Jim Marston Dissertation Chapter 4.5

4.5. Reported and Perceived Transfer-Making Behavior

Observation reveals that many vehicle drivers will disobey speed laws, change lanes, run caution lights, go around crossing gates, and certainly change routes in order to save a perceived or actual minute amount of time. Highway traffic engineers are thus very comfortable using “time savings” as the utility function being maximized to best represent typical drivers and their decision to change routes. Recker, Chen, & McNally (2001) state that “travel demand theory, whether derived from consumer demand theory or direct demand principles, is intrinsically rooted in the notion that travel time is a commodity to be saved” (p. 339). They then state that the time-savings would be transformed, by the traveler, into something of intrinsic value; i.e., more time spent on performing activities or increasing the spatial extent of available alternatives for performing activities. These observations on transportation explain automobile use but do little to explain the patterns of transit use and decision-making. In a previous study (Golledge & Marston, 1999) , it was noted that some blind people did not mind when the van service took them much longer to get home, as long as they got to their door. Unlike car drivers, it appears that time is not the prime utility to be considered for blind travelers, and any use of the time saved might be transformed into something other than the variables that conventional accessibility and traffic demand models are based on.

Little is known about the motivation or utility that affects decision making when it comes to leaving a transit vehicle in order to make a transfer to a faster route, like an express bus or rail system. Unlike simply changing lanes or turning onto an expressway, this action requires a multitude of actions. Even with perfect knowledge of the system, one must leave the vehicle, walk some distance to the new area, perhaps wait for the new mode, and board the vehicle. Even with a free transfer, one might have to go through a fare gate or access a fare machine. For these reasons, the same type of utility maximization behavior for transit use employed by drivers is not expected. The utility of saving time is confounded by these other necessary actions and efforts. Little is known about what this impedance to making a transit transfer to save time is or what it is based on. The next sections reports on transfer making behavior reported by transit users, both sighted and blind.

4.5.1. Impedance Considerations while Making a Transfer Decision

Data were collected in order to understand this impedance to making transfers and to evaluate how it differs for the sighted public and for people with vision restrictions. If the reluctance or impedance to change modes is different between the sighted and people with limited or no vision, these data could be used to measure another restriction to access and allow the computation of another accessibility measurement.

4.5.1.1. Spatial impedance or distance decay

The concept of distance decay stems from Newton’s model of planetary attraction or gravity. He discovered that the attraction between two bodies was not only based on the mass of the two bodies, but was affected by the distance between them.

Social gravity models use some form of attraction between places to determine the pull effect and some type of force decay with increased distance (distance decay) to account for the tyranny of distance or other effort. A simple gravity model would be:
I_ij=g(A₁*A₂)/d_ij^X Where

Iij is the interaction between two locations, i and j,

g is a gravitational constant,

A₁ and A₂ are some measure of the attraction of the two locations,

d_ijis the distance or effort between the two locations, and

X is an exponent that shows the effort (force) needed to overcome distance.

Calculating a coefficient to model the distance decay is more complicated than a simple exponential function of physical distance. The mode of travel must be considered, but the limitations to travel exhibited by the individual should also be considered. In a previous discussion (see Section 3.1.2.2 , Time Penalty Formulation ), the need to consider individual constraints on travel was introduced. These constraints include the mode choice and restrictions on an individual’s travel abilities and how they both affect the measurement of accessibility will be considered next.

4.5.1.1.1. Effect of Travel Mode Selection

Consider a large employment center located 10 miles across town from a large residential area. Those people using a private car would expend very little personal energy and would be able to make the trip in 15 or so minutes. Without a car, people could take a city bus that might take 40 or more minutes and require some personal energy expenditure for walking to and from the bus stop. Others might ride a bike, which could take an hour and require more energy from the user. Still others might be forced or choose to walk, which could take several hours and much energy output. Thus the resistance to overcome this 10-mile commute is no simple constant and varies greatly depending on which mode of travel is available. Although all these modes would get a person there, the work site is much more “accessible” to a driver than a bus user, a cyclist, or a pedestrian. One cannot calculate an accessibility measure from a job site to the residential area unless the mode of travel is considered.

4.5.1.1.2. Effect of the Person-Mode, or Type of Individual Constraint

Consider now two neighbors who both ride transit to the job site. One is blind and the other is sighted. The field test data previously discussed indicate that it would probably take more time and energy for the blind person to make the exact same trip. In addition, perhaps there is a faster route that requires making a transfer, but one would have to walk an additional several blocks and cross some busy streets. The sighted person might decide to expend the effort to transfer to save travel time, while the blind person might be content to spend more time on the slower bus route rather than deal with extra navigation effort, street crossings, stress, and apprehension. As in the first example, these two people also have different accessibility to the same site, and their resistance or distance impedance is different. Equation 1 , on page 94 , explains how the variable l is used to designate the person type, i.e., the specific type of access or mode for each individual.

Because of the spatial separation of one vehicle to another vehicle, there is always some effort of distance to overcome when making a mode transfer. The more effort that is needed to overcome this spatial separation, the more impedance there is to overcome and the more reluctance there is to attempt it. The increased impedance for the blind when attempting to make a timesaving transfer is addressed next.

Making a mode transfer can introduce time randomness into the equation for all riders. Will the next vehicle be waiting and ready to go, and will the streets have a WALK or WAIT signal? These and other variables are not controllable by the potential user. This is why timed and coordinated transfer stops are so helpful to users. If not synchronized, a person will wait, on average, at least ½ of the headway time for an incoming vehicle. The experiment question about transfer-making behavior was phrased in order to try and eliminate the effect of this randomness from subject responses.
The subjects were presented with the following scenario:

“For each situation, assume that you are a regular rider of a transit line and your trip home takes you one hour. You find out that a new route such as an express bus or rail service has opened up. You can save some time on your one-hour trip but will have to make a transfer from your regular route to the new route or system. For these situations, assume that there is no waiting time at the transfer site, only the walking and search time and effort. The questions ask about making this new modal transfer in both familiar and unfamiliar areas. How much time would you have to save before you would make a transfer to another mode?”

With this scenario, there is no ambiguity about the time waiting for the new mode vehicle, and, since they were asked how much time they would want to save on a trip to home, the actual walk and search times should not enter into their response. Their estimate should be based strictly on the effort, stress, and apprehensions of the transfer, search, and walk. Each subject responded to this scenario for six different types of transfers: a transfer in the same block, one block away, and three blocks away, in both familiar and unfamiliar areas. Blind subjects were asked this question during the pre-test interview to gauge their current accessibility and also after they had used RIAS in the field test. A group of 30 sighted people, matched by age and sex to the blind subjects, also reported their answers (see Section 1.6.6 , Sighted Subjects for Baseline ). The sighted group data act as a control for comparison with the two blind data sets.

4.5.2. Transfer Data Analysis

As would be expected from 30 subjects of various ages and sex, there was a wide range of answers to these questions. Figure 4.5 shows the distribution of these data for each subject for each of the six transfer tasks. Table 4.9 shows the number of people in each group, who showed the most reluctance to change vehicles for potential time-savings, and reveals that the utility of saving time is overshadowed by other factors. For the blind using their regular skills and aids, it appears that comfort, secure and known surroundings, uncertainty, apprehension, and fear are affective states or utility functions to be considered. Even for the sighted control group, some people put a very high value on other utilities than saving a small amount of time.

Figure 4. 5 : Data Points for Six Transfer Scenarios

Same Block In Familiar Area

Same Block In Unfamiliar Area

1 Block In Familiar Area

1 Block In Unfamiliar Area

3 Blocks In Familiar Area

3 Blocks In Unfamiliar Area

The horizontal line in the diamond shape on the chart shows the mean of each set of data points, and the diamond shows the 95% confidence level of that mean. At a glance, one can see that the reported time savings required is much higher, in each category of distance and area familiarity, for the blind using their regular methods. The addition of auditory and spatial information makes those estimated data quite similar to that given by sighted respondents. The means diamonds for the sighted and the blind, when they considered RIAS, there is a large overlap, showing that there is no significant difference in their data. P values are discussed later.

Many of the people reported they would require a large amount of timesavings before making a transfer, and Table 4.9 shows the percentage of responses with high amounts (30 to 60 minutes) of time that they would rather stay on a known vehicle than to make a transfer and save that time.

Table 4. 9 Percent of Subjects with High Resistance to Transfe r Vehicles.

# of Extra Minutes Would Stay on Vehicle	Percent of Subjects
# of Extra Minutes Would Stay on Vehicle	Blind Regular	Blind W/ RIAS	Sighted (control)
60 (no transfer)	18%	1%	3%
40 or more	36%	2%	5%
30 or more	71%	16%	7%

The utility function of saving time is clearly not what motivates all transit users, especially for the blind. Fully 18% of the responses to the six transfer scenarios showed that the blind would “waste” 60 minutes rather than change vehicles. Over a third, 36%, would spend an additional 40 or more minutes than attempt a transfer, and almost three out of four (71%) would rather spend an additional 30 minutes or more than make a transfer. That amount of resistance to saving time, as compared to the sighted control group, demands closer analysis.

Table 4.10 shows the mean responses from the three subject groups for the six (distance and familiarity) transfer task scenarios. For example, the sighted (control) subjects said they would not make a transfer in the same block in a familiar area unless they could save 11.6 minutes out of the 60-minute trip home. They would walk a block if it could save them 13.1 minutes from the original trip time, but they would need to save 20.8 minutes before they would walk three blocks for a transfer. In contrast, the reported mean times were much higher for blind subjects using their regular skills and aids. These subjects reported that they would have to save 18.3 minutes to make a transfer in the same block, 23.5 minutes to go one block, and a mean of 33 minutes to go three blocks in a familiar area in order to attempt the transfer.

Table 4. 10 Mean Responses for Six Transfer Scenarios

		Mean Saved Time To Make a Transfer
Area	Subject Type	Same Block	1 Block	3 Blocks
Familiar	Blind, Regular Method	18.3	23.5	33.0
	Blind, with RIAS	11.5	13.9	20.0
	Sighted	11.6	13.1	20.8
Unfamiliar	Blind, Regular Method	27.0	33.8	44.0
	Blind, with RIAS	13.7	16.9	23.8
	Sighted	12.1	14.0	23.0

The discussion that follows is focused on a set of graphs that show the reported mean times and a linear trend line for different combinations of conditions. Figure 4.6 shows the data for three subject groups making a transfer in a familiar area. In all three cases, the further people had to walk to make the transfer, the more time they wanted to save.

The trend line for the blind subjects using their regular methods of navigation was the steepest and had a much higher initial resistance. The sighted subjects show a flatter linear trend. There was a highly significant difference between transfer behavior reported by the sighted (control) and by the blind subjects using their normal technique (p<.0001 or less for all three distances—same block, 1 block and 3 blocks). After the blind subjects used RIAS in the experiment, they changed their transfer-making perception and thus the impedance to accessibility. The estimated means were much lower than what they originally reported as their regular behavior. This difference for the two blind conditions was also highly significant (p<.001 or less for the three distances). In addition, the behavior reported by the RIAS users was almost identical to the responses from the sighted control group. In fact, there was no significant difference between those two groups (P<.95, 0.65, and 0.80 for the same block, 1 block, and 3 blocks, respectively).

Figure 4. 6 Transfer Decisions in a Familiar Area

Figure 4.7 shows the data for three subject groups making a transfer in an unfamiliar area. The results look quite similar to the familiar area, although the initial resistance and slope of distance decay is higher for each group. There was a highly significant difference between transfer behavior reported by the sighted and by the blind subjects using their normal technique (p<.0001 or less for all three distances). After the blind subjects used RIAS in the experiment, they reported much different perceived transfer-making behavior.

Figure 4. 7 Transfer Decisions in an Unfamiliar Area

This difference for the two blind conditions was highly significant (p<.0001 or less for the three distances). In addition, the behavior estimated by the RIAS users was similar to the responses of the sighted control group. There was no significant difference between those two groups (P<.38, 0.12, and 0.79 for the same block, 1 block, and 3 blocks, respectively).

4.5.3. Effect of Area Familiarity on Transfer Making Behavior

Unfamiliar areas present problems when cues, paths, and locations must be learned over time. Figure 4.8 compares the mean reported times for the three groups in both the familiar and unfamiliar areas.

Sighted respondents reported little difference between familiar and unfamiliar areas, and no significant difference was found (p <.16, 0.13, and 0.12, respectively for the same block, 1 block, and 3 blocks). The effect of unfamiliar environments on the people with vision restrictions is strongly shown demonstrated by a comparison of their estimated transfer behavior. Same block times went from 18.3 minutes to 27.0, 1 block times from 23.5 to 33.8, and 3 block times from 33.0 to 44.0 minutes when comparing familiar and unfamiliar transfer areas. The difference in the two familiarity conditions, for the subjects using their regular methods, was highly significant (p <.001 or less for all three distance measures). Even with the vastly lowered estimated time for transfer behavior after using RIAS, the effect of area familiarity was still in effect, although not nearly as strong. Same block times went from 11.5 minutes to 13.7, 1 block times from 13.9 to 16.9, and 3 block times from 20.0 to 23.8 minutes when comparing familiar and unfamiliar transfer areas. The data on area familiarity differences were significant (p <.002 or less for all three distance measures).

Figure 4. 8 Effect of Area Familiarity on Perceived Transfer Decisions

4.5.4. Modeling Transfer Making Behavior

Since only three distance data points were measured, an exponential decay function was not used, but, rather, a linear model of the form:
Y = B +A*X or
R = IR + T*D where

R = Resistance (total time savings needed to attempt a transfer).

IR = Initial resistance to make a transfer

T = Time resistance per interval of distance

D = Distance in blocks

This liinearization simplifies the data so that the initial resistance to make a transfer (IR) and the distance decay in minutes as distance increases (T) can be measured. Table 4.11 shows the initial resistance to travel and the per block resistance for the six test conditions.

Table 4. 11 Linear Model for Making Transfers

	Familiar Environment			Unfamiliar Environment
		Intercept	Slope		Intercept	Slope
		Initial Time Resistance	Time per Block		Initial Time Resistance	Time per Block
Blind, Regular	R=	18.5 +	4.9D	R=	27.5 +	5.6D
Blind with RIAS	R=	11.3 +	2.9D	R=	13.7 +	3.4D
Sighted (control)	R=	11.0 +	3.2D	R=	11.3 +	3.7D

The initial resistance to make a transfer in a familiar area for blind travelers using regular methods is 18.5 minutes and 4.9 minutes for each additional block they have to walk. In contrast, the sighted subjects had a mean initial resistance of 11.0 minutes to make a transfer and 3.2 minutes per block traveled. After using RIAS, the perceived initial resistance to make a transfer for the blind dropped to 11.3 minutes and 2.9 minutes per additional block.

When comparing a familiar area to an unfamiliar area, the blind regular group reported their initial resistance to making the transfer increased nine minutes to 27.5, and the resistance or distance decay increased 0.7 minutes to 5.6 minutes per block when navigating in an unfamiliar area. For the sighted, the area effect was minimal with the initial resistance increasing only 0.3 minutes to 11.3, and the decay rate increased 0.5 minutes per block. The RIAS users estimated their initial resistance increasing by 2.4 minute to 13.7, and the per block impedance increased by 0.5 minutes while transferring in an unfamiliar area.

4.5.4.1. Impedance per Block

The initial resistance (IR) to transfer in the same block included the inconvenience of leaving the vehicle and finding the next transfer point. Any variation in walking distances further than the same block would strictly measure the effort of the extra distance, since the transfer point search was included in the same block data. Subjects considered a distance of one block (from the same block transfer to the one block transfer), the entire three blocks (the difference between the same block times and the three block times), and the last two blocks (the difference from the one block times and the three block times). Figure 4.9 shows the per-block resistance to walk in minutes for the three groups in both conditions, familiar and unfamiliar areas.

The variation in mean distance resistance between the three groups was previously examined, but there also exists a variation in how increased distance affects their perceived resistance to walk. The graph shows distinct “signatures” or patterns of the perceived effort of walking. These patterns hold for both familiarity conditions. For the sighted subjects, their smallest resistance per block was for the first block walked. The “effort” of walking three blocks resulted in a much higher resistance per block. Walking the last two blocks had the highest resistance to overcome. Of the three groups, it was the sighted that had the most reluctance to walk as the distance increased.

The pattern was reversed for the blind people using their regular methods. Their highest resistance was in walking the first block and crossing a street. The per-block resistance decreased for the three-block distance and further decreased for the last two blocks. This group seemed to not be bothered by extra walking effort as much as the sighted. A blind person, well trained in Orientation and Mobility procedures, may not seem to consider the walk tasks to be very difficult, rather, it is the task of finding vehicles, signs, boarding areas, or stations that pose the bigger problem.

After using RIAS, subjects estimated their transfer times, and the resulting resistance trend was more like a combination of the other two groups. The graph shows an almost flat line because they estimated a more constant time resistance for each block traveled. The possibility of using RIAS changed their resistance signature from a decreasing trend line in their regular method to a slightly increasing line, more like the pattern the sighted exhibited.

Figure 4. 9 Distance Impedance per Block

4.5.5. Summary of Impedance to Make Transfers

Saving time is not the only factor when people consider making a transfer of transit vehicles.

Sighted transit users reported resistance to transfer probably based on affective attitudes such as less personal effort, comfort, and accepting a “sure thing,” rather than adding any more uncertainty to the trip.

Blind subjects reported a much higher resistance to making transfers, as their resistance to uncertainty would be much higher without the benefits of visual cues.

Sighted users reported little difference in resistance to transfer based on the familiarity of the area, but unfamiliar areas elicited much higher resistance data than familiar areas for the blind users.

Blind users showed a higher resistance to overcome the walking and search effort to find a transfer point.

After using RIAS, blind subjects reported transfer making behavior that was very similar to that reported by the sighted users.

Time penalties, difficulties, and uncertainty of navigation during transit tasks were examined in Chapter 3. Those timed trials, and previous field tests on finding vehicles and making transfers (Golledge & Marston, 1999; Golledge et al., 1998b) , help confirm the perceptions reported here. The lack of full information on where the transfer point is, what route or vehicle is served, finding the vehicle, crossing streets, and navigating the walk appears to markedly increase the resistance to transfer much more than it does for the sighted.

Uncertainty is increased during navigation without sight, more mistakes can be made, and there are more barriers to overcome. This uncertainty is increased in unfamiliar areas, and, by staying on a known route or vehicle, transit users assure their seating and do not have to confront situations that might cause uncertainty, new decision tasks, or obstacles. At each decision point, a person without vision might make an error or not reach their goal, and as the number of decision points increases along a route, the probability of making a path or even trip-altering mistake increases rapidly.

The difference in the times reported by the sighted and those from the blind indicate another restriction to access for people with vision restrictions in the built and transit environment. The perceived reduction in blind users’ resistance, when considering RIAS, was similar to that of the sighted and indicates that the paucity of accessible identity and directional cues in the environment helps cause the increased resistance for that group, when using their regular methods.

These data show that the impedance to efficiently make transfers directly affects the ability of a blind traveler to take full advantage of a transit system and achieve the degree of accessibility that the system was designed to provide. The lack of information and environmental cues directly and negatively impacts the ability of those with vision restrictions. These data on transfer decision making measures another barrier to, and constraint on, transit and travel accessibility for those people with vision restrictions and the models quantify the initial and distance impedance.

BACK TO OVERVIEW
BACK TO TABLE OF CONTENTS
NEXT SECTION