• Federal Highway Administration >
• Publications >
• Research Publications >
• 05137 >
• Chapter6.Cfm >
• Evaluation of Safety, Design, and Operation of Shared-Use Paths

Evaluation of Safety, Design, and Operation of Shared-Use Paths

Final Report

PDF Version (1172 KB)

CHAPTER 6. PERCEPTION DATA COLLECTION

GENERAL RATIONALE

The purpose of this phase of the study was to quantify users' perceptions of the essential structural and operational characteristics of shared-path facilities and their interaction for the purpose of developing a better understanding of how such perceptions, along with objective measures of facility performance, might be used to construct a model of shared-path LOS.

The basic approach was to have participants view selected segments of imagery collected by a bicyclist wearing a helmet-mounted video camera as he or she rode through a shared-path environment at a set, predetermined speed. We were able to use the imagery that we collected from a number of trails across the United States as part of the operational data collection effort as described in chapters 4 and 5.

User perceptions were quantified in terms of their ratings of lateral separation, longitudinal separation, and the perceived ability to pass other trail users. Participants were also asked to provide an overall rating of how satisfied they would be using the trail segment that they were viewing.

Participants ranged in age and in the level of familiarity/use of shared-path facilities. There was also a wide range in terms of estimates of their own health status and in terms of the extent to which participants engaged in walking and/or riding for recreational and/or fitness purposes. More information is provided on user attributes below.

All data were collected during fall 2002 and winter 2003. The imagery shown to the respondents in the perception phase of the study was collected during summer and fall 2001 and winter 2002.

PARTICIPANTS

A total of 105 individuals served as volunteer participants in the perception data collection phase of the study. Participants were recruited from the Raleigh-Durham-Chapel Hill, NC, area and from the greater Washington, DC, metropolitan area. The data were collected in group settings, with the size of the individual groups based on subject availability. The volunteers were recruited primarily from bicycle user groups in the two areas. The fact that the respondents were willing to give up a portion of their evening or their lunch period for a volunteer effort to aid future bicycling indicates the level of commitment that the respondents had regarding the project.

Thirty-four percent of the participants were female and 66 percent were male. The distribution of subjects by age range is given in table 23. We were very pleased by the wide range of respondent ages.

Table 23. Distribution of subjects by age

The distribution of self-reported individual health status is given in table 24, while the distribution of the participants' estimates of their frequency of walking or bicycle riding for recreational and/or fitness purposes is given in table 25. As one might expect, those interested in shared-use paths generally believed themselves to be in good health and generally walked or bicycled often for recreational and fitness uses. Note that the walking and bicycling reported in table 25 was not necessarily on shared-use paths.

Table 24. Distribution of reports of individual health status.

Table 25. Distribution of individuals' estimated frequency of riding and/or walking for either recreational and/or fitness purposes

In response to a question on how frequently respondents used shared-path facilities (either as pedestrians or as bicyclists), the estimates are provided in table 26 below. These data confirm that our respondents were generally very experienced shared-path users.

Table 26. Estimates of shared-path use.

In response to a question on the respondents' most frequent reasons for trips (either as pedestrians or bicyclists) on shared-path facilities, the following characterizes the range of responses:

• Commuting to work: 18.9 percent.
• Commuting to school: 1.2 percent.
• Utilitarian trips: 9.4 percent.
• Socialize with friends: 5.9 percent.
• Recreation: 33.5 percent.
• Fitness: 31.2 percent.

Compared to nationwide shared-path users, this respondent sample was probably over-represented for commuters to work and under-represented for commuters to school.

COLLECTION OF USER PERCEPTION DATA

Individual user perception data were collected by responses provided to items on a paper-and-pencil survey developed by the project team. More details on the nature of the survey instrument and on the data collection procedure itself are given below.

Structure and Content of the Survey Instrument

A paper-and-pencil survey instrument was developed by the University of North Carolina Highway Safety Research Center (HSRC). The instrument was similar to instruments developed and used by HSRC to collect user perception data for the FHWA Bike Index Study⁽⁶⁾ and to that developed and used by Hughes and Harkey.⁽⁶¹⁾

For the present study, thirty-six 60-s video sequences were selected from the head-mounted video camera images collected during the operational data collection phase of the study. The project team thought that this was about the upper limit of the effort that we could expect from our volunteer respondents without fatigue having a major impact on the results. All video sequences were black-and-white and were limited to the field of view of the camera selected for use in the study. Video quality, especially after digitizing, ranged from good to marginal because some images had less than optimal camera angles and some were quite dark. Nonetheless, the project team selected real video over staged video or still images for this study because the real video best conveyed actual path operations. The images were good enough and were displayed for long enough to give the respondents a realistic view of operations on the trail at that time. After digitizing, the video sequences did not include sound.

Selection of the individual video sequences was based on a review of the structural and operational facility characteristics of the trails used in the operational data collection phase of the study. Basically, we began by selecting 10 of the trails on which we collected operational data that had the best video quality and that best spanned the range of geographic locations, trail widths, and trail geometrics. Then, for each of these trails, we selected three trials (six for the higher volume Chicago and Seattle trails) at the proper bicycle speed (see below) that represented high, medium, and low user volumes. Finally, the team selected and digitized 60-s clips from within the longer trials that best represented the volume levels desired and that did not contain unusual events that could cause bias in the response, such as a passive pass. Table 27 shows a summary of the 36 video clips that we used.

Whereas operational data were collected under three different desired bicycle speeds (trail mean, trail mean plus one standard deviation, and trail mean minus one standard deviation), the sequences used in the perception data collection phase were all from trials where the test bicycle speed was between 15.29 and 20.93 km/h (9.5 and 13 mi/h). We thus tried to remove the speed at which the bicycle was moving as a variable for the respondents.

Subjects viewed each of the 36 conditions while seated in a group setting. Video sequences were projected from a laptop computer using a liquid crystal display (LCD) projector and a projected screen size of approximately 1.8 m (6 ft) high by 3.0 m (10 ft) wide. All video sequences were stored in a CD-ROM format.

After the respondents entered the testing room, they were seated, welcomed, asked to fill out the informed consent form (appendix A), asked to read an introduction (also in appendix A), and asked to fill in their background information (appendix B). They responded to a warmup sequence of three 60-s trials. We did not analyze the data from these three warmup trials. After all respondents' questions about the format were answered, the main testing began.

Table 27. Characteristics of the 36 perception data collection video clips

1 ft = 0.305 m, 1 mi/h = 1.61 km/h
Ea = Active passes
b = Passive passes
c = In focus, slightly out of focus, moderately out of focus, or highly out of focus.

The 36 video sequences were presented in four blocks of nine trials each. The order in which the blocks were shown to different groups was varied randomly to avoid learning or fatigue biases. A trial consisted of a 60-s video presentation, followed by 30 s during which the screen just displayed the trial number and the participants recorded their responses on the printed response sheet that was provided. Overall testing time, with a short break between blocks two and three, was approximately 80 min.

Format of Participants' Responses

Participants were required to rate their perceptions of selected facility conditions on a 5-point scale, an example of which is shown in figure 20. Explanations of "lateral separation," "longitudinal separation," "ability to pass," and "overall" were provided to subjects at the outset in the introductory fact sheet (see appendix A).

Condition Number CD-4, 035

Figure 20. Representative response format.

These four response parameters were selected in order to: (1) solicit user perceptions along dimensions pertinent to the design and operation of the type of facility of interest, and (2) to obtain user perception data consistent with the variables in the equations that we developed in chapter 3. In particular, we were interested in user perceptions about the interaction between structural and operational facility characteristics and the ability to pass, since the latter is probably a key construct in estimating the perceived LOS of a facility.

In general, the respondents were able to follow the directions that we provided and to complete the rating of all 36 video sequences. Only one respondent left early, and no respondents reported motion sickness from viewing the images. Some respondents indicated that the video image quality was a problem (see chapter 7 for an analysis of that factor); however, most indicated that they were able to fairly rate the path and its operation from the images displayed. Appendix C shows a typical frame from the video clips that we used for each of the 10 paths.

FHWA-HRT-05-137