COMMENTS

The 1979 Eugene study is possibly the first community-based study that included data on bicycle use as well as crashes, and which therefore could produce figures on crash rates. It is also of interest as a window into an early community bicycle planning process in the United States. Eugene, Oregon, the location of the state university, was one of the very first communities in the USA to have such a program.

The Eugene study examined all types of bicycle crashes. It analyzed car-bike collisions using the categories developed in the national Cross/Fisher study which had been published the year before. While the Cross/Fisher study could determine the percentages of various types of car-bike crashes, it included no usage data, and so it could not evaluate risk. The Eugene study could do this. One important conclusion, for example, was that sidewalk riding is more hazardous than riding in the streets, for a general cycling population.

The Eugene study does have important limitations. A small-sample problem is inherent in any community study: any individual hotspot or crash type may not represent enough crashes during a sampling period to establish with statistical certainty that there is a problem. Incomplete reporting of crashes, and spotty usage data, also may make it difficult to identify problems. Nonetheless, the community must attempt to identify and correct them. Also, authors' preconceptions and a community's political goals may bias the conclusions of a study. For any or all of these reasons, the Eugene study provides a less than complete picture, and does reach a few conclusions which have not been borne out by other research.

Unlike the Kaplan, Bikecentennial and Cross/Fisher studies which preceded it, the Eugene study does not include any statistical analysis, and so it provides no formal evaluation of the conclusions that its numbers suggest. Nonetheless, some of the cumulative figures in the Eugene study are sufficiently robust to allow of statistically valid conclusions.

The following comments refer to specific locations in the text of the report. Links to the text are provided.

Introduction

Citizen Involvement -- [page viii]: Citizen involvement has been an important feature of community bicycle programs in the USA. Eugene pioneered in developing a model for such programs which has become more or less standard, with a bicycle committee including representatives of local government as well as citizen representatives.

Recent History of Eugene Bikeways -- [page ix]. The City of Eugene first looked to construct only "bike paths", but over a few years, developed a "bikeway plan" that took in other types of facilities. The present report also calls for other measures, including education. It would be interesting to know the history that brought about these developments. One factor is obvious enough: the non-bicycling public thinks mostly about bicycling at a child's or novice's level of skill, and of separating bicycles from motor traffic. On the other hand, a bicycle committee charged with facilitating bicycle travel to all destinations must come to realize that financial constraints and availability of rights-of-way will require that most destinations be served only by roads. In fact, separate facilities are often most useful where the road system does not go. The bridges in Eugene which are open only to non-motorized traffic provide an example.

The Eugene study's call for education followed from one of its most robust conclusions: that the cyclist was at fault in  two out of every three car/bike collisions.

Chapter I - Project Summary

Eugene's Bikeway System

[page I-1] All of the types of facilities described in the list here except for contraflow lanes have been described in multiple studies in the USA. Some European studies examine contraflow lanes. Eugene was very early in implementing them. Unfortunately, there is no separate analysis of them in the report. It is likely that they were not extensive enough to generate significant results.

Bikeway System Usage

[page I-2] The reported 76 percent increase in bicycle usage between 1971 and 1978 is important in establishing a basis for comparison of crash numbers. The factors leading to this increase may be more difficult to determine. Not only did Eugene establish its bicycle program, but also, a national "bicycle boom" occurred in the years 1971-1974, with increased bicycle sales and usage. (The bicycle boom is often attributed to the fuel availability crisis of 1974 but actually preceded it, see graph in 1978 AAA study). An examination of year-by-year bicycle usage figures might help clarify these issues; see comments on Chapter II, below.

Bicycle Accidents

[page I-3] Here it is reported that the number of crashes "nearly doubled" between 1974 and 1978, while as reported in the previous section, bicycle usage increased only 76 percent between 1971 and 1978. In other words, the reported crash rate increased moderately over the time when new bicycle facilities were being installed.

The report does not make it clear why the reported crash rate increased, but there are some likely explanations. With a rapid increase in ridership, more inexperienced bicyclists will be riding. Also, an active bicycle program may lead to reporting of more crashes. It would be jumping to a conclusion to say that the new bicycle facilities were in fact more hazardous than preexisting conditions.

The introduction does give conclusions about reasons for changes in crash rates, but evidence to support these conclusions is not yet presented.

The reported crash rates are low -- 0.7, 0.6, and 1.8 crashes per 100,000 bicycle miles for striped lanes, signed streets, and sidewalk routes respectively. [However, my recalculation from the individual items in the Eugene study's tables gave rates of 0.8, 0.3 and 1.9 -- see explanation].  Even the Kaplan study of League of American Bicyclists members, generally experienced, avid cyclists, determined a rate of 11.3 per 100,000 miles.

In the Kaplan study, 27.8 percent of the crashes, or  3.14 per 100,000 miles, were serious enough to require emergency room treatment. These 27.8 percent include 6.7%, or  0.76 per 100,000 miles, which were serious enough to require hospitalization. Only the last of these figures is in the same range as those in the Eugene study. Studies of children and college-affiliated cyclists have found rates 5 to 7 times as high as the Kaplan study.

Crash rates in different environments, with different populations and different types of reporting are not directly comparable unless these variables can be examined separately. The crash rates for Eugene may indeed be lower than in other communities, but to a considerable extent, the low reported crash rates must reflect the low rate of reporting of crashes to police.

The conclusion that bicyclist operator error was responsible for two thirds of crashes is reasonably consistent with other studies of general bicycling populations in the USA. Results from the Cross/Fisher study show slightly more than half of all car-bike crashes due to bicyclist operator error. The difference may reflect differences in behavior, and differences in reporting.

Bicycle Accident Reporting and Monitoring

[page I-4] This section confirms the low reporting rates. There was some reporting of crashes by bicyclists who answered questionnaires, but most reporting was by police.

Chapter II - Bikeway System Usage

Bikeway Usage Variation

[page 2-3] The 76% increase in usage between 1971 and 1978 is revealed here to result from measurements at only six intersections, including one where bicycle traffic decreased. The small number of intersections where counts were taken in 1971, before the start of the study period, limited the possibilities for comparison. The data are therefore not sufficiently reliable to extrapolate to the entire city or to allow a statistically reliable comparison of crash rates. The more extensive bicycle counts conducted for the study would provide better comparison with future counts.

Chapter III - Bicycle Accidents

Annual Summary of Bicycle Accidents

[page 3-1] That 18 percent of the reported accidents resulted in "severe incapacitating injury" and 53 percent resulted in "major non-incapacitating injury" provides some explanation of the low rate of reported crashes. That the analysis of crashes is only from police department reports and not based on questionnaires answered by cyclists provides another explanation for the low rates. Less-serious crashes tend to be much more numerous, as shown in other studies based on reporting by cyclists themselves, but they are rarely reported to police.

[page 3-2] The increase in crashes shown in the table on this page is mostly in the categories of "minor injury accidents" and "no apparent injury accidents", suggesting that increased sensitivity of reporting during the study period may account for the difference. There is no obvious trend for more serious injuries. The numbers for each year in all categories except, perhaps, "major non-incapacitating injury accidents" are small enough that it may not be possible with any certainty either to identify a trend or establish that there is no trend.

Bicycle Accident Types

[page 3-2 continued in Chapter 3, part 2] The Eugene study was one of the first to adopt the categories of car-bike crashes from the groundbreaking Cross/Fisher study of car-bike collisions, and probably the first community-based study to do so.

Other studies have consistently shown that car-bike collisions only account for 10 to 20 percent of all bicycle crashes. Note that the categories A through F in the Eugene study describe only car-bike collisions, though the text indicates that "[t]raffic Accident Reports for all accidents involving bicycles" were used. The Eugene study does include an "other" category, not included in the Cross/Fisher classifications, which includes the crash types not involving motor vehicles. However, the emphasis of the Eugene study is strongly colored by the data source -- police reports -- and the emphasis on the Cross/Fisher classification scheme.

The Eugene study's review of the data makes generally cogent points about the age range of bicyclists most likely to have each type of crash, and about countermeasures.

[Table III-2] The numbers correspond well to those of the Cross/Fisher study. Numbers of reported crashes for all categories are small enough that the range of statistical error is considerable; it is widest for Group D, Motorist Overtaking/Overtaking Threat, for which there are only 11 reported crashes.

The importance of the numbers for individual crash types within each category varies depending on the category. All of the crash types in each of the Groups A through D result from similar types of errors and suggest similar countermeasures. On the other hand, the individual crash types in Group E, Bicyclist Unexpected Turn/Swerve, Group F, Motorist Unexpected Turn, and in the "Other" category, are decidedly different. For example, Motorist Unexpected Turn could be a turn to the right or to the left.

As mentioned earlier, the "Other" category is underrepresented, due to the rarity with which crashes in this category are reported to police. A later section of the Eugene report includes data which quantifies this underreporting.

Accident Group C

[page III-9] The illustration shows bicyclists riding in the correct position in the roadway in all but the Type 10 crash, though the text indicates the following percentages of hazardous or incorrect cyclist actions:

[Page III-10] The claim that "bike lanes reduce the occurrence" of Type 8 crashes has often been repeated. The data here do not either substantiate or refute that assertion. Bike lane advocates claim that bike lanes reduce sidewalk riding, and that the directional arrows in bike lanes encourage travel in the correct direction. Bike lane opponents point out that competent cyclists would not ride on the sidewalk anyway, and that bike lanes can encourage incorrect turning and crossing maneuvers by bicyclists and motorists. The relative importance of these factors depends on design details not described in the report, and on the sophistication of the cycling population.

Accident Group D

[page III-12] While the report notes that car-overtaking-bike collisions are perceived as a serious threat by bicyclists and motorists, the low numbers of such collisions are consistent with those in other studies. Bicyclists riding at night without lights, and intoxicated motorists, are mentioned as causes of such crashes.

The report correctly indicates that engineering countermeasures to Type D collisions are costly. The report does not suggest education and law enforcement as countermeasures.

Accident Group E

[page III-14] Note that while all of these crash types involve cyclist error, Type 21 is unlike the others in that it involves wrong-way riding.

Accident Group F

[page III-16] The illustration shows a Type 22 collision in which a bicyclist is lawfully waiting to turn left from the center of the roadway, but this subtype represents only a minority of reported Type 22 collisions.

[page III-17] Two of the Type 22 collisions occurred when the bicyclist was on the sidewalk or crosswalk -- therefore, on the left side of the roadway. The hazard of such collisions is increased because the bicyclist is entering the intersection from an unexpected direction.

The text indicates that 7 of the 13 Type 22 collisions occurred when a bicyclist was riding in a bike lane on the left side of the street. Six of these 7 crashes were along Pearl Street, a one-way street. Therefore, the bicyclists were traveling lawfully in the direction of traffic. The text suggests countermeasures -- "relocation or elimination or elimination of the bike lane, control of driveway access, or parking removal" -- yet bike lanes on the left side of one-way streets are common elsewhere and have not been shown to cause unusual problems. The text does not indicate any unusual features of the Pearl street bike lane that led to the high rate of Type 22 crashes. Perhaps this bike lane was behind parked cars?

[page III-18] The text indicates that Type 24 crashes occur when a bicyclist is "approaching unexpectedly from the rear of a motor vehicle turning right." The fault is then with the bicyclist. However, such crashes can also occur when a motor vehicle overtakes a bicyclist and then turns right. The fault is then with the motorist.

Other Bicycle Accidents

As indicated earlier, this is a catch-all category for crashes which do not involve moving motor vehicles.

[page III-19] The credibility of the claim that "bicycle lanes reduce the frequency of occurrence of [collisions with parked cars (Type 27)]" depends on the location of the bike lanes. If parking is eliminated to construct a bike lane, or the available width between moving motor vehicles and parked vehicles is increased, then the claim is credible. If the bike lane provides inadequate width adjacent to parked cars, then the claim is not. Also, it is not entirely clear what is meant by a parked car in connection with this crash type. Is the parked car stationary, or is it exiting a parking space? In the latter case, this crash type might be categorized as Type 22, "Motorist unexpected turn/swerve."

The low-rate of car door collisions (Type 28) suggests that Eugene's streets provide adequate width between the "door zone" and moving motor traffic -- or that on-street parking is not common along routes frequented by bicyclists. 

[page III-20] "The forty-eight accident types described above, accounted for 12% of the reported accidents." I can not make sense of this statement. Other studies show that bicycle-motor vehicle collisions account for about 12% of all crashes, but the 48 types in the Eugene study also include crashes which do not involve motor vehicles. The text repeatedly indicates that it describes all reported crashes during the sample period. Perhaps the number is a typographical error? Perhaps what is meant is that the reported crashes were calculated to be 12% of all crashes?

[page 3-20, continued in Chapter 3, part 3] It is unfortunate that the lack of usage data makes comparisons of crash rates before and after installation of facilities impossible. The report can claim that facilities reduced crashes, but it can not show that they reduced crash rates.

Bicycle Facilities

Reduced Accident Frequencies

[page 3-21] The meaning of the column in the table labeled "others" can not be determined. It can not be a count of crashes before the sample period, because there are post-installation entries in this column. The word "others" might indicate crashes which did not involve motor vehicles, as in the earlier part of the chapter, except that a description on page III-24 of crashes on 15th Street indicates that non-motor vehicle crashes were counted.

[page 3-23] Here is the table of streets with reduced crash frequencies (per year) after installation of striped lanes. I have added columns showing the number of crashes on which each of the crash frequencies was based.

At least one number in every comparison here is too small for the rates to have statistical validity. Because crashes occur at random times rather than on a schedule, the number of crashes could easily have been much higher or as low as zero, due simply to random variation -- an example of the small-sample problem which bedevils community studies. Also, the numbers are in crashes per year, rather than crashes per 100,000 miles. Variations in usage affect these numbers, but were not recorded. The number of crashes following an installation may be higher, and the rate lower only because the sampling interval is longer.

"These striped lanes have channelized bicycle traffic in an expected manner, and promote riding consistent with the Rules of the Road." See comments about page III-10.

Unchanged accident frequencies

Here is the table of streets described as having unchanged crash frequencies (per year) after installation of bicycle facilities. I have added columns showing the number of crashes on which each of the crash frequencies was based.

The same comments about small samples apply as for increased crash frequencies, above. But also, the number of crashes per year actually increased in three of the five cases. Increased usage, increased risk and random variation could all increase the crash numbers. In the two cases without comparisons, the installation was made too close to the beginning or end of the sample period.

It is stated in the text that "the average annual frequency of bicycle accidents was very high along 18th Street prior to striping", and the number in the table bears this out. Unfortunately, there are no post-installation numbers for 18th Street. A review of numbers from later years would allow a comparison. The same would be true of an installation on Willamette Street made near the end of the sample period. In fact, such a comparison was carried out for 18th Avenue, but Eugene's Bicycle Coordinator, Diane Bishop, is not confident about the results:

18th [Avenue] is a weird one. I think the numbers we used in its study were too small to be statistically significant and have warned people who have asked for that study. But you make a good point. We should do some follow up counts at the same locations, check crash data, and see what has changed.

18th continues to be a street that cyclists aren't fond of. It has pretty heavy traffic, lots of turning movements from cars (I hear from cyclists this is where they have to watch especially carefully for right turners), and the surface in the bike lane is miserably bumpy to ride on.

(in e-mail to John S. Allen, March 28, 2003)

Increased accident frequencies

[page III-24] Here is the table of streets with reported increased crash frequencies (per year) after installation of bicycle facilities. I have added columns showing the number of crashes on which each of the crash frequencies was based.

As noted in the text, the increase in crash rates may be due to increased usage, but there are no usage data. Still, unlike with the comparisons of unchanged and reduced crash rates, the high post-installation numbers give some validity to these results except perhaps for Coburg Road, where the pre-installation crash count also was high. The descriptions of the causes of the crashes are credible and agree with those of other studies. Why there are problems with cyclists' swerving left across the bike lane on Harlow Road, but not elsewhere, is an interesting question. The question as to why the crash rate is so high in the left-side Pearl Street bike lane has already been discussed in comments about page III-17.

Problem Streets

[page III-25] The statement that striping projects will "allow" riding consistent with rules of the road goes beyond the statements on page III-10. Riding according to the rules of the road would be allowed except in the case of a posted prohibition requiring cyclists to ride, for example, on the sidewalk.

Bicycle Route Accident Rates

"Accident rates on 'Separate Bicycle Facilities' are not appropriate for this evaluation project, as there were no bicycle-motor vehicle collisions on Eugene's separate bikeways. The accident reporting study described in the next chapter determined that the bicycle accidents which do not involve motor vehicles are seldom reported to the City." This report acknowledges and attempts to quantify the problem in reporting of non-motor vehicle crashes. But only an evaluation of all crashes on an equal basis will catch all important deficiencies. A crash resulting from a facilities deficiency need not involve a motor vehicle, and can occur on streets, or on separate facilities.

[page III-26,] To check the crash rates given in the report, I recalculated the total bicycle mileage for each segment as length, times bicycle volume, times days of implementation.

(end_date - start_date) * length * bicycle_volume

I have also recalculated the crash rate, using the formula described in the text,

                         crashes * 100,000                    
(end_date - start_date) * length * bicycle_volume

My calculations can account only for segments on which bicycle volume was known. I have assumed that the date of implementation was at the middle of the month given. You may review the Microsoft Excel workbook which includes my calculations, if you wish. My results agree well with those in the report, with a few important exceptions (see below).

For striped routes, I calculated a crash rate of 0.80 per 100,000 miles rather than the 0.7 given in the report, for a total of 6.2 million miles of riding and 49 reported crashes.

Note that Table III-4 and III-5 are accompanied by footnotes and text commentary indicating "accidents per 100,000 miles per year". This is a nonsense quantity (accident-years per 100,000 miles). The correct units, actually embodied in the calculation, are "accidents per 100,000 miles". See also comments on Fig. IV-4, which includes the formula for calculation of accident rates.

[page III-28] I have also recalculated crash rates for segments of signed bicycle routes in which information is available. The average crash rate is only 0.32 per 100,000 miles rather than the 0.6 given in the report. I have no explanation for the nearly 2/1 discrepancy. There were 31 crashes in all, enough to have some statistical validity, in 10.4 million miles of riding.

It is interesting that the rate for signed routes is lower than for striped routes.   This difference might reflect inherent differences in the safety of the installations, and/or differences in traffic volume. It might also reflect differences in averaging. I can not explain why neither of my numbers agrees with that in the report.

The report indicates that signed routes did not reduce the crash rate, while striped routes did. The average crash rate on the signed routes was lower, then, even before they were signed -- according to the calculations in the report and even more so, according to my calculations. This result suggests that signs were placed on streets which did not, on average, have as bad a record for crashes as the streets that were striped.

Sidewalk Bicycle Routes

This report is of historic importance in having established the conclusion that sidewalk bicycle routes are hazardous for a general cycling population. The conclusion in this report is based on only 22 crashes, leaving considerable room for statistical uncertainty. The conclusion has, however, been confirmed by many other reports. As with the tables discussed earlier, I have recalculated the crash rates and mileage to check the numbers. There is one significant discrepancy; when calculated directly from the crash and mileage data, the crash rate I calculated for Hilyard street is more than twice as high as the one given in Table III-3 of the report and somewhat higher than the different number given on p. 3-28:

The average crash rate for sidewalks given in the report is 1.8 per 100,000 miles. The rate I calculated for equal weighting of all miles traveled is 1.95 per 100,000 miles, in 1.1 million miles of riding.

The report does not give a number for crash rates on streets with no treatment. There are only two such streets on which bicyclist mileage was recorded, 18th Street between Bailey Hill and Willamette, and 24th Street, Olive to Columbia. There were 8 reported crashes -- too few for statistical validity -- in 1.1 million miles of riding, for a rate of 0.7 per 100,000 miles. That this number is lower than for signed bicycle routes may reflect differences in characteristics of the streets, or else only statistical uncertainty.

Summary of Corridor Accidents

[page III-30] The assertion is made again that bike lanes reduce crashes. See comments on page III-10 but also note that the crash rate is lower in any case for the signed routes than for those with bike lanes. This difference could be due to any or all of several factors, as described earlier.

Bicycle Accident Causes

[page III-31] The list of identified driver errors raises some questions. Some of the categories such as "improper passing" and "improper sidewalk riding" are vague, though this is not of concern in determining who is at fault. Two of the categories, "wheel fell off" and "brake failure" do not necessarily reflect operator error. In spite of these minor issues, the conclusion that 2/3 of the reported crashes resulted from bicyclist operator error is robust -- probably the most robust one in the entire study.

[page III-32] The call for bicyclist education is eloquent, and well-supported by the data.

Chapter IV - Bicycle Accident Reporting and Monitoring

[Fig. IV-1] The categories of accidents are rather vague and open to interpretation, especially "bicycle hit car or "car hit bicycle", which are a matter of opinion or of reporting in many cases. It would have been better if the categories in chapter III could have been applied to this questionnaire -- though interviews of the crash participant(s) would then have been necessary to identify the categories. The short description asked for at the bottom of the form might help define the categories to some degree. The table does indicate the category of crash according to the Cross-Fisher categories, indicating that this in fact could be determined from the responses and/or from police reports.

[Table IV-1] The voluntary report forms do indicate that crashes not involving motor vehicles outnumber those which do. However, voluntary reporting leads to a problem in determining the ratio. To get a good reading on the ratio, it is important to track crashes over a sample period within a predefined population group, as the Kaplan study and Bikecentennial Study did. Problems with the reporting are described straightforwardly in the text following the table.

[page IV-5] The figure given, that 61% of crashes do not involve a motor vehicle, is low compared with the 80% to 90% in population-based studies. This difference probably reflects reporting issues; particularly, that less serious crashes which did not bring cyclists into emergency rooms were not reported, and that the police reports disproportionally reported car-bike collisions.

[page IV-6] The bicycle accident records are derived from police reports for the most part, unlike the questionnaires, which directly report bicyclists' responses. Hence, the bicycle accident records provide a way of cataloguing crash data rather than an additional source of new data.

[Fig. IV-4] This figure includes the formula used elsewhere in the report and described as counting "accidents per 100,000 bicycle miles per year",

                 (Accidents)                   
(Years) x (365 days) x (Bicycles) x    (Miles)      
            (year)      ( day )     (100,000 miles)

However, the units for years and days both cancel out of this formula, the quantity described as "100,000 miles" is really a dimensionless number, and the unit "bicycles" can be taken out of the formula, as it is only an identifying label. The formula therefore resolves to:

100,000 * Accidents
  Mile

Recalculation of the crash rates confirms that this was actually the formula used. This is an appropriate formula for calculation of actual risks. It appears that the error was not in the preparation of the formula used in the tables, but in describing that formula. The data which this study offers are therefore significantly more useful than they appear to be.

[page IV-9] The bicycle route evaluations describe a valuable monitoring function in identifying crash hotspots. Such a monitoring system completes a feedback loop by which crash reports can lead to amelioration of problems.

Chapter V - Public Involvement and Input

[page V-1] It is not clear whether the list of comments reflects one of the hearings (which one?) or both. The wording "There were only a few brief periods when the microphone was not occupied, as was the case with the original hearing" is confusing, as it is in a paragraph referring to the first of the two hearings described.

The comments reflect the usual public desire for more separate bicycle facilities. Whether these are in fact practical to serve the desired transportation goals is an issue that members of the public usually misunderstand. The comments do, however, also reflect a wide range of other concerns, indicating an unusually well-informed populace.

[page V-2] The change in percentages of comments on different topics indicates that the public became more knowledgeable during the study period. It would be interesting to know what factors led to this change: raw experience, or perhaps also informational campaigns.

The line "Classes teaching effective bicycling skills" reveals awareness of John Forester's Effective Cycling textbook (first published 1975) on the part of one or more commenters. The emphasis on education elsewhere in the report also reflects an awareness of Forester's work.

[page V-3] The responses about bicycle paths and the hazards of motor vehicle traffic reflect the usual public opinion, which in some ways conflicts with the results of scientific research including this study.

[page V-4] The differences of opinion between bicyclists and non-bicyclists are telling in how they reveal differences in perception.

[page V-5] The differences in perception of travel times are very telling. In fact, as "commuter races" have often shown, bicycle travel times often are shorter than motor vehicle travel times for short trips, due to the ability to park closer to the trip endpoints, to take shortcuts not open to motor vehicles, and to keep moving when motor vehicles are stalled in traffic jams.

The response for the percentage of riding on bicycle paths may be too high -- see bicycle volume map in Chapter 2. One possibility is that respondents confused striped bike lanes with bike paths. Such confusion is common. Unfortunately the locations of paths and other facilities can not be identified precisely from the bicycle volume map.

The numbers on transportation mode preference show that bicycle use mostly substitutes for motor vehicle use; unlike in Pasanen's Helsinki study, there is little substitution for public transportation use. The numbers make it clear that   there was little public transportation in Eugene during the study period in any case.

It should be noted that Eugene is a university town and that a large percentage of the student population would not have owned motor vehicles. There is a major difference in mode choices depending on age, and on ownership of motor vehicles. The Kaplan study addressed this issue, but the Eugene study does not discuss it. Age, car ownership and licensing were recorded on the questionnaires, but the report does not distinguish between the transportation mode choices of cyclists based on these criteria.

Chapter VI - Bikeway Improvements

[page VI-1] It is likely that the problems and solutions discussed in this chapter had a significant effect in advancing the work that led to the AASHTO (American Association of State Highway and Transportation Officials) Guide for the Development of New Bicycle Facilities, published in 1981.

[page VI-4] The report does not include photographs, plans or other detailed documentation on the conditions deemed to require improvement, and so the comments here can not include a detailed examination of the recommended improvements for specific locations. Most of the recommended improvements seem reasonable, though some are questionable: for example, to install raised reflector dots along a bike lane stripe on 350th Street. The reflector dots would pose a surface hazard to cyclists.

[page VI-9] The lighting recommendations account for pedestrians on the paths. Properly-equipped cyclists using paths without pedestrians would get by with only their own headlights. Pedestrians can be difficult to see by the light of a typical bicycle headlight, and many bicyclists do not use a headlight, even though it is required by law.

Next: Letter from Consultant