<%BANNER%>

Content and Predictive Validity of a Self-Report Safe Driving Behavior Measure for Older Adults and a Proxy Measure for ...

Permanent Link: http://ufdc.ufl.edu/UFE0041217/00001

Material Information

Title: Content and Predictive Validity of a Self-Report Safe Driving Behavior Measure for Older Adults and a Proxy Measure for Family Members or Caregivers
Physical Description: 1 online resource (139 p.)
Language: english
Creator: Winter, Sandra
Publisher: University of Florida
Place of Publication: Gainesville, Fla.
Publication Date: 2009

Subjects

Subjects / Keywords: assessment, automotive, cpbr, crash, criterion, driving, focus, group, injury, irt, measurement, occupational, older, qualitative, roc, safety, therapy, validity
Rehabilitation Science -- Dissertations, Academic -- UF
Genre: Rehabilitation Science thesis, Ph.D.
bibliography   ( marcgt )
theses   ( marcgt )
government publication (state, provincial, terriorial, dependent)   ( marcgt )
born-digital   ( sobekcm )
Electronic Thesis or Dissertation

Notes

Abstract: Objective. Older adults, family members, and professionals may benefit from a driver's self/proxy-report designed to accurately assess safe driving behaviors. The objective of this study is to report on item development and validity testing of a self-report older adult Safe Driving Behavior Measure (SDBM). Method. Based on theoretical frameworks (Precede-Proceed Model of Health Promotion, Haddon's matrix and Michon's model), existing driving measures, previous research, and guided by measurement theory, we: (1) developed items capturing safe driving behavior, establishing face validity using peer reviewers and content validity using expert raters. (2) conducted focus groups to evaluate or generate items based on respondents' driving experiences and refine the SDBM items based on feedback. Older drivers (mean age 70.5, SD = 4.5) and family members (mean age 50, SD =20) participated in three focus groups in Florida and Ontario. Using content and thematic analyses, we coded responses to existing items, new items, or revisions, and (3) evaluated the concurrent criterion validity of the SDBM using Receiver Operating Characteristic (ROC) analyses to compare ratings from older drivers, caregivers, and driving evaluators. Results. Peer review indicated acceptable face validity. Initial expert rater review yielded a scale content validity index (CVI) rating of 0.78, with 44/60 items rated > = 0.75. Sixteen unacceptable items ( < = 0.5) required major revision or deletion. The next CVI scale average was 0.84 indicating acceptable content validity. Focus group one and two findings supported 46 of 72 existing items and generated 16 new items. Focus group three findings supported 40 existing items and generated 13 item revisions. Based on ROC analyses, SDBM ratings by evaluators resulted in the highest area under the curve (AUC) of 0.99. Family members/caregivers' ratings resulted in an AUC of 0.67. Drivers' ratings resulted in the lowest AUC of .54. Conclusion. Initial findings expert review and focus group findings suggest the SDBM may be a relevant and useful driving self/proxy-report for older adults. However, based on results of the Receiver Operating Characteristic analyses, only the driving evaluator ratings on the SDBM have acceptable accuracy, as measured by AUC. Future testing will focus identifying items that are more predictive of driving ability and refining SDBM items (e.g., content or wording) to address rater bias and improve rating accuracy of the drivers and family members/ caregivers.
General Note: In the series University of Florida Digital Collections.
General Note: Includes vita.
Bibliography: Includes bibliographical references.
Source of Description: Description based on online resource; title from PDF title page.
Source of Description: This bibliographic record is available under the Creative Commons CC0 public domain dedication. The University of Florida Libraries, as creator of this bibliographic record, has waived all rights to it worldwide under copyright law, including all related and neighboring rights, to the extent allowed by law.
Statement of Responsibility: by Sandra Winter.
Thesis: Thesis (Ph.D.)--University of Florida, 2009.
Local: Adviser: Classen, Sherrilene.
Local: Co-adviser: Rosenbek, John C.
Electronic Access: RESTRICTED TO UF STUDENTS, STAFF, FACULTY, AND ON-CAMPUS USE UNTIL 2010-12-31

Record Information

Source Institution: UFRGP
Rights Management: Applicable rights reserved.
Classification: lcc - LD1780 2009
System ID: UFE0041217:00001

Permanent Link: http://ufdc.ufl.edu/UFE0041217/00001

Material Information

Title: Content and Predictive Validity of a Self-Report Safe Driving Behavior Measure for Older Adults and a Proxy Measure for Family Members or Caregivers
Physical Description: 1 online resource (139 p.)
Language: english
Creator: Winter, Sandra
Publisher: University of Florida
Place of Publication: Gainesville, Fla.
Publication Date: 2009

Subjects

Subjects / Keywords: assessment, automotive, cpbr, crash, criterion, driving, focus, group, injury, irt, measurement, occupational, older, qualitative, roc, safety, therapy, validity
Rehabilitation Science -- Dissertations, Academic -- UF
Genre: Rehabilitation Science thesis, Ph.D.
bibliography   ( marcgt )
theses   ( marcgt )
government publication (state, provincial, terriorial, dependent)   ( marcgt )
born-digital   ( sobekcm )
Electronic Thesis or Dissertation

Notes

Abstract: Objective. Older adults, family members, and professionals may benefit from a driver's self/proxy-report designed to accurately assess safe driving behaviors. The objective of this study is to report on item development and validity testing of a self-report older adult Safe Driving Behavior Measure (SDBM). Method. Based on theoretical frameworks (Precede-Proceed Model of Health Promotion, Haddon's matrix and Michon's model), existing driving measures, previous research, and guided by measurement theory, we: (1) developed items capturing safe driving behavior, establishing face validity using peer reviewers and content validity using expert raters. (2) conducted focus groups to evaluate or generate items based on respondents' driving experiences and refine the SDBM items based on feedback. Older drivers (mean age 70.5, SD = 4.5) and family members (mean age 50, SD =20) participated in three focus groups in Florida and Ontario. Using content and thematic analyses, we coded responses to existing items, new items, or revisions, and (3) evaluated the concurrent criterion validity of the SDBM using Receiver Operating Characteristic (ROC) analyses to compare ratings from older drivers, caregivers, and driving evaluators. Results. Peer review indicated acceptable face validity. Initial expert rater review yielded a scale content validity index (CVI) rating of 0.78, with 44/60 items rated > = 0.75. Sixteen unacceptable items ( < = 0.5) required major revision or deletion. The next CVI scale average was 0.84 indicating acceptable content validity. Focus group one and two findings supported 46 of 72 existing items and generated 16 new items. Focus group three findings supported 40 existing items and generated 13 item revisions. Based on ROC analyses, SDBM ratings by evaluators resulted in the highest area under the curve (AUC) of 0.99. Family members/caregivers' ratings resulted in an AUC of 0.67. Drivers' ratings resulted in the lowest AUC of .54. Conclusion. Initial findings expert review and focus group findings suggest the SDBM may be a relevant and useful driving self/proxy-report for older adults. However, based on results of the Receiver Operating Characteristic analyses, only the driving evaluator ratings on the SDBM have acceptable accuracy, as measured by AUC. Future testing will focus identifying items that are more predictive of driving ability and refining SDBM items (e.g., content or wording) to address rater bias and improve rating accuracy of the drivers and family members/ caregivers.
General Note: In the series University of Florida Digital Collections.
General Note: Includes vita.
Bibliography: Includes bibliographical references.
Source of Description: Description based on online resource; title from PDF title page.
Source of Description: This bibliographic record is available under the Creative Commons CC0 public domain dedication. The University of Florida Libraries, as creator of this bibliographic record, has waived all rights to it worldwide under copyright law, including all related and neighboring rights, to the extent allowed by law.
Statement of Responsibility: by Sandra Winter.
Thesis: Thesis (Ph.D.)--University of Florida, 2009.
Local: Adviser: Classen, Sherrilene.
Local: Co-adviser: Rosenbek, John C.
Electronic Access: RESTRICTED TO UF STUDENTS, STAFF, FACULTY, AND ON-CAMPUS USE UNTIL 2010-12-31

Record Information

Source Institution: UFRGP
Rights Management: Applicable rights reserved.
Classification: lcc - LD1780 2009
System ID: UFE0041217:00001


This item has the following downloads:


Full Text

PAGE 1

1 CONTENT AND PREDICTI VE VALIDITY OF A SEL F REPORT SAFE DRIVING BEHAVIOR MEASURE FOR OLDER ADULTS AND A P ROXY MEASURE FOR FAMILY MEMBERS OR CA REGIVERS By SANDRA MAE WINTER A DISSERTATION PRESENTED TO THE GRADUATE SCHOOL OF T HE UNIVERSITY OF FLORIDA IN PARTIAL FULFILLMENT OF THE REQUIREMENTS FOR THE DEGREE OF DOCTOR OF PHILOSOPHY UNIVERSITY OF FLORIDA 2009

PAGE 2

2 2009 Sandra Mae Winter

PAGE 3

3 To my family for all your support

PAGE 4

4 ACKNOWLEDGMENTS I thank my husband, Kim Wint er, for his faithful support, constant encouragement and willingness to adapt to many changes over the course of my PhD studies. My son Sam receives thanks for his belief in me, for being my e to play heartily and often. My parents were big supporters of my decision to pursue my PhD and told me I would make it. My best friend Martha wrote uplifting e mails, called, and helped me keep it all in perspective. I thank Sherrilene Classen, my advis or and chair for her guidance and for fostering my development as a scholar, researcher and a person during the past five plus years. She gave me a chance to pursue an important goal and provided the support I needed to overcome some tough hurdles. Her mai n interest has been promoting my development as a researcher and scholar and I am grateful for her investment in my growth. My appreciation also goes to my committee members, Dr. John Rosenbek, Dr. Barbara Lutz, and Dr. Michel Bdard. Dr. Rosenbek taught m y first Rehabilitation Science class and helped me develop the scientist mindset so necessary for this research. Dr. Lutz broadened my qualitative research knowledge and skills and helped me problem solve those aspects of this study. Dr. Bdard shared his expertise in older driver safety research and brought a fresh perspective to the work. Lastly, I thank Dr. Ellen Lopez, a former committee member, for her mentoring and support early in my PhD program I have heartfelt appreciation for Kezia Awadzi who sh owed me it could be done, provided an example of tenacity under difficult circumstances, and helped

PAGE 5

5 me in little and big ways as I pursued this degree. Nita Ferree also supported me along the way with words of encouragement. I wish to acknowledge fellow RS D and PHHP students (and graduates) Michele Woodbury, Megan Witte Bewenitz, Swathy Sundaram, Jessica Johnson, and Pey Shan Wen. The project was funded by a National Institutes of Health/ National Institute Community Participation in Develop ing a Measure for Safe Older Driver Behaviors (R21) PAR 06 247 (PI Classen). The grant team was very helpful and supportive during this process; I thank Dr. Craig Velozo, Dr. Babette Brumback, and Desiree Landford. I acknowledge the National Older Driver Research and Training Center at the University of Florida for support and research infrastructure. Research assistants Brandy Couture, Julie Riendeau, and Laura Diamond of the Interdisciplinary Research Program on Safe Driving at Lakehead University also c ontributed to recruitment and data collection. Expert raters who assisted with this study include: Dr. David Eby, Dr. Bella Dinh Zarr, Dr. Holly Tuokko and Mr. Jim Langford. Peer reviewers were from the Qualitative Data Analysis Group at the University of Florida led by Dr. Barbara Lutz and Dr. Sharleen Simpson. I also wish to recognize the grant advisory board members: Paul Boase, Dr. Jan Polgar, Dr. David Eby, Dr. Barbara Messinger Rapport, and Frank Carroll.

PAGE 6

6 TABLE OF CONTENTS page ACKNOWLEDGMENTS ................................ ................................ ................................ ...... 4 LIST OF TABLES ................................ ................................ ................................ ................. 9 LIST OF FIGURES ................................ ................................ ................................ ............ 10 LIST OF ABBREVIATIONS ................................ ................................ ............................... 11 ABSTRACT ................................ ................................ ................................ ........................ 12 CHAPTER 1 INTRODUCTION ................................ ................................ ................................ ......... 14 Background ................................ ................................ ................................ ................. 14 Older Adult Drivers ................................ ................................ ............................... 15 Research Context ................................ ................................ ................................ 17 R esearch Questions ................................ ................................ ............................. 18 Statement of Purpose ................................ ................................ ........................... 18 Rationale and Significance ................................ ................................ ................... 19 Previous Research ................................ ................................ ............................... 20 Theoretical Framework ................................ ................................ ............................... 21 Precede Proceed Model of Health Promotion ................................ ..................... 21 ................................ ................................ ................................ .... 24 ................................ ................................ ................................ ..... 25 Existing Measures ................................ ................................ ................................ 26 Approach and Methods ................................ ................................ ............................... 28 Community Based Participatory Research ................................ .......................... 28 Classical Test Theory ................................ ................................ ........................... 30 Validity ................................ ................................ ................................ ............ 30 Content validity ................................ ................................ .............................. 31 Criterion validity ................................ ................................ ............................. 32 Item Response Theory ................................ ................................ ......................... 32 Conclusion ................................ ................................ ................................ ................... 34 2 ITEM DEVELOPMENT AND VALIDITY TESTING F OR A SAFE DRIVING BEHAVIOR MEASURE ................................ ................................ .............. 35 Background ................................ ................................ ................................ ................. 35 Safe Driving ................................ ................................ ................................ .......... 35 Self Report ................................ ................................ ................................ ............ 37 Community Based Participatory Research (CBPR) ................................ ............ 40 Measurement Theory ................................ ................................ ........................... 41 Rationale and Significance ................................ ................................ ................... 42

PAGE 7

7 Purpose ................................ ................................ ................................ ................. 43 Methods ................................ ................................ ................................ ....................... 43 Setting ................................ ................................ ................................ ................... 43 Research Team ................................ ................................ ................................ .... 43 Process for Item Development ................................ ................................ ............. 44 Precede Proceed Model of Health Promotion ................................ .............. 44 ................................ ................................ ............................. 45 ................................ ................................ .............................. 45 Existing Measures ................................ ................................ ................................ 45 SDBM Tool Design ................................ ................................ ............................... 46 Item Development ................................ ................................ ................................ 47 Item Review ................................ ................................ ................................ .......... 47 Analysis ................................ ................................ ................................ ................. 49 Results ................................ ................................ ................................ ......................... 50 P eer Review ................................ ................................ ................................ ......... 50 Content Validity Index ................................ ................................ .......................... 50 Discussion ................................ ................................ ................................ ................... 53 3 CONTRIBU TION OF FOCUS GROUPS TO DEVELOP A SELF AND PROXY REPORT SAFE DRIVING BEHAVIOR MEASURE ................................ ..... 55 Background ................................ ................................ ................................ ................. 55 Rationale and Significanc e ................................ ................................ ................... 57 Purpose ................................ ................................ ................................ ................. 57 Methods ................................ ................................ ................................ ....................... 57 Procedure ................................ ................................ ................................ ............. 58 SDBM development ................................ ................................ ....................... 58 Focus groups ................................ ................................ ................................ 59 Participants ................................ ................................ ................................ ........... 60 Data Collection ................................ ................................ ................................ ..... 60 Data Analysis ................................ ................................ ................................ ........ 60 Content analysis ................................ ................................ ............................ 61 Thematic analysis ................................ ................................ .......................... 61 Iterative Process of Item Development ................................ ................................ 62 Results ................................ ................................ ................................ ......................... 63 Focus Groups 1 and 2 ................................ ................................ .......................... 63 Person Vehicle domain (PV) ................................ ................................ ......... 65 Person Environment domain (PE) ................................ ................................ 66 Person Vehicle Environment domain (PVE) ................................ ................. 66 Focus Group 3 ................................ ................................ ................................ ...... 67 Discussion ................................ ................................ ................................ ................... 69 Conclusion ................................ ................................ ................................ ................... 71 4 SELF REPORT SAFE DRIVING BEHAVIOR MEASURE AS A PREDICTOR OF OLDER ADULTS PASSING/FAILING AN ON ROAD DRIVING EVALU ATION ................................ ................................ ............................. 73

PAGE 8

8 Introduction ................................ ................................ ................................ .................. 73 Research Context ................................ ................................ ................................ 76 Use of ROC in Measu re Development ................................ ................................ 76 Rationale and Significance ................................ ................................ ................... 77 Research Questions ................................ ................................ ............................. 78 Research Questions ................................ ................................ ............................. 78 Methods ................................ ................................ ................................ ....................... 78 Recruitment / Sample ................................ ................................ ........................... 78 D esign ................................ ................................ ................................ ................... 79 Procedure ................................ ................................ ................................ ............. 79 Measurement ................................ ................................ ................................ ........ 81 SDBM ................................ ................................ ................................ ............. 81 Clinical tests ................................ ................................ ................................ ... 82 On road test ................................ ................................ ................................ ... 83 Data Collection ................................ ................................ ................................ ..... 84 Data Analysis ................................ ................................ ................................ ........ 85 Descriptive analysis ................................ ................................ ....................... 85 ROC analyses ................................ ................................ ................................ 86 Results ................................ ................................ ................................ ......................... 87 Demographics ................................ ................................ ................................ ....... 87 Drivers ................................ ................................ ................................ ............ 88 Family members/caregivers ................................ ................................ .......... 88 ROC curves ................................ ................................ ................................ .......... 92 Drivers SDBM total ................................ ................................ ........................ 92 Family member/caregiver SDBM total ................................ .......................... 93 Evaluators SDBM total ................................ ................................ ................... 95 ROC Curve Comparison ................................ ................................ ...................... 96 Post hoc Analysis ................................ ................................ ................................ 97 Discussion ................................ ................................ ................................ ................... 99 5 SUMMARY AND CONCLUSION ................................ ................................ .............. 104 Limitations ................................ ................................ ................................ ........... 106 Strengths ................................ ................................ ................................ ............. 108 Implications ................................ ................................ ................................ ......... 108 Conclusion ................................ ................................ ................................ .......... 109 APPENDIX A SAFE DRIVING BEHAVIOR MEASURE DRIVER VERSION .............................. 111 B SAFE DRIVING BEHAVIOR MEASURE CAREGIVER VERSION ...................... 124 C FOCUS GROUP TWO GUIDE ................................ ................................ ................. 127 LIST OF REFERENCES ................................ ................................ ................................ 129 BIOGRAPHICAL SKETCH ................................ ................................ .............................. 139

PAGE 9

9 LIST OF TABLES Table page 1 1 Existing driving safety measures ................................ ................................ ............ 28 2 1 Peer review results for face validity ................................ ................................ ....... 50 2 2 Items and Content Validity Index (CVI) before (initial) and after (final) team revisions and expert ratings. ................................ ................................ ......... 52 3 1 Demographic Profile of Participants from three Focus Groups. ........................... 63 3 2 Thematic analysis results listing 23 new themes. ................................ ................. 64 3 3 Focus groups 1 and 2: Quote summary listed by Person Vehicle Environment domains ................................ ................................ ............................. 64 3 4 Focus group 3: Quote summary of item feedback by category ............................ 67 4 1 2X2 Table Displaying Relationship of True Positives, False Positives, True Negatives, and False Negatives. ................................ ................................ ... 87 4 2 Descriptive Profi and Driving Variables for both Sites. ................................ ................................ ...... 90 4 3 for both Sites ................................ ................................ ................................ ........... 91

PAGE 10

10 LIST OF FIGURES Figure page 2 1 Steps in Process of Item Development and Testing for the Safe Driving Behavior Measure ................................ ................................ ...................... 48 2 2 Examples of a CVI Rating including Feedback from Expert Reviewers. .............. 49 3 1 Iterative Process for Item Development, Revisions and/or Refinement ............... 62 4 1 ROC Curve for the Drivers ................................ ................................ ..................... 92 4 2 ROC Curve for the Family Members/Caregivers ................................ ................... 94 4 3 ROC Curve for the Evaluators ................................ ................................ ............... 95 4 4 Scatter Plot and Linear Regression Line for the Drivers ................................ ....... 97 4 5 Scatter Plot and Line ar Regression Line for the Family Members/ Caregivers ................................ ................................ ................................ ............... 98 4 6 Scatter Plot and Linear Regression Line for the Evaluators ................................ 99

PAGE 11

11 LIST OF ABBREVIATION S AD Alzheime ADL Activities of daily living AUC area under the curve CAT computer adaptive testing CAOT Canadian Association of Occupational Therapists CBPR community based participatory research CDE comprehensive driving evaluation CVI content validity in dex FARS Fatality Analysis Rating System FG Focus Group IADL Instrumental Activity of Daily Living IRT item response theory LU Lakehead University NHTSA National Highway Traffic Safety Administration PE Person Environment PPMHP Precede Proceed Model of H ealth Promotion PV Person Vehicle PVE Person Vehicle Environment ROC receiver operat ing characteristic SDBM Safe Driving Behavior Measure SD standard deviation SLR systematic literature review UF University of Florida

PAGE 12

12 Abstract of Dissertation Pr esented to the Graduate School of the University of Florida in Partial Fulfillment of the Requirements for the Degree of Doctor of Philosophy CONTENT AND PREDICTI VE VALIDITY OF A SEL F REPORT SAFE DRIVING BEHAVIOR MEASURE FOR OLDER ADULTS A ND A PROXY MEASURE F OR FAMILY MEMBERS OR CA REGIVERS By Sandra Mae Winter December 2009 Chair: Sherrilene Classen Co chair: John Rosenbek Major: Rehabilitation Science Objective Older adults, family members, and professionals may benefit s self/proxy report designed to accurately assess safe driving behaviors. The objective of this study is to report on item development and validity testing of a self report older adult Safe Driving Behavior Measure (SDBM). Method Based on theoretical fra meworks (Precede Proceed Model of measures, previous research, and guided by measurement theory, we: ( 1 ) developed items capturing safe driving behavior, establishing face validity using p eer reviewers and content validity using expert raters. ( 2 ) conducted focus groups to evaluate or generate items based on feedback. Older drivers (mean age 70.5, SD = 4.5) and family members ( mean age 50, SD =20) participated in three focus groups in Florida and

PAGE 13

13 Ontario. Using content and thematic analyses, we coded responses to existing items, new items, or revisions, and ( 3 ) evaluated the concurrent criterion validity of the SDBM using Receiver O perating Characteristic (ROC) analyses to compare ratings from older drivers, caregivers, and driving evaluators. Result s. Peer review indicated acceptable face validity. Initial expert rater review yielded a scale content validity index (CVI) rating of 0 .78, with 44/60 items deletion. The next CVI scale average was 0.84 indicating acceptable content validity. Focus group one and two findings supported 46 of 72 existing items and g enerated 16 new items. Focus group three findings supported 40 existing items and generated 13 item revisions. Based on R OC analyses SDBM ratings by evaluators resulted in the highest area under the curve (AUC) of 0 .99 Family members/c ratings resulted in an AUC of 0 .67 ratings resulted in the lowest AUC of 54 Conclusion Initial findings expert review and focus group findings suggest th e SDBM may be a relevant and useful driving self/proxy report for older adults However, based on results of the Receiver Operating Characteristic analyses, only the driving evaluator ratings on the SDBM have acceptable accuracy, as measured by AUC Future testing will focus identifying items that are more predictive of driving ability and refining S DBM items (e.g., content or wording) to address rater bias and improve rating accuracy of the drivers and family members/ caregivers.

PAGE 14

14 CHAPTER 1 INTRODUCTION Background In a mobile society where older Americans are dependent on the automobile as the pri mary mode of transportation, periodic assessment of their driving safety is a key concern. Currently, this is being done through comprehensive driving evaluations (CDEs) administered by driving rehabilitation specialists; by community based programs such a s those offered by AARP, the American Automobile Association (AAA), the American Occupational Therapy Association (AOTA), and the American Society on Aging (ASA); family/caregiver assessments; and self assessments. Self report measures (based on self asses sment) and proxy measures (based on observations from family, caregiver behaviors, increase driving safety awareness and knowledge, and promote behavior change and safer driving ou tcomes (i.e., fewer crashes, crash related injuries or deaths) (Classen, Lopez, et al., 2007; Eby, Molnar, Shope, Vivoda, & Fordyce, 2003; McGee & Tuokko, 2003). Benefits of self reports include low cost, convenience, and confidentiality. However, self rep ort measures are limited by self selection bias (i.e., capable persons are more likely to complete the self report) social desirability bias (i.e., persons are more likely to give answers that will be viewed favorably by others) recall bias (i.e., abilit y to accurately affirmatively to all questions) ( Furr & Bacharach, 2008; Sundstrom, 2005; Zhou & Lyles, 1997).

PAGE 15

15 Older Adult Drivers According to the National Highway Traffic Saf ety Administration (NHTSA) (2008a), 31 million persons age 65 and older are licensed to drive in the U.S. As a group, older adults are less involved in fatal crashes in relation to drivers age 16 to 34. However, in 2008, older adults accounted for 5,569 tr affic related fatalities and over 183,000 injuries from crashes (NHTSA, 2008a). Older drivers who experience physical or cognitive declines with aging have an increased crash risk and older adults with co morbidities or frailty have an increased risk of cr ash related injuries or fatalities (Anstey, Wood, Lord, & Walker, 2005; Classen et al., 2006). These crash related injury and fatality risks, combined with projected growth in the older adult population, create a need to plan injury prevention intervention s (Bdard, Stones, Guyatt, & Hirdes, 2001; Classen, Lopez, et al., 2007). Injury prevention efforts include interventions addressing safety awareness, knowledge, and behaviors. Identification of safe driving behaviors (or deficits thereof) is the first st promote behavior change and safer driving (i.e. fewer violations, near misses, crashes, injuries and deaths) (Classen, Lopez, et al., 2007). From the existing older driver safety literature, there is support for use of educational approaches to increase awareness of driving behaviors and bring about changes in knowledge, attitudes and behavior (Marottoli, 2007). However, the effectiveness of educational interventions in the reduction of crashes, dea ths and injuries has not been established to date (Kua, Korner Bitensky, Desrosiers, Man Son Hing, & Marshall, 2007).

PAGE 16

16 Part of developing an effective intervention for older adults is to accurately examin e older driver behaviors. Compar ed to drivers younge r than age 65, we know older drivers tend to engage in fewer risk taking behaviors (such as speeding or driving under the influence)(Centers for Disease Control and Prevention, 1999; Zhang, Fraser, Lindsay, Clarke, & Mao, 1998). Findings from crash dataset difficulty judging left turns against traffic and failure to yield right of way at intersections ( Awadzi, Classen, Hall, Duncan, & Garvan, 2008; Braitman, Kirley, Ferguson, & Chaudhary, 2007 ; McGwin & Brown, 1999 ) Two measures of safety include crashes and failing of on the road driving evaluations. Crash rates derived from local, state and national datasets inform traffic safety but there are statistical limitations in using crashes as an ou tcome measure for a safety intervention. Crashes are rare event s, and crash measures often include only crashes of a defined severity (e.g. property damage or fatalities) and do not include potentially relevant unsafe incidents, near misses or minor cras hes (Goode et al., 1998) Moreover, self reports of crashes are higher than those based on state records, indicating under report ing for police or insurance purposes (Mar o t t oli Cooney, & Tinetti, 1997 ) Driving behaviors and skills can be m easured using an on the road evaluation, considered the gold standard for determining if someone can drive safely. However, limitations of on the road evaluation s include reduced validity due to either changing conditions (given an open course), lack of re al world situations (given a closed course), time needed to complete, limited availability, difficulty of

PAGE 17

17 administration and high out of pocket costs for drivers, exposure to risks, need for trained personnel to administer the test, expensive equipment (e. g., a dual brake equipped vehicle), liability issues, and limited opportunity for access (Kua, Korner Bitensky, & Desrosiers, 2007; Odenheimer et al., 1994; Wang & Carr, 2004). Research Context This article based dissertation was part of a larger planning study (CLASSEN 1 R21 AG31717 01) to develop a Safe Driving Behavioral Measure (SDBM) for older adults using community based participatory research (CBPR) and item response theory (IRT). CPBR engages community members in public health research (e.g. in det ermining the research question, recruitment, data collection or analysis) of problems related to social or environmental (resource) disparities ( Israel, Schulz, Parker, & Becker, 1998 ). I tem response theory uses probability models to study person abilities behaviors or traits and create a hierarchy of item responses (Hambleton, Swaminathan, & Rogers, 1991; Stone, 1997). Used together, a CBPR approach and IRT can facilitate development of a safe driving behavior measure for older adults. Developing behavior al items through an iterative process of item testing, refinement and re testing is a critical step contributing to the main objective of the parent grant (R21 AG31717 01, PI Classen), developing and testing (in a multi site clinical trial) a driving safet y measure that demonstrates affordability, accessibility availability and acceptability for use on a population level. The larger study involved research at two sites, the University of Florida in Gainesville (UF) and Lakehead University (LU) Thunder Ba y, Ontario. The

PAGE 18

18 primary site, UF, has a group of experienced older driver safety researchers, a National Older Driver Research and Training Center, and a driving rehabilitation and evaluation center (Independence Drive). Lakehead University (Thunder Bay, O ntario, Canada) provides expert researchers with published work in older driver safety, an Interdisciplinary Research Program on Safe Driving, a diverse participant pool, and a unique driving environment. Research Questions Question one What is the fac e and content validity of the domains, constructs, and safe driving behavioral items established during an iterative process of item refinement and re testing? Question two What are the domains, constructs, and items that comprise ing from the perspective of older drivers and families? Question three What is the concurrent criterion validity of the SDBM as measured using ROC analyses based on ratings by driver, caregiver and evaluator? Statement of Purpose The overall purpose of this dissertation is to describe the development of the SDBM using three steps: item development, item refinement, and item testing. Items were developed using a theoretically derived framework of domains, concepts and constructs and rank ed behaviors as less safe or more safe. As part of item development, the content validity of the SDBM was quantified at the item and scale level. Item refinement included use of focus safe d riving from the perspective of older drivers and families. Focus group

PAGE 19

19 findings were used to develop, revise, refine, and re test items measuring safe driving behaviors The goal was to develop items with acceptable face and content validity that incorpora ted geographic or environmental diversity. Item testing entailed obtaining SDBM ratings from drivers, family members/caregivers, and driving evaluators. A s a measure of concurrent criterion validity, t hese SDBM ratings were compared against the outcome of an on road driving evaluation using Receiver Operat ing Characteristic curve analyses In the article based chapters 2, 3, and 4 I will detail the three studies used to develop the SDBM and establish its psychometric properties. The literature review provi ded in this chapter addresses the theoretical foundation of the SDBM, provides a critical review of existing measures, and addresses two approaches used in this research, CBPR and IRT. Rationale and Significance Identification of safe driving behaviors (or deficits thereof) is the first step to and safer driving (i.e. fewer violations, near misses, crashes, crash related injuries and deaths) (Classen, Lopez, et al., 2007). To date, the development, testing and application of older driver safety assessment tools has not extensively incorporated the input of older drivers, family, caregivers and community stakeholders (national and international agents of the safe older driver communi ty); or addressed the need to refine items accounting for geographic differences (e.g., urban vs. rural or Midwestern vs. Southeastern US), cultural factors (e.g. expected driving behavior) and environmental conditions (e.g. weather, road conditions, or traffic). Moreover, from the existing older driver

PAGE 20

20 safety literature, it is clear that safety is dependent on gaining awareness of driving behaviors and using this awareness to bring about changes in knowledge, attitudes and behavior (Eby et al., 2003; Win dsor Walker, Caldwell, & Anstey, 2006). Awareness can potentially be attained through use of a measure designed to enable older adults, family or professionals to easily and accurately identify and rate the relative safety or lack of safety of older adult We propose that tools developed without stakeholder input and without attention to geographical, cultural or environmental factors may not be seen as relevant by users and may lack face or content validity. Moreover, if key domains w ere not addressed or the tool was not examined in relation to a standard such as on road testing, construct or criterion validity would also be impacted Furthermore, existing research has not produced a theoretically derived assessment tool with item sele ction guided by item response theory to ensure a variety of hierarchically organized items to enable increased measurement accuracy Previous Research S afe driving behavior items were developed working from the previous research of Classen and colleagues (including a systematic literature review [Classen et al., 2006], findings from a study of the Fatality Analysis R eporting System [ Awadzi, Classen, Garvan & Komaragiri, 2006] and a meta synthesis of qualitative studies on older driver safety [Classen, Win ter & Lopez, 2009]). The systematic literature review (SLR) study used the Precede Proceed Model of Health Promotion as a guiding framework for the analysis of over 200 U.S. older driver safety studies (from 1985 2005) (Classen et al., 2006). The SLR findi ngs

PAGE 21

21 support the importance of a systems approach considering statistically significant risk/protective factors from socio ecological domains. T he Fatality Analysis Reporting System (FARS) study findings included 11 key risk and protective factors present i n fatal crashes including person vehicle environment interactions influencing driving safety. The relationship of the 11 risk and protective factors to outcomes of crash related injuries and fatalities was quantified (Awadzi et al., 2006). From a synthesis of qualitative older driver safety studies (Classen et al., including motivations for driving behaviors, social reinforcers of safe driving and barriers to improved driving s afety. Classen Lopez and colleagues (2007) developed a conceptual plan for a safe driving intervention (addressing crash/injury/fatality reduction), specifically noting a need for a safe driving behavior measure suitable for population based use. Collecti vely, the findings from these previous studies indicated necessary domains for the Safe Driving Behavior Measure and illustrated behaviors that may influence driving safety. Theoretical Framework The framework for item development included three theoreti cal models emphasizing public health and health promotion, injury prevention and driving which are discussed below. Precede Proceed Model of Health Promotion The Precede Proceed Model of Health Promotion (PPMHP) is socio ecological model that evolved i n response to the need for a cost benefit evaluation framework for public health programs. The PPMHP was based on research from public health, medical care, family planning, psychology, and

PAGE 22

22 social factors in health behavior. Influencing theories include he alth belief models, change models including diffusion and adoption theory, and models addressing health behavior change (Green & Kreuter, 2005). The PPMHP has two components, the Precede component guides a social and ecological assessment of the health pro blem and intervention planning while the Proceed component directs steps for program evaluation. In Precede, there are five phases for intervention planning; social assessment, epidemiological assessment, educational and ecological assessment, administrat ive and policy assessment and intervention alignment and intervention implementation. In the Proceed phase the planned interventions are implemented and evaluated in terms of process, impact of the intervention on behavior and outcomes. Assumptions of the PPMHP include a socio ecological orientation to a health problem that looks at the person or community interacting with the environment and the factors that create or modify health or illness and related behaviors. From this view, it is important to includ e the persons affected in defining and addressing the health problem being studied through both social assessment intervention process. A review of behavioral and social scienc e theories and models used in injury prevention research found the PPMHP to be the most commonly cited model, primarily by studies addressing bicycle helmet use in children (Trifiletti, Gielen, Sleet, & Hopkins, 2005) I have discussed older driver safety studies by Classen and colleagues using the PPMHP in previous research. Other motor

PAGE 23

23 safety related PPMHP research studies include driving under the influence of alcohol (Cotton & Stewart 1983), health education and motor vehicle injuries (Sleet, 1987), and child pedestrian injuries (Stevenson, Iredell, Howat, & Cross, 199 9 ). PPMHP domains address safety influences from the person (e.g., knowledge, behavior, attitudes or health status), the physical environment (e.g., physical factors such as highway design or vehicle features) and the social environment (e.g., driving laws or assistance with transportation) (Green & Kreuter, 2005). The PPMHP establishes the importance of examining person, vehicle and environment interactions. Furthermore, the PPMHP supports a process for item development that involves the community (older drivers, family, caregivers and partners from the older driver safety network) in defining and addressing the health problem being studied. Within this study, the PPMHP provided a conceptua l framework for identification, categorization and explanation of driving behaviors contributing to older driver safety. As part of the Precede phase, factors are assessed across domains addressing behavior, environment (physical and social), predisposing factors (e.g. knowledge, attitudes or values), reinforcing factors (i.e. consequences from behavior that contribute to it being continued or extinguished), enabling factors (e.g. laws, rules or policies), and health education (i.e. programs to impact k nowledge and behavior). Knowledge of these domains and their components was essential to refinement of the measure and content validity (i.e. the tool has sufficient domains and items to measure the phenomena under study).

PAGE 24

24 x is a conceptual framework illustrating injury as a result of interactions between people and their environment (Haddon, 1972). William H. the reduction of crash related injur ies and fatalities and uses a systems approach and an emphasis on the etiology of crashes (i.e. root causes). The primary are specific, measurable and preventable events. A ccident factors can be studied, systematically outline d and targeted with prevention strategies. e factors related to the host (person), agent (means of force transmission), environment (physical and social driving co ntext) and vectors (physical forces present in the interactions among the person, vehicle and physical environment). Haddon also conceptualized a time continuum for injury risk and protective factors as pre event ( e.g., before a crash), event ( e.g., during a crash) and post event ( e.g., after a crash). Runyan ( 1998, M atrix and the development of ten injury counter measures. Two earlier works connecting epidemiology to accident prevention (Gordon, 1949) and specifying injury causing factors (Gibson, 196 4 M atrix. First, Gordon outlined injury control as part of public health and established a framework illustrating injury resulting from host, agent and environment interactions. Subseq uently, Gibson further characterized injury causing agents based on their physical force as thermal, radiant, chemical, mechanical or electrical. Haddon also drew from public health efforts to combat polio ( e.g.,

PAGE 25

25 Matrix has been used in research in motor vehicle injury prevention ( ZaZa et al., 2001 ), as well as prevention of other unintentional injuries in the home ( Runyan, Marshall, Coyne Beasley, & Casteel, 2005), and the workpl ace ( Peek Asa & Zwerling, 2003). In this dissertation c M atrix were part of the theoretically derived framework for identification, description, and categorization of safe driving behaviors in older adults and contributed domains that were assessed during the conten t validity analysis. behavior on three levels, strategic, tactical, and operational. Strategic behaviors can occur before or during driving, include general goal f ormation decisions (e.g., trip planning) and occur in minutes to hours. Tactical behaviors are outputs reflect ing conscious decisions made while operating the vehicle relating to interaction with the immediate driving environment (e.g., maneuvers or car ha ndling) and occur in seconds to minutes. Operational behaviors are outputs that happen primarily by subconscious action (e.g. maintaining your lane) and part of a historical revi ew of driver behavior modeling and had a foundation in the cognitive modeling work of Janssen (1979) who looked at route planning by of driving behavior. Cognitive models descri be behavior as output based on the how we perceive the environment). Behavior is classified in order to better understand interactions between the

PAGE 26

26 person the car and the environment and to predict driver behavi or. Such an understanding contributes to further modeling of driving behavior as well as interventions to improve safety whether related to car design or driving development of vehicle cont rols or safety systems ( Cacciabue, 2007), driving after brain injury (Handler & Patterson, 1995) and driving assessment ( Unsworth, Lovell, Terrington, & Thomas, 2005). Existing Measures We identified and reviewed eleven existing measures as part of establ ishing the content domain for the Safe Driving Behavior Measure. Since the SDBM was being developed as a self report measure, our review of existing measures included tools and questionnaires using self report from older drivers (Hunt, personal communicati ons, April 30, 2007; Lawton, Parker, Manstead, & Stradling,1997; Stalvey & Owsley, 2000; Owsley, Stalvey, Wells, & Sloane, 1999 ; Yee & Melichar, 1992). We also investigated items from one set of self report measures and two Behind the wheel evaluations tha t were analyzed using Rasch methodology (Kay, Bundy, Clemson, & Jolly, 2008; Justiss, 2006; Myers, Paradis, & Blanchard, 2008). The measures reviewed are listed in Table 1 1, with author information and a brief description of the measure. The existing meas ures M M atrix, items were coded as addressing person factors (skills, knowledge, attitudes, or health status), vehicle factors ( selection and use of car features) or environment factors (physical environment such as weather or social environment such as

PAGE 27

27 passengers in car). We coded items according to when the behavior occurs (pre driving or driving) and the level of the behavior ac cording to (strategic, tactical or operational). Overall, the existing item measures were found to have a bias toward person items with minimal representation of environmental factors and a gap in items pertaining to the vehicle. Further more, the majority of items were written to capture driving deficits by older adults, thus the items did not represent the broader range of driving behaviors demonstrated by older adults. If the existing self report measures have floor or ceiling effects, or scores score lacks the rigor desired for informing driving decisions. The driving measures listed that are on road evaluations have the necessary spread of items and meaning ful score, but professional driving evaluations are not accessible to the majority of older drivers who may desire to assess their driving skills. An appropriately designed self report measure could be used by older drivers as well as family member s careg ivers and professionals working with older adults further action (e.g., medical appointment or on road evaluation).

PAGE 28

28 Table 1 1. Existing driving safety measures Meas ure Description Driver Perceptions and Practices Questionnaire (Stalvey & Owsley, 2000) 74 item self report measure capturing driver knowledge, attitudes, perceptions, self regulation and perceived self efficacy for driving and driving changes Driving H abits Qu estionnaire (Owsley et al.,1999) 34 item self report measure capturing current driving status, driving exposure, driving dependence, driving avoidance., crashes, citations and driving space Driving Survey (Hunt unpublished) 28 item questionnaire capturing demographics, driving dependence, health care, ADL limitations, driving status, driving limitations, activities, unsafe driving, driving history, d riving rehabilitation, self rated driving ability, and mobility needs Manchester Driving Behaviou r Questionnaire (Lawton et al 1997) 28 items self report capturing demographics, aggressive violations, ordinary violations, errors and lapses Comprehensive Older Driver Assessment (Yee & Melichar, 1994) 208 item questionnaire capturing demographics, dr iv ing history and pattern s performance, environment, mobility, social network, activities, well being & outlook, and health indicators Knowledge Assessment Test (Yee & Melichar, 1994) 49 items capturing driving knowledge and traffic safety knowledge Att itudes Assessment Test (Yee & Melichar, 1994) 18 items capturing driving attitudes and attitudes about traffic safety Older Driver Self assessment Inventory (Yee & Melichar, 1994) 15 items capturing driving history, driving knowledge, and driving behavior s On road driving assessment for senior drivers (Kay et al., 2008) 19 items capturing driving behaviors and errors on maneuvers tested in an occupational therapy on road driving assessment Driving Comfort Scales (Myers et al., 2008) I tems capturing driv er co mfort for driving situations in the day 13 items and night 16 items. Behind the wheel evaluation (Justiss et al., 2006) 91 items (maneuvers) scored capturing driver errors for driving of varying complexity including straight drive, lane changes, turn s, and merging. Approach and Methods Community Based Participatory Research Community Based Participatory Research (CBPR) is a research approach based on acknowledging the community, tapping their expertise and knowledge,

PAGE 29

29 and partnering with the commun ity in the creation and implementation of research to address a public health problem (Israel et al., 1998). A CBPR approach has been advocated as a way to understand real life phenomena, as the approach utilizes the knowledge and perceptions of persons wi th first hand experience (Minkler & Wallerstein, 2003). Using CBPR, participants may be involved in all aspects of a project including defining the research question, selecting the methods and development of a protocol, providing data, and data interpretat ion. CBPR addresses an issue of concern to the target community, and knowledge building, empowerment and capacity building (though connection to resources) are integrated in the research (Green et al., 1995). Potential p ure may include identifying key behaviors or giving feedback on the relevance or understandability of items (Stone, 1997; Streiner & Norman, 2003). Processes that engage community participants such as focus groups or an advisory committee can be used along with measurement theory principles during item development to improve the acceptability, relevance, clarity, understandability, cultural relevance and formatting of the items and measure (Vogt, King, & King, 2004). For this dissertation to study driving behavior of older adults, we sought information from older drivers, families and other community members with an interest in the driving safety of older adults. Th is study design incorporates aspects of community based participatory research (CBPR) princip les with involvement of local stakeholders (via focus groups) and national and international stakeholders (via advisory committee) at multiple stages of the study, as well as item response theory (IRT) principles

PAGE 30

30 Focus groups Focus group studies involv e a series of discussions with groups of six to nine participants (ideally) on a topic of interest. Groups are moderated to provide an atmosphere conducive to participants sharing their knowledge, perceptions and experiences (Krueger & Casey, 2000). The us e of focus groups is compatible with IRT. Suggested uses of focus groups during item development include obtaining participant descriptions of the phenomena under study to inform item revisions, asking participants to critique and refine items (e.g., re wo rd for better understanding), and having participants rate items (i.e. compare relative difficulty of two items)(Stone, 1997). Classical Test Theory Validity Face validity is an initial judgment of whether a tool sufficiently assesses behaviors in the t argeted domains (Streiner & Norman, 2003). More rigorous Norman, 2003, p 19). As part of conten t validi ty, we will establish a priori validity for the SDBM items, constructing items based on team experience and previous research ( Classen et al., 2008; Classen et al., 2006; Classen et al., 2009) and comparing the areas covered to items from 11 existing drivi ng measures for older adults. The goal is to develop a set of items with no floor or ceiling, provide meaningful description s and capture safe driving on a continuum of behaviors. Content Validity will be further examined using a content validity index wi th 4 expert reviewers experienced in older driver safety research and measurement.

PAGE 31

31 Content validity Following development of an initial item set, judgment of the items by outside raters is an essential step to establish the relevance of the items and to obtain suggestions on revisions to improve the items (e.g., increase clarity). The ratings can be provided by an informal process (e.g., peer review by colleagues) or a more formal procedure, a content validity index (CVI). A CVI is a quantification of it em and measure relevance based on ratings by multiple expert raters (Lynn, 1986). Three or more raters are needed to provide a rigorous rating (Lynn, 1986), and the raters should have expertise (i.e. practice or research experience) in the content area un der investigation (Grant & Davis, 1997). A CVI uses a Likert scale rating of relevance as follows: 1 = not relevant, 2 = relevant with major revisions, 3 = relevant with minor revisions and 4 = very relevant. C ontent validity index raters also provide subj ective ratings of item accuracy, purpose, organization, clarity, appearance, concision, understandability, and adequacy. Content validity index analyses include calculation of an item level relevance rating and an average relevance rating for the item set. Analyses also include determining rater consensus and discrepancies in subjective feedback of item accuracy, purpose, organization, clarity, appearance, concision, understa ndability, and adequacy. Content validity index results are used to refine the items and the CVI process is repeated until an acceptable level of content validity is reached (average CVI of 0.90) (Waltz, Strickland, & Lenz, 2005).

PAGE 32

32 Criterion validity Cri terion validity can be evaluated by comparison of one measure such as a tool under development against a standard measure or criterion (Carmines & Zeller, 1979). Concurrent criterion validity is assessed if both measures at taken at the same point in tim e. Predictive criterion validity is assessed if outcomes of a measure are used to predict future performance measured by a criterion. Receiver Operating Characteristic (ROC) curves is a method of measuring criterion validity by examining the ability of th e measure to correctly identify positive (sensitivity ) and negative (specificity) cases (McDowell, 2006) Receiver Operating Characteristic curves can be used to compare different measures or to compare one measure under varied conditions (i.e. different raters or different cut point scores). When ROC curves are used to compare two measures at one point in time, in our case the SDBM ratings against the on road test outcome, concurrent criterion validity is being measured. Item Response Theory Item r espon se t heory (IRT) uses probability models to study abilities, behaviors or traits and create a hierarchy of item responses (Hambleton, Swaminathan, & Rogers, 1991). Examples of health applications of I RT include the NIH study PROMIS, using IRT to build measu res for 5 patient reported outcomes including physical functioning and social role participation (Castel et al., 2008). Item r esponse t heory facilitates measurement of a construct with items that describe a continuum of the construct under study, forming a n item hierarchy with no floor or no ceiling (Bond & Fox, 2007). Item sets developed with classical test theory typically have a tradeoff between item coverage (spread) and

PAGE 33

33 precision of measurement H owever, using IRT and Computer Adaptive Testing (CAT) th e items best suited for a respondent can be pulled from a larger bank providing both precision and spread. response theory i s summarized as follows. First, develop the construct/ trait to be tested on a less to more continuum, determining the characteristics of someone ideas of an items for a person at the low and high end of the scale then fill in gaps with additional item comparative approach, decide about item placement along continuum, comparing item germs whose location is known to items whose location needs to be determined. The initial set of item ge validity of this set of items comes from rater consensus. The next step is to use the item map to create full items, with wording that is descriptive and understandable. Several people should read and review each i tem and provide feedback on revisions to improve item clarity and reduce ambiguity. The final step is to administer the set of items to a sample from the target population and to conduct the Rasch analysis obtaining the measures and fit statistics. The Ras ch measures enhance the map of the variable and provide evidence for (or against) construct validity. Comparison of the initial map (item hierarchy hypothesized before testing) and the map generated from testing a sample provides evidence that the theory b ase for the items was correct and the items

PAGE 34

34 were well written, in this instance the hypothesized map and the map obtained from testing subjects will be close. Conclusion Self report tools are a potentially beneficial tool for use by older adults in ratin g their driving. However, current measures lack the necessary precision, accuracy depth and breadth of items. This dissertation presents three phases of a study to develop and test the Safe Driving Behavior Measure. Item development for the measure requir ed a theoretical base and a priori content validity of domains and items which were established in relation to existing measures and research. In addition, content validity was quantified with use of expert raters. Further development and item refinement w as informed by focus groups and participation of an advisory panel. Subsequently, ROC analyses described the s ensitivity, specificity and accuracy of the SDBM Following this structured process provided the necessary information to refine the SDBM at each step

PAGE 35

35 CHAPTER 2 ITEM DEVELOPMENT AND VALIDITY TESTING FOR A SAFE DRIVING BEHAV IOR MEASURE Background In a mobile society where older Americans are dependent on the automobile as the primary mode of transportation, periodic assessment of their driving saf ety is a key concern. Currently, this is being done through comprehensive driving evaluations (CDEs) administered by driving rehabilitation specialists; by community based programs such as those offered by AARP, the American Automobile Association (AAA), t he American Occupational Therapy Association (AOTA), and the American Society on Aging (ASA); family/caregiver assessments; and self assessments. Self report measures (based on self assessment) and proxy measures (based on observations from family, caregiv driving behaviors, increase driving safety awareness and knowledge, and promote behavior change and safer driving outcomes (i.e., fewer crashes, injuries or deaths) (Classen Lopez, et al., 2007; Eby et al., 2003; McGee & Tuokko, 2003). However, self report measures are limited by self selection bias (i.e., capable persons are more likely to complete the self report) and social desirability bias (i.e., persons are more likely to give ans wers that will be viewed favorably by others) (Sundstrom, 2005; Zhou & Lyles, 1997). The purpose of this study is to report on item development and validity of a new self report measure for older adults: the Safe Driving Behavior Measure (SDBM). Safe Driv ing Safe driving is the outcome of a sequence of events, and interactions among events occurring at the person, vehicle and environment levels. Safety can be characterized by the absence of near misses, errors, violations, crashes, and crash

PAGE 36

36 related injur ies or deaths. Safe driving (or lack thereof) for older adults has traditionally been measured by examining citation or violation history, crash rates, and crash related morbidity and mortality rates (Centers for Disease Control and Prevention [CDC], 1999; CDC, 2007; National Highway Traffic and Safety Administration [NHTSA], 2008 a ). Understanding driving behavior sets the foundation for developing a self report tool for older adult drivers. Driving behavior is complex with multiple underlying components su ch as attention, cognition, decision making and vehicle control (Michon, 1985). Older driver assessments need to capture specific driving characteristics including patterns of driving behaviors and driving errors. For example, compared to younger drivers, older drivers (65+ years) tend to engage in fewer risk taking behaviors (e.g., speeding, driving under the influence of alcohol, or not using a seatbelt). However, findings from crash datasets indicate that older adults have some salient areas of limitatio ns, primarily in relation to left turns against oncoming traffic and failure to yield right of way at intersections (McGwin & Brown, 1999). Furthermore, if involved in crashes, older drivers are at greater risk of injury than younger motorists (Awadzi, Cl assen, Hall, Duncan, & Garvan, 2008; Bdard et al., 2002; CDC, 1999; Dellinger, Kresnow, White, & Sehgal, 2004; Zhang et al., 1998). The accepted industry standard in North America and elsewhere for assessing driving errors and determining safety, especia lly for older adults with health impairments, is the comprehensive driving evaluation (CDE) (American Occupational Therapy Association, 2005; Canadian Association of Occupational Therapists, 200 9 ; Di Stefano & Macdonald, 2005; Korner Bitensky, Gelinas, Man Son Hing, & Marshall, 2005). The CDE is a battery of sensory, motor, and cognitive clinical tests and an on

PAGE 37

37 road evaluation conducted by a driving rehabilitation specialist (DRS), usually an occupational therapist with specialty training. CDEs have severa l limitations such as time needed to complete, high out of pocket costs for drivers, exposure to risks, need for trained personnel to administer the test, expensive equipment (e.g., a dual brake equipped vehicle), liability issues, and limited opportunity for access (Kua, Korner Bitensky, & Desrosiers, 2007; Wang & Carr, 2004). Furthermore, depending on the state or province, the driving evaluator may be legally or ethically required to report unsafe performance to the licensing authority, with the potentia l negative outcome of the driver losing his/her license. Moreover, the CDE occurs on the level of the individual, and is therefore not a viable option to serve the safety needs of the growing numbers of aging drivers, currently estimated at over 33 million in the U.S. and Canada (NHTSA, 2008 a ; Transport Canada, 2007). Self R eport Due to these limitations, a valid self report/proxy measure may provide an excellent opportunity for targeting older drivers with unsafe behaviors, and to solicit the feedback fr om their caregivers. A concern in using self reports to assess driving is the issue of selection bias and social desirability bias inherent to this method of assessment. Selection bias occurs when persons completing the self report differ from those who do not (e.g., persons confident about their driving may be more likely to complete a driving self report) (Zhou & Lyles, 1997). Social desirability bias impacts self reports if respondents answer according to how they would like to be perceived, typically ov erstating desirable behavior and understating undesirable behavior (Fowler, 1995; Lajunen & Summala, 2003). For example, Marottoli & Richardson (1998) reported that older adults showed high confidence for driving and tended to over rate their driving

PAGE 38

38 abili ty compared to the results of a driving evaluator. Bias and error in self reports can remembering the behavior well enough to respond accurately (recall bias), and concern about discl osing undesirable or unsafe behaviors (Fowler, 1995). These issues can be addressed by using modern statistical adjustments that account for anticipated response bias, by obtaining proxy report in addition to self report, and by using measures that are val Lin, Morgan, & Magaziner, 2005; Sundstrom, 2008). Valid self report measures may yield multiple benefits such as ease of use, maintaining confidentiality, and immediate feedback in a safe env ironment (such as the home). A driver can gain insight on his or her driving safety status without risking the report can be made available with little to no cost to the broad older driver populat ion in the U.S. and abroad (Classen Lopez, et al., 2007; Eby et al., 2003; McGee & Tuokko, 2003). Furthermore, from a measurement standpoint, self reports have shown positive correlations with an on road evaluation (Pachana & Petriwskyj, 2006; West, Frenc h, Kemp, & Elander, 1993). Among currently available self use outside of a driver improvement course or research setting. These measures include the Driving Decisions Workbook (Eby et al., 2003), Ol der Driver Skill Assessment and Resource Guide (AARP, 1992), Drivers 55 plus: Check Your Own Performance (AAA Foundation for Traffic Safety, 1994) and the computer based Roadwise Review (American Automobile Association, 2004). Strengths of these

PAGE 39

39 measur es i nclude assessment of driving related medical conditions and medication use (Driving Decisions Workbook), measurement of ph ysical and cognitive abilities associated with crash risk (Roadwise Review) and education for drivers about risk factors and strategie s for driving safely (Driving Decisions Workbook, Drivers 55 plus: Check Your Own Performance, Older Driver Skill Assessment and Resource Guide and Roadwise Review). Of these measures, only the Driving Decisions Workbook development entailed comparison wit h a criterion measure to determine the influence of self report bias (e.g., social desirability) (Eby et al., 2003). The Driving Decisions Workbook and Roadwise Review both incorporated focus group feedback from stakeholders during measure development (Eby et al., 2003; Staplin & Dinh Zarr, 2006). this measure has been criticized for a lack of published psychometric testing in validity (Eby, Trombley, Molnar, & Shope, 1998). Limitations of paper measures designed with an educational focus (i.e., emphasizing safety knowledge such as driving strategies) include their length (20 47 pages) which increases respondent burden (AAA Foun dation for Traffic Safety, 1994; AARP, 1992; Eby et al., 2003). The computer based Roadwise Review takes approximately 40 minutes to complete, may be challenging for older adults with low computer fluency to use, and requires assistance of another person t o complete ( Myers, Blanchard, MacDonald, & Porter, 2008; Staplin & Dinh Zarr, 2006). Lastly, none of these measures was developed from a socio ecological framework; the current measures have a person focus and lack adequate vehicle and environment items.

PAGE 40

40 Self report challenges include making the tool available for the target population, designing a tool that supports self administration, and ensuring the behavioral measures are relevant for a broad group of older drivers. A need emerges for a valid and rel iable self report/proxy report to : 1 ) identify safe/unsafe driving behaviors in older adults 2 ) provide feedback on strategies for driving safer and longer 3 ) indicate any necessary referrals to address problems. This need can be more comprehensively fulfilled by using community based methods (building a valid and reliable instrument); and using item response theory (providing meaningful and measureable descriptions at the level of the person and the item). The contribution of Community Based Participatory Research (CBPR), traditional measurement methods, and item response theory are discussed next. Community Based Participatory Research (CBPR) CBPR is a research approach b ased on acknowledging the community, tapping their expertise and knowledge, and partnering with the community in the creation and implementation of research to address a public health problem (Israel et al. 1 998). Using CBPR, participants may be involved in all aspects of a project including defining the research question, selecting the methods and development of a protocol, providing data, and data interpretation. CBPR addresses an issue of concern to the target community, and knowledge building, empower ment and capacity building (though connection to resources) are integrated in the research (Green et al., 199 5 ). Potential contributions of participants in developing a measure include identifying key behaviors and providing feedback on the understandabili ty of items (Stone, 1997; Streiner & Norman, 2003). Use of a CBPR approach captures the real life phenomena under study

PAGE 41

41 by utilizing the knowledge and perceptions of persons with first hand experience, persons such as older drivers, families and other comm unity members with an interest in the driving safety of older adults (Minkler & Wallerstein, 2003). In developing a driving self report, description of the real life phenomena accounts for driving differences across settings due to geographic variations (e .g., terrain or weather) and cultural norms. During development of the measure, processes that engage community participants such as focus groups and feedback from an advisory committee can be integrated with measurement theory principles to improve the ac ceptability, relevance, clarity, understandability, cultural relevance and formatting of the items and measure; all of which enhance the validity of the measure (Vogt, King, & King, 2004). Measurement Theory Items developed for a measure must have accept able face and content validity. Face validity is an initial judgment of whether the tool assesses the behaviors it is supposed to (Streiner & Norman, 2003). Carmines and Zeller (1979) define a measure with content validity as one whose collective items ade quately represent the construct under investigation. Defining the construct and related domains for inclusion in a measure using theoretical frameworks is a critical step in establishing face and content validity (Lynn, 1986). Following development of an i nitial item set, judgment of the items by outside raters is essential for further item revisions and improvements. Content validity is assessed by a quantification of item and measure relevance obtained from expert raters, using a content validity index (C VI) (Lynn, 1986). traditional methods have limitations when the goal is development of accurate items to ample, a precise and

PAGE 42

42 objective measurement of driving function requires assessing the difficulty a person has with driving situations of varying complexity such as parking or merging. Therefore, in addition to traditional measurement methods addressing val idity, there are advantages to using an item response theory approach (Bond & Fox, 2007). Item r esponse t heory (IRT) is a measurement model that relates item difficulty for a measure to person ability, thereby guiding more precise measurement (Bond & Fox, 2007). Item response theory has been used in driving studies to develop driving scales and to analyze behaviors observed in on road testing (Kay, Bundy, Clemson, & Jolly, 2008; Myers, Paradis, & Blanchard, 2008). In constructing a measure using IRT, b ehav iors representing the construct under consideration (e.g., safe driving) are outlined on a continuum of item difficulty with items for persons with a greater or lesser capacity (Stone, 1997). Items are designed with a range and specificity to maximally sep arate individuals based on ability. For example, we propose that individuals who are designated lane when driving. On the other hand, individuals who are safer dri vers will as controlling their car on a snowy road. Therefore, given the benefits of IRT, such as obtaining increased precision, the IRT method becomes particularly us eful when measuring a functional behavior such as driving. Rationale and Significance We do not have an existing valid and reliable self report/proxy measure of older driver safety behaviors that is accessible, relevant, culturally sensitive, geographica lly representative and appropriate for a broad population of older drivers in North America (the long term focus of this project). Clearly, such a safe driving behavior measure has

PAGE 43

43 utility for older drivers, concerned family members, caregivers, aging advo cates, and professionals on a community and population wide level. Purpose Using traditional measurement methods and IRT, as well as CBPR principles, our objective is to report on item development and validity of a new self report measure for older adu lts: the Safe Driving Behavior Measure (SDBM). The use of this measure will position occupational therapists, and agents of the aging networks to influence older driver safety at the population level. Methods Setting Two geographically diverse and cultural ly rich sites (in Florida and Ontario) participated in this study. The participating universities, one at each site, both have a driving evaluation / research center with an established record of older driver research. Research Team Across the two sites, t he research team was comprised of five PhD level researchers with combined research expertise in older driver safety, public health, measurement theory, item development, CBPR and biostatistics. Two additional team members participated in the development, refinement and testing of the SDBM, one an occupational therapist and Certified Driving Rehabilitation Specialist, and the other an occupational therapist and doctoral candidate in Rehabilitation Science. For this phase of the study, we included community participation from four expert raters representing U.S, Canada and Australia; and the feedback from five advisory panel members representing, the Canadian Association of Occupational Therapists (CAOT), AARP, Transport Canada, University of Michigan Transpo rtation Research Institute, and

PAGE 44

44 opinions of 14 focus group participants in Thunder Bay, Ontario, and 18 focus group participants in Gainesville, Florida. Process for Item Development Our methods for item development included the CBPR approach, measurement principles, guiding frameworks including three theoretical models, previous research, published driving self report measures, published driving evaluations from the older driver safety literature, and expert opinion, each discussed below. Precede Proceed Model of Health Promotion This team has an established publication record of research on older driver safety issues using the Precede Proceed Model of Health Promotion (PP MHP) (Green & Kreuter, 2005). The PPMHP domains address safety influences on the level of the driver (e.g., knowledge, behavior, attitudes or health status) ( Awadzi et al., 2008; Classen et al., 2006, Classen Lopez, et al., 2007); the vehicle (e.g., safet y features or equipment)( Awadzi et al., 2008; Classen Lopez, et al., 2007); the physical and political environment (e.g., day time conditions, or state licensing policies) ( Awadzi et al., 2008; Classen Lopez, et al., 2007); the social environment (e.g., driving laws or assistance with transportation) (Classen et al., 2009 ); and person vehicle environment interactions (Classen et al., 2008) Collectively our research demonstrated that behavioral and environmental factors are understudied, yet importan t determinants for older driver safety; and that a comprehensive assessment must include person, vehicle and environmental factors, as well as the interactions among these factors ( Classen et al., 2008; Classen Lopez, et al., 2007).

PAGE 45

45 M atrix Ha M atrix (1972) assumes that injury is a preventable occurrence and provides a framework to target risk or protective factors that can occur at the person level, means of injury transmission level, environment level (physical and social driving contex t) and vector level (physical forces). Haddon conceptualizes behaviors related to injury on a time continuum. For the purpose of this study, we defined a time continuum for behaviors focusing on two points: before driving (e.g., checking vehicle), and duri ng driving (e.g., driving maneuvers). behavior on three levels, strategic, tactical, and operational. Strategic behaviors can occur before or during driving, incl ude general goal formation decisions (e.g., trip planning) and occur in minutes to hours. Tactical behaviors reflect conscious decisions made while operating the vehicle relating to interaction with the immediate driving environment (e.g., maneuvers or car handling) and occur in seconds to minutes. Operational behaviors are outputs that happen primarily by subconscious action (e.g., maintaining your lane) and occur in seconds. Existing Measures We reviewed several existing driving self report and evaluati on measures (Hunt, personal communications, April 30, 2007; Justiss, 2006; Kay et al., 2008; Myers Paradis, et al., 2008; Reason et al., 1990; Stalvey & Owsley, 2000; Owsley et al., 1999 ; Yee & Melichar, 1992). We extracted and coded items from these measu res according M M atrix, items were coded as addressing person factors

PAGE 46

46 (skills, knowledge, attitudes, or health status), vehicle factors (selection and use of car features) or environment factors (physical environment e.g., weather; social environment e.g., passengers in car). We coded items according to when the behavior occurred (pre (strategic, tactical or operational). Overall, we found the items on the existing measures were biased toward person factors with minimal representation of environment and vehicle factors or person vehicle environment interactions. Moreover, most items ass essed driving deficits (e.g., failure to maintain proper speed for conditions), thus not representing the broader range of driving behaviors demonstrated by older adults. SDBM Tool Design We designed the SDBM as a self report tool for use by older adults in assessing driving behavior. Modified versions of the tool were created for use by family members, caregivers and professionals. The current paper and pencil version of the SDBM is an older driver self report measure, including instructions, demographic profile (e.g. age, gender), driving history profile (e.g., driving habits and patterns) and 68 questions on driving ability (Appendix A [driver] and B [caregiver] ). For each driving behavior item, the respondent was asked to rate their driving difficulty with that behavior over the last three months. To minimize recall bias (e.g., failure to remember previous driving behavior accurately) we selected three months as the response time period (Fowler, 1995). For most respondents, a three month driving period would incorporate a variety of driving situations while allowing for easy recall of driving behavior and difficulties encountered. considering difficulty to be more inf ormative of safety and congruent with IRT principles. Based on team judgment and peer review feedback we used five response levels for

PAGE 47

47 the driving behavior items: cannot do, very difficult, somewhat difficult, a little difficult Item Development Working from our earlier definition of driving behavior, team expertise, previous research findings and existing measu res, we generated item ideas representing a continuum of driving behavior from least difficult (e.g., locate controls such as lights or 2 1 outlines the step by step process for item development. Team members worked consistently (2 3 hours per week x 6 months) and collaboratively to initiate, discuss, refine, reword, add, or delete ideas for items, and to determine item placement along the continuum. We developed driving behav ior items to provide adequate representation of person vehicle, person environment and person vehicle (before and during driving ) and including strategic, tactical and operational concepts. Item Review The item review process was completed in two stages. First, a group of peer reviewers (five doctoral students and four PhD level trained qualitative researchers) from within the uni versity provided input and recommendations on the face validity of the items and the utility of the measure (e.g., ease of completing items and time needed). Following the guidelines of Lynn (1986) who suggested that three or more raters are adequate to pr ovide a rigorous rating, we asked four expert raters from our advisory committee to complete a content validity index (CVI). They rated the relevance

PAGE 48

48 of each SDBM item on a four point Likert scale (1 = not relevant, 2 = relevant with major revisions, 3 = r elevant with minor revisions and 4 = very relevant) and gave feedback on item accuracy, purpose, organization, clarity, appearance, understandability, and adequacy (Grant & Davis, 1997). A sample CVI form item is shown in Figure 2 2. Figure 2 1. Steps in Process of Item Development and Testing for the Safe Driving Behavior Measure Add items capturing vehicle and environmental factors based on previous research Brainstorm to generate item ideas Assess face validity thro ugh peer review Review existing driving measures using 3 theoretical frameworks Revise item set Re write items to capture safe driving behaviors Acceptable face and content validity (CVI of 0.80) reached? Yes, accept item set No, repeat ste ps from set" Frameworks: 1) Precede Proceed Model Assess content validity through Content Validity Index (CVI) Previous research: 1) Systematic Literature Review 2) Actuarial data (FARS Fatality A nalysis Rating System) 3) Metasynthesis 4) Other published self report studies Place items on continuum of hierarchical difficulty Add or delete items, revise item descriptions, revise rating scale Hierarchical difficulty obtained from principl es of Item Response Theory (IRT) Focus groups: Add items from field notes

PAGE 49

49 Item Expert Ratings CVI Rater Comments Item level CVI 58. Drive confidently without periodic formal testing (such as vision, hearing or driving a bility tests)? 2 1 1 4 AA= still seems unclear, is the testing supposed to make the driver more confident? You are bring ing in a new a person to judge how much difficulty they have driving confidently. This is just too confusing. CC= meaningless "confidently" can be justified or unjustified DD = no comment 0.25 Figure 2 2. Examples of a CVI Rating including Feedback from Expert Reviewers Legend: Expert rating 1 = not relev ant, 2 = relevant with major revisions, 3 = relevant with minor revisions and 4 = very relevant; AA, BB, CC, DD = each of the four raters; Item level CVI indicates the level of agreement on relevance between four raters, with 1= agreement among 4 raters, 0 .75= agreement by 3 raters, 0.5 agreement by 2 raters and 0.25= relevance rating from one rater. Analysis Our analysis plan included calculation of both an item level CVI and the scale average CVI. Based on CVI procedures (Lynn, 1986), rater scores were c ollapsed with a item and comments in a spreadsheet, we calculated item level CV I scores (i.e., the scale average CVI. Item level CVIs of 0.75 or 1.00 were acceptable (0.75 = the item was rated as relevant by 3 raters; 1.00 = the item was rated as relevant by 4 raters) while scores of 0.5, 0.25 or 0.0 (0.5 = the item was rated as relevant by 2 raters, 0.25 =

PAGE 50

50 the item was rated as relevant by 1 rater and 0.0 = no raters indicated relevance) were unacceptable. Th e statistical package R: A Language and Environment for Statistical Computing, version 2.5.0 (R Development Core Team, 2008) was used to comput e the CVI. Following the analysis, items with a low item CVI (< 0.5) were revised by the research team. The CVI process and item refinement were repe ated until a cceptable Results Peer Review P eer review feedback indicated overall favorable face validity and utility Reviewers suggest ed revisions for tool format and 11 (18%) of the 60 SDBM items. Specific peer review results are shown in Table 2 1 according to the type of feedback (e.g., pertaining to instructions, wording, relevance, or response level). Table 2 1 Peer review results for face validity Revision categor y Peer reviewer comment Instructions (2 recommendations) Item clarity (4 recommendations) Allow of medication effects targeted by item. for intersection item Enhance description of item describing maki ng a left turn with Response levels (2 recommendations) well as extreme, moderate, somewhat, and not difficult. Use numbers o r codes for responses instead words. Relevance (1 recommendation) drivers as being important Content Validity Index First round CVI results indicated 44 (73%) of the 60 SDBM items had an acce ptable it em

PAGE 51

51 relevant). Sixteen (27%) of the items had an item lack of relevance or the need for a major revision of that item. The scale average CVI was 0.78, falling short of the acceptable scale average of 0.80. Collectively, the item level CVIs and the scale average CVI indicated a need for revisions and a repeat of the expert review and content validity index (CVI). Therefore, we revised the 16 items with a CVI rating of <0.5 as follows: 6 items with low relevance were deleted, 1 item was moved to Section A (demographic questions), 1 item was grouped in the scale with items of similar difficulty, 6 items were re worded, and 2 items were retained without revision based on team judgment and clinical or theoretical significance (Table 2 2). We obtained a second round of ratings for the CVI to re assess 5 revised items and to obtain initial ratings of 21 new items added following 3 focus groups condu cted at the Florida and Ontario sites. Second round CVI results indicated the 5 revised SDBM items had an acceptable item acceptable item level CVI of 0.5 o r less, indicating either a lack of relevance or the need for a major revision of that item. Therefore, we revised the 6 items with a CVI rating of <0.5 as follows: 1 item with low relevance was deleted, 2 items were moved to Section B: Driving history pro file, and 3 items were re worded following rater feedback. Following these revisions the final scale average CVI, on the retained items from the first and second round ratings, was 0.84. Thus the final scale average was above the acceptable scale average l evel of 0.80 (House, House, & Campbell, 1981), indicating relevance of the SDBM items.

PAGE 52

52 Table 2 2. Items and Content Validity Index (CVI) before (initial) and after (final) team revisions and expert ratings Item Initial CVI Item revision Final CVI Fas ten your seatbelt 0.5 Moved item to Section B: Driving history profile Not rated Check mirrors when backing out from driveway or parking space 0.5 Revised wording to incorporate both check mirror AND turn head 1.0 Locate the control for your hazard lig hts 0.5 Deleted in concurrence with rater feedback of low relevance Not rated Keep both hands on the wheel while driving 0.5 Deleted in concurrence with rater feedback of low relevance Not rated Keep wheels straight while waiting to turn 0.0 Deleted in c oncurrence with rater feedback of low relevance Not rated Pull out smoothly from a stop sign with no harsh acceleration 0.25 Deleted in concurrence with rater feedback of low relevance Not rated Pull in or back out of tight parking spots 0.5 Re worded to follow CVI rater 0.75 Maintain car speed within 5 miles/8km over or under the posted speed 0.5 Deleted addressed by SDBM Not rated Drive with tractor trailers (transport tr ucks) around you 0.5 Retained based on team judgment. 0.5 Scan left to right at intersection 0.5 Re 1.0 Complete a 100 mile (160km) drive on your own 0.5 Deleted in concurrence with rate r feedback of low relevance Not rated Drive up or down a steep hill 0.5 Re 0.75 Drive when you are upset 0.5 Re anxious, worried, sad or angry 0.75 Drive after taking medication 0.5 Moved item to Section B: Driving history profile Not rated Drive in an unfamiliar area 0.5 Retained based on team judgment 0.5 Drive through a railroad crossing 0.5 Deleted in concurrence with rater feedback of low relevance Not rated The item pool is currently distributed across three main domains (Appendix A): Section A Demographic p rofile, Section B Driving h istory p rofile and Section C Driving

PAGE 53

53 behaviors The item pool in Section C yields a combination of items capturing behaviors across the span of person, vehicle, and environment. For example, Section C captures 68 safe driving behaviors classified in the person vehicle domain (11 items), person environment domain (42 items), and person vehicle environment domain (15 items). Di scussion There is a growing population of older drivers who would benefit from a self report measure designed to provide targeted feedback on their driving ability. The purpose of this study was to describe the item development, and to report on validity a nalyses and rater agreement for the SDBM, a self report measure for older drivers. In this study, we illustrate the use of theoretical frameworks for developing a self report for measuring safe driving behavior and indicated limitations of existing self re port measures in meeting the need for precise measurement tied to driving ability. The SDBM is unique among self report/proxy measures in that we used a combined measurement approach with traditional measurement strategies addressing validity and IRT strat egies to increase the precision of items. Furthermore, we employed a CBPR approach to inform item development and contribute to geographic and cultural relevance of the items. Development of the SDBM was informed by feedback from a peer review group and t wo rounds of expert rater review. In describing the process for development of the SDBM, we intend to further the body of knowledge on self/proxy reports for use in promoting awareness of safe driving behavior by older adults or their caregivers. The face and content validity of this tool, as well as the relevance determined by expert raters during two rounds of reviews, provide early support for the utility of this item pool and for further developing the measure.

PAGE 54

54 Limitations of self report will be accou nted for in the three following ways. We are implementing steps to minimize the effects of selection bias by providing a tool that can be used privately and confidentially. We will manage social desirability bias by comparing results with a companion proxy report from a caregiver, and by conducting formal on road driving assessments from which we will discern concurrent criterion validity between the self report, proxy report and on road test. Finally, we are managing recall bias by asking participants abou t their driving behaviors in the last three months. We did not conduct rater reviews on all the items during the second round of reviews, but only on the revised items (from previous rater feedback), and new items (from focus groups). Strengths of the SDB M include: theoretical support from three frameworks (Precede Proceed Model of Health Promotion collaborative item development process solicited feedback with revisions completed at several time points, and a final r ater agreement of 84% on the validity of the tool. Finally the items have been developed to capture behaviors across the span of the person, vehicle, environment, and their interactions. With future research, including testing in a multi site study, we aim to provide a valid and reliable tool informing older adults, family members, caregivers and professionals about driving behavior and safety. We assert that, given accurate s referrals for driving evaluation or remediation will be more readily accomplished.

PAGE 55

55 CHAPTER 3 CONTRIBUTION OF FOCU S GROUPS TO DEVELOP A SELF AND PROXY REP ORT SAFE DRIVING BEHAVIO R MEASURE Background ty mobility, which is the p. 631). Specialist occupational therapists evaluate driving skill s and safety (American Occupational Therapy Association, 2005; Canadian Association of Occupational Therapists, 2009). An occupational therapy evaluation of driving may include a self report, or a proxy report of driving completed by family or professional s, addressing an Self reports have the potential to enhance safety by increasing awareness of driving behaviors (Eby, Molnar, Shope, Vivoda, & Fordyce, 2003; National Hi ghway Traffic Safety Administration [NHTSA], 2008 b ). Self reports may be used by older drivers or family members outside of the clinic to assess driving deficits and inform decision making about seeking professional evaluation or treatment (NHTSA, 2008 b ). Research on self reports and their effectiveness is a necessary step in improving driver safety (Dickerson, et al., 2007). In developing a new measure, focus groups may be used to help researchers access the knowledge and perceptions of persons with first hand experience (Minkler & Wallerstein, 2003). Moreover, focus group methodology can inform revisions and promote the acceptability, relevance, clarity, understandability, cultural relevance and formatting of the existing items and measure (Vogt, King, & King, 2004). Client participation may involve identifying key behaviors or giving feedback on the relevance

PAGE 56

56 or understandability of items (Stone, 1997; Streiner & Norman, 2003). In our review of existing self reports, described further below, we ascertaine d whether client participation was used during tool development. As a first step in development of a driving tool, we reviewed 11 existing driving self report and evaluation measures. We examined items from these measures according to a theoretical frame work based on the Precede Proceed Model of Health Specifically, our review of measures found person and environmental factors were emphasized with a gap in items addressing the vehicle. Furthermore, items of existing tools primarily focused on deficits rather than presenting a range of behaviors from unsafe to safe to more precisely measure drivin g behaviors. To our knowledge, only two existing self report measures incorporated client participation during measure development, the Driving Decisions Workbook (Eby et al., 2003) and Roadwise Review (Staplin & Dinh Zarr, 2006). The Driving Decisions Wor kbook is a paper and pencil measure including driving behavior questions with related educational content about driving risks and recommended behaviors. The Driving Decisions Workbook has been used as part of educational interventions with older drivers, w ith a recent enhancement of health conditions information (NHTSA, 2008 b ). The Roadwise Review software is a computer based self assessment that includes versions of tests predictive of safe driving (e.g., visual field, gait speed) and is distributed by the American Automobile Association (2004). Based on our review of existing measures, and the limitations

PAGE 57

57 identified (few vehicle items and focus on deficits), we proceeded with development of a new driving self report. Rationale and Significance We assert of the SDBM. Our development of a new self/proxy report is based on identification of domain ga ps (e.g., lack of vehicle items), the need for items measuring a broader range of driving behaviors, and the lack of self reports developed with input from older drivers and family members. Thus, we conducted focus groups with older drivers and their famil y members to develop and evaluate items for a self/proxy report tool, the Safe Driving Behavior Measure (SDBM). Purpose We used three focus groups conducted at two sites to: 1) expand on or generate new items reflecting safe driving behaviors in the phys ical and social environment; and 2) obtain feedback for refining the initial item set. We therefore solicited comments from participants on item relevance, difficulty, clarity, and revisions needed. Methods This focus group study is part of a broader proj ect that utilized conceptual theoretical frameworks, community based participatory research (CBPR) principles and item response theory (IRT) in developing the SDBM. This study included focus groups conducted at the University of Florida (UF) in Gainesville Florida and Lakehead University (LU) in Thunder Bay, Ontario. The research team includes five PhD level researchers with backgrounds in occupational therapy, driving evaluation/rehabilitation,

PAGE 58

58 rehabilitation science, public health, psychology, measuremen t, and biostatistics; one doctoral level research assistant, two graduate level research assistants, and two driving evaluators. Team members at both sites followed IRB approved research protocol to promote consistency in recruitment, consent, focus group moderation, and data collection. Procedure SDBM development Three theoretical models formed the basis for the instrument: the Precede Proceed Model of Health Promotion (PPMHP) (Green & Kreuter, 2005) matrix for injury prevention (1972) and Mich (1985) Informed by these models, we established three domains for item generation: Person Vehicle (PV), Person Environment (PE), and Person Vehicle Environment (PVE). The PV domain includes behaviors related to use of car co use of emergency brake). The PE domain includes behaviors in response to physical factors (e.g., terrain or weather) or social factors (e.g., interactions with passengers). The PVE domain includes behaviors combining a pe behaviors in the use of vehicle features or controls and a response to environmental report, with domain based items, has three sections: section A demographic profile, section B driving history profile, and section C driving behaviors. The proxy report includes revised sections A and C. Using IRT principles, driving behavior items ranged from easy to challenging items. Based on the last thre e months, respondents must rate the difficulty of each driving behavior on a 5

PAGE 59

59 process and findings are reported elsewhere ( Classen et al., in press). Focus groups To generate items, two focus groups (UF, LU) examined driving situations (e.g., challenging situation). The third group (UF) revi sed and refined the items and commented on formatting, clarity, relevance, and utility. Institutional review board approval was granted for this project. Participants provided telephone and written informed consent prior to focus group involvement and were paid $50.00 to participate in this portion of the study. Focus group guides. We developed focus group guides for Focus Group 1 (FG1) and Focus Group 2 (FG2) with the same 12 questions and related prompts for both sites, exce pt for local street references (Appendix C ) We organized questions according to the PV, PE, and PVE domains. For example, to address the PV domain we asked respondents to describe car features that contributed to their comfort, safety, or driving ease; for the PE domain we asked for d riving descriptions of different geographic conditions (e.g., driving in a hilly vs. flat terrain) as well as social factors (e.g., driving while negotiating pedestrians and scooters); for the PVE domain we asked for a description of challenging driving si tuations (e.g., driving in stormy weather). We concluded groups by inviting participants to share any additional information. For Focus Group 3 ( FG3 ) we administered SDBM sections A demographic profile, B driving history profile, and C driving behavior s. We asked participants about their experience of answering the questions, the difficulty, ease or understandability of the items, and

PAGE 60

60 whether the items captured relevant driving behaviors. We also asked for item feedback related to wording, formatting, a nd response options. Participants The participants were drivers aged 65 85 and their family members aged 18 85. We recruited participants by distributing flyers, through community presentations (e.g., at senior meal sites), advertisements, word of mouth referrals, and using participant registries. Inclusion criteria were: 1) personal experience with older driver safety, 2) physical and mental endurance to participate in two hour focus group, 3) ability to communicate in English, and 4) community dwelling. Exclusion criteria were physical or mental conditions (based on participant self report) that would impair the ability to participate in a two hour meeting and/or answer questions. Data Collection The focus group sessions were all approximately 90 minutes moderated by research team members, and documented with audio recording and field notes. Following each focus group, we uploaded the audio files to a secure server and had the content professionally transcribed. The doctoral research assistant verified t he verbatim transcripts for accuracy against the audio recordings and field notes before importing the transcripts into ATLAS.ti qualitative data analysis software (Muhr, 2004). Data Analysis We used content and thematic analyses, and an iterative team r eview process, to code responses from all three focus groups. Content analysis consists of identifying, coding, and counting specific descriptions (e.g., behaviors) (Miles & Huberman, 1994) Thematic analysis involves both identifying a priori themes (esta blished before the analysis) and generating new themes by labeling and categorizing recurrent responses

PAGE 61

61 (Ryan & Bernard, 2003) The doctoral research assistant had prior training and experience on use of ATLAS.ti qualitative software. We coded focus group responses (i.e., labeled text sections) that described a driving behavior and the relevance of a items (content validity), 2) new themes/items, 3) item revisions, or 4) general feedback. Content analysis We used content analysis to identify, code, and count specific driving behavior descriptions pertaining to existing SDBM items. For example, the respondent comment g in and out of the vehicle is very There is inherent overlap in the domains as driv ing behaviors all involve the person, vehicle, and environment to some extent. Therefore, in order to improve data analysis, person adapting the vehicle and was coded to the Person Vehicle domain. Thematic analysis We used thematic analysis to identify a priori themes (i.e., behaviors covered in the existing item set); and to identify and c ode driving behaviors and concerns of participants that were not previously captured with our existing items. Coding for new behaviors and concerns generated additional themes and items. For example, we five respondent comments (FG1 and FG2) emphasizing of the importance of vehicle maintenance on safe driving.

PAGE 62

62 Iterative Process of Item Development Figure 3 1 illustrates our iterative process in selecting and refining items. Working from an initial it em set and item hierarchy (A and B in figure), we held three focus groups (C) to discuss driving and review the SDBM. We analyzed focus group data to generate items and inform item revisions (D). The multi site research team reviewed the findings, interpre ted the data, revised and refined the items (E). We arranged items according to the hypothesized level of difficulty (e.g., we placed driving in daylight hours as easier than driving at night) (A and B). The cycle of data analysis and interpretation (A thr ough E) was on going from FG1 to FG3 (three month period). Following focus group analysis (three iterations of A through E), a revised item set was prepared for validation with on road testing. Figure 3 1. Iterative P rocess f or I tem D evelopment, R evisions and/or R efinement (E) Team discussion / data interpretation (D) Content and thematic analysis (A) SDBM item set development including revisions and/or additions (C) Focus groups with old er drivers and family members (B) Rate each item according to a h ierarchy of difficulty using Item Response Theory (IRT) principles

PAGE 63

63 Results Table 3 1 outlines the demographic characteristics of participants in each of the three focus groups. FG1 (n=14) had 7 older drivers and 7 family members, FG2 (n=6) had 5 older drivers, and 1 family member, and FG3 (n=11) had only older drivers. We excluded five persons reporting medical reasons (e.g., cognitive issues or poor health) questions. Table 3 1. Demogra phic Profile of Participants from three Focus Groups. Demographic Variables Group 1 Group 2 Group 3 Focus Group Respondents (N=31) Older Driver (n=23) Family Member (n=8) n=14 7 (50%) 7 (50%) n=6 5 (83%) 1 (17%) n=11 11 (100%) 0 Age Mean an d Standard deviation (SD) Older Driver Family Member 72.43 (3.26) 50.43 (21.61) 68.4 (3.97) 47 (0) 70.27 (5.1) N/A Gender Older Driver Female (n=14) Older Driver Male (n=9) Family Member Female (n=6) Family Member Male (n=2) 4 (29%) 3 (21%) 5 (36%) 2 (14%) 3 (50%) 2 (33%) 1 (17%) None 7 (64%) 4 (36%) NA NA Race African American (n=3) Caucasian (n=28) 0 14 (100%) 1 (17%) 5 (83%) 2 (22%) 9 (78%) Education Less than high school (n=2) Hig h School/ Vocational school (n=13) Some college after high school (n=4) 2 (14%) 9 (64%) 0 1 (7%) 2 (14%) 0 3 (50%) 0 1 (17%) 2 (34%) 0 2 (18%) 4 (36 %) 4 (36%) 1 (9%) Focus Groups 1 and 2 Respondents provided in depth driving descriptions, discussing behaviors they perceived as safe (e.g., slowing down) as well as risky (e.g., crossing lanes of a highway). Our content analysis of 202 respondent quote s identified driving behaviors in

PAGE 64

64 the PV (15 quotes), PE (129 quotes), and PVE (58 quotes) domains. Based on the content analysis, we identified comments pertaining to 46 of the 72 existing SDBM items. Thematic analysis led to 23 new themes as listed in Ta ble 3 2. Table 3 2. Thematic analysis results listing 23 new themes. Themes in alphabetical order Alternate transportation Bicyclists, scooters and motorcycles Driving another vehicle Erratic behavior of others Health factors Interpreting traffic light s Keeping up with road rules Long distance driving Maintaining your vehicle Merging Navigation and map use Parallel parking Passing lanes Passengers Persons running red light Reacting to animals Road features Safety Concerns Self regulation Signage Training/defensive driving Vehicle features Visual scanning Of 23 themes, 16 remained after we collapsed themes with a similar meaning ( ) These 16 themes led to 5 new items for section B D riving history profile, and 11 new items for section C D riving behaviors. Table 3 3 summarizes focus group findings by item domains. Quotes from older drivers and family members, exemplifying the item domains, are inte grated below. Table 3 3. Focus groups 1 and 2: Quote summary listed by Person Vehicle Environment domains Item quotes by domain Focus Group1 ( FG1) Drivers FG1 Family Focus Group 2 ( FG2) Drivers FG2 Family Total quotes of Drivers and Family Members Person Vehicle 5 3 6 1 15 Person Environment 51 20 52 6 129 Person Vehicle Environment 22 20 15 1 58 TOTAL 78 43 73 8 202

PAGE 65

65 Person Vehicle domain (PV) features, car fit, ma intenance, driving other cars, adjusting mirrors, turn signal or headlight use, and seat adjustment. For example, five participants said car maintenance influenced their safe driving, as exemplified in this quote: You can't do anything about it [slush/spra y from trucks], so you have to slow down and make sure you get your windshield cleared off. About the only safety factor out there is to slow down and make sure that your vehicle is mechanically fit and your wipers are working and your windshield washer's working. (FG1 Participant 14, Older driver, Female) Car design features also influenced perceptions of safe driving. Twenty respondents discussed car features including blind spots, seat adjustment, mirrors, headrests, and technology options like Global Positioning Systems (GPS). One driver reported her new car negatively impacted her safety due to blind spots and the headrests: I just purchased a new car [with] blind corners... really quite pronounced, as compared to my old one [car]. Also, the head re st they have on the back finding my old car was safer. (FG1 Participant 7, Family member, Female) PV included fitting in your seat so you could see over the steering wheel. One driver stated the relevance of including car fit on the SDBM explaining: her, simple thing like that. (FG2 Participant 1021, Older driver, Female)

PAGE 66

66 Person Environment domain (PE) Weather, road features, and the b ehaviors of other drivers were common topics in the PE domain. Quotes described factors in the physical environment like road debris as well as social factors such as passengers. Respondents discussed how the presence of passengers could either hinder safe ty (e.g., too much talking) or help (e.g., passenger navigated). One driver related his safety concern about talking with his wife: We don't talk that much because I find it can be distracting too, so I just crank up the rock tunes a little bit. I fi nd I can get carried away if I'm talking too; I start to lose my concentration, so it's not that we don't like each other; we just try to stay safe. (FG1 Participant 2, Older driver, Male) lity to interpret lights and react appropriately as described by one family member: Well, a lot of people don't understand if there's a green arrow, obviously you make a left. If they don't want you to do that, then they put a red arrow, but if the that you can still make a left hand turn if traffic is clear. But, I see the old ladies who sit there. (FG2 Participant 1017, Older driver, Male) Person Vehicle Environment domain (PV E) Behaviors we coded as PVE generally occurred in more complex driving situations and incorporated knowledge and car handling. Examples included, avoiding dangerous situations (e.g., car pulling in front of you), driving in severe weather (e.g., thunderst orms, fog, snow, or ice) and driving in a complex urban area. One driver described the challenges of driving in a busy highway in Canada: That situation in Toronto on 401 when you don't know where you're going and you're traveling a hundred and twenty [km ], if you're not going a hundred and twenty they'll drive right over you. You have to have a good second pe rson sitting to watch the signs. (FG1 Participant 3, Older driver, Male)

PAGE 67

67 A new theme from PVE was passing a car or truck. As described by one f amily member, situations involving passing could be complex with many environmental factors: went into a dip, and I was going over the hill. So I'd just seen the car passing but didn't know when we met on top of the hill, if he would be in his lane. So there is not a lot of reaction time that one lane, the transports, the curves, and al so when there is a passing lane the person that's supposed to end their passin g lane suddenly can see someone coming up very fast and they either run out of shoulder or someone's trying to get ahead at that last moment. (FG1 Participant 12, Family member, Female) Two respondents talked about their experiences with road debris (on e blowout and one crash). One driver explained how she and her husband changed their behaviors (e.g., being attentive and ready to react) after experiencing a blowout: [ T of metal, out there acro ss the road, and my husband, we saw it, but he couldn't avoid it, and had a blowout . But there is a lot of debris, I was amazed; every time we go down there now, we're extra careful looking for it. (FG2 Participant 1021, Older driver, Female ) Focus Group 3 Focus Group 3 reviewed item m eaning s and provided both overall and item specific feedback. Feedback related to the domains that the items are covering, item difficulty, demographics, driving history, formatting, clarity, wording, and respo nse options as summarized in Table 3 4. T able 3 4. Focus group 3: Quote summary of item feedback by category Item feedback Number of quotes from Focus Group 3 ( FG3) drivers Domains covered 2 Difficulty of items 12 Formatting of questionnaire 5 Clarity / wording 13 R esponse options 5 TOTAL 56

PAGE 68

68 Based on the SDBM review (FG3) our final item set for section C driving behaviors had 68 items (50 from the original item set, 3 revised and 15 new items). Revisions also included additions, deletions and wo rding changes for section A demographic profile and section B driving history profile. Collective feedback indicated items were easy to use and covered a sufficient range of safe driving behaviors. The SDBM was perceived as a potentially useful tool fo assessment. One driver commented on the range of behaviors included in the items of Participant 1025, Older driver, Female). Twelve comments wer e about item difficulty caused her to reflect on her driving: S ome of [the items] I knew exactly what my limitations were because over the years I'm aware of what I can' t do . But some of them are more minor things that we still have to stop and think. Did I [have difficulty] because we haven't thought about them that much. That's why you have the questionnaire, and we're thinking about it. (FG3 Partici pant 1027, Older driver, Female) Five respondent comments addressed item formatting, leading to two revisions. For example, some respondents suggested the item for maintaining your car fit best in section B driving history profile. Respondents made 13 c omments on item clarity leading to revisions for 6 items. Revisions included changing wording, adding examples, and changing the type of question (e.g., from section C driving behavior to section B driving history profile). One example of a wording cha Quote 33, Older Driver, M ale).

PAGE 69

69 Discussion The purpose of this study was to demonstrate how focus groups contributed to item development for the SDBM, a safe driving self/proxy report measure for older drivers, family, caregivers and professionals. This study contributes to a bod y of research on driving self reports for older adults. From our analysis, both older driver and family member respondents were active and contributed to the group process. Our focus group sample sizes (from six in FG2 to 14 in FG1) provided the opportuni ty for respondents to share their driving experience as well as comment on the experiences of other respondents. The 90 minute focus group length was sufficient, as in all three groups the moderators were able to ask all focus group guide questions, adding prompts and explanations as needed. We expanded on an existing item set and our previous research, integrating the knowledge and lived experiences of older drivers and family members. To our knowledge, the SDBM is the first self report measure developed using theoretical geographically diverse sites. rich item set reflecting the complex nature of driving behavior. Two publis hed self reports reporting positive aspects of client participation are the Driving Decisions Workbook (Eby et al., 2003) and Roadwise Review (Staplin & Dinh Zarr, 2006). Unlike our process in developing the SDBM, developers of the Driving Decisions Workbo ok and Roadwise Review included older drivers in evaluation of the measures rather than item development/refinement. For the Driving Decisions Workbook, drivers viewed the tool as informative about driving behaviors and useful for self assessment. For Road wise Review, input from users during development led to

PAGE 70

70 refinement of the computer interface and instructions. In terms of measurement precision, t he Driving Decisions Workbook results were significantly correlated (r = 0.30, p< 0.05) with an on road evalu ation of driving (Eby et al., 2003). However, the Driving Decisions Workbook has primarily an educational focus, and driver results do not include a summary score to inform drivers about risk. Results from Roadwise Review tasks do present a research based estimate of risk based on prior studies and normative data but, to our knowledge, the criterion validity of Roadwise Review has not been established in terms of the gold standard, passing or failing an on road driving evaluation. Furthermore, Roadwise Revi ew requires a computer and a partner for administration, with computer tasks that may be too complex for some users ( Myers, Blanchard, MacDonald, & Porter, 2008). Similar to the Driving Decisions Workbook and Roadwise Review, the SDBM focuses on key drivi ng behaviors for older adults and driving difficulty. However, the SDBM targets behavioral items without an educational narrative, producing a shorter length (12 pages versus 47 pages for the Driving Decisions Workbook) and reducing related user burden. Co benefit that family involvement can be obtained (through proxy SDBM) but is not necessary to complete the self assessm ent. Thus, compared to the Driving Decisions Workbook and Roadwise Review our approach yielded a more concise item set in a format that is easily completed by an older adult in 20 minutes with a summary score Moreover, our focus group respondents reported increased awareness of key driving behaviors, an important step in modifying driving behavior to increase safety (NHTSA,

PAGE 71

71 2008 b ). NHTSA also outlined the need for driving assessment tools older drivers and family members could use outside of a medical sett ing, a need the SDBM meets. Furthermore, based on the research recommendations of Dickerson et al. (2007), we criterion validity. Next steps, in progress, are to quantify th e SDBM as a predictor of passing/failing on road tests and to identify the rater reliability/ severity of the SDBM among a group of drivers, evaluators, and caregivers. Limitations of this study include limited generalizability; the driving descriptions provided are specific to both the locations as well as the participants (drivers of the young old [65 74 years] and old group [75 84 years]). We used a convenience sample for this study and men and racial minorities were under represented. While the overa ll sample represented subjects with a variety of education levels, the UF group had greater than 80% of respondents with some college education. The data supported the majority of the SDBM items with development of 23 additional themes. However, some drivi ng behaviors on the SDBM, (e.g., looking left and right before crossing an intersection) were not discussed by any of the groups. This was not surprising as some driving behaviors are considered to be habitual and may not facilitate discussion. While focus group findings supported the relevance of the SDBM items, we acknowledge the need for further psychometric research, and have work in progress to establish the construct validity, criterion validity, and reliability of the measure. Conclusion A need exis ts for a self/proxy report measure of key driving behaviors yielding a high level of measurement precision. Our use of focus groups provided stakeholder input, facilitating the addition of items and revision of an initial item set. Occupational

PAGE 72

72 therapists and professional s addressing community mobility may benefit from having a theoretically based self/proxy report that incorporates the perspective of older drivers and family. The SDBM study expands on existing older driver self report research by using a t wo pronged approach of CBPR to develop/refine an item set capturing PV, PE and PVE behaviors, and IRT to increase measurement precision. As such, when further validated, it may have potential to assist older drivers, family members, and professionals in ev aluating the driving of an older adult. The next step is to test the SDBM ratings of older drivers and family members/caregivers and evaluators against the outcome of an on road test.

PAGE 73

73 CHAPTER 4 SELF REPORT SAFE DRIVING BEHAVIOR MEASURE AS A PREDICTOR OF OL DER ADULTS PASSING/F AILING AN ON ROAD DRIVING EVALUAT ION Introduction Self assessment is one method older adults use to examine their driving and make decisions about when, where and how to drive (National Traffic Highway Safety Administration, 2008). In N orth America, automobiles are the primary mode of transportation, and all projections suggest that this mode of transportation will continue to be the primary one to maintain independent community mobility. As such, driving safety is and will continue to b e an important consideration for older adults. The current method to evaluate safe driving is through the comprehensive driving evaluations (CDEs). The CDE is administered by a driving evaluation specialist and considered the industry gold standard (Ameri can Occupational Therapy Association, 2005; Canadian Occupational Therapy Association, 2005). It usually includes a battery of clinical tests of cognition, vision, sensory and motor skills, as well as an on road test. These evaluations are time and labor i ntensive, require specialized equipment and training, have high out of pocket costs, limited geographic availability, are not generally reimbursed by third party insurances, and are most appropriate for older driver when deficits exist that potentially imp act driving safety (Kua, Korner Bitensky, & Desrosiers, 2007; Wang & Carr, 2004). A self report tool can overcome some limitations of a CDE. It can be completed in less time than a driving evaluation (e.g., 30 minutes instead of four hours), requires mini mal instruction, and can be made widely available at low to no cost to the driver. Self reports also meet the need of older adults for convenient and confidential assessment of some aspect of their driving abilities. Self report driving measures

PAGE 74

74 (based on self assessment) and proxy measures (based on observations from family, behaviors, increase driving safety awareness and knowledge, and promote behavior change (Classen e t al., 2007; Eby, Molnar, Shope, Vivoda, & Fordyce, 2003). However, self report measures are limited by selection bias (i.e., capable persons are more likely to complete the self report) and social desirability bias (i.e., persons are more likely to give a nswers that will be viewed favorably by others) (Sundstrom, 2005; Zhou & Lyles, 1997). Previous studies comparing self report and on road driving performance have not provided conclusive evidence supporting self report as a reliable and valid measure of d riving performance. As part of a longitudinal study with adults aged 72 and older, Marottoli & Richardson (1998) found self rated driving ability was not associated with on road driving performance based on results from 35 drivers who completed self repor ts and a driving performance assessment. Another self report, the computer based Roadwise Review is based on clinical measures predictive of driving ability (Staplin & Dinh Zarr, 2006). However, to date no published studies have compared Roadwise Review ra tings against on road performance. NHTSA (2008) conducted a study of the SAFER Driving instrument, with a focus on health concerns impacting driving. Findings based on sixty eight drivers over age 65 that completed the SAFER Driving instrument and an on ro ad driving test showed low but significant correlations between self reported health conditions and on road performance (r= .26, p< 0.05). In clinical practice, proxy reports from family or caregivers have been used to supplement self report information f rom drivers. Studies of everyday performance show

PAGE 75

75 caregiver reports tend to be closer to actual performance scores than self report from older adults (Diehl, 1998). For driving research, comparison of self report and proxy report has been more common in st udies of drivers with early stage dementia or mild cognitive impairment. Wild and Cotrell (2003) studied 30 older drivers (15 of whom had and 15 healthy controls ) and 30 family members or informants. Both sets of drivers (mild AD and healthy elderly) over rated their driving in comparison to the driving evaluator. However, the drivers with AD rated themselves better than the evaluator rating on 7 of the 10 areas measured; whereas the healthy drivers rated themselves better in o nly one area. Family members or informants ratings were more accurate when compared to drivers, but were still different from evaluators in two areas based on discrepancy scores. Proxy reports, despite offering potential improvement in accuracy of driving performance measurement, are not without limitations. Family members, especially those who are dependent on the driver for transportation, and other proxy reporters may be resistant to providing a negative performance rating (Wild & Cotrell, 2003). Score discrepancies may be greater for instrumental activities of daily living such as driving, as opposed to physical activities of daily living, based on findings from a study comparing self report, proxy report and observation of functional status in 233 olde r adults (Magaziner, Zimmerman, Gruber Baldini, Hebel, & Fox, 1997). Moreover, for driving assessment, there is a lack of empirical evidence correlating caregiver/proxy reports with on road performance ratings (Dobbs, Carr, & Morris, 2002). Overall, based on findings in the literature, we expected that older adults self report of their driving

PAGE 76

76 performance will be the least accurate, with some improvement in accuracy for caregivers, compared to the most accurate observations by the driving evaluator. Resear ch Context This study was part of a larger National Institutes of Health sponsored R 21 grant # 1R21AG031717 01 (Classen PI) with the primary aim of creating a valid and reliable self report measure (SDBM) indicative of safe driving behaviors and applica ble to the older adult driver population in North America. This study was conducted at two sites, the University of Florida, Gainesville, Florida (primary site), and Lakehead University, Thunder Bay, Ontario, Canada. To date, working with diverse community partners -older drivers/ families/ caregivers, and (inter)national agents of the safe older driver community -we have established face, content and construct validity of the SDBM (Classen, et al in press; Winter et al., manuscript under review; Classe n et al., manuscript under review). Use of ROC in Measure Development A Receiver Operat ing Characteristic curve (ROC curve) is one way to address concurrent criterion validity of a measure. In the ROC curve plot, the true positive rate (i.e., sensitivity) of a test against the false positive rate (i.e., 1 specificity) Sensitivity is the ability to obtain a positive test (failing an on road test) when performance deficits are truly present that is, the ability of a driving screening tool to identify a d river who is unsafe based on an unsafe outcome after conducting an on road evaluation. Specificity is the ability to obtain a negative test (passing an on road test) when performance deficits are truly absent that is the ability of a driving screening to ol to properly classify those drivers who would pass an on road evaluation. By illustrating the sensitivity and specificity of test scores, the ROC provides a measurement of how well the test picks

PAGE 77

7 7 up the signal (true positives) against the amount of noise (false positives) (Centor, accuracy or ability to discriminate between true positives and true negatives and appropriately classify subjects. Higher AUC values indicate higher accuracy or discriminability of a test, and values above 0.70 indicate acceptable accuracy given the null hypothesis or chance line of 0.5 (Streiner & Cairney, 2007). The point on the curve closest to the upper left corner (where Sensitivity= 1.0 and Specificity = 1.0) is theoretically the most optimal cut point. Decisions regarding acceptable sensitivity and specificity for a test are based on trade offs depending on the costs and outcomes that are associated with false positives (i.e., more tests and/or treatment) and false negatives (i.e., failure to identify or treat a condition) For driver screening, a diagnostic decision can be made to favor sensitivity in order to identify a greater number of unsafe drivers. Drivers who were falsely identif ied as unsafe will face additional testing of their driving but no potential for loss of license based on screening outcome. Rationale and Significance Criterion validity of driving self report tools can be established through comparison with an on road driving test, providing critical information on the predictive accuracy of the self report. The lack of self report measures which have been validated against an on road test creates a gap in the literature. To address this gap, we sought to establish th e criterion validity of the Safe Driving Behavior Measure, using ROC curves, to evaluate the predictive accuracy of SDBM ratings completed by drivers, family members/caregivers and evaluators to passing or failing an on road test.

PAGE 78

78 Research Questions Our p rimary question was: What is the concurrent criterion validity of the SDBM completed by the (a) driver, the (b) family member/ caregiv er, and the (c) evaluator, when tested against the on road driving evaluation Global Rating Score (i.e. outcome of pass o r fail)? Our secondary question was: Were there significant differences between the research sites in the drivers, family members/caregivers and evaluators that may have influenced study results? Research Question s The primary question s measured b y the ROC area under the curve, what is the accuracy of driver statistically significant differences in participants between the sites fo r key demographic, clinical, health or Methods This study received Institutional Review Board approval from the University of Florida and Lakehead University. Before engaging in study tasks, participants provided both verbal telephon e consent and signed written informed consent. Older driver participants and family members/caregivers received $50 for their study participation. Recruitment / Sample Older drivers and their caregivers/ family members or friends were recruited from North Florida and Thunder Bay, Ontario, Canada during a seven month period. To recruit older drivers and family members/caregivers, we distributed flyers, placed local of part icipants who had previously enrolled in aging studies, and who had provided

PAGE 79

79 permission for follow up contact. Participants at both sites represented a convenience 85 years of age; 2) current driver with a valid license, 3) cognitive ability to complete SDBM based on judgment of telephone screener, and 4) cognitive and physical ability to participate in an on n medically advised not to drive, 2) uncontrolled seizures in the last year, or 3) a medication regime causing central nervous system impairment. Family on the drivin g behavior of the older adult and exclusion criterion was the presence of a physical or mental condition with potential to impair completion of the proxy observation. Design This prospective pilot study used a convenience sample of 55 drivers and 55 famil y members/caregivers from two sites to test the concurrent criterion validity of the SDBM (driver, family member/caregiver and evaluator versions) against the outcome (pass/fail) an on road test. Procedure Prior to subject testing, we conducted an on si te training at Lakehead University to standardize research procedures. Standardization of research procedures included training research assistants at Lakehead on all clinical measures as well as data collection procedures. The evaluator for the on road te st at UF was an occupational therapist and certified driving rehabilitation specialist with over six years experience. The evaluator for the on road test at Lakehead was a trained driving evaluator approved by the Ministry of Transportation. Both evaluator s also participated in training to standardize road test scoring and determine the inter rater reliability between the two

PAGE 80

80 evaluators on scoring three healthy volunteers on passing or failing the Lakehead on road driving course. Inter rater reliability ind icated a perfect agreement, or 100%, between the two raters who had simultaneously rated the driving performance of the healthy volunteers tested. Following the training, recruitment was initiated at both sites and drivers were scheduled. All older driver s completed the SDBM questionnaire and a brief clinical test battery followed by a standardized on road driving evaluation. At the Florida site, the SDBM questionnaire was administered by a research assistant, the clinical battery and on road test were com pleted by the occupational therapist/certified driving rehabilitation road testing occurred on the same day as the SDBM and clinical testing, except in cases of poor weather, when the test was scheduled as soon as possible thereaft er At the Ontario site, research assistants administered the SDBM questionnaire and conducted all tests in the clinical battery. On road testing by the driving evaluator occurred shortly after the SDBM and clinical testing. The two evaluators (one per s ratings or proxy ratings. Evaluators completed the on road test with the drivers and then completed a SDBM section C to rate the driver on the 68 driving behavior items. Driving evaluators were reimbursed for their services as project staff. Each on road test included driving of varied complexity with residential, suburban, urban and highway driving. Tests were held in daylight and good weather conditions. The Florida site used a standardized fixed route co urse with components of progressive difficulty and a drive time of about one hour. Drivers used a dual brake equipped test vehicle, with the evaluator in the passenger seat. Details on the course have been published extensively

PAGE 81

81 (Classen, Shechtman, et al., 2007; Justiss, Mann, Stav, & Velozo, 2006; Stav, Justiss, McCarthy, Mann, & Lanford, 2008). The Ontario site also used a standardized fixed route course with sections of varying difficulty and a drive time of about 30 minutes, details have been published previously ( Bdard, Isherwood, Moore, Gibbons, & Lindstrom, 2004). Drivers used a dual brake equipped test vehicle, with the evaluator in the passenger seat. On road test results, including the type of errors made, were discussed with drivers and in some c ases the accompanying family member. Drivers who failed the on road test were counseled regarding the potential for remediation (i.e. evaluation (in Florida). Family members/ca regivers completed SDBM questionnaire section A (Demographics) to provide information on themselves and their relationship with the driver (e.g., how often did they ride with the driver). They also completed section C (Driving behavior) of a proxy SDBM to rate the driver on 68 driving behaviors, based on their observations over the last three months. Measurement Explanatory variables included demographic, driving, and health variables for the drivers and family members/caregivers. The independent variable s were the SDBM ratings for each rater group (drivers, family members/caregivers, and evaluators). The dependent variable was the outcome of the on road test, specifically the Global Rating Score of passing or failing the on road test. SDBM In short, the 6 8 item SDBM (Appendix A) was informed by three theoretical models the Precede Proceed Model of Health Promotion (PPMHP) (Green & Kreuter, 2005),

PAGE 82

82 and modern measurement theory, also known as item response theory (IRT) (Hambleton, Swaminathan & Rogers, 1991). In its current form, the driver self report SDBM has three sections: section A Demographic profile, section B Driving history profile, and Section C Driving be haviors. The proxy report includes sections A and C (Appendix B) Section A items included gender, race, and level of education. Section B items include days per week of driving, as well as crashes and violations in the past three years. For section C, bas ed on the last three months of driving, the individual rates his/her difficulty in executing a driving task by using a Likert scale from 1 to 5; 1= Cannot do and 5=Not difficult. Not applicable was an option for some items and was he final SDBM score represents the reported level of difficulty for 68 items with the highest raw score being 340 points, if no difficulty is reported across 68 behaviors. Psychometric testing is ongoing and initial results on the construct validity via Ra sch analysis are reported (Classen, Wen under review). Clinical tests Tests completed prior to on road test included a health history, Mini Mental State Examination (Folstein, Folstein, & McHugh, 1975), Optec 2500 vision screening (Stereo Optical Company 2007), and motor screening. As part of the health history, drivers reported on medications (over the counter and prescription) and co morbidities, both variables with potential impact on driving. The MMSE was used to measure cognitive function in terms o f orientation, registration, attention and calculation, recall, and language with an overall score below 24 indicative of impairment. In a study of drivers age 65 90, normative ranges were identified as 24 30 (Eby, Molnar, Shope, & Dellinger, 2007). The Optec 2500 vision testing machine was used to test visual acuity with

PAGE 83

83 b inocular vision of 20/20 or 20/40 considered within normal limits (Owsley et al., 1998). Th e Optec 2500 is commonly used in driving research and has reported high accuracy and reliabil ity when administered according to protocol (Stereo Optical Company, 2007). The sensory/motor screening provides a measure of sensation, range of motion, muscle strength and coordination. Motor screening was used to evaluate functional range of motion and muscle strength in the neck, trunk, upper extremities and lower extremities; hand and foot coordination; and gait speed (i.e., rapid pace walk). Eby and colleagues (2007) found a n ormative Rapid Pace Walk value of 7 seconds for drivers age 65 90. These sc reenings have acceptable reliability and validity when conducted by research staff trained in these procedures (Classen, Shechtman, et al., 2007; Justiss, Mann, Stav, & Velozo, 2006; Stav, Justiss, McCarthy, Mann, & Lanford, 2008). On road test The on roa d test consisted of a standardized road course in each location and the outcome of the road test was measured with (1) Global rating Score, and (2) Driving error score (number and type of errors) made. ( 1 ) The Global Rating Score had four levels: pass, pass with recommendations, fail with recommendations, and fail. ( 2 ) Driving error scores were calculated based on eight types of errors listed below, with additional description published previously (Justiss, 2006): a vehicle position, that is position of the vehic le in relation to other vehicles, objects, or pavement markings b speed regulation that is maintaining speed limit as well as controlled acceleration and braking c lane maintenance that is lateral positioning of the vehicle in the lane during driving or wh ile stopped d yielding that is giving right of way to other vehicles as appropriate

PAGE 84

84 e signaling that is, proper use and timing of turn signals f visual scanning that is, checking blind spots and intersections g adjustment to stimuli that is responding to d riving situations such as road sign information, vehicle movements, pedestrian movements, or potential hazards and h gap acceptance that is, demonstrating safe time and or spacing distance to cross in front of oncoming traffic While the road courses at bo th sites were similar composition, with suburban, urban, and expressway sections, there were differences in time, length and the way to the types and number of errors f road test tally sheet to mark errors by category for the entire drive. Data Collection Data were collected by research assistants at each site. As such, research assistants supervised completion of SDBM, and f or Lakehead site, they obtained clinical measures. Upon completion of the SDBM questionnaire, clinical tests, and on road tests, research assistants entered the data from drivers, family members/caregivers, and evaluators into a password protected database on a secure server. Databases for each site were maintained on a UF server. The principal investigator and the research assistant at UF conducted regular audits of the data to identify and correct data entry errors, to identify missing data and to ensure accuracy, completion and following of the protocol for data collection and entry.

PAGE 85

85 Data Analysis Descriptive analysis The descriptive analysis for the drivers and family members/caregivers addressed key socio demographic, health, clinical and driving variab les (driver only). Socio demographic variables analyzed were age, gender, race, and education. Health variables analyzed for the driver were MMSE, vision acuity, number of medications, number of co morbidities and rapid pace walk time. Driving variables an alyzed were number of days per week of driving, crashes, violations, SDBM driving behavior total, and on road test outcome of passing or failing. For the family member/caregiver, driving related variables analyzed were number of days per week the rater rod e with the driver and the family member/caregiver SDBM ratings. The descriptive analysis of drivers and family members/caregivers included comparisons between groups using SPSS 17(SPSS Inc., 2008). We conducted an independent samples t test of age, MMSE, n umber of medications, number of co morbidities, rapid pace walk, SDBM driving behavior totals, and for family indicated whether the assumption of equal variance in the t wo samples was met, and if necessary the t test for unequal variances was used. We used Chi square tests to compare gender differences, and road test outcomes. For the analyses of race, education, number of crashes, number of violations, and relationship w ith the driver (i.e. family member or friend), the assumption for a Chi square analysis (i.e., five 2000). sed into two levels representing a high school education and some college beyond high school. Similarly,

PAGE 86

86 due to the low number of minority participants, race was collapsed into Caucasian and Non Caucasian categories. ROC analyses All ROC analyses were ru n in SPSS version 17.0. In order to determine sensitivity (true positive rate) and specificity (true negative rate) with an ROC curve, an inverse of the SDBM score was used so that a high value would be associated with the outcome of failing the road test, while low values would be associated with the outcome of passing the road test. The ROC analysis provided sensitivity and specificity for each SDBM rating based on the number of drivers the SDBM correctly categorized according to on road test outcomes. Ba sed on sensitivity and specificity a cut point could be selected for each set of raters. The concurrent criterion validity analyses entailed ROC curves plotting the sensitivity and specificity of the SDBM total score for each rater group. For any score, te st outcomes for a given sample can be shown in a 2X2 table, as illustrated in Table 1 below. In addition to determining sensitivity and specificity, positive predictive value and negative predictive value can also be calculated. Positive predictive value ( PPV) is the proportion of clients testing positive who were correctly classified (i.e., they were rated as unsafe based on SDBM score and by driving evaluator based on the on road test). Negative predictive value (NPV) is the proportion of patients testing negative who were correctly classified (i.e., they were rated as safe based on SDBM score and by driving evaluator based on the on road test). We calculated an estimate of the area under the curve (AUC) using SPSS 17. The formula for AUC is a probability formula P(Y_1 > Y_2), where Y_1 is the score of a randomly selected driver who passed, and Y_2 is the score of a randomly selected driver who failed. Calculations from the 2X2 table in Table 1 are as follows:

PAGE 87

87 sensitivity = a/a + c specificity = d/ b + d PPV = a/ a + b NPV= d/ c + d. Table 4 1. 2X2 Table Displaying Relationship of True Positives, False Positives, True Negatives, and False Negatives. Fail on road test (unsafe) Pass on road test (safe) Positive test (SDBM identified as unsafe) T P= True positive (a) FP=False positive (b) Negative test (SDBM identified as safe) FN=False negative (c) TN=True Negative (d) The ROC curve for each set of raters is shown with f ive SDBM data points (scores) and the associated specificity and sensitivit y We presented the AUC of each ROC curve and examined PPV and NPV using a potential SDBM score Subsequently, we used a Wald type Chi square test to compare the AUC of the ROC curves for the drivers, family members/caregivers and evaluators (Seber, 1977). A Bonferroni correction was used to adjust for the three way comparison among the driver, caregiver and evaluator ROC curves, with an a priori p value of 0.0167 (Portney & Watkins, 2000). T he ROC curves are shown below. Results Demographics Study partici pants included 55 drivers and 55 family members/caregivers from Florida (35 drivers and 35 family members/caregivers) and Ontario (20 drivers and 20 family members)

PAGE 88

88 Drivers Table 4 2 presents results for key demographic, clinical, health, and driving var iables for the drivers. Drivers had a mean age of 73.9, were predominately Caucasian and most had some training or college after high school. There were no significant differences in driver ages, race, or on road test outcomes between the groups. The Flori da group had a significantly higher number of participants with at least some training or college education after high school compared to Ontario ( p =0.02). MMSE was similar between the groups with mean scores above 27 indicating intact cognition for most d rivers. For physical health and motor performance the number of co morbidities and rapid pace walk (seconds) respectively, was also similar between driver groups. However, the number of medications was higher for Florida. Days driven per week was higher f or Ontario drivers ( p< 0.01). For both groups, the mean number of crashes and violations reported was low (less than 1). Of all 55 drivers, only 3 drivers (2 in Florida, 1 in Ontario) reported a crash. A total of 10 drivers reported 1 or 2 violations (6 in Florida and 4 in Ontario). While the SDBM mean score was significantly higher for Ontario (320.35) than for Florida (305.06), p< 0.01, there was no significant difference in the proportion of persons failing the on road test between sites ( 2 = 0.003, df [1], p =0.88). Family m embers/ c aregivers Table 4 3 presents results for key demographic and driving variables for the family members/caregivers. In both groups, family members/caregivers were predominately women and Caucasian. There were sta tistically significant differences in age (Florida 66.0 versus 56.4 in Ontario, p =0.02), and education past high school (85% for Florida and 55% for Ontario, p =0.02). Also significant was relationship with the driver I n

PAGE 89

89 Florida 20% of the raters were fri ends while in Ontario all raters were family members ( p =0.04). The days per week the rater rode with the driver was similar. The mean SDBM scores were higher for Ontario (312.9) than for Florida (305.49), but this was not statistically significant (indepen dent samples t test, p = 0.86)

PAGE 90

90 Table 4 2 Demographic Variables All Florida Ontario Mean group differences (p<0.05) Age Mean (SD) 73.9 (5.6) 73.1 (5.3) 75.3 (5.9) t (53) = 1.42, p =0.16 Gender Female Male n=24 n=31 15 (43%) 20 (57%) 9 (45%) 11 (55%) 2 = 0.024, df (1), p= 0.88 Race Drivers Caucasian Drivers Non Caucasian / Other n=53 n=2 34 (96%) 1 ( 4%) 19 (95%) 1 ( 5%) p= 1.0 Education < High School Some college / training af ter high school n=14 n=41 3 ( 9%) 32 (91%) 11 (55%) 9 (45%) p <0.01 MMSE Mean (SD) Minimum Maximum 27.35 (1.76) 23 30 27.31 (1.86) 23 30 27.4 (1.64) 25 30 t(53) = 0.17, p = 0.86 Visual acuity Vision 20/20 or 20/40 Vision 20/50 or less 49 (89%) 6 (11%) 29 (83%) 6 (17%) 20 (100%) 0 (0%) p = 0.08 Number of medications Mean (SD) 6.8 (3.8) 7.94 (4.03) 4.85 (2.3) t(52) = 3.58, p < 0.01 Number co morbidities Mean (SD) 4.16 (1.91) 4.46 (1.99) 3.65 (1.69) t(53) = 1.52, p = 0.13 Rapid Pace Walk Mean (SD) 6.12 (1.48) 5.99 (1.32) 6.33 (1.74) t(53) = 0.81, p = 0.42 Days driving per week Mean (SD) 5.7 (1.6) 5.23 (1.83) 6.45 (0.83) t(53) = 2.81, p <0.01 Crashes reporte d No crashes 1 crash report 52 (94%) 3 ( 6%) 33 (94%) 2 ( 6%) 19 (95%) 1 ( 5%) p= 1.0 Violations reported No violations 1 2 violations 45 (82%) 10 (18%) 29 (83%) 6 (17%) 16 (80%) 4 (20%) xact test, p= 1.0

PAGE 91

91 Table 4 2 continued SDBM Driving Behavior Total Mean (SD) 310.62 (20.48) 305.06 (20.03) 320.35 (17.79) t (53) = 2.83, p =0.01 Road test Outcomes Passed road test Failed road test 41 (75%) 14 (25%) 26 (74%) 9 (26%) 15 (75%) 5 (25%) 2 = 0.003, df (1), p =0.88 Legend: SD = standard deviation; t = independent sample t test; 2 = Chi square; df = degrees of freedom Table 4 3 Variables All Florida Ontario Mean differences between groups (p<.05) Age Mean (SD) 62.5 (14.8) 66.0 (12.7) 56.4 (16.6) t (53) = 2.41 p =0.02 Gender Female Male n=45 (81%) n=10 (19%) 29 (83%) 6 (17%) 16 (80%) 4 (20%) 2 = 0.07, df (1), p= 0.79 Race Caucasian Non Caucasian / Other n=54 (98%) n= 1 (2%) 35 (100%) NA 19 (95%) 1 (5%) p= 0.36 Education < High School Some college / training after high school n=14 (75%) n=41 (25%) 5 (15%) 30(85%) 9 (45% ) 11 (55%) p= 0.02 Relationship with Driver Family member Friend n=48 (87%) n= 7 (13%) 28 (80%) 7 (20%) 20 (100%) 0 (0%) p= 0.04 Days per week you ride with driver Mean (SD) 2.4 (2.2) 2.4 (2.0) 2.4 (2.5) t(53) = 0.05, p = 0.96 Caregiver SDBM total Mean (SD) 308.18 (29.6) 305.49 (30.21) 312.9 (28.65) t(53) = 0.89, p = 0.38 Legend: SD = standard deviation; t = independent sample t test; 2 = Chi square; df = degrees of freedom

PAGE 92

92 ROC curves Drivers SDBM total Figure 4 1 ROC C urve for the D rivers: the solid line represents the SDBM ratings of drivers plotted as sensitivity (y axis) and 1 specificity (x axis) based on five select data points. To illustrate, at a score of 320 the sensitivi ty is 71% and 1 specificity (the false positive rate) is 59%. The ROC curve, with an AUC of 0.54, is close to the dashed reference line (drawn for AUC = 0.50), indicating a performance similar to chance (50:50) in predicting passing or failing a road test. sensitivity and specificity. For the 55 drivers studied the SDBM mean score was 311 (SD= + p = 0.68 95% c onfidence Score (sensitivity, 1 specificity) Point 1 312 (.50, .51) Point 2 316 (.57, .59) Point 3 320 (.71, .59) Point 4 324 (.71, .68) Point 5 328 (.86, .73)

PAGE 93

93 Interval (0.36, 0.7 1). For illustrative purposes, I will examine using 320 (out of 340) as a cut off. At a score of 320, sensitivity = 0.71, specificity = 0.42, PPV= 0.30, and NPV= 0.81. T he 55 drivers would be classified as follows: true positives (10), false positives (24) true negatives (17), and false negatives (4). The AUC is poor at 0.54 as it approaches the chance level (0.50) for correctly discriminating unsafe and safe drivers. Using a score of 320 as a cut off would result in fair sensitivity and poor specificity. Sensitivity of 0.71 means 71% of the drivers identified as unsafe, based on SDBM results, also failed the on road test (true positive). A specificity of 0.42 means only 42% of the drivers classified as safe also passed the on road test (true negative). PP V, the probability that a person who tests positive is truly positive, is very low. If 320 were used as a cut off, close to half of all drivers would be subjected to additional testing to determine their driving fitness. Although the NPV is high, approxima tely 20% of unsafe drivers would not be detected at this level (i.e., they test negative safe when in fact they are unsafe). Family m ember/ c aregiver SDBM total The ROC data points show five family member/caregiver SDBM scores, and the associated sens itivity and specificity. The caregivers SDBM mean rating was 308 (SD=30). The AUC is 0.67, p = 0.65 95% confidence interval (0.48, 0.85) For the family members/caregivers I will use 312 (out of 340) as an example cut off. For a score of 312, sensitivity = 0.64, specificity = 0.68, PPV= 0.41, NPV= 0.85. Using 312 for a cut off would result in: true positives (9), false positives (13), true negatives (28), and false negatives (5).

PAGE 94

94 Figure 4 2 ROC C urve for the F amily M embers/ C aregivers: the solid line r epresents the SDBM ratings of caregivers plotted as sensitivity (y axis) and 1 specificity (x yield a ROC curve that is above the dashed reference line (drawn for AUC = 0.50) and ap proaches a fair performance level of AUC = 0.70. The AUC of .67 approaches, but does not reach, an acceptable AUC level of .70. Considering a potential family member/caregiver cut off score of 312, sensitivity and specificity are fair. Given a sensitivity of 0.64, 64% of the drivers identified as unsafe based on SDBM results would also fail the on road test (true positive). Given a specificity of 0.68, only 68% of the drivers classified as safe based on the SDBM rating also passed the on road test (true ne gative). PPV is low, given a result of unsafe on the Score (sensitivity, 1 specificity) Point 1 304 (.50, .24) Point 2 308 (.57, .27) Point 3 312 (.64, .32) Point 4 316 (.71, .37) Point 5 328 (.71, .49)

PAGE 95

95 family member/caregiver SDBM, there is a 41% probability that the driver would also be classified as unsafe based on the on road test. NPV is high, given a result of safe on the family member/caregiver S DBM, there is an 85% probability that the driver would also be classified as safe based on the on road test. Evaluator s SDBM total Figure 4 3 ROC C urve for the E valuators: the solid line represents SDBM ratings of the evaluators plotted as sensitivi ty (y axis) and 1 specificity (x axis) based on five select data points. The AUC = 0.99 and the ROC curve is well above the reference line indicating superior accuracy in predicting passing or failing a road test. Score (sensitivity, 1 specificity) Point 1 290 (.88, .08) Point 2 295 (.93, .09) Point 3 300 (1, .12) Point 4 305 (1, .20) Point 5 310 (1, .27)

PAGE 96

96 The ROC data points show five evaluator SD BM scores, and the associated AUC = 0.99, p < 0.00 01 95% confidence interval (0.00, 1.00) Considering a potential cut off of 300 (out of 340), sensitivity = 1.00, specific ity = 0.88, PPV= 0.74, and NPV= 1.00. Based on a cut off score of 300 drivers would be classified as follows: true positives (14), false positives (5), true negatives (36), and false negatives (0). When used by a trained evaluator following observation of on road driving, the SDBM displays high accuracy with an AUC of 0.99 ( p < 0.001). Given a cut off of 300 on the SDBM total, sensitivity is perfect and specificity is high. Sensitivity of 1.0 means that all the drivers identified as unsafe, based on SDBM res ults, also failed the on road test (true positive). Specificity is high at 0.88 with 88% of the drivers classified as safe passing the on road test (true negative). PPV is moderate, with 74% probability that a driver who rated as unsafe based on the evalua tors SDBM rating would be classified as unsafe based on the on road test. NPV is high at 1.00 with a 100% probability that a be classified as safe to drive using an on roa d test. ROC Curve Comparison Results of the Wald type Chi squared test demonstrate the Driver and Evaluator ROC curves were significantly different ( p < 0.01); as were the Caregiver and Evaluator ROC curves ( p <0.02). No significant difference was found bet ween the Driver and Caregiver ROC curves ( p < 0.31). The most significant difference was between the driver and evaluator AUC (.54 vs. 0.99). There was also a significant difference between the caregiver and evaluator AUC (.67 vs. 0.99), although the careg iver AUC was closer to the evaluator AUC. The evaluator was, as expected, the most accurate rater.

PAGE 97

97 Post hoc Analysis Based on the ROC results, scatter plots were used to further examine the rating patterns of drivers, family members/caregivers, and evalu ators. A z score was used to adjust for scoring differences between the Florida and Ontario sites. Variables used for the scatter plots were the z score for the number of errors on the on road test (y axis) and the number of deductions from the SDBM sectio n C driving behavior rating (x axis). Figure 4 4. Scatter P lot and L inear R egression L ine for the D rivers: The negative trend of the linear regression line and the R 2 of < 0.00 show the SDBM score did not predict the number of errors on the on road test.

PAGE 98

98 Figure 4 5 Scatter Plot and Linear Regression L ine for the Family Members/ C aregivers: The linear regression line and the R 2 of 0 10 show the SDBM score was a slight predictor of the number of errors on the on road test.

PAGE 99

99 Figure 4 6 Scatter P lot a nd L inear R egression L ine for the E valuator s : The linear regression line and the R 2 of 0.42 show the SDBM score was a moderate predictor of the number of errors on the on road test. Discussion Our primary goal was to assess the concurrent criterion validit y of the SDBM based on ratings by drivers, family members/caregivers and driving evaluators. This was accomplished by comparing ratings against the on road driving evaluation Global Rating Score (i.e., outcome of pass or fail) using ROC analyses. We used R OC curves examining the accuracy or discriminability of the SDBM for three sets of raters (drivers caregivers, and evaluators )

PAGE 100

100 Future selection of a cut off score for the SDBM will consider the personal, economic and social costs of misclassifying drive rs. For any particular cut point on the SDBM, there is a probability that unsafe drivers will not be detected (i.e., drivers who would fail an on road test will be misclassified as safe). Moreover, for the same cut off there is a probability that safe driv ers (i.e., drivers who would pass on on road test) will be misclassified as unsafe. Changing the cut off alters the sensitivity, specificity, PPV, and NPV. For example, changing the cut point for the drivers from a score of 320 to a score of 328 improves t he sensitivity of the measure from 0.71 to 0.86 while specificity decreases from 0.41 to 0.27; PPV of 0.29 is less than prior value of 0.30 obtained at the cut point of 320, while the NPV increases from 0.81 to 0.85. Given the decrease in PPV a change to a cut point of 320 would not be recommended. Unsafe drivers who are classified as safe and continue to drive, present a risk to themselves and others, for being crash involved. On the other hand, safe drivers who are classified as unsafe may face additional testing to determine their driving fitness (Shechtman, Awadzi, Classen, Lanford, & Joo, in press). Therefore, we deduce that for the SDBM we prefer a high positive predictive value (PPV) over a high negative predictive value (NPV). In effect, we want the SDBM to correctly identify unsafe drivers with a low level of misclassification of safe drivers. passing or failing the on road test. The family members/caregivers were more accu rate in their SDBM ratings; however their ROC AUC, at .67, was below the acceptable level of .70 or higher. The evaluators ROC analysis showed excellent accuracy for predicting passing or failing the on road test. Based on scatter plot findings, the driver

PAGE 101

101 ratings did not predict the number of errors during the on road test (R 2 < 0.0 1 ) The were slightly predictive ( R 2 = 0.10 ratings were moderately predictive with a R 2 of 0. 42 However, given th at their ratings were performed directly after the road test that may have influenced scoring. Our secondary question was to examine between site differences in drivers and caregivers. Results of analyses to examine between group differences indicated th e driver groups were similar in gender distribution and race, with a statistically significant difference in level of education. Days driven per week, while statistically significant, were only slightly higher for Ontario than Florida (6.45 versus 5.23 day s). The mean number of medications was higher for Florida (7.94 versus 4.85) While the mean road test was the same as Florida, indicating a greater tendency to over ra te their driving. Family members/caregivers in Florida were older and more educated than in all family. However, given the lack of statistical ly significan t differenc es in family and relationship differences (which were significant) impacted the SDBM ratings. The goal of the parent grant was to develop a driving self report for older ad ults that is informative of safe driving, with proxy versions for family, caregivers and evaluators. Based on the results of this study, neither the driver nor caregiver SDBM had an ROC with an acceptable AUC of 0.70 or higher. Additionally, it has been su ggested that a useful screening tool will correctly identify 80 90% of drivers (Bdard, Weaver, Darzins, & Porter, 2008). While the evaluator SDBM total score correctly

PAGE 102

102 g the on road test. In contrast, the caregiver SDBM total score identified 64% of drivers correctly whereas the driver SDBM total score correctly identified 49%. Limitations of this study include lack of an advance power analysis to establish sample size, differences in road tests and scoring between the sites, the presence of and the failure to identify and eliminate outliers (i. e., persons whose SDBM scoring deviated grea ter than two standard deviations) While providing initial information on the SDBM through ROC analyses, the ROC findings lack statistical or clinical significance. Positive predictive value (PPV) and negative predictive value (NPV) were a secondary focus of this study. However, Sackett and Haynes (2002) suggest that in conditions with a low prevalence, such as the inability to drive safely in older adults, PPV and NPV are speci ficity. Strengths of this study include findings on accuracy of the SDBM ratings by the drivers, family members/caregivers and evaluators. This knowledge will help us evaluate bias and sources o f error in the SDBM and contribute to future refinement. Fu rthermore, this study provides a foundation for future research a ddress ing sensitivity and specificity of the SDBM as well as PPV and NPV Similar to Wild and Cotrell (2003) we found that both drivers and family members/caregivers differed from evaluator s in their driving ratings using the SDBM. Biases described earlier, including selection bias, social desirability bias, acquiescence

PAGE 103

103 ratings. In addition, the fact tha t the evaluators completed a SDBM for each driver following an on road test may have biased their ratings. This study was a pilot study with 55 drivers and 55 family members/caregivers from a convenience sample. Further testing is needed to evaluate the a ccuracy of the driver and family member/caregiver ratings including the p otential biases. Future research will be planned to review the SDBM items with drivers. Strategies such as cognitive interviewing can be used to better determine how drivers define s afe driving and how they interpret the SDBM items. In depth interviews and focus groups with drivers could potentially add to the diversity of driving behaviors addressed in the SDBM and allow testing of any SDBM item revisions. In this study we present th e concurrent criterion validity testing of the SDBM. Our findings contribute to a body of research on the use of self reports to measure the driving of older adults and contribute to establishing the psychometric properties of the SDBM.

PAGE 104

104 CHAPTER 5 SUMMAR Y AND CONCLUSION Community mobility, including older driver assessment, is an important practice and research area for occupational therapists (American Occupational Therapy Association, 2005; Canadian Association of Occupational Therapists, 2009) Older d rivers may experience physical or cognitive declines with aging creating an increased crash risk, and older adults with co morbidities or frailty have an increased risk of injuries or fatalities when a crash occurs (Anstey, Wood, Lord, & Walker, 2005; Cla ssen et al., 2006). In 2008, older adults accounted for 5,569 traffic related fatalities and over 183,000 injuries from crashes (NHTSA, 2008a). Injury prevention efforts for older drivers include interventions addressing safety awareness, knowledge, and be haviors. Identification of safe driving behaviors (or deficits thereof) is a first step to increasing older (Classen, Lopez, et al., 2007). Self report is one method of identify ing safe driving be haviors Self reports of driving performance are common in both occupational therapy clinical practice and driving research. However, p revious research on the validity of self reports and the degree to which self reports correlate with on road performance has been mixed. Overall, I identified the existing self report measures as hav ing limitations related to design, content, administration, or validity. Self reports also have limitations including forms of rater bias such as selection bias social desirab ility bias and recall bias P roxy report s (such as by family members or caregivers) have been suggested as a way to overcome self driving performance. However, proxy reports, while o ffering potential improvement in

PAGE 105

105 report, are subject to some of the same biases and limitations. Given the status of existing self reports, this dissertation sought to describe the item development, refinement, a nd validity testing for a self report Safe Driving Behavior Measure for older adults, and the proxy report for family members or caregivers. The first research aim was to develop items for the SDBM reflecting safe driving by older adults and establish f ace and content validity The research q uestion was: behavioral items established during an iterative process of item refinement and re A priori content validity was established b ased on theoretical frameworks (Precede M M odel), existing driving measures and previous research G uided by the theoretical fra me works, as well as me asurement theory (especiall y IRT) we developed items capturing safe driving behavior. We established face validity using peer reviewers and content validity using expert raters. Peer review results indicated acceptable face validity. Initial expert rater review yielded a scale cont ent validity index (CVI) rating of 0.78, with 44/60 items Subsequently, the second round CVI scale average was 0.84 indicating acceptable content validity. The second r e sear ch aim was to conduct focus groups with older drivers and family members/caregivers to expand the item bank and inform item revisions. The research question was: What are the domains, constructs, and items that comprise m the perspective of older drivers and families I conducted

PAGE 106

106 experiences, and 2) refine the SDBM items based on feedback. Older drivers (mean age 70.5, SD = 4.5) and family membe rs (mean age 50, SD =20) participated in three focus groups in Florida and Ontario. Using content and thematic analyses, I coded responses to existing items, new items, or revisions. Findings from F ocus G roup s O ne (Ontario) and T wo (Florida) supported 46 o f 72 existing items and generated 16 new items. F indings from F ocus G roup T hree (Florida) supported 40 existing items and generated 13 item revisions. Following revisions, the team prepared a draft SDBM for use in validity testing with drivers family memb ers/caregivers and driving evaluators The third research aim was to use ROC analyses to examine the sensitivity and specificity of the SDBM based on three sets of ratings, drivers, family members/ caregivers, and driving evaluators. The research question concurrent criterion validity of the SDBM as measured using ROC analyses based on SDBM ratings by driver, family member/ caregiver and evaluator in comparison to outcomes of an on found drivers w ere less accur ate raters than family members/ caregivers but the difference was not statistically significant. Based on AUC drivers and family members were less accurate that the evaluators and those differences were significant. The evaluators demonstrated high accura cy. These results indicate the item set could be used by driving evaluators to correctly classify drivers. However, biases were present in the drivers and family members/ ratings that require further examination and potential revision of items. Limitations Limitations of this study will be discussed for each stage of the dissertation Item development is an iterative process and it can be difficult to know when a behavior

PAGE 107

107 measure has a sufficient number and breadth of items to adequately measur e ability. Furthermore, measurement theory and questionnaire development guidelines require items that measure discrete behaviors. However, the act of driving is complex and often involves simultaneous performance of several behaviors in response to the dr iving environment and demands. These issues were raised not only during the item development, but also during face validity and content validity testing As part of validity testing, we conducted a peer review of the SDBM focusing on face validity and an e xpert review focusing on content validity. A limitation for the content validity work was brought up by the reviewers. Following the first round CVI and item revisions, raters were only given the revised items to review and any new items, and did not re ra te the entire measure. The expert reviewers indicated a collective preference for reviewing the entire measure at all stages, in order to better evaluate the SDBM. For the focus groups, while older drivers and family members at both Ontario and Florida si tes participated in a safe driving discussion (focus groups one and two), only the Florida location (focus group three) reviewed the draft SDBM and gave feedbac k on the measure and the items. The use of SDBM review in Ontario was considered by the team, an d a decision made that data saturation was reached in focus group three (i.e., findings sufficiently addressed all review needs). While focus groups were a recommended strategy to enhance item development and refinement, one on one interviews incorporating review of the SDBM could have been used to further explore participants understanding and interpretation of the items as well as their responses. L imitation s for the ROC analyses include a small sample of 55 drivers ( most who were judged safe based on the on road test), and the potential for bias in the self report and

PAGE 108

108 proxy report measuring driving behavior. A larger sample with a greater proportion of drivers with deficits would have increased our ability to determine the accuracy of the SDBM ratings in predicting passing or failing an on road test. The effect of bias on self report, including selection bias, social desirability bias and recall bias, may all partly explain score discrepancies between the drivers, family members/caregivers, and evaluators ratings. Further testing will be used to identify and adjust for these biases. Strengths Strengths of this dissertation include development of a self report measure developed using both C ommunity B ased Participatory Research and item response theory appro aches. The SDBM items were developed with a strong theoretical and evidence base as part of a broader line of research in public health, injury prevention, and driving safety. Our use of peer review and expert review provided not only ratings of the SDBM items but also in depth feedback that was used to refine the SDBM. While focus groups one and two helped generate new items for the SDBM, the findings provided strong support for the initial item set and provided rich descriptions of driving behaviors F ur thermore, for the ROC analyses, the SDBM ratings were compared against the out comes of an on road evaluation allowing comparison of rater accuracy in using the SDBM among the drivers, family members/caregivers, and evaluators. Implications These dissertat ion findings suggest several implications for future development o f the SDBM. First, road test should be further investigated. As part of post doctoral work, f uture research will be plann ed using strategies such as cognitive interviewing and focus groups to gain additional information on how drivers define driving safety and interpret SDBM items.

PAGE 109

109 Second, interview and focus group findings will be used to revise items to better match driver conceptual understanding potentially improving their interpretation of SDBM items and response accuracy. Third, o nce a suitable le v el of driver response accuracy is reached, as defined by an AUC 0.70, the SDBM could be t est ed i n a multi site study to further examine and refine the geographic diversity and cultural sensitivity of the SDBM Finally, f ollowing this refinement of the SDBM, and based on IRT, algorithms can be developed and implemented via computerized adaptive testing ( CAT ) to deliver targ eted questions to drivers and family members/caregivers All of the above approaches will help reach the long term goal of this project, to develop, test, and implement a SDBM for use among older drivers and their family members/caregivers on a population based level. Conclusion The purpose of this dissertation was three fold: 1) to develop a theoretically and empirically based SDBM item set with face and content validity 2) to conduct focus groups in order to expand the item set and inform i tem revisions and 3) to investigate concurrent criterion validity of the SDBM using receiver operating characteristic (ROC) curve analyses with three rater groups (driver, family member/caregiver, and driving evaluator) Face and content validity testing indicated an acceptable item set with relevant items (based on CVI score relevance of the item set, expanded our knowledge of the person vehicle environment domains, and guided subsequent item revisions. Following SDB M revisions, and using results of an on road test, concurrent criterion v alidity testing was tested for each of the

PAGE 110

110 By addressing driving safety, an aspect of community mobility, the findings of this dissertation a dd to the pr actice of occupational therapy These findings demonstrate the SDBM has high sensitivity and specificity when used by a trained driving evaluator such as a certified driving rehabilitation specialist As such, the SDBM has usefulness in driving evaluation However, f urther testing of the SDBM must be performed with occupational therapy generalists to determine the utility of the SDBM. This dissertation also contributes to rehabilitation science The SDBM has a theoretical base in public health (Precede Pr oceed Model of Health Promotion), injury Moreover, this dissertation builds on prior se lf report research addressing the needs of older drivers. My use of community based participatory resea rch (CBPR) strategies classical test theory, and item response theory (IRT) was consistent with rehabilitation science principles I i ncorporat ed CBPR strategies (e.g., focus groups and an advisory committee ) to acknowledge and utilize the perspectives a nd knowledge of older drivers family members/caregivers and national/international community partners Classical test theory informed measurement of the face, content, and concurrent criterion validity of the SDBM. Item response theory provided a n organiz ing framework for safe driving behaviors im prov ing measurement accuracy. My findings indicate the SDBM is relevant for older drivers and family members/caregivers and yields high predictive validity when used by driving rehabilitation specialists. Future research will include in depth interviews and focus groups with drivers family members/caregivers and community partners to improve the utility of the SDBM.

PAGE 111

111 APPENDIX A SAFE DRIVING BEHAVIO R MEASURE DRIVER VERSION A. Demographic Profile 1. What is your birth year? _______ 2. What is your gender? Male Female 3. What is your ethnicity? Do you consider yourself to be: Hispanic or Latino (A person of Cuban, Mexican, Puerto Rican, South or Central American, or other Spanish culture or origin, regardless of race) Not Hispanic or Latino 4. What is your race? Would you say you are: American Indian / Alaska Native / First Nations / Aboriginal or Inuit : having origins in any of the original peoples of North, Central, or South Americ a, and who maintains tribal affiliation or community attachment. Asian : having origins in any of the original peoples of the Far East, Southeast Asia, or the Indian subcontinent including, for example, Cambodia, China, India, Japan, Korea, Malaysia, Pakis tan, the Philippine Islands, Thailand, and Vietnam. Black or African American : having origins in any of the black racial groups of Africa. Native Hawaiian or Other Pacific Islander : having origins in any of the original peoples of Hawaii, Guam, Samoa, o r other Pacific Islands. White : having origins in any of the original peoples of Europe, the Middle East, or North Africa. Other: specify ___________ Go to question # 8) No Yes Mostly (for part of the year) Instructions: 1 Please answer all 9 questions to the best of your ability. 2 Answer by checking the box or filling in the blank.

PAGE 112

112 6. Who lives with you? Spouse or partner Child Family/Other relative: specify: ________ Friend(s) Paid caregiver Other: specify __________________ 7. How many other licensed drivers are in your household? _______ 8. What is your highest level of educati on? Did not go to school Completed Grade school (5th grade) Completed Middle school (8th grade) Completed High School/G.E.D. (12th grade) Completed Vocational Training Some College after High School Graduation Associate D egree Some Professional School after College Graduation Doctoral Degree 9. Do you use any of the following assistive devices? Corrective lenses (such as eyeglasses or contacts) Hearing device \ he aring aid Mobility device (such as cane, walker, wheelchair) Car devices (such as seat pad, pedal assist, spinner knob) Other: (list)______________________________________ B. Driving History Profile 1. How many days a week do you ty pically drive? 0 1 2 3 4 5 6 7 Instructions: 1 P lease answer all 18 questions to the best of your ability. 2 Answer by checking the box or filling in the blank.

PAGE 113

113 2. When you drive, who usually rides with you? (Please check all that apply) Spouse / Partner Family member Friend Caregiver Other No one 3. Has a health condition limited your ability to drive? No Yes 4. Ha s taking medications limited your ability to drive (over the counter or prescribed)? No Yes 5. Did you get any of the following tested in the last year? (Please check all that apply) Vision Hearing Physical exam / checkup Other tests (list)__________ ____________________________ 6. In the past year, did you complete any of the following car maintenance? (Please check all that apply) Oil change Checking tires Checking fluid levels Checking headlights, brake lights and parking lights 7. Do you avoid (when possible) any of these driving situations? (Please check all that apply) Rush hour/heavy traffic Interstate/ highway driving Rain Night time driving Left hand turns against traffic Other (list)_______________________________________ None

PAGE 114

114 8. Do you use alternative transportation (such as taking a bus or taxi)? Always Often Sometimes Rarely Never 9. Would you consider alternative transportation if it were available? No Yes 10. As the driver on a long trip, how frequently do you take breaks? Every 1 to 2 hours Every 3 to 4 hours Every 5 to 6 hours Rarely or Never 11. Is it difficult for you to fasten your seatbelt? Always Often Sometimes Rarely Never 12. As a driver, have you been involved in a crash in the past 3 years? No Yes 13. As a driver, how many crashes were you involved in during the past 3 years? 1 2 3 4 5 or more 14. How many moving violations, citations or traffic tickets have you had in the past 3 years? (If y 0 1 2 3 4 5 or more

PAGE 115

115 15. What moving violations, citations or traffic tickets did you receive in the past three years? (Please check all that apply) Failure to yield Going too slowly Not obeying traffic lights Not obeying traffic signs (such as stop sign) Improper passing Improper turning Careless driving Reckless driving Driving under influence of drugs or alcohol (DUI/DWI) Speeding Tailgating Other (list)______________________________________ 16. When did you last attend a driver education, training or retraining course? (If you Within the past year 1 3 years ago More than 3 years ago Never 17. If you have attended a driver education class, trainin g or re training, what type was it? (Please check all that apply) On line class Classroom course for all drivers Classroom course for mature drivers Course with classroom and behind the wheel instruction Other (list)_____________________________________ 18. How do you keep up with changes in road rules or laws? (Please check all that apply) Driving class Newspaper TV Friends or family Computer Police or law enforcement Other (list)_______________ _______________________ None of the above

PAGE 116

116 C: Driving Behavior Items Note the example below: A. Put your key in the ignition? Cannot Do Very Difficult Somewhat Difficult A Little Difficult Not Difficult Instructions: 1 Please answer all 68 questions to the best of your ability. 2 Based on your driving in the last three (3) months, tell us how much difficulty you have with the driving behaviors on the following pages. 3 Mark one of these answers: Cannot Do too difficult to manage Very Difficult doing it is a major challenge Somewhat Difficult doing it is a moderate challenge A Little Difficult doing it is a minor challenge Not Difficult you can do it with ease Not Applicable question does not apply

PAGE 117

117 1. Open your car door? Cannot Do Very Difficult S omewhat Difficult A Little Difficult Not Difficult 2. Get in your car? Cannot Do Very Difficult Somewhat Difficult A Little Difficult Not Difficult 3. Turn the steering wheel? Cannot Do Very Difficult Somewhat Difficult A Little D ifficult Not Difficult 4. Adjust your car mirrors? Cannot Do Very Difficult Somewhat Difficult A Little Difficult Not Difficult 5. Stay awake while driving? Cannot Do Very Difficult Somewhat Difficult A Little Difficult Not Diffic ult steering wheel? Cannot Do Very Difficult Somewhat Difficult A Little Difficult Not Difficult 7. Stop for pedestrians crossing the roadway? Cannot Do Very Difficult Somewhat Diffic ult A Little Difficult Not Difficult 8. Drive in good weather? Cannot Do Very Difficult Somewhat Difficult A Little Difficult Not Difficult 9. Stay in your own lane? Cannot Do Very Difficult Somewhat Difficult A Little Difficult Not Difficult 10. Drive during daylight hours? Cannot Do Very Difficult Somewhat Difficult A Little Difficult Not Difficult

PAGE 118

118 11. Remember to turn on your headlights before driving in the dark? Cannot Do Very Difficult Somewhat Difficult A Little Difficult Not Difficult 12. Check for a clear path when backing out from a driveway or parking space? Cannot Do Very Difficult Somewhat Difficult A Little Difficult Not Difficult 13. Reach the gas pedal (accelerator) and brake p edal? Cannot Do Very Difficult Somewhat Difficult A Little Difficult Not Difficult 14. Press the gas or the brake when intended? Cannot Do Very Difficult Somewhat Difficult A Little Difficult Not Difficult 15. Use your car controls (s uch as the turn signals, windshield wipers, or headlights)? Cannot Do Very Difficult Somewhat Difficult A Little Difficult Not Difficult 16. Place your car in the correct gear (such as drive or reverse)? Cannot Do Very Difficult Somewhat Dif ficult A Little Difficult Not Difficult 17. Operate your emergency brake? Not Applicable Cannot Do Very Difficult Somewhat Difficult A Little Difficult Not Difficult 18. Check your mirrors when changing lanes? Cannot Do Very Difficul t Somewhat Difficult A Little Difficult Not Difficult 19. Read road signs far enough in advance to react (such as make a turn)? Cannot Do Very Difficult Somewhat Difficult A Little Difficult Not Difficult 20. Obey varied forms of traffi c lights (such as green arrow for turn lane or flashing lights)? Cannot Do Very Difficult Somewhat Difficult A Little Difficult Not Difficult

PAGE 119

119 21. Drive and hold a conversation with one or more passengers? Cannot Do Very Difficult Somewhat Difficult A Little Difficult Not Difficult 22. Drive with a passenger who is providing driving directions or assistance? Cannot Do Very Difficult Somewhat Difficult A Little Difficult Not Difficult 23. Drive in light rain? Cannot Do Very Difficult Somewhat Difficult A Little Difficult Not Difficult 24. Drive on a highway with two or more lanes in each direction? Cannot Do Very Difficult Somewhat Difficult A Little Difficult Not Difficult 25. Keep up with the flow of traffic? Cannot Do Very Difficult Somewhat Difficult A Little Difficult Not Difficult 26. Keep distance from other vehicles when you change lanes? Cannot Do Very Difficult Somewhat Difficult A Little Difficult Not Difficult 27. Ch ange lanes in moderate traffic? Cannot Do Very Difficult Somewhat Difficult A Little Difficult Not Difficult 28. Drive cautiously (to avoid collisions) in situations when others are driving erratically (such as speeding, road rage, crossing lan e lines or driving distracted)? Cannot Do Very Difficult Somewhat Difficult A Little Difficult Not Difficult 29. Brake at a stop sign so car stops completely before the marked line? Cannot Do Very Difficult Somewhat Difficult A Little Dif ficult Not Difficult 30. Maintain lane when turning (not cut corner or go wide)? Cannot Do Very Difficult Somewhat Difficult A Little Difficult Not Difficult

PAGE 120

120 31. Back out of parking spots? Cannot Do Very Difficult Somewhat Difficult A Little Difficult Not Difficult 32. Enter the flow of traffic when turning right? Cannot Do Very Difficult Somewhat Difficult A Little Difficult Not Difficult 33. Share the road with vulnerable road users such as bicyclists, scooter drivers motorcyclists? Cannot Do Very Difficult Somewhat Difficult A Little Difficult Not Difficult 34. Drive on graded (unpaved) road? Not Applicable Cannot Do Very Difficult Somewhat Difficult A Little Difficult Not Difficult 35. Che ck blind spots before changing lanes? Cannot Do Very Difficult Somewhat Difficult A Little Difficult Not Difficult 36. Drive with tractor trailers (transport trucks) around you? Cannot Do Very Difficult Somewhat Difficult A Little Difficul t Not Difficult 37. Merge onto a highway? Cannot Do Very Difficult Somewhat Difficult A Little Difficult Not Difficult 38. Use a map while driving? Cannot Do Very Difficult Somewhat Difficult A Little Difficult Not Difficult 39 Make a left hand turn crossing multiple lanes and entering traffic (with no lights or stop signs)? Cannot Do Very Difficult Somewhat Difficult A Little Difficult Not Difficult 40. Parallel park? Not Applicable Cannot Do Very Difficult Somewhat Difficult A Little Difficult Not Difficult

PAGE 121

121 41. Stay within the lane markings unless you have to make a lane change? Cannot Do Very Difficult Somewhat Difficult A Little Difficult Not Difficult 42. Stay within your lane in the abs ence of road features such as clearly marked lane lines, reflectors or rumble strips? Cannot Do Very Difficult Somewhat Difficult A Little Difficult Not Difficult 43. Keep distance between your car and others (allow time to react to hazards)? Cannot Do Very Difficult Somewhat Difficult A Little Difficult Not Difficult 44. Look left and right before crossing an intersection? Cannot Do Very Difficult Somewhat Difficult A Little Difficult Not Difficult 45. Drive in a construc tion zone? Cannot Do Very Difficult Somewhat Difficult A Little Difficult Not Difficult 46. Drive in dense traffic (such as rush hour)? Cannot Do Very Difficult Somewhat Difficult A Little Difficult Not Difficult 47. Pass (overtake) a car in the absence of a passing lane? Cannot Do Very Difficult Somewhat Difficult A Little Difficult Not Difficult 48. Pass (overtake) a larger vehicle such as a RV, tractor trailer (transport truck), or dump truck in the absence of a passin g lane? Cannot Do Very Difficult Somewhat Difficult A Little Difficult Not Difficult 49. Drive in an unfamiliar urban area? Not Applicable Cannot Do Very Difficult Somewhat Difficult A Little Difficult Not Difficult 50. Control yo ur car when going down a steep hill? Not Applicable Cannot Do Very Difficult Somewhat Difficult A Little Difficult Not Difficult

PAGE 122

122 51. Exit an expressway, or inter state from a left hand lane? Not Applicable Cannot Do Very Difficult Somew hat Difficult A Little Difficult Not Difficult 52. Drive in a highly complex situation (such as a large city with high speed traffic, multiple highway interchanges and several signs)? Not Applicable Cannot Do Very Difficult Somewhat Difficult A Little Difficult Not Difficult 53. Control the car (brake hard or swerve) to avoid collisions? Cannot Do Very Difficult Somewhat Difficult A Little Difficult Not Difficult 54. Drive a different car (such as ntal car)? Not Applicable Cannot Do Very Difficult Somewhat Difficult A Little Difficult Not Difficult 55. Alter your driving in response to changes in your health (such as vision, reaction time, fatigue, thinking, joint stiffness, medications )? Cannot Do Very Difficult Somewhat Difficult A Little Difficult Not Difficult 56. Drive when you are upset (anxious, worried, sad or angry)? Cannot Do Very Difficult Somewhat Difficult A Little Difficult Not Difficult 57. Stay focus ed on driving when there are distractions (such as radio, eating, drinking, pet in the car)? Cannot Do Very Difficult Somewhat Difficult A Little Difficult Not Difficult 58. Drive in an unfamiliar area? Cannot Do Very Difficult Somewhat Diff icult A Little Difficult Not Difficult 59. Drive at night? Cannot Do Very Difficult Somewhat Difficult A Little Difficult Not Difficult

PAGE 123

123 60. Avoid dangerous situations (such as car door opening, car pulling out, road debris, or an animal darting in front of you)? Cannot Do Very Difficult Somewhat Difficult A Little Difficult Not Difficult 61. Drive when there is fog? Cannot Do Very Difficult Somewhat Difficult A Little Difficult Not Difficult 62. Drive at night on a dark road with faded or absent lane lines? Cannot Do Very Difficult Somewhat Difficult A Little Difficult Not Difficult 63. Drive when there is glare or the sun is in your eyes? Cannot Do Very Difficult Somewhat Difficult A Little Dif ficult Not Difficult 64. Turn left across multiple lanes when there is no traffic light? Cannot Do Very Difficult Somewhat Difficult A Little Difficult Not Difficult 65. Drive in a thunderstorm with heavy rains and wind? Cannot Do Very Difficult Somewhat Difficult A Little Difficult Not Difficult 66. Control your car on a wet road? Cannot Do Very Difficult Somewhat Difficult A Little Difficult Not Difficult 67. Control your car on a snow covered road? Not Applicable Cannot Do Very Difficult Somewhat Difficult A Little Difficult Not Difficult 68. Control your car on an icy road? Not Applicable Cannot Do Very Difficult Somewhat Difficult A Little Difficult Not Difficult

PAGE 124

124 APPENDIX B SAFE DRIVIN G BEHAVIOR MEASURE CAREGIVER VERSION Safe Driving Behavior Measure A. Demographic Profile/ Caregiver 1. What is your birth year? _______ 2. What is your gender? Male Female 3. What is your ethnicity? Do you consider yourself to be : Hispanic or Latino (A person of Cuban, Mexican, Puerto Rican, South or Central American, or other Spanish culture or origin, regardless of race) Not Hispanic or Latino 4. What is your race? Would you say you are: American Indian / Alaska Native / Fi rst Nations / Aboriginal or Inuit : having origins in any of the original peoples of North, Central, or South America, and who maintains tribal affiliation or community attachment. Asian : having origins in any of the original peoples of the Far East, South east Asia, or the Indian subcontinent including, for example, Cambodia, China, India, Japan, Korea, Malaysia, Pakistan, the Philippine Islands, Thailand, and Vietnam. Black or African American : having origins in any of the black racial groups of Africa. Native Hawaiian or Other Pacific Islander : having origins in any of the original peoples of Hawaii, Guam, Samoa, or other Pacific Islands. White : having origins in any of the original peoples of Europe, the Middle East, or North Africa Other: specify ___ ________ Instructions: 1 Please answer all 15 questions to the best of your ability. 2 Answer by checking the box or filling in the blank.

PAGE 125

125 5. What is your highest level of education? Did not go to school Completed Grade school (5th grade) Completed Middle school (8th grade) Completed High School/G.E.D. (12th grade) Completed Vocational Training Som e College after High School Graduation Associate Degree Some Professional School after College Graduation Doctoral Degree No Yes 7. How many days a wee k do you typically drive? 0 1 2 3 4 5 6 7 Go to question # 10) No Yes Mostly (for part of the year) 9. Who lives with you? Spouse or partner Child Family/Other relative: specify: ________ Friend(s) Paid caregiver Other: specify __________________ 10. What is your relationship with the driver we are testing? Spouse or partner Child Family/Other relative: specify: ________

PAGE 126

126 Friend(s) Paid caregiver Other: specify __________________ 11. How many other licensed drivers are in your household? _______ 12. Do you rely on the driver for any of the following trips or activities? Shopping Grocery store Social activities See friends or family Church See doctor or get medical care Work related acti vities Other ( please list)______________________________ 13. How many days a week do you ride with the driver for whom you are completing the checklist? 0 1 2 3 4 5 6 7 14. If the driver reduced or stopped driving would it significantly impact your c urrent lifestyle? No Yes ____________________________________________________________

PAGE 127

127 APPENDIX C FOCUS GROUP TWO GUID E Focus Group Introduction Script From our research, we know that driving is important and people generally want to continue to drive as they age. We are studying safe and unsafe driving to develop a self report checklist for either a self rating or rating the driving of someone you live with or care for. We would like your input on what to include on the checklist. We will discuss your experience with everyday driving situations including both safe driving behaviors and unsafe driving behaviors. Situations: 1 ) What challenges do you face on local trips? a What are your safety concerns? b What do you do about your safety concerns? 2 ) Can you describe a trip you have had on a busy local road (multiple lanes with high traffic and businesses such as Archer Road or University Avenue)? 3 ) Do you plan your driving to drive on certain types of roads or avoid certain situations like high speed roads, heavy traffic or having to make a left turn without a traffic light? a What do you do to adjust your driving in these si tuations? 4 ) What are your safety concerns when taking a state road, like 441? a What do you do about your safety concerns? b c What ar e those conditions? d What do you do instead? 5 ) What are your safety concerns when taking an interstate, like I 75? a What do you do about your safety concerns? b 75? c What are those conditions? d What do you do instead? 6 ) For any of the roads we have discussed, have you noticed features of the road that help you drive? For example, have you driven on a road with wider lane markings (stripes) or with reflectors? a Can you describe the road features that h elped you drive?

PAGE 128

128 7 ) What weather conditions do you consider unsafe for driving? a If you had to drive in those conditions, what would you do to stay safe? 8 ) How do you handle distractions when you drive, such as ea ting, drinking, talking with a passenger or having a pet in the car? 9 ) Describe the most challenging situations you face when driving? a How do you handle those situations? 10 ) Do you have any driving difficulty relate d to your health? a How do you handle those challenges? 11 ) Do certain features of your car make driving easier (i.e. more comfortable or safer)? For example, does your car have wide angle mirrors, adjustable seats or an adjus table steering wheel? a Can you describe any car features make driving more comfortable? b Can you describe any car features that make driving safer? c Are there any car features you would change to improve either the comfort or safety of your driving? 12 ) Do you have other thoughts or opinions you would like to share about mature drivers and safe or unsafe driving?

PAGE 129

129 LIST OF REFERENCES AAA Foundation for Traffic Safety. (1994). Drivers 55 plus: check your own performance: a self rating form of questions, fa cts and suggestions for safe driving. Washington, D.C.: AAA Foundation for Traffic Safety. AARP (1992). Older driver skill assessment and resource guide (Creating mobility choices). Washington, DC: AARP. American Automobile Association. (2004). Roadwise R eview [Computer software]. Heathrow, FL: AAA Public Affairs. American Occupational Therapy Association. (2005). Statements: Driving and community mobility. American Journal of Occupational Therapy,59 (6), 666 670. American Occupational Therapy Association (2008). Occupational therapy practice framework: Domain and process (2nd ed.). American Journal of Occupational Therapy, 62 (6), 625 683. Anstey, K. J., Wood, J., Lord, S., & Walker, J. G. (2005). Cognitive, sensory and physical factors enabling driving sa fety in older adults. Clinical psychology review, 25 (1), 45 65. Awadzi, K., Classen, S., Garvan, C. W., & Komaragiri, V. (2006). Determinants of older driver safety from a socio ecological perspective. Topics in Geriatric Rehabilitation, 22 (1), 36 44. Awad zi, K. D., Classen, S., Hall, A., Duncan, R.P., & Garvan, C.W. (2008). Predictors of injury among younger and older adults in fatal motor vehicle crashes. Accident Analysis & Prevention, 40 1804 1810 Bdard, M., Guyatt, G. H., Stones, M. J., & Hirdes, J. P. (2002). The independent contribution of driver, crash, and vehicle characteristics to driver fatalities. Accident Analysis & Prevention, 34 717 727. Bdard, M., Isherwood, I., Moore, E., Gibbons, C., & Lindstrom, W. (2004). Evaluation of a re train ing program for older drivers. Canadian Journal of Public Health, 95 (4), 295 298. Bdard, M., Stones, M. J., Guyatt, G. H., & Hirdes, J. P. (2001). Traffic related fatalities among older drivers and passengers: Past and future trends. The Gerontologist, 41 751 756. Bdard M., Weaver, B., Darzins, P., & Porter, M. M. (2008). Predicting driving performance in older adults: We are not there yet! Traffic Injury Prevention, 9 (4), 336 341.

PAGE 130

130 Bond, T. G., & Fox, C. M. (2007). Applying the Rasch model. Fundamenta l measurement in the human sciences. (2nd ed.). Mahwah, NJ: Lawrence Erlbaum Associates. Cacciabue, P. C. (2007). Modelling driver behaviour in automotive environments: Critical issues in driver interactions with Intelligent Transport Systems. New York: Sp ringer Verlag Inc. Canadian Association of Occupational Therapists (2009). CAOT position statement: Occupational therapy and driver rehabilitation. Retrieved June 30, 2009, from http://www.caot.ca/default.asp?pageid=1353 Carmines, E. G., & Zeller, R. (1979 ). Validity. In Reliability and validity assessment (pp. 17 27). Thousand Oaks, CA: Sage. Castel, L. D., Williams, K. A., Bosworth, H. B., Eisen, S. V., Hahn, E. A., Irwin, D. E., et al. (2008). Content validity in the PROMIS social health domain: a qualit ative analysis of focus group data. Quality of life research : an international journal of quality of life aspects of treatment, care and rehabilitation, 17 (5), 737 749. Centers for Disease Control and Prevention. (1999) Surveillance for Injuries and Viole nce Among Older Adults. MMWR CDC Surveillance Summary, 48 (SS08), 27 50 Centers for Disease Control and Prevention. (2007). Older adult drivers: Fact sheet. Retrieved August 25, 2008, from http:/ /www.cdc.gov/ncipc/factsheets/older.htm Centor, R. M. (1991). Signal detectability: The use of ROC curves and their analyses. Medical Decision Making, 11 (2), 102 106. Classen, S., Awadzi, K. D., & Mkanta, W. M. (2008). Person vehicle environment inte ractions predicting crash related injury among older drivers. The American Journal of Occupational Therapy, 62 (5), 5 80 587. Classen, S., Garvan, C. W., Awadzi, K., Sundaram, S., Winter, S., Lopez, E. D. S., et al. (2006). Systematic literature review and s tructural model for older driver safety. Topics in Geriatric Rehabilitation, 22 (2), 87 98. Classen, S., Lopez, E. D. S., Winter, S., Awadzi, K., Ferree, N., & Garvan, C. W. (2007). Population based health promotion perspective for older driver safety: Conc eptual framework to intervention plan. Clinical Interventions in Aging, 2 (4), 677 693. Classen, S., Shechtman, O., Stephens, B., Davis, E., Justiss, M., Bendixen, R., et al. (2007). The impact of roadway intersection design on driving performance of young and senior adults. Traffic Injury Prevention, 8 (1), 69 77.

PAGE 131

131 Classen, S., Winter, S., & Lopez, E. D. S. (2009). Meta synthesis of qualitative studies on older driver safety and mobility. The Occupational Therapy Journal of Research: Occupation, Participation and Health, 29 (1), 24 31. Classen, S., Winter, S. M., Velozo, C. A., Bdard, M., Lanford, D., Brumback, B., et al. (in press). Item development and validity testing for a Safe Driving Behavior Measure. American Journal of Occupational Therapy Cotton, D., & Stewart, A. (1983). Focus on planning: A drink driving programme needs assessment. In P. A. Howat, J. A. Bunbury & K. J. Fisher (Eds.), Resources and responsibility: Proceedings of the 53rd Australian and New Zealand Association for the Advancement of S cience Congress, Health Education Section (pp. 118 129). Perth: School of Community Health and the Centre for Advanced Studies in Health Sciences. Dellinger, A. M., Kresnow, M. J., White, D. D., & Sehgal, M. (2004). Risk to self versus risk to others: How do older drivers compare to others on the road? American Journal of Preventive Medicine, 26 (3), 217 221. Di Stefano, M., & MacDonald, W. (2005). On the Road Evaluation of Driving Performance. In J. M. Pellerito (Ed.), Driver rehabilitation and community mo bility: Principles and practice (pp. 255 274). St Louis, Missouri: Elsevier Mosby. Dickerson, A. E., Molnar, L. J., Eby, D. W., Adler, G., Bedard, M., Berg Weger, M., et al. (2007). Transportation and aging: a research agenda for advancing safe mobility. G erontologist, 47 (5), 578 590. Diehl, M. (1998). Everyday competence in later life: current status and future directions. Gerontologist, 38 (4), 422 433. Dobbs, B. M., Carr, D. B., & Morris, J. C. (2002). Evaluation and management of the driver with dementia Neurologist, 8 (2), 61 70. Eby, D. W., Molnar, L. J., Shope, J. T., & Dellinger, A. M. (2007). Development and pilot testing of an assessment battery for older drivers. Journal of Safety Research, 38 (5), 535 543. Eby, D. W., Molnar, L. J., Shope, J. T., Vivoda, J. M., & Fordyce, T. A. (2003). Improving older driver knowledge and self awareness through self assessment: the driving decisions workbook. Journal of Safety Research, 34 (4), 371 381. Eby, D. W., Trombley, D. A., Molnar, L. J., & Shope, J. T. (199 8). The assessment of (No. UMTRI 98 24). Ann Arbor, MI: The University of Michigan Transportation Research Institute.

PAGE 132

132 Folstein, M. F., Folstein, S. E., & McHugh, P. R. (1975). "Mini mental state". A p ractical method for grading the cognitive state of patients for the clinician. Journal of Psychiatric Research, 12 (3), 189 198. Fowler, F., Jr. (1995). Improving survey questions: Design and evaluation Thousand Oaks, CA: Sage Publications. Furr, R. M., & Bacharach, V. R. (2008). Psychometrics: An introduction Thousand Oaks, CA: Sage Publications. Gibson, J. J. (1964). The contribution of experimental psychology to the formulation of the problem of safety a brief for basic research. In W. Haddon, E. A. Su chman & D. Klein (Eds.), Accident research: methods and approaches New York, NY: Harper & Row. Goode, K. T., Ball, K., Sloane, M., Roenker, D. L., Roth, D. L., Myers, R. S., et al. (1998). Useful Field of View and other neurocognitive indicators of crash risk in older adults. Journal of Clinical Psychology in Medical Settings, 5 (4), 425 440. Gordon, J. E. (1949). The epidemiology of accidents. American Journal of Public Health, 39 504 515. Grant, J. S., & Davis, L. L. (1997). Selection and use of content experts for instrument development. Research in Nursing and Health, 20 (3), 269 274. Green, L. W., George, M. A., Daniel, M., Frankish, C. J., Herbert, C. J., Bowie, W. R., et al. (1995). Study of participatory research in health promotion: Review and recom mendations for the development of participatory research in health promotion in Canada Ottawa, Ontario: Royal Society of Canada. Green, L. W., & Kreuter, M. W. (2005). Health program planning: An educational and ecological approach (4th ed.). New York: M cGraw Hill. Haddon, W. J. (1972). A logical framework for categorizing highway safety phenomena and activity. The Journal of Trauma, 12 (3), 193 207. Hambleton, R. K., Swaminathan, H., & Rogers, H. J. (1991). Fundamentals of item response theory. London: Sa ge. Handler, B. S., & Patterson, J. B. (1995). Driving after brain injury. Journal of Rehabilitation, 61 43 49. House, A., House, B., & Campbell (1981). Measures of interobserver agreement: Calculation formulas and distribution effects. Journal of Psychop athology and Behavioral Assessment, 3 37 57.

PAGE 133

133 Israel, B. A., Schulz, A. J., Parker, E. A., & Becker, A. B. (1998). Review of community based research: assessing partnership approaches to improve public health. Annual Review of Public Health, 19 173 202. Janssen, W. H. (1979). Routeplanning en geleiding: Een literatuurstudie. [Route planning and guidance: A literature study] (No. IZF 1979 C 13). Soesterberg, Netherlands: Institute for Perception TNO. Justiss, M. (2006). Development of a Behind the wheel dr iving performance assessment for older adults. Unpublished Dissertation, University of Florida, Gainesville. Justiss, M., Mann, W., Stav, W., & Velozo, C. (2006). Development of a Behind the Wheel Driving Performance Assessment for Older Adults. The Older Driver, Part 2. Topics in Geriatric Rehabilitation, 22 (2), 121 128. Kay, L., Bundy, A., & Clemson, L. (2009). Validity, reliability and predictive accuracy of the Driving Awareness Questionnaire. Disability and Rehabilitation, 31 (13), 1074 1082. Kay, L., B undy, A., Clemson, L., & Jolly, N. (2008). Validity and reliability of the on road driving assessment with senior drivers. Accident Analysis & Prevention, 40 (2), 751 759. Korner Bitensky, N., Gelinas, I., Man Son Hing, M., & Marshall, S. (2005). Recommenda tions of the Canadian Consensus Conference on driving evaluation in older drivers. In W. Mann (Ed.), Community mobility: Driving and transportation alternatives for older persons. Binghamton, NY: Haworth Press. Krueger, R. A., & Casey, M. A. (2000). Focus groups: A practical guide for applied research (3rd ed.). Thousand Oaks, CA: Sage. Kua, A., Korner Bitensky, N., & Desrosiers, J. (2007). regarding driving: Focus group findings. Physical & Occupational Therapy in Geriatrics 25 (4), 21 40. Kua, A., Korner Bitensky, N., Desrosiers, J., Man Son Hing, M., & Marshall, S. (2007). Older driver retraining: A systematic review of evidence of effectiveness. Journal of Safety Research, 38 (1), 81 90. Lajunen, T., & Summala, H. (2003). C an we trust self reports of driving? Effects of impression management on driver behaviour questionnaire responses. Transportation Research Part F: Traffic Psychology and Behaviour, 6 (2), 97 107. Lawton, R., Park er, D., Manstead, A. S. R., & Stradling, S. G. (1997). The role of affect in predicting social behaviors: The case of road traffic violations. Journal of Applied Social Psychology, 27 (14), 1258 1276.

PAGE 134

134 Lynn, M. R. (1986). Determination and quantification of content validity. Nursing Research, 35 (6), 382 385. Magaziner, J., Zimmerman, S. I., Gruber Baldini, A. L., Hebel, J. R., & Fox, K. M. (1997). Proxy reporting in five areas of functional status: Comparison with self reports and observations of performance. American Journal of Epidemiology, 146 (5), 418 428. Marottoli, R. (2007). Enhancement of driving performance among older drivers : AAA Traffic Safety Foundation. Marottoli, R. A., Cooney, L. M., Jr., & Tinetti, M. E. (1997). Self report versus state records for identifying crashes among older drivers. Journals of Gerontology Series A: Biological Sciences and Medical Sciences, 52 (3), M184 187. Marottoli, R. A., & Richardson, E. D. (1998). Confidence in, and self rating of, driving ability among older drivers. Accident Analysis & Prevention, 30 (3), 331 336. McDowell, I. (2006). Measuring health: a guide to rating scales and questionnaires (3rd ed.). New York: Oxford University Press. McGee, P., & Tuokko, H. (2003). The older & wiser driver: A self assessment pr ogram Victoria, British Columbia: Centre on Aging, University of Victoria. McGwin, G., & Brown, D. B. (1999). Characteristics of traffic crashes among young, middle aged, and older drivers. Accident Analysis & Prevention, 31 (3), 181 198. Michon, J. A. (19 85). A critical view of driver behavior models: What we know, what should we do? In E. L. Evans & R. Schwing (Eds.), Human behavior and traffic safety (pp. 485 520). New York: Plenum. Miles, M. B., & Huberman, A. M. (1994). Qualitative data analysis (2nd e d.). Thousand Oaks, CA: Sage. Minkler, M. & Wallerstein, N. (Eds). (2003). Community based participatory research for health San Francisco: Jossey Bass. Muhr, T. (2004). ATLAS.ti (Version 5.0). Berlin: ATLAS.ti Scientific Software Development GmbH. Myers, A. M., Blanchard, R. A., MacDonald, L., & Porter, M. M. (2008). Process evaluation of the American Automobile Association Roadwise Review CD ROM: Observed and reported experiences of older drivers. Topics in Geriatric Rehabilitation: Special Issue on Publ ic Health, Disability, and Aging 24 (3), 224 238.

PAGE 135

135 Myers, A. M., Paradis, J. A., & Blanchard, R. A. (2008). Conceptualizing and measuring confidence in older drivers: development of the day and night driving comfort scales. Archives of Physical Medicine & Rehabilitation, 89 (4), 630 640. National Highway Traffic Safety Administration. (2008a). Traffic Safety Facts 2008 Data: Older Population (No. DOT HS 811 161). Washington, D.C: National Highway Traffic Safety Administration. National Highway Traffic Safety Administration (2008b). Older driver self screening based on health concerns (No. DOT HS 811 046A ). Washington, D. C. : National Highway Traffic Safety Administration. Odenheimer, G. L., Beaudet, M., Jette, A. M., Albert, M. S., Grande, L., & Minaker, K. L. (1994). Performance based driving evaluation of the elderly driver: safety, reliability, and validity. Journal of Gerontology, 49 (4), M153 159. Owsley, C., Ball, K., McGwin, G., Jr., Sloane, M. E., Roenker, D. L., White, M. F., et al. (1998). Visual pro cessing impairment and risk of motor vehicle crash among older adults. JAMA, 279 (14), 1083 1088. Owsley, C., Stalvey, B. T., Wells, J., & Sloane, M. E. (1999). Older drivers and cataract: Driving habits and crash risk. Journals of Gerontology: Series A: B iological Sciences & Medical Sciences, 54 (4), M203 M211. Pachana, N., & Petriwskyj, A. M. (2006). Assessment of insight and self awareness in older drivers. Clinical Gerontologist, 30 (1), 23 38. Peek Asa, C., & Zwerling, C. (2003). Role of environmental in terventions in injury control and prevention. Epidemiology Review, 25 (1), 77 89. Portney, L., & Watkins, M. P. (2000). Foundations of clinical research: Applications to practice (2nd ed.). Upper Saddle River, N.J: Prentice Hall Health. R Development Core T eam. (2008). R: A Language and Environment for Statistical Computing. Vienna, Austria: R Foundation for Statistical Computing. Reason, J., Manstead, A., Stradling, S., Baxter, J., & Campbell, K. (1990). Errors and violations on the roads: a real distinctio n? Ergonomics, 33 (10), 1315 1332. Runyan, C. W. (1998). Using the Haddon's matrix: Introducing the third dimension. Injury prevention, 4 (4), 302 307. Runyan, C. W. (2003). Introduction: Back to the Future -Revisiting Haddon's Conceptualization of Injury Epidemiology and Prevention. Epidemiology Review, 25 (1), 60 64.

PAGE 136

136 Runyan, C. W., Marshall, S. W., Coyne Beasley, T., & Casteel, C. (2005). Commentary: Recommendations for home safety research and intervention. American Journal of Preventive Medicine, 28 (1), 116 118. Ryan, G. W., & Bernard, H. R. (2003). Techniques to identify themes. Field Methods, 15 (1), 85 109. Sackett, D. L., & Haynes, R. B. (2002). Evidence base of clinical diagnosis: The architecture of diagnostic research. BMJ, 324 (7336), 539 541. Seber G. A. F. (1977). Linear Regression Analysis New York: John Wiley & Sons. Shechtman, O., Awadzi, K. D., Classen, S., Lanford, D., & Joo, Y. (in press). Validity and critical driving errors of on road assessment for older drivers. American Journal of Occu pational Therapy Sleet, D. A. (1987). Health education approaches to motor vehicle injury prevention Public Health Reports, 102 606 608. Snow, A. L., Cook, K. F., Lin, P. S., Morgan, R. O., & Magaziner, J. (2005). Proxies and other external raters: metho dological considerations. Health Services Research, 40 (5, Part 2), 1676 1693. SPSS Inc. (2008). SPSS for Windows, Release 17.0.0. Chicago, Illinois: SPSS Inc. Stalvey, B. T., & Owsley, C. (2000). Self perceptions and current practices of high risk older dr ivers: Implications for driver safety interventions. Journal of Health Psychology, 5 (4), 441 456. Staplin, L., & Dinh Zarr, T. B. (2006). Promoting rehabilitation of safe driving abilities through computer based clinical and personal screening techniques. Topics in Geriatric Rehabilitation, 22 (2), 129 138. Stav, W. B., Justiss, M. D., McCarthy, D. P., Mann, W. C., & Lanford, D. N. (2008). Predictability of clinical assessments for driving performance. Journal of Safety Research, 39 (1), 1 7. Stereo Optical C ompany Inc. (2007). Reference and instruction manual: Optec 2500 Vision Tester Chicago, Illinois. Stevenson, M., Iredell, H., Howat, P., & Cross, D. (1999). Measuring community/environmental interventions: The Child Pedestrian Injury Prevention Project. I njury Prevention, 5 26 30. Stone, M. H. (1997). Steps in item construction. Rasch Measurement Transactions, 11 (2), 559.

PAGE 137

137 Streiner, D. L., & Cairney, J. (2007). What's under the ROC? An introduction to Receiver Operating Characteristics curves. Canadian Jou rnal of Psychiatry, 52 (2), 121 128. Streiner, D. L., & Norman, G. R. (2003). Devising the Items. In Health measurement scales: A practical guide to their development and use (pp. 15 27). Oxford: Oxford University Press. Sundstrom, A. (2005). Self assessmen t of knowledge and abilities (No. EM No 54). Ume, Sweden: Ume University. Sundstrom, A. (2008). Self assessment of driving skill A review from a measurement perspective. Transportation Research Part F: Traffic Psychology Behavior, 11 1 9. Trifiletti, L. B., Gielen, A. C., Sleet, D. A., & Hopkins, K. (2005). Behavioral and social sciences theories and models: are they used in unintentional injury prevention research? Health Education Research, 20 (3), 298 307. Transport Canada (2007). Canadian motor vehi cle traffic collision statistics: 2006 (Report No. TP3322), Ottawa, Ontario: Transport Canada. Unsworth, C. A., Lovell, R. K., Terrington, N. S., & Thomas, S. A. (2005). Review of tests contributing to the occupational therapy off road driver assessment. Australian Occupational Therapy Journal, 52 (1), 57 74. Vogt, D. S., King, D. W., & King, L. A. (2004). Focus groups in psychological assessment: enhancing content validity by consulting members of the target population. Psychological Assessment, 16 (3), 231 243. Wadley, V. G., Okonkwo, O., Crowe, M., Vance, D. E., Elgin, J. M., Ball, K. K., et al. (2009). Mild cognitive impairment and everyday function: An investigation of driving performance. Journal of Geriatric Psychiatry and Neurology, 22 (2), 87 94. Walt z, C. F., Strickland, O. L., & Lenz, E. R. (2005). Measurement in nursing and health research (3rd ed.). New York: Springer Publishing. Wang, C. C., & Carr, D. B. (2004). Older driver safety: a report from the older drivers project. Journal of the American Geriatrics Society, 52 (1), 143 149. West, R., French, D., Kemp, R., & Elander, J. (1993) Direct observation of driving, self reports of driver behaviour, and accident involvement. Ergonomics, 36 (5), 557 567. Wild, K., & Cotrell, V. (2003). Identifying driving impairment in Alzheimer Disease: A comparison of self and observer reports versus driving evaluation. Alzheimer Disease & Associated Disorders, 17 (1), 27 34.

PAGE 138

138 Windsor, T. D., Walker, J. G., Caldwell, H., & Anstey, K. J. (2006). Health knowledge rel ated to driving, and driving associated anxiety: A public health approach to driving in later life. Canberra, Australia: Ageing Research Unit, Centre for Mental Health Research, Australian National University Yee, D., & Melichar, J. F. (1992). Accident pr evention through driving skills assessment and interventions for older drivers: A programmatic research project San Francisco: San Francisco State University, Department of Health Education. Zaza, S., Carande Kulis, V. G., Sleet, D. A., Sosin, D. M., Elde r, R. W., Shults, R. A., et al. (2001). Methods for conducting systematic reviews of the evidence of effectiveness and economic efficiency of interventions to reduce injuries to motor vehicle occupants. American Journal of Preventive Medicine, 21 (4, Supple ment 1), 23 30. Zhang, J., Fraser, S., Lindsay, J., Clarke, K., & Mao, Y. (1998). Age specific patterns of factors related to fatal motor vehicle traffic crashes: focus on young and elderly drivers. Public Health, 112 (5), 289 295. Zhou, M., & Lyles, R. W. (1997). Self selection bias in driver performance studies. Transportation Research Record: Journal of the Transportation Research Board, 1573 86 90.

PAGE 139

139 BIOGRAPHICAL SKETCH Sandra Winter graduated from Halifax High School, Halifax, PA, in 1985. In 1989, she received a dual Bachelor of Science degree in Parks and Recreation/Therapeutic Recreation and Anthropology from Slippery Rock University. From 1989 to 1995 she worked as a Certified Therapeutic Recreation Specialist in Florida, North Carolina, and Michiga n. She attended Western Michigan University in Kalamazoo, Michigan, from 1994 to 1997, graduating with a Master of Science in Occupational Therapy. Sandra practiced occupational therapy at the ARC of Palm Beach County from 1998 to 2004, working primarily i disabilities. From 2004 to 2009 she attended the University of Florida in Gainesville, graduating with a PhD in rehabilitation science and a concentration in disability science.