Once they provided a list of these candidates, a follow-up recruitment letter was
emailed to them to ask for confirmation. Focus group instructions and a map to the meeting
place were attached to the email as well. Two focus group sessions were held with the truck
drivers as follows:
* 5 people including 4 FTA Road Team members, 2.5 hours of discussion, on November
* 4 people including 3 FTA Road Team members, 2.5 hours of discussion, on December 8th,
It was initially planned to also hold two focus group sessions with truck managers, but the
FTA found the recruitment of these individuals to be much more difficult. Ultimately, just one
2-hour focus group session was held with three managers on November 17th, 2005.
3.1.2 Participant Selection
When the FTA agreed to help recruit the participants for the focus group meetings, the
guidelines for selecting the participants were developed and provided to the FTA (Appendix C).
This was intended to ask the FTA to consider the guidelines to obtain a representative sample of
the Florida trucking community. They describe characteristics of eligible participants and the
desirable participant composition for each focus group. This section explains the reasoning
behind the development of the guidelines.
Initially, a total of four focus group interviews were proposed (two with drivers and two
with managers), balancing the scope and resources available for this study. The two distinct
groups of participants were separately invited to different meetings because they were expected
to have different perspectives with regard to the discussion topics. Truck drivers may show more
concerns about the traffic, roadway, and/or control factors affecting truck driving comfort and
amount of earning while truck company managers may be more concerned with the factors
contributing to their trucking businesses. The separation of the two groups not only made the
participants comfortable to share their opinions, but also helped the research team to clearly
identify the perceptions of each group. The desired number of participants in each focus group
was originally 8-10. However, it proved challenging for the FTA to obtain this number of
participants. Thus, about half this number of participants for each focus group was obtained.
Nonetheless, these group sizes worked out well given the depth and breadth of the participants in
each session, and that each of the participants had much to contribute to the discussions.
There are several maj or socio-economic or working characteristics of the trucking
community that may be highly correlated with their perceptions on the factors affecting LOS on
transportation facilities. It is not realistic to take into account all the characteristics of the focus
group participants with the several focus group meetings, but consideration would help recruit a
more representative sample of the Florida trucking community. The following characteristics of
the trucking community were considered in the guidelines:
Hauling distance: Long-haul trucking with frequent travel on the Florida' s SIS facilities
was of main interest in this exploratory study as opposed to local or short-haul trucking, which
only use small portion of the Florida' s SIS facilities.
Carrier type: It was desired to include the participants from both for-hire and private
truck companies. Private trucks moved 32.7 percent, and for-hire trucks transported 43.3 percent
of the total freight value originating in Florida in 2002 (BTS, 2004). The rest of the percent (24
percent) was delivered by other modes such as train, ship, plane, or multiple modes. One survey
study (Golob and Regan, 2002) verified the perceptional difference between private and for-hire
truck companies about various sources of traffic information.
Load type: It was desired to include truck drivers from both Truckload (TL) and Less-
than-Truckload (LTL) carriers. It is likely that the perceptions of TL drivers on truck trip quality
issues are different from those of LTL driver due to the weight of the equipment and goods. The
TL and LTL operations are defined as follows:
* TruckLoad (TL): The quantity of freight required to fi11 a truck. Usually in excess of
10,000 pounds. When used in connection with freight rates, the quantities of freight
necessary to qualify a shipment for a truckload rate.
* Less-than-TruckLoad (LTL): A quantity of freight less than that required for the
application of a truckload rate. Usually less than 10,000 pounds and generally involves the
use of terminal facilities to break and consolidate shipments.
Truck type: It is likely that perceptional difference among the drivers operating different
types of trucks exists due to their discrete size and operational characteristics. The majority of
the trucks on the roads were categorized into the following three types of trucks in terms of truck
* Straight Truck (single unit truck)
* Truck-Trailer (truck-trailer, truck-double trailer)
* Tractor-Trailer (tractor-semitrailer, twin-trailer, rocky mountain double, turnpike double)
It was desired that each truck driver focus group include at least three truck drivers, each
operating different three types of the trucks.
Fleet size: It was desired to include truck drivers that represented companies with a
variety of fleet sizes. About 87 percent of the carriers in the U. S. operated 6 or fewer trucks, 9
percent operated the fleet size between 7 and 20, and 4 percent operated more than 20 trucks in
2004 (ATA, 2005).
Others: ATA (2005) showed that 29.4% of the total drivers in the U.S. were minorities in
2003 and the percentage has been going up gradually from the year 1996. Only 4.6 % of the
total truck drivers in the U.S. were women in 2003, but the percentage has been going down
since 2001. So, inclusion of the minority or woman truck drivers in focus group sessions was
desired, but this was obviously difficult to attain, given their limited proportions.
Overall, a good composition of participants was recruited for the two truck driver focus
groups in terms of carrier type and primary load type. The participants were all from major
trucking companies with at least 16 years of truck driving j ob experience. The truck types they
operate include various tractor-trailers. However, no straight truck driver, truck trailer driver,
female driver, or minority driver was present on the meetings. Only one truck company manager
focus group was conducted with three managers. They were all from the TL carriers with at least
5 years of truck company manager experiences. One was from private company while the other
two were from for-hire companies. A detailed description of the focus group participants'
backgrounds is provided in Appendix E.
3.1.3 Focus Group Questionnaires
For each focus group interview, several discussion topics were introduced to the
participants by the moderator through several open-ended questions. Then, the participants
talked about the topics amongst each other with a little guidance of the moderator. An identical
set of questions was presented for each meeting to compare and contrast the perceptions of truck
drivers with those of truck company managers. The issues to be covered were selected according
to the overall obj ectives of this focus group study, which is to investigate which factors
transportation service providers should focus on to better accommodate truck traffic. Following
four topics were considered as most relevant to this study, therefore were covered during each
focus group meeting.
Truck travel route and departure time selection:
* Who is responsible for selecting a truck travel route and departure time for a delivery?
* When selecting a travel route and departure time for a delivery, what factor are considered
and what is their relative significance?
Truck trip quality on various transportation facilities:
* When you drive a truck on Florida's roadway facilities (i.e., freeways, arterials, and two-
lane highways) for a delivery, what factors affect the quality of a truck trip and how
significant is each of those factors for it? (for truck driver groups)
* When you consider truck operations on Florida' s roadway facilities (i.e., freeways,
arterials, and two-lane highways), what factors affect the quality of a truck trip and how
significant is each of those factors for it? (for truck company manager groups)
Transportation service improvement priority for trucking industry:
* What types of transportation facilities (e.g., freeways, urban arterials) would you
emphasize most for improving truck operations in Florida?
* What are your top priorities for improving trip quality/travel condition for commercial
Truck delivery schedule reliability:
* What factors affect the truck drivers' ability to reach a destination on time?
* How often does a late delivery take place?
* What are the typical consequences for you/your company of a late delivery?
The most important contributors to quality of a truck trip perceived by the trucking
community were thought to be the factors that they consider for truck route and departure time
decision, so they were asked in the first place. The participants provided their valuable insight
about in what respect certain routes and times of day are preferred or avoided by them and which
specific factors contributes to each of those aspects. The performance levels of those factors
have most significant impacts on their trucking business and truck trip quality perceived by them.
Secondly, the participants were directly asked to list the factors affecting quality of a truck
trip. The topic was introduced by each roadway type, in that it was generally admitted in the first
focus group meeting that the importance of a factor on truck trip quality vary by which type of a
roadway it is on. For example, it was discussed that the importance of shoulder width and
condition on truck trip quality for two-lane highways was considered to be much bigger than that
on freeways. The participants also indicated how and how much each factor influences truck trip
quality based on their direct or indirect experiences.
Improvement priorities among various transportation issues and facilities were asked next.
It was intended to search for the facilities or specific factors which are getting relatively less
attention by transportation service providers compared to their relative importance on truck trip
quality. The participants' comments on this issue will be a good reference for prioritizing
transportation improvement plans.
As the truck volume and the demand for just-in-time deliveries have increased in Florida,
the importance of truck delivery schedule reliability has become critical. Thus, at the last part of
the discussion, the participants were asked to list the factors contributing to the issue and indicate
their level of impact on it. Truck company manager participants explicitly mentioned that their
customers assess the performance level of their truck companies based primarily upon on-time
delivery performance. The factors affecting on-time performance, frequency and consequences
of a late delivery were explored within this topic.
The factors identified as important to any of the topics covered in the focus group
discussion were used as a primary input for a follow-on survey so that the importance of each of
those factors could be verified by a larger audience and the relative significance among the
factors on truck trip quality could be quantitatively investigated.
3.1.4 Conducting Interviews
The focus group meetings were held at the Civil Engineering Conference Room or
Transportation Research Center Conference Room at the University of Florida. All the meetings
were moderated by a researcher who is knowledgeable to direct the discussions for the purpose
of this study. Upon the arrival of the participants, they were asked to fi11 out an informed
consent form and a two-page participant background survey (Appendix F and G). During the
main session, the moderator introduced the selected topics to the participants with several
general open-ended questions (Appendix D). Each question was displayed on a big screen to
help keep the participants focused on the issue being discussed. The participants discussed each
topic amongst each other with just a little guidance from the moderator. Although all the issues
were planned to be covered within the two hours, it turned out to not be enough time because
they had so much to say about each topic. The truck driver focus groups were extended by about
30 minutes with the permission of the participants. Participation in the focus group meetings
The audio from the focus group meetings was recorded on a laptop computer through
external microphones and a specialized software product that enabled every comment to be
associated with the corresponding speaker. The recorded audio mies (.way mies) were
transcribed into an electronic text format. The transcriptions were reviewed in detail by the
research team to ensure their accuracy. After the transcriptions were complete and accurate, the
focus group discussions were summarized and used as guidance in the development of the
3.1.5 Summary of Discussions
The results of the focus group studies were summarized by each issue covered in the
discussions. The participants did not always focus on the issue being discussed. They
sometimes jumped onto the previously discussed issues or the issues to be discussed later. As
long as the discussions remained on relevant topics, the moderator did not redirect the
conversation. Thus, it was required to collect participants' comments about each issue spread
over the transcriptions. Some of the comments were paraphrased for clarity and brevity, but care
was taken not to place any personal perceptions on the summary. The factors perceived to affect
truck trip quality were categorized by each transportation facility type (e.g., freeways). Each of
the factors was listed and described with the participants' comments about its contribution to the
truck trip quality and their direct or indirect experiences with it. The perceptions of truck
company managers were separately summarized to be compared with those of truck drivers.
The focus group meetings covered many issues in a limited amount of time. The
discussions were directed by the moderator to draw out a list of the factors important to truck trip
quality, not spending too much time on one or two specific factors. Thus, it was not appropriate
to make quantitative implications of the discussions on the relative importance of each factor. It
was investigated quantitatively in a follow-on survey, which was developed based on the focus
group discussion summary.
3.2 Survey Studies
The primary obj ective of this survey study was to verify the importance of each factor
identified in the previous focus group study with a larger audience and to quantitatively measure
the relative significance of each of those factors on truck trip quality on Florida' s roadway
system. Preference of the respondents on truck driving time of day was also investigated to
explore how quality of truck driving environment varies by time of day. The background
characteristics of the respondents that were correlated with their perceptions on truck trip quality
or preference on truck driving time of day were identified.
The survey included following three parts: working and socio-economic backgrounds of
the respondents; their perceptions on the relative importance of each factor on truck trip quality;
and their opinions about the importance of each hypothetical truck LOS performance measure.
Two sets of survey forms were distributed; one for truck drivers and the other for truck company
managers. Some questions about the backgrounds of the respondents differed in the two survey
forms, but identical sets of factors and hypothetical performance measures were evaluated in
both surveys. They were carefully selected based on the previous focus group study results.
Only truck trip quality determinants on the following three roadway types were investigated
considering the lengths and complexity of the surveys: freeways; urban arterials; and two-lane
A total of 459 truck drivers and 38 truck company managers responded to the written
surveys collected at Florida Truck Driving Championship (FTDC) event or the postage-paid
mail-back surveys distributed at four agricultural inspection stations. The survey responses were
analyzed with various statistical methods such as descriptive statistics, Exploratory Factor
Analysis (EFA), multiple comparisons of the means, non-parametric tests, and chi-squared tests
3.2.1 Question Types
When the issues to be questioned in the survey were selected, a questions type for each of
those issues was carefully determined to effectively carry out the study. It involved such various
considerations as applicability of statistical analyses, respondent burden, respondent error (or
complexity level), measurement accuracy, etc. Following fiye question types were utilized in
this survey study: interval-rating questions; ratio-scale questions; forced-ranking questions;
discrete-choice questions; and Eixed-sum questions. This sections describes characteristics
(advantages and disadvantages) of those question types based on which the question types for
each survey issue of this study were determined.
220.127.116.11 Interval-rating questions
This question type is often used to ask for the perceptions of the respondents. Respondents
are asked to present their perceptions on the level of importance, satisfaction, or agreement for
each item on an interval-rating scale (e.g., 1 = Not at all Important, 7 = Extremely Important).
The accuracy of the participants' opinions may be improved as the number of selectable points in
a rating scale increases. However, more points require the respondent to think about the
differentiation along the extended scale, increasing response burden and time to complete the
survey. It is typical to have at least Hyve points on the scale, since fewer than five points
generally do not generate normally distributed data which are usually obtained when the data are
truly interval in nature. Seven-and ten-point scales are most common. Even-numbered scale
may be used when there is a need to force the respondents to commit to one side or the other, but
it typically creates a downward bias as they tend to choose five as a neutral point in a ten-point
A critical differentiator between interval-rating and ordinal scale questions is that equal
intervals exist between each adj oining pair of response options. If all response options are
literally presented (e.g., 1 = Not At All Important, 2 = Little Important, 3 = Important, 4 = Very
Important, 5 = Extremely Important), the difference between each adj oining pair of response
options are not the same, resulting in ordinal scale data. Thus, the response values from an
ordinal scale question do not have numerical meaning, and categorical or nonparametric
statistical analyses can only be applicable to the data. It is also true for a forced-ranking question,
which is one type of the ordinal scale questions. The responses from an interval-rating scale
question are appropriate for the numerical interpretation (although not perfectly), enabling
further statistical analyses (e.g., parametric statistical analyses). However, an interval-rating
scale question allows respondents to indicate that everything is almost equally
important/satisfied, making the distinction among the factors difficult, while a forced-ranking
question require them to make clear distinctions among the factors.
18.104.22.168 Ratio-scale questions
Respondents are asked to provide their answer as a ratio to a basic unit of measurement
(e.g., year, dollar, hour, etc.) assigned to each question. The ratio scale of measurement is the
most informative scale and widely used to obtain participants' background data (e.g., age,
number of children, how long their phone call was on hold, etc.). It has zero position indicating
the absence of the quantity being measured and is an open-ended interval-rating scale, in that
there is no designated upper-end point.
The use of a ratio-scale question for respondents' perceptions or opinions (e.g., perceived
importance or satisfaction levels) may not be a good idea. It may decrease a response rate,
increase respondent error, and bring about unreasonably large variances among the responses or
participants, rather than increase level of precision in their responses with numerous numbers of
points in the scale.
22.214.171.124 Forced-ranking questions
Respondents are given a number of factors and asked to place them in order based on a
certain criterion (e.g., importance level). For example, six factors might be presented and the
respondent is asked to place a "1" next to the most important factor, a "2" next to the second
most important factor, and so on. Again, the response values from this scale do not have a
meaning numerically, so the data should be analyzed categorically or nonparametrically. These
data are often analyzed with a cumulative frequency distribution for each factor. For example,
"90 percent of the respondents perceived that factor A is at least secondly important among the
listed factors." This approach forces them to clearly distinguish one factor from others, usually
producing a larger variance among the responses than an interval-rating scale question.
A forced ranking question is heavily prone to respondent error. Respondents might
interpret it as an interval-rating question, use the top and bottom rankings more than once, or
rank the top and bottom items skipping the middle ones. In a telephone survey, you could train
the interviewers to prompt the respondent for full and correct answers. For a web-based survey,
you could insert an error feedback system, not letting the respondents move through the survey
without completing the question correctly. These practices are as likely to annoy the respondent
as to get real answers. It could also require the respondents to make a considerable effort to
complete the survey as the number of factors to be ranked increases. Rank ordering of 10 factors
is, in fact, extremely difficult.
126.96.36.199 Discrete-choice questions
Respondents are given a list of items and asked to choose one answer that best applies, or
specific number of answers that best applies, or all the choices that apply. A survey developer
decides which type of discrete-choice question is appropriate for a survey issue, considering the
characteristics of the issue, study purpose, and respondents' burden accompanied by the question.
This question type is widely used in a survey to ask for the respondents' background
characteristics as well as their perceptions on certain issues. The responses are nominal (or
categorical), so it is easier for the respondents to reply to the question than other question types.
The discrete-choice data are often presented by (cumulative) frequency distributions of each
option category. If the respondents' perceptions are asked with a discrete-choice question, logit
modeling technique can be applied where the responses data are used as a response variable and
other background characteristic data as explanatory variables. Relationships between the
perceptions and background characteristics of the respondents are identified by the logit
modeling technique, which provides an equation to predict the response once a set of background
characteristics are given. This question type is often used to ask about the topics that may be
potentially sensitive to the respondents (e.g., income level) because it is less personally obtrusive
than a ratio-scale question to the respondents. In that case, the response data can be used in logit
modeling process as a potential explanatory variable.
Multiple choices, multiple responses: Respondents are given multiple response
categories and asked to select either all the response choices that apply, or specific number of
response options that apply. The former question type is usually used for some background
characteristics (e.g., choose all the types of foods you can cook) and the latter is often used to
investigate the perceptions of the respondents (e.g., select 2 types of foods you like the most).
The response data are often analyzed by (cumulative) frequency distributions. When the former
question type is used, the data are often converted to binary data for each response category and
used as potential explanatory variables for various statistical modeling (e.g., logit, probit, or
regression modeling). The latter question type requires less respondent burden than other
question types for respondents' perceptions. The respondents only have to choose a certain
number of factors (e.g., 2, 3, or 5 factors) from a list of factors according to their perceptions.
Two or three factors are most commonly asked to be selected, but the number of factors to be
selected basically should be determined based on the purposes of a study. One study by
Washburn, et al. (2004) used this question type to identify most important user-perceived trip
quality determinants on rural freeways. The respondents were asked to select 3 factors most
important to their perception of trip quality among 16 listed factors. The importance of each
factor was presented by the percent of the respondents who selected the factor in their top 3 most
Multiple choices, single response: Respondents are asked to select only one most
appropriate response out of multiple response options. A survey designer should be careful to
direct the respondents to select only one most appropriate response, not just any responses that
may apply. If the perceptions of the respondents are asked with this question type, the
relationships between their perceptions and background characteristics can be modeled by multi-
category logit or probit modeling technique, which provides a means of predicting response
probability, given a set of background characteristics. This question type can also be used to ask
for some background characteristics of the respondents (e.g., race). In that case, the responses
are often converted to a set of binary data to be used as explanatory variables for other models.
Binary choice: The respondent is asked to select one out of two choices. Typically, these
choices are true or false, or yes or no (e.g., existence of any dependent). If the perceptions of the
respondents are asked with this question type, the relationships between their perceptions and
background characteristics can be modeled by binary logit or probit modeling technique, which
provides a means of predicting response probability, given a set of background characteristics.
This question type can also be used to ask for some background characteristics of the
respondents (e.g., gender) and defined as an explanatory variable for other models. In all cases,
the data can be easily represented by frequency distributions.
188.8.131.52 Fixed-sum questions
Fixed-sum or fixed-allocation questions are combination of the interval-rating scale and
the forced-ranking questions. Respondents are given a set of factors and asked to allocate a total
of 100 points to the factors for a certain aspect (e.g., importance level). They are encouraged not
only to consider rankings of all the listed factors before the allocations, but also to present level
of each factor for the aspect by a number. This question type is also used to obtain some
background characteristics of the respondents (e.g., percent of your trip purposes: business;
leisure; social). The block of points to allot is typically 100 since most people are comfortable
thinking in percentages, but allocated scores often tend not to add up to 100 as the number of
factors in question increases. Time-consuming data cleansing is required for the responses not
summing to 100. For this reason, it is common to present not more than 10 factors and state
what an equal weighting would be. Typically, 4, 5, or 10 factors are listed to be evaluated
because the equal weighting (i.e., 25, 20, and 10) is easily recognized by the respondents. The
response data are often presented with descriptive statistics and can be considered as a response
variable or an explanatory variable for various statistical analyses.
3.2.2 Survey Development
The survey issues covered in this study were selected for the ultimate obj ective of this
study which is to find out what should be focused on by transportation service providers to better
accommodate truck traffic on current roadway systems. They are presented in Table 3-1 with
their corresponding question types and analysis methods utilized in the survey to address those
issues. Two different survey forms were prepared (Appendix H for truck driver survey and
Appendix I for truck company manager survey). Identical sets of factors were evaluated in the
two surveys to compare the perceptions of the two groups, but most questions regarding
respondents' background characteristics necessarily differed between the two surveys.
Participant's background: Background characteristics of each participant were asked in
the first part of the survey. It was intended to discover the relationships between their
backgrounds and perceptions on truck trip quality. Thus, the characteristics suspected to explain
the potential variances in their perceptions were included in the first place.
The background survey section includes the questions about socio-economic status and
working characteristics of the respondents. Various question types were used for this section
according to the characteristics of the selected issues. Discrete-choice questions were used to
ask for socio-economic backgrounds (e.g., age, annual income) to be less personally intrusive to
the respondents than ratio-scale questions. All the response options for working characteristics
questions (e.g., types of goods hauled, truck types used, duties of truck company managers, etc.)
were determined through extensive literature search and discussions with members of Florida
trucking industry to reflect the current state of Florida trucking industry.
Preference on truck driving time of day: Truck driver respondents were asked to
present current and preferred truck driving times of day, while only preferred truck driving times
of day could be asked in the manager survey. It was intended to investigate how quality of truck
driving condition varies by time of day. This offered valuable information about the preference
of the trucking community on night-time delivery and how the preference of truck drivers on
truck driving time of day is different by their background characteristics. Time of day was
divided into 6 time periods and preference on each of the periods was asked with a binary choice
question to reduce respondent burden.
Relative importance/satisfaction of each factor on truck trip quality: Relative
importance of each factor on truck trip quality was asked simultaneously with relative
satisfaction of each factor on overall Florida roadway facilities. Importance-sati sfaction (or
importance-performance) analysis approach is often used in the field of marketing research to
prioritize attributes of a product for improvement (Martilla and James, 1977). That is,
manufacturers should firstly focus on improving the attributes of a product that are perceived to
be important, or are unsatisfied by the customers. In this study, this approach provided
transportation service providers with valuable insights about what should be firstly focused on to
improve LOS for trucks (i.e., the factors that are perceived to be important, but are not well
satisfied by truck drivers).
Basically, the factors identified in the previous focus group studies were presented in the
survey to be evaluated, but a couple of other factors that the research team considered to be
important were also included (e.g., frequency of faster vehicles following your truck). Weather
factors (e.g., thunder storms, heavy rain falls) were excluded from this survey study even though
they were perceived to be fairly important in the focus group studies, because they cannot be
controlled by transportation service providers. The relative significance/sati sfaction of each
factor on following three roadway facilities was evaluated: freeways; urban arterials; two-lane
highways. Truck trip quality issues on multilane highways were not covered in this study to
keep the proper length of the surveys. It was indicated in the focus group studies that the
trucking community was least concerned about truck trip quality on multilane highways among
those on various roadway facilities (i.e., freeways, urban arterials, two-lane highways, and
The importance/satisfaction of each factor was asked on a 7-point relative interval rating
scale (-3 = Least Important (or Satisfied), O = As Important (or Satisfied) As Others, +3 = Most
Important (or satisfied)). It was not appropriate to use typical interval-rating scale, ordinal scale,
or ranking scale types of questions. As mentioned in a previous section, a typical interval-rating
scale question allows respondents to indicate that everything is almost equally important, making
the distinction among the factors difficult. An ordinal scale or forced-ranking questions do not
allow mathematical interpretation of the survey responses, restricting the applicability of various
statistical analyses. The number of factors to be evaluated ranged from 18 to 19 by roadway type,
so the use of a forced-ranking scale question was strongly discouraged because it would
significantly increase respondents' burden and error, or decrease the response rate. Bubble-
shaped option boxes were displayed for participants to reply to this survey question easily. A 7-
point scale was used for this question, balancing the precision in measuring respondents'
perceptions and the respondent burden. A small number of points in scale decrease level of
measurement precision, while a large number of scale increase the respondent burden.
Improvement Priority Score (IP~S): Although not asked directly in the survey,
improvement priority of each of the listed factors could be assessed through a combination of
Relative Importance Score (RIS) and Relative Satisfaction Score (RSS). Improvement priority is
proportional to RIS and inversely proportional to RS~S. That is, the more important or the less
satisfied a factor was, the more the factor is in need of improvement. Based on this reasoning,
Equation 3.1 was devised to calculate the Improvement Priority Score (IPS) for each factor.
IPS =I x (RIS -RSS) (3.1)
IPS = Improvement Priority Score (-42 +42)
RIS = Relative Importance Score (1 7)
RSS = Relative Satisfaction Score (1 7)
a = +1 if RIS >= RS~S, otherwise -1
RIS and RS~S of each factor collected on an interval-rating scale of -3 to +3 were converted
to a scale of 1 to 7 for calculating IPS for each factor. The greater the IPS of a factor is, the more
improvement on the factor is anticipated. When the importance level of a certain factor is equal
to the satisfaction level, IPS will be zero. A total of 49 (7 RIS x 7 RS~S) possible responses and
their corresponding IPS are tabulated in Appendix J. One survey response presents one IPS for
each factor. An average IPS of each factor from all the responses is used to estimate its relative
need of improvement.
Relative importance of each factor on overall trucking business: Relative importance
of each factor on Operating Cost (OC), On-time Performance (OP), and truck drivers' Trip
Satisfaction (75) was asked individually on a 7-point relative interval rating scale (-3 = Least
Important, O = As Important As Others, +3 = Most Important) in the manager survey. The
percent allocation of the managers' concerns on each of these issues was also asked with a fixed-
sum question. OC, OP, and TS values were converted to a scale of 1 to 7 and relative importance
of each factor on Overall Trucking Business (OTB) was calculated by summing the OC, OP, and
TS values weighted by their corresponding portion of the managers' concerns.
Relative importance of each category of factors on truck trip quality: All the listed
factors for each roadway type were divided into four categories for each roadway type and the
relative importance of each of the categories on truck trip quality was evaluated with a forced-
ranking question (1-4, 1 = Most Important, 4 = Least Important). This was intended to
investigate which types of transportation service are relatively important to LOS perceived by
truck mode users for each roadway type and how this importance priority varies for various
roadway facilities. The relative importance of four identical categories were evaluated for
freeway and two-lane highway facilities (i.e., physical roadway condition, traffic condition,
traveler information systems, and other drivers' behavior), but 'signal condition' replaced
'traveler information systems' category to be evaluated for urban arterial facilities. A forced-
ranking question was used for this issue to focus on the distinctions among the importance of the
categories of factors. The use of an interval-rating scale question was not appropriate because
there is much chance that most respondents simply state that every category is equally extremely
Applicability of single hypothetical performance measure to estimate truck trip
quality: Based on the focus group studies, several hypothetical truck LOS performance measure
were developed by the research team for each roadway facility (e.g., Ease of Driving at or above
the Speed Limit for freeways). This was intended to investigate whether it is possible to use only
one or two performance measures to adequately evaluate truck trip quality on each roadway type.
The hypothetical performance measures for each roadway type were selected in a way they are
independent one another. That is, each performance measure presents different aspect of truck
driving condition. Each performance measure can be considered as a function of multiple
specific factors that were evaluated in a former part of the survey. The applicability of each
hypothetical performance measure solely to evaluate truck LOS was asked on a typical 7-point
interval-rating scale (1 = Not at all Applicable, 7 = Perfectly Applicable). It was expected to
obtain large enough variances for the distinction amongst the factors since it is not reasonable for
the respondents to indicate that most of the different performance measures are perfectly
applicable solely to assess truck LOS. Again, a 7-point scale was used for this question,
balancing the precision in measuring respondents' perceptions and the respondent burden.
Relative improvement priority for each roadway type: The relative improvement
priority among various roadway types was asked at the last part of the survey. It was intended to
find out which roadway facility types are more in need of improvement for the trucking
community among the four roadway facility types (i.e., freeways, urban arterials, two-lane
highways, and multilane highways). A forced-ranking question was used for this issue to focus
on the distinctions among the roadway facility types while keeping the respondent burden at a
reasonable level (1-4, 1 = Most in Need of Improvement, 4 = Least in Need of Improvement).
The survey data is beneficial for prioritizing the improvement needs of various roadway facility
types for truck traffic.
3.2.3 Data Collection
Truck drivers and truck company managers, as the maj or truck mode users on the Florida' s
SIS facilities, are the target population in this study. A truly random sample is impossible to
obtain considering expected difficulties with recruitment, time, and budget. However, an effort
has been made to get a reasonably representative sample for this study.
Based on the review and considerations of truck driver survey methods discussed in
Chapter 2, two different approaches were used for survey data collection of the truck drivers.
The first method involved the distribution of the written surveys at the Florida Truck Driving
Championship (FTDC) event. The second method consisted of distributing the postage-paid
surveys at several agricultural inspection stations.
The first truck driver survey was conducted at the FTDC event (on June 1-3, 2006 in
Tampa). This event is co-sponsored by the FTA and they assisted the research team with the
distribution of the surveys. They were given to the drivers while they were in a session where
they were required to fill out other paper work as well.
There were some concerns expressed by the research team about the original length (six
pages) of the survey, and the research team discussed this with the FDOT and the FTA.
However, the FTA representatives felt that the length would not be a problem because they
would require the drivers to fill them out as part of their participation in the event, so it was
decided to leave it at that length.
The first two pages of the survey included the questions regarding the socio-economic and
working characteristics of the respondents. The relative importance/sati sfaction of each factor
on truck trip quality on freeways, urban arterials, and two-lane highways were asked on pages 3,
4, and 5 of the survey, respectively. The last page was used to ask about the applicability of
several different performance measures being suitable as a single performance measure to
estimate truck trip quality on each roadway type (Appendix H for the truck driver survey form).
As FTA recommended, a total of 220 truck driver survey forms were provided for
distribution at the driving competition. This number was based on the number of registered
participants in the event. Of this total, 148 surveys were returned to the researchers.
Unfortunately, only 3 8 respondents out of the total of 148 respondents (25.7%) completed all
sections of the survey as directed. Most participants chose to not fill out the survey in its entirety,
even though they were instructed to do so. Most respondents may have not taken the surveys
seriously, or they might have thought that the survey was a bit long or hard for them to complete,
especially the survey sections for relative importance/satisfaction of each factor on each roadway
Given these results, it was decided to also conduct an in-field survey data collection effort.
With the assistance of FDOT personnel, postage-paid mail-back surveys were distributed at
several agricultural inspection stations. For this survey effort, a reduced version of the surveys
was prepared in hopes of obtaining a good response rate and a higher level of completion of the
surveys. The response rate statistic from the FTDC survey data indicated that the survey sections
for relative importance/sati sfaction of each factor required the most amount of respondent burden
among all the sections in the survey. Most survey participants did not have problems or
difficulties filling out the last page on applicability of a single performance measure. Thus, the
survey was reduced to three pages in length; the first page for background information, the
second page for relative importance/sati sfaction of each factor on one of the 3 roadway types
(freeways, arterials, or two-lane highways), and the last for applicability of a single performance
measure. This survey (one with freeway related factors on the second page) is included in
Appendix K. A total of 4000 postage-paid surveys were supplied to the FDOT for distribution at
the inspection stations. A total of 1000 surveys were distributed at each of four inspection
The four stations were located at northern border locations, and included:
* Pensacola station on I-10 (West)
* Suwannee station on I-10 (near Live Oak)
* Hamilton station on I-75
* Yulee station on I-95
The previous focus groups indicated that most truck drivers' concerns are on freeways or
two-lane highways as alternatives to freeways. A significant portion of their trips are also on
freeways, distance-wise as well as time-wise. Thus, of the 1000 surveys distributed at each site,
500 surveys were freeway related (i.e., the second page asked questions about relative
importance and satisfaction of factors affecting freeway quality of service), 250 surveys were
two-lane highway related, and 250 were arterial related. The research team was informed by
FDOT that only about 3 percent of all the truck drivers traveling on the Interstates could bypass
the agriculture stations. These vehicles generally include pre-pass users, car haulers, and empty
flat beds. The surveys were distributed during the week of August 14-20, 2006. A return date
of September 1st was put on the surveys. Most surveys were returned by that date, but some
were still received up to a month later. Surveys received after October 1st were not included in
the data set because data reduction was complete at that point and analysis had started. A total of
311 surveys were returned by October 1st, yielding a response rate of 7.8%. As shown in Table
3-2, the response rate for the freeway-related surveys was much higher than those for arterials or
two-lane highways. It implies that this particular population of truck drivers was more
concerned or interested about transportation services on freeways. The completion rate of these
surveys was much higher than those from the FTDC. It is likely that the reduced length and
greater flexibility of when they could fill them out contributed to this.
According to the advantages and disadvantages of the various survey methods reviewed
and discussed in Chapter 2, the phone-based survey method was considered as the most efficient
way to survey truck company managers. However, this method was not well suited for this study
because it does not allow the respondents to complete the second section of the survey without
bias. This section requires the respondents to present their perceptions on the relative
importance/sati sfaction of each factor among a total of ~18-19 factors. It would be possible only
when the respondents can skim through all the factors simultaneously.
With these considerations, truck company managers were surveyed with two different
approaches. One method was the in-field survey during the FTDC event. The other method was
the postage-paid mail-back survey with a number of trucking companies listed in the FTA
A survey data collection effort for truck company managers was made on June 1, 2006 at
the Fairgrounds in Tampa, where the truck driving championship was taking place (Figure 3-2).
There was a modest crowd present on this day, with a mix of company managers, other truck
company employees, as well as family and friends. The research team set up a table with several
chairs, covered by a large umbrella. A survey poster describing the purpose and background of
this research proj ect was placed on the table to solicit participation of truck company managers
(Figure 3-3). The announcers for the driving competition also made several announcements
throughout the day about our effort. Eleven truck company managers filled out the surveys, with
seven of them answering most of the survey questions as directed (Appendix I for the truck
company manager survey form).
For the second survey data collection effort of truck company managers, potential
recipients were identified from the FTA membership directory, which contains a list of all carrier
and allied members with their affiliation and contact information. All the Florida-based carriers
in the directory (a total of 180 trucking companies) were considered as potential survey
participating trucking companies. All the allied members were removed from consideration
because they do not operate trucks. They support trucking companies by offering various
services such as accident investigation, insurance, truck driver training, truck sales or rentals, etc.
In the FTA directory, the carriers can be divided into 5 chapters (geographical locations), or 6
conferences (carrier types). A total of 50 Florida-based carriers were selected for the survey
using the stratified random sampling procedure that preserves the proportional composition of all
the Florid-based carriers in terms of conference and chapter (a total of 30 strata).
A reduced version of the manager survey (one with freeway related factors on the second
page) was used (Appendix L). The background section of the manager survey is different from
that of the driver survey. For the rest of the sections, the same sets of roadway, traffic, and/or
control factors are presented, but questioned differently. For instance, the relative importance of
each factor on operating cost, on-time performance, and truck drivers' trip quality were asked
respectively in the manager survey, while the importance and satisfaction of each factor on truck
trip quality were asked in the driver survey.
Five postage-paid surveys were mailed to each of the 50 FTA carrier members (5 x 50 =
250 surveys) with a cover letter asking them to distribute the surveys to the managers
(transportation, safety, dispatch, logistics, etc.) at their companies (Appendix M for the cover
letter). As can be seen in the cover letter, it was mentioned that this study was being conducted
with the cooperation of the FTA. It was hoped that this would encourage managers to respond to
the survey. Follow-up phone contacts were made to each of the companies to whom surveys
were mailed (about one week later) to confirm that they received them and ask for their support
in filling them out. Initially, a total of 300 surveys were prepared in case some the 50 carriers do
not receive the surveys. Of 50 additional surveys, 5 surveys were resent to one carrier that did
not receive the survey and 45 surveys were mailed to 9 different newly-selected carriers. As a
result, 27 surveys were obtained from all the Florida regions except for West Florida. Nineteen
surveys were from common (for-hire) carriers, 6 from private carriers, and 2 from tank carriers.
Table 3-3 shows the number of carriers that participated in the survey from each conference by
each chapter, out of the total of 59 carriers. Table 3-4 indicates the number of surveys received
from each conference by each chapter, out of the total of 300 surveys.
3.2.4 Data Reduction
It turned out that most respondents completed only parts of the surveys and there was also
some evidence that some of the respondents did not pay enough attention to fi11 out the surveys
as directed. The length, or the perceived complexity, of the survey may have kept the
participants from completing them correctly. Considering that the purpose of this survey is to
potentially identify ways of improving the working environments of the participants, there is no
other source of non-response bias expected in this study. Thus, given an overall low response
rate, it was decided that in addition to complete surveys, partially completed surveys would be
utilized for data analyses once surveys with unreliable responses were screened out according to
certain criteria. The usability of survey responses for data analysis was determined for each
survey question. Survey data filtering criteria were developed to assess the validity of the survey
responses (Appendix N). However, for a few surveys, it was still difficult to determine their
validity with the filtering criteria. The research team had to make decisions in those cases
through discussions. In that process, the researchers were stricter on the validity of survey
responses collected during FTDC than of the postage-paid surveys due to the larger respondent
burden to complete the surveys distributed at the FTDC event. Tables 3-5, 3-6, and 3-7 illustrate
the number of valid surveys out of total number of returned truck driver surveys from the FTDC
event, the postage-paid surveys, and both combined. Table 3-8 describes the number of valid
surveys out of the total number of returned manager surveys from all the survey data collection
3.2.5 Data Analysis
A variety of statistical methods were used to analyze the survey data to satisfy the
obj ectives of this study. All the variables observed in the survey were first summarized by
descriptive statistics and/or (cumulative) frequency distribution to investigate overall
distributions of the responses. The following five statistical methods were applied to the survey
data for further analyses: Exploratory Factor Analysis (EFA); Games-Howell multiple
comparison test; Kruskal-Wallis test; Mann-Whitney test; and chi-squared test. Exploratory
Factor Analysis (EFA) was performed with Relative Importance Score (RIS) of all the factors to
look for common latent factors that are important to truck trip quality. Each pair of the mean
importance of hypothetical truck LOS performance measures for each roadway type was
compared with Games-Howell tests to find out which performance measure is more important
than others statistically. Potential relationship between each background characteristics of the
respondents and their perceptions on each potential truck LOS performance measure was
investigated through nonparametric tests such as Kruskal-Wallis test and Mann-Whitney test.
Chi-squared test, in particular, was performed to discover potential relationship between each
background characteristics of the respondents and their preference on each truck driving time of
184.108.40.206 Descriptive statistics
Two most important descriptive statistics were calculated for all the interval-rating and
ratio-scale questions; mean and standard deviation. A central tendency of each variable was
presented by its mean, while its standard deviation was presented to measure a typical degree of
spread of the variable. For nominal (categorical) data, (cumulative) frequency distributions were
presented to describe overall distribution of the survey responses. Histograms and scatter plots
were used to display the results.
220.127.116.11 Exploratory factor analysis (EFA)
Exploratory Factor Analysis (EFA) is a statistical method to explain a large number of
metric variables in terms of their common, underlying dimensions, that is, latent factors. Latent
factors are unobserved entities that influence a set of measures (variables) and are extracted from
the correlations among the variables. EFA results provide how the variables are grouped into a
small number of latent factors from the respondents' perspectives and how much each latent
factor is correlated with each of the variables. Factor analysis is a multivariate interdependence
technique with which all the variables are simultaneously considered, rather than multiple
regression, discriminate or canonical analyses, in which one or more variables are explicitly
considered as dependent variables (Hair, et al., 2005). This technique is often used in social
science research to summarize the data by identifying a set of latent factors that influence each
set of variables, which correlate highly amongst each other. The following seven steps are
typically taken to perform EFA (Field, 2005).
Calculation of correlation matrix: The starting point of factor analysis is to create a
correlation matrix, in which the inter-correlations between each pair of observed variables are
presented. A basic assumption of a factor analytic procedure is that a group of variables that
significantly correlate with each other do so because they are measuring the same common,
underlying dimension. Thus, if a group of variables seem to correlate highly with each other
within the group, but correlate very badly with variables outside of that group, they are
considered to well measure a common, underlying dimension, which is called a 'latent factor'.
The ultimate obj ective of the EFA procedure is to reduce the correlation matrix to a factor matrix,
which provides the correlations between the latent factors and each of the observed variables (i.e.,
factor loadings). This can be done by various factor extraction methods that are introduced later
in this section.
Factorability investigation: To identify common underlying dimensions that explain
the patterns of collinearity among the variables, the observed variables have to be inter-
correlated enough to be factorable, but they should not correlate too highly. It is important to
avoid multicollinearity (i.e., variables that are very highly correlated) and singularity (i.e.,
variables that are perfectly correlated) as these would cause difficulty in determining the unique
contribution of the variables to a factor.
The communality of a variable is the sum of the loadings of the variable on all extracted
factors. This represents the proportion of the variance in that variable that can be accounted for
by all extracted factors. Thus, if the communality of a variable is high, the extracted factors
account for a large proportion of the variables' variance. This means that this particular variable
is reflected well via the extracted factors, and hence the factor analysis is reliable. When the
communalities are not high, the sample size has to be large enough to compensate for this. To
examine whether the sample is large enough to elicit a meaningful factor solution, the Kaiser-
Meyer-Olkin (KMO) measure of sampling adequacy is used. The KMO statistic varies between
0 and 1. A value of 0 represents that the sum of partial correlations is large relative to the sum of
correlations, indicating that factor analysis is likely to be inappropriate. A value close to 1
represents that patterns of correlations are relatively compact, so factor analysis should yield
distinct and reliable factors. Typically, a KMO statistic value of greater than 0.5 is acceptable to
perform factor analysis.
Bartlett' s test of spherity tests the null hypothesis that the original correlation matrix is an
identity matrix. When the correlation matrix is an identity matrix, there would be no correlations
between the variables, eliminating the need for a factor analytic procedure. Thus, this test has to
be significant. A significance value (p value) less than 0.05 is usually necessary to justify the
factor analytic procedure.
Multicollinearity and singularity problems can be detected by investigating the
determinant of the correlation matrix. Usually, it is considered to not be a problem if the
determinant is greater than 0.00001.
Extraction of factors: A variety of statistical methods have been developed to extract
latent factors from an inter-correlation matrix of the observed variables. They include the
principal component extraction method, the principle axis extraction method, the maximum
likelihood extraction method, the unweighted least-squares extraction method, the generalized
least squares extraction method, the alpha extraction method, and the image factoring extraction
The two most commonly used extraction methods are the principle component and
principle axis methods. There are three types of variance in the variables: common, specific, and
error. Common variance is the variance in a variable which is shared with all other variables in
the analysis. Specific (unique) variance is the variance associated with only a specific variable.
Error variance is the inherently unreliable random variation. The principle component method
Einds latent factors that maximize the amount of total variance (i.e., sum of common, specific,
and error variances) that is explained, while the principle axis method Einds latent factors that
maximize the amount of common variance that is explained. The main difference between the
two types of methods lies in the way the communalities are used. Communality of a given
variable is the proportion of its variance that can be accounted for by extracted factors. In the
principle component method, it is assumed that all the communalities are initially one (unities
are inserted in the diagonal of the correlation matrix). That is, the total variance of the variables
can be accounted for by the extracted factors. On the other hand, with the principle axis method
the initial communalities are not assumed to be one (it does assume error variance). They are
usually estimated by taking the squared multiple correlations of the variables with other variables.
These estimated communalities are then represented on the diagonal of the correlation matrix,
from which the eigenvalues are determined and factors are extracted.
Theoretically, when the analyst is primarily concerned about determining the minimum
number of factors needed to account for the maximum portion of the variance represented in the
original set of variables, and has prior knowledge suggesting that specific and error variance
represent a relatively small portion of the total variance, the principle component method is
appropriate. In contrast, when the primary objective is to identify the latent dimensions or
constructs represented in the original variables, and the analyst has little knowledge about the
amount of specific and error variance and therefore wishes to eliminate these variances, the
principle axis method is appropriate. Practically, however, both methods are widely used and the
solutions generated by each usually do not differ significantly.
Figure 3-4 shows an example of a path diagram for an exploratory factor analytic model
by the principle component extraction method. As discussed, this method finds latent factors
that maximize the amount of total variance (of the observed variables) that is explained, and it is
assumed that there is no error variance. An EFA model is constructed in the way that series of
regression equations are set up to summarize its configuration. Equation 3.2 describes an EFA
model by the principle component extraction method. Each of the variables is defined as a linear
combination of the factors (i.e., sum of the products of each latent factor variable and the factor
loading of each observed variable on the corresponding factor). Results from the EFA include
derived loadings of each variable on each factor and calculated factor scores for each subj ect on
Az, = a factor loading of the ih Variable on the fh latent factor
F, = fh latent factor variable (a common, underlying dimension, j= 1 to k, k = number of
observed variables). These correspond to Fl, F2, and F3 in Figure 3-4.
Determination of number of factors to be retained: The maximum number of factors
that can be extracted is equal to the number of observed variables. However, the purpose of an
EFA is to adequately explain a relatively large number of variables with a small number of
factors. Thus, the analyst seeks to identify the smallest number of factors that explain a
considerably large amount of variance in the observed variables.
There are several criteria for the number of factors to be extracted, but these are just
empirical guidelines rather than an exact quantitative solution. In practice, most factor analysts
seldom use a single criterion to decide on the number of factors to extract. Some of the most
commonly used guidelines are latent root, percentage of variance, and scree test criteria. With
the latent root (eigenvalue) criterion, only the factors having an eigenvalue greater than one are
each factor. The factor scores are a composite measure that can be used for subsequent analyses.
When an orthogonal rotation method is used, the scores of the factors can be considered to be
independent of each other, and thus can be used as explanatory variables in a multiple regression
analy si s.
V, = [ ( A z, x F, )
ith observed variable (i = 1 to k, k = number of observed variables). These correspond to Vl,
V2, ..., V10 in Figure 3-4.
retained. It should be noted that the total sum of eigenvalues from the data is equal to the total
number of variables and the variance of a single variable is considered as the eigenvalue of one.
Thus, the rationale for the eigenvalue criteria is that any individual factor should account for at
least the variance of a single variable if it is to be retained for interpretation. The percentage of
variance criterion is a different approach. Using this method, the cumulative percentages of the
variance extracted by successive factors is the criterion. It is common in social science research
to consider a solution that accounts for at least 60% of the total variance as a satisfactory solution.
Another common approach is the scree test criterion. The scree test is derived by plotting the
latent roots (eigenvalues) against the number of factors in their order of extraction. The scree
plot illustrates the rate of change in the magnitude of the eigenvalues for the factors. The rate of
decline tends to be fast for the first few factors, but then levels off. The "elbow", or the point at
which the curve bends, is considered to indicate the maximum number of factors to extract.
Rotation of factors: Once the number of factors to be retained is decided, the next
logical step is to determine the method of rotation. The fundamental theorem of factor analysis
is invariant within rotations. That is, the initial factor matrix is not unique. There are an infinite
number of solutions, which produce the same correlation matrix, by rotating the reference axes
of the factor solution. A primary obj ective of the rotation is to make each variable load highly
on only one factor and have nearly zero loadings on the other factors. This simplifies the factor
structure and helps to achieve a more meaningful and interpretable solution. The simplest case
of rotation is an orthogonal rotation in which the angles between the reference axes of factors are
maintained at 90 degrees. Thus, there is no correlation between the extracted factors. A more
complicated form of rotation allows the angle between the reference axes to be other than a right
angle and is referred to as an oblique rotation. The factors are allowed to be correlated with each
other in this type of rotation. Orthogonal rotation procedures are more commonly used than
oblique rotation procedures because researchers often try to obtain an independent set of factors
to clarify the meaning of each factor. Three maj or orthogonal approaches are varimax,
quartimax, and equamax rotation methods, and two maj or oblique approaches are direct oblimin
and promax rotation methods.
Criteria for the significance of factor loadings: If there are variables that load highly
on two or more factors, or do not load highly on any factor, they are excluded from a factor
solution because it is not clear which factors) has an influence on the variables and this
ambiguous relationship between the factors and the variables blur the interpretation of a factor
solution. Whether a factor loading of a variable is significant or not depends on the sample size,
the total number of observed variables, and the total number of extracted factors. The larger the
sample size, the smaller the loading is considered to be significant. The larger the total number
of variables, the smaller the loading is considered to be significant. The larger the number of
factors, the larger the size of the loading on latent factors is considered to be significant. As a
rule of thumb, factor loadings greater than f 0.5 are considered to be significant when the
sample size is 120 or more, and factor loadings greater than f 0.65 are considered to be
significant when the sample size is 70 or more.
Naming of factors: Once the latent factors to be retained, and the variables associated
with each of those factors are identified, the analyst attempts to assign some meaning to the
factors based on the patterns of the factor loadings. It should be noted that the factor loadings
represent the correlation, or linear association, between a variable and the latent factors. Thus,
the analyst makes a determination as to what an underlying factor may represent, investigating
all the variables' loadings on the factor in terms of their size and sign. The larger the absolute
magnitude of the factor loading for a variable, the more important the variable is in interpreting
the factor. The sign of the loadings also need to be considered in labeling the factors. It may be
important to reverse the scoring of the negatively worded items in Likert-type instruments to
prevent ambiguity. That is, in Likert-type instruments some items are often negatively worded
so that high scores on these items actually reflect low degrees of the attitude or construct being
As the importance level of each traffic, roadway, and/or control variable was evaluated
on a 7-point interval-rating scale, EFA was applied to search for a set of latent factors that
accounts for the patterns of collinearity among the variables. The extracted factors and their
correlations with observed variables reflect in what respect each variable contributes to truck trip
quality and the degree to which the importance of each variable is explained by the underlying
latent factors. The principle component extraction method was used to find the set of latent
factors that accounts for the maximum amount of variance of the observed variables. The results
of the EFA presented a set of extracted latent factors, the percent of trace, that is, the portion of
the total variance (of the observed variables) that is explained by each latent factor, and the
correlations between each observed variable and latent factor (i.e., factor loadings). It was not
possible to apply Confirmatory Factor Analysis (CFA) or Structural Equation Modeling (SEM)
statistical techniques to the survey data because no previous hypothetical model construct exists
for the truck trip quality issue.
18.104.22.168 Multiple comparison test
When an Analysis of Variance (ANOVA) test verifies that the means of multiple variables
(i.e., more than two) are statistically different, multiple comparison procedures are widely used
to determine which means are different from one another. Fisher's Least Significance Difference
(LSD), Tukey's W, Student-Newman-Keuls (S-N-K), and Duncan's tests are often used (Ott and
Longnecker, 2006), assuming each sample from the groups is selected from a normal population
with an equal variance. However, if it is not reasonable to assume equal variance, pair-wise
multiple comparison procedures such as Tamhane's T2, Dunnett's T3, Games-Howell, or
Dunnett' s C tests can be used (Dunnett, 1980).
As the importance of each hypothetical performance measure was evaluated in a 7-point
interval rating scale, the Games-Howell pair-wise multiple comparison test was performed to
investigate if the differences among the mean importance levels of the performance measures are
statistically significant. Normal Quantile-Quantile (Q-Q) plots showed that each sample was
approximately normally distributed (Figure 3-5), but the equal variance assumption did not hold
according to Levene' s tests (Levene, 1960).
Thus, it was necessary to perform the multiple comparison tests with methods that do not
rely on the equal variance assumption. The Games-Howell test was performed among the
multiple comparison tests with unequal variances to investigate every possible statistical
difference in the pair-wise mean comparisons among the groups. The Games-Howell test is the
most liberal, meaning that differences between group means are identified as being significant
more readily with this test than the other tests. The Games-Howell test is a modification of S-N-
K procedure, using the q test statistic (i.e., studentized range statistic). Equation 3.3 is used to
calculate the test statistic, reflecting heterogeneous variances and sample sizes in the error term
in the denominator.
q = (3.3)
q = studentized range test statistic
pu, = calculated mean of the group i
s, = calculated standard deviation of the group i
nl = sample size of the group i
Equation 3.4 is also used to calculate the degrees of freedom for each pair-wise comparison to
adjust the error term. The calculated test statistics are compared with critical q values with the
corresponding degrees of freedom found in the studentized range tables.
df = n n(3.4)
n -1 n -1
df = degrees of freedom for each pair-wise comparison
22.214.171.124 Non-parametric test
The survey respondents' backgrounds were collected from various question types (i.e.,
ratio-scale, discrete choice, forced ranking, etc.) and their perceptions on the importance of each
hypothetical performance measure were evaluated on a 7-point interval rating scale. Thus,
ordered probit modeling technique (Greene, 2000) was first applied in an attempt to explore
respondents' backgrounds that may explain the variance in their perceptions. However, with the
ordered probit modeling, only a small number of potential explanatory variables were found to
be statistically significant, resulting in generally poor model fits. The Kruskal-Wallis test and
Mann-Whitney test (non-parametric version of Analysis of Variance (ANOVA) and t-test) were
applied to the data. The use of parametric ANOVA and the t-test was not appropriate even
though it is more powerful, because it was not reasonable to assume normality and equal
variances with the data.
The Kruskal-Wallis test (Conover, 2001) is a non-parametric one-way ANOVA by ranks.
It is used with k independent groups, where k is equal to or greater than 3, and measurement is at
least ordinal. The sample sizes across the groups can vary because the samples are independent.
The null hypothesis is that the k samples come from the same population. The alternative
hypothesis states that at least one sample comes from a different population. Following H test
statistic is used to test the hypothesis.
H = 12 '2 3 x (N + 1) (3.5)
N (N + 1) nl
H= Kruskal-Wallis test statistic
N = total sample size
k = number of independent samples
RI = sum of the ranks of group i
n, = sample size of group i
The calculated Htest statistic approximately follows a Chi-Squared distribution (X2) with k-1
degrees of freedom. Thus, for a specified value ofa the null hypothesis is rejected when
calculated H value exceeds the critical value of X2 for k-1 degrees of freedom.
The Mann-Whitney test (Conover, 2001) is a non-parametric t-test by ranks. It is used
specifically with two independent groups, and measurement is also at least ordinal. The sample
sizes between the two groups can vary because the samples are independent. The null hypothesis