Conceptualizing and Measuring Confidence in Older Drivers: Development of the Day and Night Driving Comfort Scales

630

ORIGINAL ARTICLE

Conceptualizing and Measuring Confidence in Older Drivers: Development of the Day and Night Driving Comfort Scales Anita M. Myers, PhD, Josee A. Paradis, MSc, Robin A. Blanchard, MSc ABSTRACT. Myers AM, Paradis JA, Blanchard RA. Conceptualizing and measuring confidence in older drivers: development of the Day and Night Driving Comfort Scales. Arch Phys Med Rehabil 2008;89:630-40. Objective: To examine and measure driving confidence from the perspective of older adults. Design: Used focus groups for construct examination, item generation, and ratings; conducted psychometric testing using Rasch analysis for scale refinement; examined test-retest reliability and associations with driver characteristics and driving habits. Setting: Retirement complexes and seniors’ housing and centers in Ontario, Canada. Participants: Convenience samples of current drivers (n⫽143) (range, 66 –92y) and 7 counselors. Interventions: Not applicable. Main Outcome Measure: The Day (DCS-D) and Night (DCS-N) Driving Comfort Scales developed inductively with older drivers. Results: Older drivers believed that it was important to consider confidence in their own abilities and discomfort caused by other drivers, to separate day and night driving, and to specify the driving context (eg, traffic flow, speed). Rasch analysis showed that the final 13-item DCS-D and 16-item DCS-N were both hierarchic and unidimensional, with good person (.89, .96) and item (.98, .97) reliabilities, respectively. Test–retest reliability was adequate for the DCS-D (intraclass correlation coefficient [ICC]⫽.7) and good for the DCS-N (ICC⫽.88). Scores were significantly associated with reported driving frequency, situational avoidance, and perceived abilities (P⬍.001). Conclusions: The Driving Comfort Scales are promising new tools for research and practice. Key Words: Aged; Aged 80 and over; Automobiles; Rehabilitation; Reproducibility of results. © 2008 by the American Congress of Rehabilitation Medicine and the American Academy of Physical Medicine and Rehabilitation LTHOUGH OLDER DRIVERS are disproportionately involved in fatal collisions (per miles driven) and represent A the fastest growing segment of the driving population, efforts 1-3

to restrict older drivers such as age-based licensing require-

From the Department of Health Studies and Gerontology, University of Waterloo, Waterloo, ON, Canada. Presented to the Gerontological Society of America, November 2006, Dallas, TX. Supported by the Canadian Driving Research Initiative for Vehicular Safety in the Elderly (CanDRIVE), Canadian Institutes of Health Research. No commercial party having a direct financial interest in the results of the research supporting this article has or will confer a benefit upon the authors or upon any organization with which the authors are associated. Correspondence to Anita M. Myers, PhD, University of Waterloo, 200 University Ave W, Waterloo, ON N2L 3G1, Canada, e-mail: [email protected]. Reprints are not available from the authors. 0003-9993/08/8904-00455$34.00/0 doi:10.1016/j.apmr.2007.09.037

Arch Phys Med Rehabil Vol 89, April 2008

ments must be weighed against the impact on quality of life.2,4 Driving is the preferred mode of transportation in North America, and many seniors, particularly those in rural areas, rely on driving to maintain their mobility and independent lifestyles.1,4-6 Seniors who forfeited their licenses have expressed regret, social isolation, and loss of self-worth.6,7 Prospectively, driving cessation has been associated with both increased depression8 and reduced out-of-home activities.9 Not surprisingly, most seniors want to keep driving for as long as possible and want control over the decision to stop.6 For some, driving cessation may be a gradual process of self-imposed restrictions and compensations.10 In a large survey, nearly half of older drivers said that they drove less than they did 10 years ago, and as age increased they were more likely to avoid driving in peak hours, on highways, in bad weather, and at night.5 Although findings are mixed concerning the extent to which older drivers are aware of declining abilities or willing to acknowledge such limitations,11-16 it is becoming increasingly apparent that self-perceptions play a critical role in determining why some people adjust their driving but others do not.4,6,10-16 In particular, confidence may be an important mediator between declining abilities, associated problems (such as night blindness), and ensuing self-regulation.6,11,12,14 Driving confidence is worthy of investigation given the proven importance of self-efficacy in other domains, such as the mediating role of balance confidence in self-imposed activity restriction.17,18 Rooted in Bandura’s social cognitive theory, self-efficacy is a stronger determinant of behavior than one’s actual skills or abilities.19 People who lack confidence in a particular domain will avoid challenging situations (specific to that context) as much as possible and are less likely to persist in the face of obstacles. Self-efficacy is primarily influenced, either positively or negatively, by (1) performance, (2) vicarious experience, (3) verbal persuasion, and (4) physiologic cues (eg, feeling anxious).19 There have been a number of attempts to measure this construct—for example, by asking older drivers to rate their stress level in 16 driving situations.10 Parker et al11 created a nervousness/confidence subscale (␣⫽.87) using 12 questions from their Driving Behavior Questionnaire: 7 on the extent of nervousness in various driving situations and single items on how relaxed, stressed, confident, flustered, or calm people usually feel when driving. Another group of researchers, meanwhile, adapted items from the Driving Habits Questionnaire16 from a difficulty to a confidence-rating format but without further psychometric testing.20 When this project began, only 1 scale designed specifically to measure driving confidence had been published. The Driving Confidence Rating Scale asks respondents to rate their levels of confidence from 0 (not at all confident) to 10 (completely confident) driving in each of the following conditions: at night, in bad weather, in rush hour or heavy traffic, on the highway, on long trips, changing lanes on a busy street, reacting quickly, pulling into traffic from a stop, making a left turn across traffic, and parallel parking into a space between cars.14 Unfortunately, the developers did not provide any psychometric evidence for this scale.

631

DAY AND NIGHT DRIVING COMFORT SCALES, Myers Table 1: Characteristics of Older Driver Samples Study 1

Study 2

Ratings (n⫽49)

Pilot (n⫽16)

Characteristics

Focus Groups (n⫽42)

Validation (n⫽100)

Retest Group (n⫽ 27)

Region Urban Rural Women Age (y) College education Worry about car expenses Lives alone Other driver in household Others rely on them Cataract surgery Use cane or walker Able to walk .25 mile (.40km) No. of days driven past week

24 (57.1) 18 (42.9) 27 (64.3) 78.0⫾5.5 19 (45.2) 19 (46.3) 24 (57.1) 16 (38.1) 16 (40.0) 14 (35.0) 9 (23.1) 33 (84.6) 5.0⫾2.0

27 (55.1) 22 (44.9) 34 (69.4) 79.0⫾5.9 16 (32.6) 22 (44.9) 31 (63.3) 21 (43.8) 15 (31.9) 21 (42.9) 9 (18.8) 9 (60.0) 4.4⫾2.5

16 (100.0) 0 (0.0) 9 (56.3) 78.0⫾4.6 9 (56.3) 5 (31.3) 10 (62.5) 5 (31.3) 5 (35.7) 5 (31.3) 6 (40.0) 13 (86.7) 4.0⫾2.3

68 (68.0) 32 (32.0) 61 (61.0) 79.7⫾6.4 32 (32.0) 38 (38.0) 58 (58.0) 38 (38.0) 36 (40.9) 42 (42.8) 32 (32.9) 77 (79.4) 4.8⫾2.2

22 (81.5) 5 (18.5) 18 (66.7) 78.6⫾6.6 12 (50.0) 21 (77.8) 17 (63.0) 7 (25.9) 10 (40.0) 8 (29.6) 10 (37.0) 22 (81.5) 5.7⫾1.9

NOTE. Values are n (% based on valid cases) or mean ⫾ standard deviation (SD).

Subsequently, another tool—the Adelaide Driving Self-Efficacy Scale (ADSES)— has been published, with some accompanying psychometric support (internal consistency, ability of scores to discriminate between older adults with stroke and younger hospital staff, and those who passed or failed on-road driving tests).21 The ADSES consists of 12 items rated from 0 (not confident) to 10 (completely confident). Content is similar but not identical to Marottoli and Richardson’s scale.14 Both developers used a deductive approach for item generation (ie, based on the literature and their judgment).14,21 Together, these studies suggest that perceived stress, nervousness, and confidence appear to be associated with reported driving frequency11,14 and avoidance,10,20 as well as perceptions of driving abilities.11,14 Findings are mixed, however, concerning the association with adverse driving events and on-road performance.11,14,20,21 In any case, the findings are difficult to interpret because the construct itself has not been well defined, consistently measured, or thoroughly examined from a psychometric standpoint. The present studies were conducted to further examine this construct from the perspective of older drivers themselves and to develop a scale based on Bandura’s theory.19 A mixedmethods, inductive approach (ie, involving the intended recipients in the scale development process) was used to enhance the tool’s conceptual relevance and content validity.22,23 Rasch analysis was used to examine the tool’s structure and help interpret the numeric ratings in relation to the descriptive content and underlying construct.24 Thus, this project afforded the opportunity to compare the resulting content of scales, which purportedly measure the same underlying construct (ie, driving confidence) using distinct approaches to item generation (inductive vs deductive) and structural examination (use vs nonuse of Rasch analysis). METHODS The current project involved 2 studies. The first study (tool development) comprised 3 sequential steps: construct exploration and item generation, item verification, and pilot-testing. The second study (validation) examined the new tool’s structural properties, test-retest reliability, and associations with driver characteristics and driving habits. All procedures were approved by the university ethics board, and informed consent was obtained from all participants.

Convenience samples of English-speaking persons aged 65 years and older, who held a valid driver’s license and currently drove, were recruited through local seniors’ centers, retirement complexes, and subsidized seniors’ housing. To capture regional diversity (types of roads and distance to services), particularly for study 1, recruitment took place in 2 communities (urban and rural) situated in different areas of the province (south and north, respectively). All participants were asked to complete a background questionnaire to obtain demographic and health information and general driving habits. Table 1 shows the characteristics of the samples. Study 1: Conceptualization and Development Step 1: Exploration and item generation. Our first objective was to gain a better understanding of driving confidence from the perspective of older drivers themselves. We conducted 4 separate focus groups—2 in each community with drivers aged 65 to 79 years and those aged 80 years or above. This age grouping was considered important because, in Ontario, all drivers over the age of 80 years are required to undergo mandatory re-licensing every 2 years (comprising vision and rules testing, review of driving records, and a group education session). Semi-structured scripts were used to guide the discussion. Embedded in a general discussion about driving, the key probes were as follows: (1)

Do you ever feel concerned while driving (and how would you describe this feeling)? (2) In what situations do you tend to feel _____ (using their terms from question 1)? (3) Do you try to avoid driving in ______ (from question 2)? (4) What if you had an appointment or needed groceries, what do you do? Step 2: Item verification. Situations emerging from the focus group data that appeared to evoke concerns among older drivers were selected for further examination. The objectives of this step were to obtain ratings and feedback on the pool of potential items or driving situations, as well as further clarification of the construct. Once again, 4 groups (2 in each community, segregated by age) were assembled for this purpose. Participants from the previous phase were invited to return and constituted 55% (n⫽27) of the total sample (N⫽49). Arch Phys Med Rehabil Vol 89, April 2008

632

DAY AND NIGHT DRIVING COMFORT SCALES, Myers

Participants were asked to rate the various situations and complete the background questionnaire on their own. The ratings comprised frequency driving in each situation (now and compared with 10 years ago), confidence (0%–100%), and identifying which 3 situations made them most uncomfortable. The groups then reviewed the tools for clarity of instructions, response options, and content (whether any important situations had been missed). In addition, driving improvement counselors from the Ontario Ministry of Transportation were asked to rate these situations from their perspective as “experts.” The driving improvement counselors deliver the education session and conduct the testing for re-licensing drivers 80 years and over. Through their supervisors, driving improvement counselors were e-mailed the letter of information and rating form. They were asked to rate the extent to which each situation was problematic for older drivers, as well as whether any important driving situations had been missed. Seven counselors returned the rating form. Step 3: Pilot testing. Explained in the results, the ratings and feedback from step 2 led to the refinement of various items, removal of others, and creation of a preliminary new scale— labeled the Driving Comfort Scale (DCS). Subjects were asked to rate how comfortable they were driving in the daytime (25 items) and at night (15 items), using an 11-point scale from 0% (not at all comfortable) to 100% (completely comfortable). Two of the previous rating forms (frequency of current driving, 10-y comparison) with some modifications were also administered, and a new rating form (situational avoidance) was added. The objectives of this step were to determine the ease and time for completion of the new DCS, obtain feedback on all aspects of the tool, examine preliminary scale and item properties, and modify the tool accordingly. The preliminary DCS and other rating forms were pilot-tested with 2 groups of older drivers, both in southern Ontario for convenience. Sixteen volunteers, all of whom had previously participated in either step 1 or 2, were separated by sex (7 men, 9 women) to compare perspectives. Study 2: Psychometric Testing and Scale Refinement In this study, the modified 17-item Driving Comfort Scale for day driving (DCS-D) and the 18-item Driving Comfort Scale for night driving (DCS-N), based on pilot-test results, were subjected to more thorough psychometric testing with a larger sample. The specific objectives were to examine the scale’s structural properties and test-retest reliability, as well as explore patterns of association between driving comfort, characteristics of older drivers, self-reported driving patterns (frequency, restriction, avoidance), perceived driving abilities, and problems (accidents, getting lost, etc). As in study 1, participants had to be aged 65 years or older, hold a valid driver’s license, and currently drive. Two of the facilities (1 in the south, 1 in the north) that took part in the initial focus groups (⬇3mo earlier) were invited to participate again; 23 people volunteered. To recruit 77 additional participants (100 current drivers), 12 new retirement complexes and seniors’ centers in adjacent communities in southern Ontario (both rural and urban) were approached. Similar to study 1, tool administration took place in a smallgroup format (8 –10 people a session). Participants were asked to self-complete, in order, the DCS-D and DCS-N, the background questionnaire (to serve as a distracter), and 3 rating forms: situational driving, frequency (now and compared with 10 years ago), and situational avoidance. Once everyone had completed the questionnaires, the group was asked for feedback on the tools. Arch Phys Med Rehabil Vol 89, April 2008

Finally, people interested in further study participation were asked to sign consent forms giving us permission to contact them in 1 to 2 weeks and 1 year later. To examine test-retest reliability, the DCS-D and DCS-N were re-administered to a subsample of 27 drivers 1 to 2 weeks after initial scale completion (average, 10.9⫾2.8d). Data Analysis Audiotapes of the focus groups were transcribed verbatim for coding and independent analysis by 2 researchers. Thematic analysis was used to address construct relevance (ie, importance and meaning of driving-related concerns), and content analysis was used to determine situational relevance (ie, driving situations that evoked these concerns) and identify a pool of potential items (step 1). Each group was analyzed separately before comparing groups, following a sequential process of open, axial, and selective coding.25 Quantitative analyses were performed using SPSSa for Windows, with significance level set at .05. Descriptive statistics included frequencies, means, standard deviations (SDs), and ranges, with related 95% confidence intervals (CIs). Comparative statistics consisted of chi-square testing for categoric variables and t tests or analysis of variance (ANOVA) for continuous variables. Total comfort scores (for day and night driving, respectively) were calculated by summing responses and then dividing by the number of items answered. Possible scores could range from 0% to 100%. Respondents had to answer at least 75% of the items to receive a total score on each scale. This procedure was used for both the preliminary version of the DCS (used in pilot testing) and the refined version used in study 2. Preliminary scale examination looked at the number of missing items, normality of distribution, item-total correlations, and internal consistency (Cronbach ␣). Rasch analysis was used to examine structural properties of the DCS-D and DCS-N. Using Winsteps,26,b raw scores were converted into interval-based, log-odds metric (or logits), and mean square standardized residuals were used to identify misfit items—for example, those that fail to contribute or produce erratic or variant responses. Mean square values from .60 to 1.4, with associated standardized z values of less than 2, are considered acceptable for rating scales.27 The person reliability index indicates how well the measure discriminates respondents. The item reliability index denotes the extent of item discrimination. Reliability estimates can range between 0 and 1, with higher values (closer to 1) preferred.24 Mapping is used to detect floor or ceiling effects, and probability curves indicate adequacy of response options.24 Too few options restrict one’s ability to communicate the experience, but too many options can cause confusion.24 Test–retest reliability of the DCS-D and DCS-N, respectively, was examined using the intraclass correlation coefficient, model 2,1 (ICC2,1), calculated using a 2-way ANOVA, considering administrators as a random effect.28 ICC values above .90 were considered high; .80 to .89, good; and .70 to .79, acceptable.29 To examine associations with driving comfort, composite scores were computed for ratings of situational frequency, restriction, and avoidance. Pearson correlation coefficients were used to examine patterns of association, because both the DCSs and the situational rating scores were normally distributed. Three other variables (perceived driving abilities, driving problems, nervousness) were also created from the background questionnaire. Because these scores were not normally distributed, associations with driving comfort were examined using the Spearman correlation coefficient (␳).

DAY AND NIGHT DRIVING COMFORT SCALES, Myers

RESULTS Study Population Characteristics of the various samples are shown in table 1. As previously noted, some people took part in more than 1 aspect of this multiphase project. Specifically, in study 1, 27 (55%) of the subjects in the rating group (step 2) had been in the focus groups (step 1), and all 16 people in the pilot sample (step 3) had taken part in either the focus or rating groups, but not both. Twenty-three percent of the validation sample (study 2), meanwhile, was composed of prior participants from study 1. Age, sex distribution, and driving patterns, however, were comparable in those who had previously participated (n⫽33) and those who had not (n⫽77). A total of 143 distinct subjects took part in these studies. Overall, participants ranged in age from 66 to 92 years, with drivers aged 80 years or over accounting for 45% to 54% of studies 1 and 2, respectively. The most commonly reported diagnosed health problems were arthritis (47%); high blood pressure or heart conditions (46%); cataracts, glaucoma, or macular degeneration (33%); back or foot problems (31%); hearing problems (31%); diabetes (25%); osteoporosis (22%); and neurologic problems (10%). Over a third had had cataract surgery (80% over a year ago), although only 3% of the total sample rated their eyesight as “worse than most.” Sixty percent of participants took prescribed medications (average, 3.3⫾2.1). Over 30% used a cane or walker. Conceptualization, Item Generation, and Verification Participants in all 4 focus groups unanimously agreed that confidence or comfort level was important to driving behavior. Both terms were used frequently, sometimes interchangeably. As 1 woman said, “I decided no more 401 [major highway], I knew I just didn’t feel comfortable or confident.” Although many people said that their confidence had diminished over the years (eg, “I am now more cautious as I am aware of my reactions slowing”), there were a few exceptions. For instance, as one woman explained, “I had to become more of a driver because my husband was sick and I had to make the trips myself, so that increased my confidence.” Participants spontaneously described situations that made them uncomfortable, nervous, or anxious (although the latter descriptors were used far less often than comfort or confidence). Some regional differences emerged. For instance, both groups from the north discussed “making left turns without traffic lights or stop signs,” whereas those from the south talked about “completing left hand turns on a yellow or red light.” Dealing with the erratic or poor driving practices of “other drivers” was mentioned by all 4 groups, citing instances of tailgating, not signaling in advance, and talking on cell phones. All groups were emphatic that the combination of driving at night in bad weather (eg, heavy rain) was particularly challenging. Based on the content analysis of the focus group data, 30 driving situations were identified for further examination. Situations were not included if mentioned by only 1 of the 4 groups (eg, being honked at, bicycles, wildlife). Each situation was worded verbatim. For instance, participants talked about getting “caught in a sudden (or unexpected) storm,” which usually entailed high winds, but could take a variety of forms depending on the season (eg, downpours, hail, or blowing snow). For ease of ratings, situations pertaining to weather, turning, highway driving, and other drivers were grouped together, as shown in the first column of table 2.

633

The rating sample (step 2) indicated that the most uncomfortable driving situations were fog (16%), caught in an unexpected or sudden storm (10%), glare or reflections (8%), and other drivers tailgating (7%). These items also had the lowest mean confidence ratings. The counselors, meanwhile, rated numerous situations as problematic for older drivers, including night, heavy rain, winter, unexpected storms, glare or reflections, unfamiliar areas, left turns, highways, changing lanes, high speeds, and multiple transport trucks. Counselors agreed that the list of 30 situations was inclusive; no additional items were suggested. The ensuing group discussions with the older drivers were very informative. Conceptually, older drivers believed that “comfort level” was more encompassing, capturing various aspects of self-confidence, including the ability to deal with or respond to situations created by other drivers. In addition, the groups were unanimous in their opinion that driving in the daytime was very different from driving at night. Several people in each group remarked that many situations (eg, bad weather, unfamiliar areas, seeing street, or exit signs) were more challenging at night. The situation itself might also differ, such as dealing with glare or reflection from the sun during the day versus from lights (of other cars or street and/or highway lamps) at night. Item 21 (at high speeds) generated a great deal of discussion, with comments such as: “Does this mean over the speed limit? Over 100 km per hour? Over 120 km per hour? Keeping up with the flow of traffic?” Because traffic flow pertained to several situations, participants suggested that this be clarified in the instructions. Other items they thought needed to be clarified (with illustrative remarks) were completing a left-hand turn on a yellow or red light (“Does this mean when already at midintersection?”), driving long distances (“How long?”), and parking in tight spots (“Does this mean pulling in or backing up? Parallel parking or in lots?”). Several people said they never parallel park. Similarly, item 29 (when you have not driven for a while) did not apply to most people and they found it difficult to imagine themselves in this situation. Over half the sample indicated they “never” drive in big cities, and some queried, “What’s big?” The ratings and feedback from step 2 led to refining various items, removing others (eg, 12, 21, 29, and 30), modifying the instructions, and creating a new DCS. Pilot Testing As shown in table 2, the preliminary DCS subjected to pilot-testing (step 3) included separate ratings of comfort level when driving in the daytime (25 items) versus at night (15 items). Item 6 (glare or reflection) was tailored to day versus night driving (ie, from the sun vs from lights). The first item on the night section was created to ascertain general level of comfort, because some people (in step 2) had said that they were uncomfortable driving at night even in ideal conditions. Other items (eg, heavy rain) were simply replicated for the night ratings, although not all situations were compared (day vs night) at this preliminary juncture. Preliminary scale analysis, as well as feedback from the pilot groups, supported the separation of day and night driving. Comfort level at night was significantly lower overall (t⫽4.09, P⬍.001) and for each of the paired items, with one exception (tailgating). Although each subscale showed good internal consistency (␣⬎.9), certain items were problematic. Fog had the lowest comfort ratings (both day and night) and the highest item-total correlation (.86), and it was rated as the situation most often avoided (87%), indicating this item should be removed. As some people remarked, “Thick fog would make anyone uncomfortArch Phys Med Rehabil Vol 89, April 2008

634

DAY AND NIGHT DRIVING COMFORT SCALES, Myers Table 2: Items and Rating Instructions for Initial Pool and Iterations of the Driving Comfort Scales Ratings of Initial Item Pool*

Preliminary DCS†

17-Item DCS-D† and 18-Item DCS-N‡

How confident are you driving? 1. At night? 2. In light rain? 3. In heavy rain? 4. In fog? 5. In rain at night? 6. In the first snow storm? 7. In winter conditions (snow, ice)? 8. There is glare or reflections? 9. Caught in an unexpected or sudden storm? 10. In unfamiliar routes (different areas), detours or sign changes? 11. In heavy traffic or rush hour? 12. Making a left hand turn at a turning lane in traffic? 13. Making a left hand turn with no lights or stop signs in traffic? 14. Completing a left turn on a yellow or red light? 15. On a 2-lane highway? 16. On a highway with 3 or more lanes? 17. Driving long distances? 18. Changes lanes in traffic? 19. Parking in a tight spot with large vehicles on either side? 20. Seeing street or exit signs with little warning? 21. At high speeds? 22. Passengers distract you or tell you how to drive? 23. Multiple transport trucks are around you? 24. Other drivers do not signal at all or with little warning? 25. Other drivers tailgate or drive too close behind you? 26. Other drivers pass on the right (or inside) lane of a highway? 27. Other drivers pass on a nonpassing lane or shoulder? 28. Other drivers appear distracted (loud music, talking)? 29. You have not driven for a while? 30. In a big city?

How comfortable are you driving in the daytime? 1. In light rain? 2. In heavy rain? 3. In fog? 4. In the first snow storm? 5. In winter conditions (snow, ice)? 6. When there is glare or reflection from the sun? 7. Caught in an unexpected or sudden storm? 8. In unfamiliar routes (different areas), detours or sign changes? 9. Making a left hand turn with no lights or stop signs? 10. Completing a left hand turn on a yellow or red light when already at mid-intersection? 11. On 2-lane highways? 12. On highways with 3 or more lanes? 13. On trips lasting more than 2 hours? 14. Changing lanes in traffic? 15. Pulling in or backing up from tight spots in parking lots with large vehicles on either side? 16. Seeing street or exit signs with little warning? 17. Keeping up with the flow of highway traffic when the flow is over the posted speed limit of 100km/h (60 miles/h)? 18. With passengers who distract you or tell you how to drive? 19. With multiple transport trucks around you? 20. Other drivers do not signal at all or with little warning? 21. Other drivers tailgate or drive too close behind you? 22. Other drivers pass on the right (or inside) lane of a highway? 23. Other drivers pass in a nonpassing lane or shoulder? 24. Other drivers appear distracted (loud music, talking)? 25. In heavy traffic or rush hour?

How comfortable are you driving in the daytime? 1. In light rain? 2. In heavy rain? 3. In winter conditions (snow, ice)? 4. When there is glare or reflection from the sun? 5. Caught in an unexpected or sudden storm? 6. In unfamiliar routes (different areas), detours or sign changes? 7. Making a left hand turn with no lights or stop signs? 8. Completing a left hand turn on a yellow or red light when already at mid-intersection? 9. Pulling in or backing up from tight spots in parking lots with large vehicles on either side? 10. Seeing street or exit signs with little warning? 11. On 2-lane highways? 12. Keeping up with the flow of highway traffic when the flow is over the posted speed limit of 100km/h (60 miles/h)? 13. With multiple transport trucks around you? 14. Merging with traffic and changing lanes? 15. Other drivers tailgate or drive too close behind you? 16. Other drivers pass on a nonpassing lane? 17. Other drivers do not signal or seem distracted?

How comfortable are you driving at night? 1. In good weather and traffic conditions? 2. In light rain? 3. In heavy rain? 4. In fog? 5. In winter conditions (snow, ice)? 6. When there is glare or reflection from lights? 7. In unfamiliar routes (different areas), detours or sign changes? 8. On 2-lane highways? 9. On highways with 3 or more lanes? 10. On trips lasting more than 2 hrs? 11. Seeing street or exit signs with little warning? 12. Keeping up with the flow of highway traffic when the flow is over the posted speed limit of 100km/h (60 miles/h)? 13. With multiple transport trucks around you? 14. Other drivers tailgate or drive too close behind you? 15. In heavy traffic?

How comfortable are you driving at night? 1. In good weather and traffic conditions? 2. In light rain? 3. In heavy rain? 4. In winter conditions (snow, ice)? 5. When there is glare or reflection from lights? 6. Caught in an unexpected or sudden storm? 7. In unfamiliar routes (different areas), detours or sign changes? 8. Making a left hand turn with no lights or stop signs? 9. Completing a left hand turn on a yellow or red light when already at mid-intersection? 10. Pulling in or backing up from tight spots in parking lots with large vehicles on either side? 11. Seeing street or exit signs with little warning? 12. On 2-lane highways? 13. Keeping up with the flow of highway traffic when the flow is over the posted speed limit of 100km/h (60 miles/h)? 14. With multiple transport trucks around you? 15. Merging with traffic and changing lanes? 16. Other drivers tailgate or drive too close behind you? 17. Other drivers pass on a nonpassing lane? 18. Other drivers do not signal or seem distracted?

*Instructions: Please rate your level of confidence when driving in each of the following situations by choosing a number from the scale below (0%, no confidence; 50%, moderate confidence; 100%, complete confidence). If you do not normally drive in the situation, try and imagine how confident you would be if you absolutely had to go somewhere 1 day and found yourself in the situation. † Instructions: Using the scale below, please rate your level of comfort by choosing a number from the scale below (0%–100%) and writing it in the blank beside each situation. If you do not normally drive in the situation, imagine how comfortable you would be if you absolutely had to go somewhere and found yourself in the situation. In your ratings, consider confidence in your own abilities and driving skills, as well as the situation itself (including other drivers). Assume normal traffic flow unless otherwise specified. ‡ Now we would like you to rate your level of comfort when driving in the following situations at night. Even if you do not normally drive at night, imagine that you were out in the afternoon, got delayed, and it was dark on your way back. In your ratings, consider confidence in your own abilities and driving skills, as well as the situation itself (including other drivers). Assume normal traffic flow unless otherwise specified.

Arch Phys Med Rehabil Vol 89, April 2008

DAY AND NIGHT DRIVING COMFORT SCALES, Myers

Fig 1. Probability curve for the 5-category collapsed DCS-D. Graphic representation of the probability of response for each of the 5 categories (0%, 25%, 50%, 75%, 100%). The y axis represents the probability of responding to 1 of these categories (0.0ⴚ1.0), and the x axis represents the person measure minus the item measure in logits.

able.” Comfort level “in the first snow storm” correlated (r⫽.78) with “caught in an unexpected or sudden storm.” Whether snow, heavy rain, or hail, they believed that the main issue was whether the storm came up unexpectedly. Driving on highways with 3 or more lanes highly correlated with trips lasting more than 2 hours (r⫽.94) and with keeping up with highway traffic flow when over the posted speed (r⫽.91). To reduce redundancy, the latter item (which subjects believed was the clearest or most specific) was retained. The item “heavy traffic or rush hour” presented substantial confusion (eg, “Does this refer to in town or on the highway?”), especially because traffic flow was addressed in the instructions and in item 17. Regarding changing lanes in traffic, some people asked, “Where? In town or on the highway?” Others said, “Does this include merging?” For clarification, this item was reworded as merging with traffic and changing lanes on the highway. Two of the items concerning other drivers (appear distracted and do not signal) also highly correlated (r⫽.92) and subsequently were combined into 1 item. Items 22 and 23 were also considered very similar (especially because the shoulder is a “nonpassing” lane) and combined. Finally, item 18 was removed because many people said they never or rarely drive with passengers, and others argued that passengers could also be helpful (as opposed to distracting). Some of the items removed from the comfort scales were subsequently incorporated into the various driving rating forms (frequency, avoidance, perceived abilities). Psychometric Examination and Scale Refinement The revised 17-item DCS-D and 18-item DCS-N were then administered to a larger sample of 100 drivers. Content was identical, other than the item on glare/reflection (tailored for day vs night driving) and item 1 on the DCS-N (see table 2); both supported by the pilot sample. Some people who did not normally drive at night noted that they sometimes “got stuck” (ie, went out in the afternoon, got delayed, and had to drive home in the dark); thus, this caveat was incorporated into the DCS-N rating instructions. Participants took, on average, about 6 minutes to complete both the DCS-D and the DCS-N (range, 5–11min). Only 9 of

635

the 100 subjects failed to answer all the items, usually only 1 or 2 items. The maximum number of items missed across the 2 scales was 7 (by 2 people). All subjects met the requisite 75% completion for calculation of total DCS-D and DCS-N scores. Good internal consistency was found for both the DCS-D (␣⫽.94) and the DCS-N (␣⫽.98) and did not change appreciably as items were sequentially deleted. Item-total correlations ranged from r equal to .49 to .80 on the DCS-D and from .76 to .92 on the DCS-N, indicating some potential item redundancy. Although no problems with scale completion were reported, Rasch probability curves showed that respondents did not reliably distinguish among the 11 response options (eg, 0 vs 10). Collapsing the responses into 5 categories— 0 (0 and 10), 1 (20 and 30), 2 (40, 50, and 60), 3 (70 and 80), and 4 (90 and 100)—as illustrated in figures 1 and 2 (for each scale) produced a more logical sequence (from low to high level of comfort increasing along the x axis), distinct peaks, and adjacent categories crossing at or near the 50% probability line.24 Accordingly, ratings were collapsed and recoded as 0%, 25%, 50%, 75%, and 100%. Stepwise Rasch analysis also identified a number of misfit items, for example, those falling outside acceptable infit parameters.27 On the DCS-D, the 4 problematic items were number 4, glare or reflection from the sun (standardized z⫽⫺2.7); number 6, unfamiliar routes (standardized z⫽⫺2.3); number 8, left turn on yellow or red light (mean square, 1.44; standardized z⫽2.7); and number 14, merging or changing lanes on the highway (standardized z⫽⫺2.6). On the DCS-N, meanwhile, the 2 problematic items were number 6, unexpected or sudden storm (mean square, .56; standardized z⫽⫺3.6); and number 9, left turn on yellow or red light (standardized z⫽2.2). The left turn item (on yellow or red light) highly correlated on both scales (DCS-D, .72; DCS-N, .89) with the other item on left turns (with no lights or stop signs) and fell on the same line of the person/item map. The item “caught in an unexpected or sudden storm” emerged as the most challenging situation on both scales (ie, associated with the least comfort); however, on the DCS-N, this item did not discriminate well (ie, made most people very uncomfortable). A few other items produced erratic responses only on the day scale (eg, glare). It is possible

Fig 2. Probability curve for the 5-category collapsed DCS-N. Graphic representation of the probability of response for each of the 5 categories (0%, 25%, 50%, 75%, 100%). The y axis represents the probability of responding to 1 of these categories (0.0ⴚ1.0), and the x axis represents the person measure minus the item measure in logits.

Arch Phys Med Rehabil Vol 89, April 2008

636

DAY AND NIGHT DRIVING COMFORT SCALES, Myers

Fig 3. Person and item map for the DCS-D. Person ability is shown on the left and item difficulty on the right (measured in logits or equal intervals). Abbreviations: M, mean; S, 1 SD; T, 2 SDs. Legend is 1 respondent (located as having a 50% probability of endorsing the item).

that glare from the sun is more variable (based on time of the day or cloud cover) than glare from lights at night. People also use sunglasses, visors, or both, to deal with driving in the sun. In any case, when the problematic items (4 on the DCS-D, 2 on the DCS-N) were removed, all mean square and standardized z values fell within acceptable limits. The resulting 5-category, 13-item DCS-D and 16-item DCS-N were unidimensional and showed good person (.89, .96, respectively) and item reliabilities (.98, .97, respectively). The graphic relationships among person ability and item difficulty are illustrated in figures 3 and 4. Viewed hierarchically (in order of progressive difficulty or challenge), item number 4 (caught in an expected or sudden storm) on the DCS-D was associated with the least comfort, and item number 1 (light rain) was associated with the most comfort. On the DCS-N, meanwhile, item number 16 (others don’t signal or distracted) was associated with the least comfort, and item number 1 (good weather and traffic conditions) was associated with the most comfort. Although some items fell on the same line (which could justify further removal), infit statistics were acceptable. A further justification for retention of these items is that, conceptually, different situations (eg, heavy rain and others tailgating) could be expected to evoke similar levels of discomfort. Scores on both the DCS-D and DCS-N were normally distributed, although more people had high values on the DCS-D, Arch Phys Med Rehabil Vol 89, April 2008

as would be expected. The Cronbach ␣ indicated good internal consistency for both the 13-item DCS-D (.92) and the 16-item DCS-N (.97). The collapsed 5-category, 13-item and 16-item scales were used in all subsequent analyses. Total scores were computed for respondents who answered at least 75% of the items (9 of the 13 on the DCS-D, 12 of the 16 on the DCS-N). This criterion was met for all 100 subjects on the DCS-D and for 99 subjects on the DCS-N (1 person who answered only 11 of the items was excluded from further analysis on this measure). Total scores on the DCS-D and DCS-N could potentially range from 0% to 100% (theoretic mean, 50), with higher scores indicating higher levels of comfort. Overall, the sample scored significantly higher (t⫽7.91, P⬍.001) on the DCS-D (59.7⫾17.9; 95% CI, 56.1– 63.2) than the DCS-N (47.9⫾23.8; 95% CI, 43.2–52.7). Scores on the DCS-D and DCS-N significantly correlated (r⫽.79, P⬍.01). On the DCS-D, the item with the lowest comfort rating was being caught in a sudden or unexpected storm (46.2⫾25), and the item with the highest comfort rating was driving in light rain (82.3⫾19.2). On the DCS-N, the item with the lowest comfort was others not signaling or appearing distracted (36.5⫾27.3), and the item with the highest comfort was driving in good weather and traffic conditions (71.2⫾25). People who scored above the sample mean on this item also scored significantly higher on the DCS-D (t⫽⫺4.76, P⬍.001).

DAY AND NIGHT DRIVING COMFORT SCALES, Myers

637

Fig 4. Person and item map for the DCS-N. Person ability is shown on the left and item difficulty on the right (measured in logits or equal intervals). Abbreviations: M, mean; S, 1 SD; T, 2 SD. Legend is 1 respondent (located as having a 50% probability of endorsing the item).

In general, the items pertaining to other drivers (tailgating, passing, not signaling) received low comfort ratings (DCS-D means, 48 – 49; DCS-N means, 36 –39). Average comfort level was substantially lower driving in heavy as opposed to light rain, both during the day (51.5 vs 82.3) and at night (40.7 vs 61.1). Test-Retest Reliability Characteristics of the subgroup who completed the DCSs a second time are shown in table 1. The reliability analysis of the DCS-D and DCS-N consisted of 27 and 25 people, respectively (because 2 people answered ⬍75% of the DCS-N items at the second administration). At both time points, average driving comfort scores were higher for the DCS-D (time 1, 69.0⫾15.7; time 2, 64.8⫾20.7) than for the DCS-N (time 1, 62.0⫾23.2; time 2, 58.8⫾24.1). Scores on the DCS-D were fairly consistent (ICC2,1⫽.70; 95% CI, .44 –.85) over 1 to 2 weeks, and the

reliability of the DCS-N was good (ICC2,1⫽.88; 95% CI, .75–.95). Two people had an unexplained difference in DCS-D scores (⬎25 scale points at time 2), and another person had such discrepancies on both the DCS-D and DCS-N. Associations With Driving Comfort Scores Scores on the DCS-D and DCS-N were then related to sample characteristics and reported driving patterns. Age was inversely related to driving comfort scores but significant only for the DCS-N scores. Men had higher comfort scores (table 3) on both scales; however, this difference was only significant for the DCS-N. Both DCS-D and DCS-N scores were significantly lower (P⬍.01) for those who were the only driver in the household. No associations emerged with respect to education, number of health problems, or medications. However, those who had had cataract surgery had lower DCS-N scores (P⫽.06). Mean comfort scores (both day and night) were Arch Phys Med Rehabil Vol 89, April 2008

638

DAY AND NIGHT DRIVING COMFORT SCALES, Myers Table 3: Associations With Driving Comfort Scores Variables

Age No. of days driven in past week Situational frequency (14 items) Restriction (14 items) Avoidance (20 items) Perceived abilities (6 items) Nervousness (5 items) Drives at night Occasionally/often (n⫽64) Never/rarely (n⫽34) Sex Men (n⫽39) Women (n⫽61)

DCS-D Score

⫺.17 .36* .55* ⫺.18 ⫺.56* .34* ⫺.25‡

DCS-N Score

⫺.22‡ .37* .53* ⫺.22‡ ⫺.49* .43* ⫺.27†

64.5⫾16.8 (60.3–68.7)* 52.0⫾17.0 (46.1–58.0)

57.1⫾19.3 (52.2–61.9)* 31.9⫾23.3 (23.7–40.0)

63.7⫾15.7 (58.7–68.8) 57.1⫾18.9 (52.3–61.9)

56.2⫾19.6 (49.8–62.6)† 42.6⫾24.9 (36.2–49.0)

NOTE: Values are Pearson correlation coefficients (age, number of days driven, situational frequency, restriction, avoidance), Spearman ␳ correlation coefficients (perceived abilities, nervousness), or mean ⫾ SD (95% CI). Possible range: situational frequency, 0 – 42; restriction, 0 –14; avoidance, 0 –20; perceived abilities, 0 – 6; and nervousness, 0 –5. *P⬍.001. † P⬍.01. ‡ P⬍.05.

highest for those who rated their eyesight as “better than most” and lowest for those who rated their eyesight as “worse than most” (although only 3 people put themselves in this category). With respect to mobility, those who indicated that they were unable to walk a quarter of a mile had significantly poorer DCS-D and DCS-N scores (P⬍.05). Comfort scores correlated with ratings of nervousness and driving abilities in the expected direction. With respect to specific abilities, respondents were most likely to admit they were “worse than others” with respect to seeing signs (22%), followed by parking and shoulder checking (9%). When asked about their “overall driving ability,” however, only 1% thought they were worse than other drivers (28% said better, 71% said the same). Drivers acknowledged relatively few driving problems over the past year: 21% admitted to near misses or getting lost, 14% to backing into things, and 12% to collisions involving another vehicle. Those who admitted to accidents involving another vehicle had lower comfort scores but not significantly. With respect to driving behavior, the number of days driven in the past week was significantly related to higher comfort scores on both the DCS-D and DCS-N. Less than half the sample was able to estimate the number of kilometers driven in the past year. With respect to the 3 driving ratings—situational driving frequency, driving restriction (vs 10y ago), and avoidance, internal consistency (␣) ranged from .90 to .93. Both comfort scores were strongly associated with situational frequency and avoidance, in the expected direction (see table 3). Restriction correlated with the DCS-N, but not the DCS-D, score. Respondents who said they (or reportedly) drove at night occasionally or often (n⫽64) were compared with those who rarely or never drove at night (n⫽34). Mean comfort scores (see table 3) were significantly higher (both day and night) for the former group. Further examination of people who said they rarely or never drive at night indicated that this group was significantly older (P⬍.001), more likely to have had cataract surgery (P⬍.05), and more likely to be women (71% vs 29%), although the sex difference was not significant. In addition to lower driving comfort scores, this group had higher avoidance (P⬍.001), lower frequency (P⬍.001), and poorer perceived driving abilities scores (P⬍.01). Arch Phys Med Rehabil Vol 89, April 2008

DISCUSSION The new DCSs differ in several important respects from other tools purportedly measuring the same construct. From their perspective, older drivers thought that “comfort level” best captured this phenomenon. A recent qualitative study by Rudman at al6 supports this conceptualization: both drivers and ex-drivers highlighted their personal comfort level (which included confidence) as a key factor in regulating their driving behavior. Consistent with Bandura’s framework,19 collisions or near-accidents, and feedback from others intensified discomfort.6 Night driving also emerged as a particular concern in their focus groups.6 Our respondents were adamant that many driving situations were more challenging at night, leading to the creation of separate day and night comfort scales. Comparatively, other tools have included only a single item pertaining to night driving.10,14,16,21 The tool of Parker et al11 did not include any items on night driving, which may explain why their sample (555 drivers; mean age, 69y) reported such low levels of nervousness and high confidence. The content of the DCSs also differs appreciably. Our drivers thought that some items (found on other tools, eg, “high speeds”) were ambiguous without further clarification. Conversely, other situations generated by our focus groups (and verified by older drivers and counselors) do not appear on other tools (eg, “caught in a sudden storm,” “seeing street or exit signs with little warning,” “glare or reflections”). The DCSs have a high degree of context (traffic flow) and situation specificity (eg, actual speed), both of which are important in measuring self-efficacy.19 Similar to others,14,21 we began with an 11-point confidence rating format. Rasch analysis showed, however, that respondents did not reliably distinguish between all these options and identified items failing to discriminate or producing erratic responses. Unlike internal consistency, Rasch addresses additivity, the basic tenet of measurement.24 In this case, the metric (calibrated on a true interval scale) represents the level of driving comfort expressed by the respondent and the amount of discomfort each situation evokes. The resulting 5-category, 13-item DCS-D and 16-item DCS-N showed good person and item reliabilities, unidimensionality, and hierarchiality. Test-retest reliability was very good for the DCS-N and acceptable for the DCS-D.

DAY AND NIGHT DRIVING COMFORT SCALES, Myers

Hierarchiality or the progressive challenge of scale items is important to establish.22,23 According to Bandura’s theory,19 the most challenging situations are the most likely to be avoided. According to Rudman et al’s6 self-regulation model, voluntary driving cessation is likely to occur when one reaches a “personally unacceptable level of comfort.” Comfort regarding night driving might be expected to progressively decline to the point where people are uncomfortable even in good weather and traffic conditions (item 1 on the DCS-N), possibly before daytime driving comfort is appreciably affected. Consistent with this supposition, self-reported driving restriction was more strongly related to DCS-N scores. Prospective studies are needed to examine this hypothesis and identity whether there is a critical level of discomfort at which voluntary cessation tends to occur. Barriers to self-restriction,15 particularly those affecting driving persistence (despite high levels of discomfort), must also be investigated. Our findings provide additional evidence that driving comfort level (confidence) is related to self-reported driving frequency,11,14 situational avoidance,20 and perceived driving abilities.11,14 Consistent with other studies, few people considered themselves worse than other drivers (when using a single item on overall ability)14-16; however, they were more likely to acknowledge specific problems (eg, seeing when raining or dark).11-13,16 Parker et al11 showed that driving confidence affects self-rated ability independent of personality traits (extroversion, neuroticism) and social desirability (lie scale). There is little consensus on how best to quantify actual driving abilities. Various studies have examined specific abilities such as vision,12,13,15,16 adverse driving events or lapses,11,14 and/or on-road performance.14,20,21 Driving involves multiple abilities, and drivers may adjust to compensate for deficits in certain areas. Crashes are also complex and can be affected by factors other than driver error.30 Drivers may not think they were at fault or may be reluctant to report adverse events for fear of losing their license. On-road evaluations (typically conducted during the day and in good weather), meanwhile, may not detect decrements.4,20 To identify discrepancies between actual and perceived driving-related abilities, the measures themselves must correspond. Ultimately, it is important to establish the influence of perceptions on driving behavior itself. Effective driving self-regulation depends on several factors: accurate appraisals of one’s driving abilities and skills (including awareness and acknowledgement of functional declines), knowledge of high-risk driving situations, and appropriate changes to one’s driving in an attempt to avoid accidents. Increasingly, efforts are being made to educate older drivers through voluntary and mandatory programs, self-assessment tools,4 and targeted interventions.30 Increased awareness of driving problems or limitations, together with lack of confidence in the situation, should prompt people to restrict or curtail their own driving.14 Decreased exposure alone, however, may not reduce risk. In fact, less practice could led to a deterioration in driving skills.6 Fortunately, skills can be improved through remediation and training.4 That being said, driver improvement programs can also backfire by making some people overconfident.30 Drivers who are overconfident in relation to their actual driving abilities or skills may put themselves and others at increased risk, whereas those who are underconfident may prematurely restrict their driving or relinquish their licenses. A critical role for rehabilitation specialists who work with older drivers is to assist clients in developing realistic appraisals to foster concordance (or a match) between their perceived and actual capabilities.

639

Study Limitations Study limitations must be acknowledged. The convenience sample was restricted to English-speaking older drivers, some of whom took part in multiple aspects of the project. People who volunteer for such studies may be more likely to consider themselves good drivers. The studies were cross-sectional (except for the 2-wk retest), driving abilities were not objectively measured, and indicators of driving behavior were limited to self-report. CONCLUSIONS This article describes the development and psychometric properties of a new tool to assess confidence, conceptualized as comfort level, in older drivers. The DCSs were inductively developed with older drivers (from both urban and rural areas), and the content is representative of progressively challenging driving situations in the daytime and at night, respectively. Together, the scales take less than 10 minutes for self-completion. Studies are underway to examine relationships between perceptions (comfort level and abilities), objectively measured driving-related abilities and driving behaviors (using recent technologies31). Prospective follow-ups are being used to investigate the predictive properties of the DCSs (ie, whether lower comfort levels lead to decreased exposure and possibly cessation) and responsiveness to change as a result of various types of interventions. Translation and cultural adaptation of the DCSs are required to examine driving comfort in non–English-speaking older adults. Designed for North Americans, the tool does not capture location-specific situations (such as roundabouts in the United Kingdom11 or Australia21), and the item on winter conditions (ice, snow) is not applicable to southern climates. Only 75% of the items need to be completed to compute total scores for the DCS-D and DCS-N. Scale items, however, should not be modified without further psychometric testing.22,23 Studies with new samples are required to replicate and extend the present findings, as well as develop benchmarks or norms for healthy drivers and various clinical populations such as those with stroke, Parkinson’s, visual disorders, or early dementia. In the absence of a criterion standard (validity), subjective rating scales must amass support through an ongoing process of hypothesis testing regarding the underlying construct.22,23 Acknowledgment: We acknowledge the assistance of Nancy Pearce. References 1. Burkhardt JE, McGavock AT. Tomorrow’s older drivers: who? how many? what impacts? Transport Res Rec 1999;1693:62-70. 2. Lyman S, Ferguson SA, Braver ER, Williams AF. Older driver involvements in police reported crashes and fatal crashes: trends and projections. Inj Prev 2002;8:116-20. 3. Hopkins RW, Kilik L, Day D, Rows C, Tseng H. Driving and dementia in Ontario: a quantitative assessment of the problem. Can J Psychiatry 2004;49:424-38. 4. Eby DW, Molnar LJ, Shope JT, Vivoda JM, Fordyce TA. Improving older driver knowledge and self-awareness through selfassessment: the driving decisions workbook. J Safety Res 2003; 34:371-81. 5. Benekohal RF, Michaels RM, Shim E, Resende PT. Effect of aging on older drivers’ travel characteristics. Transport Res Rec 1994;1438:91-8. Arch Phys Med Rehabil Vol 89, April 2008

640

DAY AND NIGHT DRIVING COMFORT SCALES, Myers

6. Rudman DL, Friedland J, Chipman M, Sciortino P. Holding on and letting go: the perspectives of preseniors and seniors on driving in later life. Can J Aging 2006;25:65-76. 7. Johnson JE. Rural elders and the decision to stop driving. J Community Health Nurs 1995;12:131-8. 8. Marottoli RA, Mendes de Leon CF, Glass TA, et al. Driving cessation and increased depressive symptoms: prospective evidence from the New Haven EPESE. Established populations for epidemiologic studies of the elderly. J Am Geriatr Soc 1997;45: 202-6. 9. Marottoli RA, de Leon CF, Glass TA, Williams CS, Cooney LM Jr, Berkman LF. Consequences of driving cessation: decreased out-of-home activity levels. J Gerontol B Psychol Sci Soc Sci 2000;55:S334-40. 10. Hakamies-Blomqvist L, Wahlstrom B. Why do older drivers give up driving? Accid Anal Prev 1998;30:305-12. 11. Parker D, MacDonald L, Sutcliffe P, Rabbitt P. Confidence and the older driver. Ageing Soc 2001;21:169-82. 12. Satariano WA, MacLeod KE, Cohn TE, Ragland DR. Problems with vision associated with limitations or avoidance of driving in older populations. J Gerontol B Psychol Sci Soc Sci 2004;59: S281-6. 13. Holland CA, Rabbitt PM. People’s awareness of their age-related sensory and cognitive deficits and the implications for road safety. Appl Cogn Psychol 1992;6:217-31. 14. Marottoli RA, Richardson ED. Confidence in, and self-rating of, driving ability among older drivers. Accid Anal Prev 1998;30: 331-6. 15. Stalvey BT, Owsley C. Self-perceptions and current practices of high-risk older drivers: implications for driver safety interventions. J Health Psychol 2000;5:441-56. 16. Owsley C, Stalvey B, Wells J, Sloane ME. Older drivers and cataract: driving habits and crash risk. J Gerontol A Biol Sci Med Sci 1999;54:M203-11. 17. Myers AM, Fletcher PC, Myers AH, Sherk W. Discriminative and evaluation properties of the Activities-specific Balance Confidence (ABC) Scale. J Gerontol A Biol Sci Med Sci 1998;53: M287-94. 18. Jorstad EC, Hauer K, Becker C, Lamb SE. Measuring the psychological outcomes of falling: a systematic review. J Am Geriatr Soc 2005;53:501-10.

Arch Phys Med Rehabil Vol 89, April 2008

19. Bandura A. Social foundations of thought and action: a social cognitive theory. Englewood Cliffs: Prentice-Hall;1986. 20. Baldock MR, Mathias JL, McLean AJ, Berndt A. Self-regulation and its relationship to driving ability among older adults. Accid Anal Prev 2006;38:1038-45. 21. George S, Clark M, Crotty M. Development of the Adelaide driving self-efficacy scale. Clin Rehabil 2007;21:56-61. 22. Streiner DL, Norman GR. Health measurement scales: a practical guide to their development and use. New York: Oxford Univ Pr; 1989. 23. Williams JI, Naylor CD. How should health status measures be assessed? Cautionary notes on procrustean frameworks. J Clin Epidemiol 1992;45:1347-51. 24. Bond T, Fox C. Applying the Rasch model: fundamental measurement in the human sciences. Mahwah: Lawrence Erlbaum Associates; 2001. 25. Strauss A, Corbin J. Basics of qualitative research: grounded theory procedures and techniques. Newbury Park: Sage; 1990. 26. Linacre JM, Wright BD. A user’s guide to Winsteps, Bigsteps, Ministeps: Rasch-model computer programs. Chicago: Mesa Pr; 1999. 27. Wright B, Linacre M. Reasonable item mean-square fit values. Rasch Meas Trans 1994;8:370. 28. Shrout PE, Fleiss JL. Intraclass correlations: uses in assessing rater ability. Psychol Bull 1979;86:420-8. 29. Kachingwe AF, Phillips BJ. Inter- and intrarater reliability of a back range of motion instrument. Arch Phys Med Rehabil 2005; 86:2347-53. 30. Owsley C, McGwin G, Phillips JM, McNeal SF, Stalvey BT. Impact of an educational program on the safety of high-risk, visually impaired, older drivers. Am J Prev Med 2004;26: 222-9. 31. Huebner KD, Porter MM, Marshall SC. Validation of an electronic device for measuring driving exposure. Traffic Inj Prev 2006;7:76-80. Suppliers a. Version 13.0; SPSS Inc, 233 S Wacker Dr, 11th Fl, Chicago, IL 60606. b. Winsteps, PO Box 811322, Chicago, IL 60681-1322.