BICYCLE ACCIDENTS AND USAGE AMONG YOUNG ADULTS:
A PRELIMINARY STUDY
 

Stuart A. Schupack
and
Gerald J. Driessen
Report No. 051-7
Research Department
National Safety Council
425 North Michigan Avenue
Chicago, Illinois 60611

July 1976

home                

Acknowledgements

This report came about through the converging and constructive energies of many persons and organizations. Ken Licht and Jack Green, Sr. of the School and College Department of the National Safety Council were especially helpful in making contacts for data collection at the various colleges and universities. Harold Heldreth of the Youth Department was highly cooperative and effective in acting as liaison with various organizations. Tom Chlapecka did much of the exploratory work on earlier attempts to collect data, organized the final survey, and supervised data collection from the cooperating schools. The local school personnel, many of them safety supervisors, were a necessary and vital link in the data chain. The students who provided the self-reports that are the basis of the results were, from one point of view, the most helpful of all. Thanks are also due to Joan Planek for editing the manuscript and to Norma Melnick for typing it and organizing the myriad of details necessary to complete the report. Finally, we are grateful for the generous (and patient) support of the Schwinn Bicycle Company whose grant served as partial support for this project. The corporate allocation of funds for research to determine the causes of accidents and find effective countermeasures is an enlightened, constructive decision. Such action deserves the explicit acknowledgment and clear expression of gratitude that these words are intended to convey.

Note: Footnotes have been incorporated into the text lines, between curly brackets { ... }

Note: The Horizontal Lines indicate a page break in the original. The page number is shown at the bottom center of the page. Some short pages have been combined with the previous page. Therefore, some page numbers are not used.

i

TABLE OF CONTENTS

LIST OF TABLES

ii

LIST OF FIGURES AND APPENDICES

Appendices

iii

PREFACE

This document represents the fourth major research report on bicycle safety issued by the Research Department of the National Safety Council in the last 9 years. {Earlier studies were by Vilardo, Nicol, and Heldreth (1968), Vilardo and Andersen (1969), and Chlapecka, Schupack, Planek, Klecka, and Driessen (1975).} The original initiative for this work began with the energy, enthusiasm, and concern of the Youth Conference of the National Safety Council for what they saw as a major accident problem affecting young persons, especially children. Their perception of the seriousness of bicycle accidents and the pressing need for effective countermeasures have been recently reaffirmed by the high placement of bicycles on the hazardous product list of the Consumer Product Safety Commission. While there is some controversy about whether bicycles "deserve" their current position at the top of the list (Flora, et al, 1975), it is likely that bicycles will continue to be listed as a highly hazardous product because of their widespread usage and the inherent risk of falls and collisions associated with regular use.

The rapid increase in the use of bicycles in the past decade has been well documented. As one would expect, this increased usage has led to a parallel rise in the number of bicycle accidents. The safety community has been especially concerned about children's bicycle accidents, and consequently much research and program development activity has dealt with this problem. Based on some of the aforementioned research, the National Safety Council developed the All About Bikes program as a specific countermeasure to help reduce bicycle accidents among elementary school aged children. This program is currently being used by school systems throughout the country. Adult bicycle driving has undergone an even more dramatic increase in popularity, but there has been little research effort spent in examining the usage and accidents of the adult bicyclist. This study is a preliminary step in remedying this oversight.

2

INTRODUCTION

The primary obstacle to be overcome in examining adult bicycling is the gathering of a suitable sample from which to collect reliable data. Neither bicycles nor bicycle drivers are licensed by official agencies. Most bicycle accidents are not entered in any regular, official reporting system except for the few injury-related motor-vehicle involvements. Traffic rules are poorly enforced for bicycle drivers and violations are rarely, if ever, entered on official records. Fuel consumption figures used to estimate automobile mileage are, of course, meaningless for bicyclists. Thus, several of the important sources of information concerning automobile usage and accidents are not available when studying bicycles. The absence of these sources makes it necessary to contact bicyclists directly in order to collect the required data.

The possible methodologies for contacting bicyclists were restricted by both theoretical and financial limitations. First, the study was intentionally restricted to young adults, aged 16 to 30. Second, we wished to gather a broad base of general information rather than detailed answers to a few specific questions, since the scarcity of previous research made it difficult to form any specific, testable hypotheses. Third, the level of funding placed limitations on the size of the sample and on the time that could be allowed for data collection.

Two initial methods of collecting data were tried, but proved unsatisfactory. One involved gathering data through cooperating bicycle clubs, and the other consisted of direct interviewing of Chicago area bicyclists. The primary problem with each of these methods was the inability to generate a sample of sufficient size for reliable analysis. While a small amount of data was collected (some 200 cases), our judgment was that any analysis would be foolhardy. Consequently, none of those data are presented in this report.

3

The third data gathering method had more favorable results. By circulating questionnaires on college campuses throughout the country under the auspices of a professor or administrator involved in campus bicycling activities, information on over 1,000 subjects was collected. The quality and quantity of these data were sufficient to allow some meaningful analysis, which is presented in this study.

The preliminary nature of this study severely limits its scope. There are two areas in which these limitations are most clearly seen. First, the group of young adult subjects was one of convenience rather than a scientifically selected representative sample, making it difficult, if not impossible, to generalize the results to a larger population. Second, the data were collected without reference to any specific hypothesis; rather, variables were included that seemed likely to shed some light on adult bicycle usage or accidents on either a common sense basis or on the basis of the earlier research on children's bicycling. While this data collection method yields a large amount of univariate descriptive data, which was previously unavailable, it also makes any multivariate analyses, even simple cross-tabulations, a haphazard process. Considering these two limitations, any conclusions drawn from this study must be viewed cautiously. The conclusions should be used as aids for further research rather than as facts with immediate program implications.

Despite these limitations, the reasons for conducting and reporting this research are clear. The methodology involved requires this type of pilot test before refinements can be added and before greater amounts of money are spent for more extensive studies. The basic data concerning certain parameters of adult bicycling usage and accidents have not been adequately documents in current survey literature, and these results, while limited, can serve as starting points for further research. It is hoped that this study will, indeed, stimulate further research in the area of adult bicycling.

5

METHOD

Survey Design

Questionnaires were sent to a total of 55 colleges and universities representing each of the nine standard U. S. Geographical areas. Schools were selected on the basis of two criteria--geographic location and the availability of a professor or administrator known to be involved in cam-pus bicycling activities. This contact on each campus was requested to oversee the distribution and collection of the survey forms and to return the completed forms to the National Safety Council.

Survey Instrument

The questionnaire (Appendix A) consisted of over 100 items divided into five sections: You, Your Bicycle, Most Recent Bike Use, Bicycle Accidents, and Local Bicycling Conditions. The items were selected partly from an NSC form used to investigate children's bicycle accidents (Chlapecka, Schupack, Planek, Klecka, and Driessen, 1975) and included additional variables to help focus on the experience of young adults.

Questions concerning the bicyclists included sex, age, bicycle driving experience, and several measures of exposure. The exposure measures were, months of regular bicycle driving, bicycle mileage driving time, and amount of driving on various roadway types. Information on the bicycle included length of time owned, frame size, wheel size, gearing, brake type, handle-bar configuration, and maintenance history. Questions concerning the respondent's most recent bicycle use were asked in order to obtain accurate details of the characteristics of a typical bicycle outing. These details included day of the week, time of the day, weather, driving surface, traffic density, type of roadway, and purpose of drive.

6

Accident details were requested for the most serious accident within the past 12 months, or, if there were none that recent, within the past 5 years. Some of this information concerning the accident paralleled the topics related to most recent bicycle use. Additional accident information was collected on activity and speed before the accident, familiarity with accident location, type of accident, and several items concerning injury severity and treatment. The final section of the questionnaire was concerned with the adequacy of local bicycle routes or lanes.

Data Collection

Standard data collection procedures were given to the campus contacts. The procedures are described in a letter included as Appendix B. In general, it was hoped that the questionnaire would be given to entire class-rooms in order to reduce the bias of having only highly interested bicyclists providing information. While the level of compliance with these instructions is unknown, the general cooperation of the campus contacts and internal evidence to be presented in the results section suggest that this source of bias was indeed reduced. On the other hand, clearly no claim can be made of having a representative sample from any of the campuses that returned forms. As stated before, however, due to the preliminary nature of the study, this is not a totally debilitating flaw.

Questionnaires were distributed in October, 1972. This meant that the experience of the high bicycling summer months would still be fresh in the minds of the respondents. It also meant that much of the reported accident experience would be from home rather than from the college campus--a fact that increases the geographic spread by some unknown amount and makes any generalizations even more problematic.

7

Data Analysis

The data analyses had three planned phases. The first was to present and discuss the basic information in the form of simple frequency distributions and proportions and involved the first estimates of several previously unknown parameters of young adult bicycling.

The second phase of the analysis involved several calculations and comparisons. Variables concerning months of driving, driving distances, and accidents were combined to calculate accident rates on a "per person" basis and, separately, on a mileage basis. These rates were further examined by sex and type of bicycle. Then the relationship of certain basic explanatory variables (sex, type of bicycle, exposure) to important accident and usage details was explored. This was done with a series of two-way cross-tabulations. The basic statistical tool was chi-square with p < .05 as the level of significance. The same basic method was used to explore differences between accident and no-accident groups.

The third phase of the analysis compared the results of these adult data with previous results from a study of children's bicycle accidents and usage. The similarities and differences between the two groups help de-fine needed areas of further research and provide some general direction for young adult bicycle programs and countermeasures.

Data editing was a preliminary step to all of these analyses. Three criteria were used in editing: completeness of information, current age of the respondent and, if accident details were reported, age of the respondent at the time of the accident. Questionnaires with large amounts of missing data were not keypunched. After keypunching, computer editing removed those cases where the respondents were not young adults, defined as aged 16 to 30. Finally, accidents to those respondents in the youngest remaining age group were removed if they occurred more than 1 year ago. This editing was done to insure that all accident details were indeed about accidents that happened to young adults and not children 11 to 15 years of age. The exposure and usage data of these cases, being based on the most recent bicycle use, were kept and included in all analyses, since all of these people were young adults when providing this information. Analyses were done primarily with the National Safety Council's IBM 1130 computer with some details completed using a Wang 520 programmable calculator.

9

RESULTS

Response Rates

Survey forms were sent to 55 colleges and universities. Returns were received from 27 schools for a 49% institutional return. Forms were sent in sets of 100 per school. One school received 2 sets, however, so a total of 5,600 forms were distributed. Returns totaled 1,370 or 24% of 5,600. Appendix C contains the names of all cooperating schools and the number of forms they returned. Editing for completeness and age led to a removal of 74 and 64 forms, respectively. The final sample contained 1,232 bicyclists.

Univariate Descriptions

Responses to all questionnaire items can be found in Appendix D. In order to facilitate use of this Appendix, items will often be referred to by both name and number, e.g., Sex (Variable 1), Age (Variable 2), etc. The Appendix reports the frequency distributions of the responses for the total group and for two groups made up of those people who reported an accident (A) and those who did not (NA), i.e., "no-accident." The total percentages reported in this Appendix for any specific variable include nonresponders, usually less than 5%. The percentages cited in the text for a given variable will not include these nonrespondents. Not all of the items in the survey form will be discussed in this report, and the reader is urged to consult Appendix D for information on any specific variable in which he may have a special interest.

10

Demographic characteristics of the sample.

 Returns were received from colleges in all nine U. S. Census areas. The number of returns from an area ranged from one (New England and Pacific) to five (East North Central and South Atlantic). While the sample cannot be considered strictly representative of the United States, it is important to note the wide geographic distribution of the returns.

The final sample of 1,232 was 62% male (Variable 1). The ages of the subjects were restricted to 16 to 30 years with 52% between 16 and 20, 40% between 21 and 25, and 8% between 26 and 30 (Variable 2). Reflecting the methodology employed, 96% of the sample reported their education (Variable 5) as some college or more.

Description of bicycle usage.

 Variables 20-44 in Appendix D give a general description of the type of bicycles these young adults were driving. The "modal" bicycle was a 10-speed (38%) with a 26" frame (49%) and 26" wheels (51%), hand brakes (64%), regular handlebars (59%), no front light (62%) but a rear reflector (82%), bought new and assembled (56%) 1 to 6 months ago (20%), and not registered with either the city or insurance company (67%). The most common variations were in terms of gears, (28% 1-speed, 26% 3-speed), handlebars (41% drop-style), and length of time owned (17% 1 to 2 years, 16% 5 to 10 years).

Variables 46-64 report on the details of the most recent bicycle use by the respondent. Since this information on most recent use was collected in the month of October, much of it pertains to driving on the college campuses. Almost one-third of the respondents, however, reported their most recent drive (Variable 47) to be more than 1 month ago, probably indicating summer driving while at home. Over half of the subjects reported that their most recent drive was 1 week ago or sooner. Weekends accounted for 35% of the last driving days (Variable 46). Bicycle driving was generally done alone {69%), on dry pavement (91%), with moderate to light motor-vehicle traffic (81%), in a residential area, (64%). Three-quarters of the

11

respondents drove farther than 1 mile, usually driving on the street with the traffic (71%). Most of their bicycle driving was on secondary streets (50%) or main streets (20%). The most common purposes for bicycling were "just for fun" (39%), exercise or health (13%), and commuting (27%). Of all the respondents, 6% reported falling on their most recent driving day.

As mentioned earlier, one potential source of bias was that only highly interested bicyclists would respond to the survey. This appears not to have been the case. When asked if they had driven a bicycle more than 100 miles in 1971, 72, or 73, (Variable 7) only 13% of the respondents answered affirmatively to all three years, and 47% did not report driving 100 miles in any of those years. Only 17% of the sample owned and drove more than one bicycle (Variable 19). When asked how many months they regularly drove a bicycle (Variable 8) and how far they drove (Variable 9), 51% of the respondents indicated they drove 4 or fewer months, and 54% estimated they drove less than 50 miles per month. Thus, it does not appear that this is a sample of extra-ordinarily active bicyclists.

Description of accidents.

Subjects were asked to supply accident details for their most serious accident within the past 12 months, or, if there were none that recent, the most serious within the past 5 years. While 390 (32%) of the subjects {An additional 220 subjects in the 16 to 20 year old age group reported details of an accident that occurred more than 1 year ago. Since these accidents could have happened when the subjects were less than 16, they have been excluded from this report. Analysis of the excluded accidents supported the assumption that they may have been children's accidents by showing, for example, an increased ratio of falls to collisions, which is more characteristic of children's bicycle accidents than of adult's.}reported having at least 1 accident in this time frame, details are available for only approximately 366 accidents because of some respondents who gave few or no details for their reported accidents. The percentages reported in this section will use as a base the number of subjects who responded to each question.

12

Variables 67 - 102 in Appendix D give a general description of the accidents of these young adults. As one would expect, the majority of accidents occurred in June (18%), July (23%), and August (11%). The 6 months from May to October included 78% of all reported accidents. Weekends, which account for 29% of the week, were not as over-represented as one might expect, with 32% of the accidents (Variable 69). Surprisingly, over twice as many accidents were reported on Saturday as on Sunday. Sixty-four percent of the accidents were reported as happening between noon and 6 p.m. (Variable 70). Most accidents occurred in the first 20 to 25 minutes of the drive (54%) and within 1 mile of home (74%) (Variable 81 and 82).

Examination of the bicycling environment did not disclose any startling conditions. Driving tended to be in good weather (74%) and was most often on secondary (46%) or main streets (23%) with light or no traffic (69%). Only 14% of the driving was against traffic or in the middle of the street, the remainder being either "with" traffic or not on a street (Variable 79). The accident area was residential in 56% of the cases (Variable 83) and was a place where the bicyclist had driven often before, 69% of the time (Variable 86). Intersections of some kind were the place of occurrence in 43% of the accidents (Variable 90) with almost half of these being the intersection of two streets.

Type of accident (Variable 89) may be categorized into falls and collisions. Using this dichotomy, 33% of the accidents were classified as falls and 67% as collisions. Crashes involving motor vehicles made up 32% of the collisions (21% of all accidents). Considering the activity just before the accident (Variable 84), the largest single category was driving straight ahead (49%), with turns making up another 22%, and getting on, starting, and
getting off accounting for 12%. Of all the respondents, 77% were sitting on the bicycle seat (Variable 87) and 48% were attempting some evasive maneuver (Variable 88) as the accident happened.

13

Some personal injury was the result of 79% of these reported accidents (Variable 95). Minor scrapes, cuts, and bruises were the most common results, accounting for 79% of the injuries. Injury treatment (Variable 99) was provided by a doctor or hospital in 12% of the cases, while 73% of the injuries were either self-treated or not treated at all. Arms and legs were the parts of the body most often injured, each accounting for 38% of the reported injuries. Injury resulted from direct contact with the ground in 66% of the reported accidents and from contact with bicycle parts in 15% (Variable 98). Some bicycle part failed in 18% of the accidents (Variable 91), though it is impossible to determine from these data whether the failures caused or resulted from the accidents. The police or insurance companies were notified in only 7% of these accidents.

Multivariate Comparisons

While the preceding section gives a general overview of the data, other questions require comparisons among two or more variables. One of the problems of survey research is that the number of possible comparisons among variables is often very large. In this study, only a few of the possible multivariate comparisons were examined. These were chosen either because past research (often on children) had shown these variables to be important, or because of general interest in the bicycling community.

Accidents, mileage, and accident rates.

Variable 65 in Appendix D gives the number of bicycle accidents involving any damage or injury in the past 12 months. There were 248 subjects (21%) who reported having had at least 1 accident in the past year. These subjects reported 373 accidents for an average of 1.5 accidents. The whole group of 1,197 subjects who responded to the question therefore experienced a rate of .31 accidents per person, per year.

In Table 1, Section A, accident rates per person were broken down by sex and type of bicycle. The male rate of .32 accidents was slightly higher than the female rate of .30. Among bicycle types (which for the multivariate comparisons were divided into three categories based on their number of gears), the higher speed bicycles (5-,10-,15-gears) had a much higher per-person rate (.42 accidents) than the 3-speed (.26) or 1-speed (.21) bicycles. When sex and bicycle type were examined simultaneously, however, the picture changed. Here females had higher rates for both 3-speed and 5-,10-,15-speed bicycles (.35 and .41 respectively) while the male rate increased only for the 5-,10-,15-speed bicycles.

Variables 8 (number of months of regular driving) and 9 (miles driven per month) were used to estimate annual mileage. These estimates, shown in Table 1, Section B, were broken down by sex and type of bicycle. While the estimates are useful in making comparisons between groups in this study, one must be very cautious in generalizing these mileage figures. Estimated average annual mileage for the entire group was 607 miles per person. Male mileage was 55% higher than female mileage (705 miles and 456 miles, respectively). Mileage also increased steadily with number of gears, 5-,10-,15-speed bicycle mileage of 882 miles being 80% higher than the 3-speed bicycle mileage of 491 miles which, in turn, was 70% higher than the 1-speed bicycle mileage of 287 miles per person, per year. The pattern was similar to that shown for accidents when examining both sex and type of bicycle. Female mileage was equally high for both 3-speed and 5-,10-,15-speed bicycles, while male mileage increased steadily with bicycle gearing.

15

Table 1

*{These numbers are approximate because each of the variables involved (1, 8, 9, 24, and 65) have slightly varying numbers of nonrespondents. See Appendix D.}


Mileage and accident information were combined to calculate accident rates per 1,000 miles of bicycle driving for sex and type of bicycle groups in Table 1, Section C. Again, although these rates are most useful in making comparisons between groups in this study, their absolute values may or may not be reasonable estimates of the "true" rate for young adults across the nation. The data collection method used in this study did not control potential sources of bias in a way that allowed precise generalizations to a larger universe. For the same reasons, tests of statistical significance were _not appropriate for this section. Rather the direction of large differences were noted, and the possible importance of these differences discussed.

16

The overall accident rate was .51 accidents per 1,000 miles. The pat-tern for males and females on the three bicycle types was very different as can be seen in Figure 1. Males, whose overall rate was .45 accidents per 1,000 miles, had a much higher rate (.76 per 1,000 miles) on 1-speed bicycles. Females whose overall accident rate of .67 per 1,000 miles was almost 50% higher than the male rate, had approximately equal accident rates for all three bicycle types. The combined accident rates for the three bicycle types showed a steady decrease with increasing number of gears, the 1-speed accident rate of .72 per 1,000 miles being 50% higher than the 5-,10-,15-speed rate of .48 accidents per 1,000 miles.

Figure 1 Accidents per 1,000 miles for males and females by type of bicycle.

17

The important difference between accidents per person and the accident rate per 1,000 miles is shown dramatically in Figure 2. Here one can see that if only calculations for accidents per person were available it would be assumed that the more gears a bicycle has, the more dangerous it is. When exposure (mileage) is controlled for, however, by calculating a rate per 1,000 miles, the trend is exactly the opposite. In other words, in these data, 5-,10-,15-speed bicycles had more accidents only because they were driven so many more miles.

Fig. 2 Accident and mileage rates by type of bicycle.

18

Sex and type of bicycle comparisons.

 Two important explanatory variables explored in this study were sex and type of bicycle. Differences between males and females have been examined in research both on children's bicycle accidents and adult automobile driving. Safety differences between various bicycle types are of interest to both bicyclists and bicycle manufacturers. In the present data, there was a significant relationship between sex and bicycle type (Table 1, Section D, X² = 77.6, df = 2, p < .001) with males having proportionately more bicycles with 5 or more speeds (56%) than females (30%). Because of this relationship, associations of sex and type of bicycle with selected usage and accident variables were more fully explored. Several of these variables were recoded for use in these analyses, as is shown in Table 2.

There were several important and internally consistent differences in the usage pattern of the three bicycle types. Table 3 indicates that higher speed bicycles were used in heavier motor-vehicle traffic (Variable 58), more frequently than lower speed types. While 53% of the 5-,10-, 15-speed bike usage was reported as being in heavy or moderate traffic, 49% of the 3-speed bicycle usage and only 38% of the 1-speed bicycle usage was so reported (Table 3). Other statistically significant, though less dramatic, differences among bicycle types were found with Variable 56 (1-speed bicycles used proportionately less than other types on pavement), Variable 59 (5-,10-,15-speed bicycles used proportionately more in business areas and less in residential areas), Variable 60 (5-,10-,15-speed bicycles driven proportionately more on main streets and less on secondary streets), and Variable 61 (1-speed bicycles driven proportionately more off the street, whereas 5-,10-,15-speed bicycles were driven more with traffic). The 3-speed bicycles fall between the higher and lower speed types in all of these analyses.

Table 2

-20-

Table 3

X² = 23.13, df = 6, p < .001

Variable 50, primary purpose of drive, may provide the best summary of these differences. When dichotomized into driving for transportation or recreation, the higher speed bicycles tended to be used more for transportation (54%), and lower speeds for recreation (61%), with the two use categories split almost evenly for 3-speed bikes (X² = 19.47, df = 2, p < .001).

Usage of different types of bicycles varied when sex of driver was considered, however. The aforementioned relationship between usage and bicycle type held for males as shown in Table 4 but as Table 5 indicates, was not significant for females.

It should be emphasized that there were no overall differences in bicycle use between males and females in this sample when type of bike was not considered. Looking at purpose of drive on the last driving day (Tables 4 and 5) for example, we see that 49% of the males and 48% of the females used their bicycles for transportation. The significant association between type of bike and usage that was present for males and not for females provides a possible explanation for the finding that male accident-mileage rates differed among bicycle types, while female rates did not, i.e., the males tended to use the different types of bikes for different purposes and females did not.

21

Table 4

X² = 20.84, df = 2, p < .001

Table 5

X² = 2.47, df = 2, NS

22

Examination of accident details also failed to provide a clear explanation for females having higher accident rates than males. Of 25 accident detail variables examined, only one (Variable 72) was related to type of bicycle, four (Variables 79, 89, 95, 97) were related to sex, and one (Variable 85) was related to both type of bicycle and sex. Variable 72, purpose of drive at time of accident, reflected the exposure difference shown in Tables 4 and 5. Table 6 shows Variable 72 by type of bike for males, and the pattern was very similar to the exposure pattern for males, i.e., with 1-speed bicycles being used mostly for recreation and 5-,10-,15-speed bicycles being used more for transportation. For females there was no significant relationship between purpose of drive at time of accident and type of bike. The other related accident variable with an exposure counterpart was Variable 79, orientation of bicycle in traffic. There was no exposure difference between males and females, but at time of accident males were driving proportionately more on the street with traffic and females proportionately more in off-street locations (X² = 10.45, df = 3, p < .02).

Table 6

X² = 12.60, df = 2, p < .01

23

Table 7 shows the significant relationship between type of accident (Variable 89) and sex. This relationship remained unchanged even when examined separately for the three types of bicycles. Males had proportionately more motor-vehicle related accidents and females more "other collisions," i.e., collisions with objects other than motor-vehicles. The proportion of falls was about equal for both. Sex was also significantly related to result of accident (Variable 95). When broken down by type of bicycle, however, the relationship was significant only for 1-speed bicycles. Overall, 79% of the reported accidents resulted in some personal injury. {While 79% may appear high, this is partly a result of the "definition" of accident that appears in Variable 65.}

Table 7

X² = 19.68, df = 2, p < .001

This held true for both males and females driving 3-speed or 5-,10-,15-speed bicycles and for males driving 1-speed bicycles. Accidents reported by females on 1-speed bicycles, however, resulted in personal injury 100% of the time (Table 8). The significant relationship between sex and part of body injured (Variable 97) was largely caused by males reporting proportionately more injuries to arms while females reported proportionately more injuries to legs (Table 9). Type of bicycle had no effect on this relationship.

24

Table 8

X² = 9.52, df = 1, p < .01

Table 9

X² = 37.33, df = 2, p < .001

Finally, the reported speed of the bicycle before the accident (Variable 85) was related to both sex and type of bicycle with males and higher speed bicycles tending to be traveling at higher speeds at the time of the accident. Males reported higher speeds than females at time of accident for each of the three bicycle types, though the association between speed and bicycle type as seen in Tables 10 and 11 was significant only for males. This finding is consistent with the other patterns in that only males reported differences in use of the three bicycle types.

25

Table 10

X²= 10.72,df=4,p<.05

Table 11

 X² = 6.22, df = 4, NS

26

Accident group vs. no-accident group comparisons.

Knowledge of differences between accident (A) and no-accident (NA) groups of bicyclists in exposure and usage patterns could help point out driving habits that are especially likely to result in accident occurrence. For these analyses, the A Group (N = 390) was composed of those subjects who reported details of an accident they had as an adult within the past 5 years, and the NA Group (N = 832) was made up of all other subjects. Comparisons between these groups were made with sex, type of bicycle, exposure (Variables 7, 8, 9, and 11) and all 19 of the variables concerning most recent bicycle use (Variables 46 - 64). These comparisons can be seen clearly in Appendix D, since the frequency distributions were reported separately for the A and NA Groups.

There was no significant relationship between sex and accident group membership. Thirty-three percent of the males and 29% of the females were in the A Group (X² = 2.27, df = 1, NS). This was in agreement with the similarity between male and female in accidents per person (Table 1). A significant relationship was evident between type of bicycle and accident group membership (Table 12). Highest speed bicycles made up 55% of the A Group but only 42% of the NA Group, a difference of 13%. Both 1-speed and 3-speed bicycles were overrepresented in the NA Group.

Table 12