Initial Goals

In our continuing fascination with mass transportation, we decided to use this project to complete an ethnographic study of Georgia Tech's Stinger system. Our initial intent was to observe and record impressions of the context in which riders use the bus system. A secondary goal: to determine whether some sort of technological application could improve the experience for the participant.

Our interest stemmed from a project developed by group of students in our CS6751 class held during Fall Quarter 1998. The group proposed and created an information application meant to improve the bus system. The "BuzzBus" prototype was a kiosk-based system that provided just-in-time information (route maps, wait times, current location of the bus) for the riders waiting at a bus stop. While the group conducted both surveys and informational interviews prior to the development of the prototype, we felt there was more information that could be gathered through a detailed (albeit quick) ethnographic study.

Observations

Our primary method of data collection included observations of the riders participating in the Stinger experience. The initial goals of the observation were to develop some sense of the usage patterns of the Stinger system. The chief goal of the "BuzzBus" prototype was to provide the rider with enough information to make an informed decision on whether or not to use the system. Our preliminary goals were less specific and narrow. We began with simple observations of how the riders used the system. In essence, we were non-intrusive observers of the system and allowed the data to direct further observations. As we progressed through the study, we gathered more substantive data through conversations and interviews (both formal and informal).

After we observed rider patterns, we began to realize that there were other participants in the Stinger experience besides the passengers. The drivers, for instance, were integral members of the process. We soon discovered that drivers do a great deal more than just simply navigate the routes – they make a variety of decisions that can impact the passenger experience profoundly. This notion led us to spend the latter stages of the observation focusing on driver behavior and work practice.

Analysis

After completing approximately 25 observation hours of the Stinger Bus System, we noticed several recurring patterns and behaviors among all the participants. In the first review of our field notes, we compiled our data and developed six categories (or classifications) of the system experience. The categories were:

Waiting Passenger Behavior

Riding Passenger Behavior

Pedestrian Behavior/Opportunistic Riders

Driver Behavior

System Fluctuations

Physical Layout

As we categorized the observed behaviors along these lines, we were influenced by the methodology of Contextual Design. Although we didn’t realize it, we were beginning to develop models of behavior that reflected some of Beyer and Hotzblatt’s work models. For the final analysis, we continued to utilize Contextual Design and its methods of analysis. Specifically, we focused on creating consolidated flow, sequence, cultural and physical models. By understanding these models of behavior, we hope to gain insight into how we might redefine (or redesign) the work practice. The following pages contain a brief description of the model, the consolidated work models themselves, and some interesting observations informed by those models.

 

Flow Model – Stinger Bus System

The flow model represents the responsibilities of the participants in the Stinger experience. It also maps the communication we observed between the participants. D1 is the focus of this model because the driver became the primary focus of our observations.

 

The communication between D1 and other drivers is the primary means for route coordination. Route coordination allows drivers to space buses evenly to reduce wait time for passengers. This communication link, however, relies upon a CB radio system that breaks up when the buses are spaced far apart.

The communication between D1 and riders is primarily nonverbal. This can lead to problems, as it is often difficult to discern the intent of a nonverbal gesture. For instance–raising your hand or sitting on a bus stop bench or nodding your head can all signify a desire to ride the bus.

Sequence Model – Stinger Bus System

This model identifies the order of specific tasks needed to complete a given action. It invokes the concepts of intent, abstract steps, and triggers in listing these chronological steps. The intent defines why the work represented by a sequence matters at all—in essence, what is driving the user initiate an action? An abstract step is a generalization across different observations of individuals completing a task. And, finally, a trigger causes the sequence of events, notifying the user to take action.

We found the first two cells of the passenger sequence model to be particularly interesting. First, the passenger must decide whether or not to ride the bus. Second, they may or may not make an accurate decision because of lack of information. This lack of information has some implications for a redesign. For instance, the introduction of an information appliance (like the "BuzzBus" prototype) at the stop could mitigate some of this information failure. We discuss additional solutions in the design implications section below.

Passenger Sequence Model

Activity	Intent	Abstract step
• Decide to walk or ride	• Get to where I am going	Trigger: see bus or bus stop • Formulate opportunity costs
• Wait for the bus	• Ride the bus	• Decide where to situate at the stop • Look for the bus or in direction of where bus is expected • Gather knowledge from others (schedules, wait times) • Decide to keep waiting

Passenger Sequence Model (cont’d)

Activity	Intent	Abstract step
• Prepare for the ride	• Make myself ready	Trigger: see bus coming • Choose appropriate bus to ride • Rise/ready yourself when the bus is near • Decide on which door to enter • Decide on how to treat disembarking passengers
• Ride the bus	• Ride the bus until my stop	• Board the bus • Decide where to sit • Decide to acknowledge driver or not • Decide what to do (read, check watch) • Clarify route with driver or other passengers
• Get off the bus	• Get off at my stop	Trigger: see the stop • Choose method of driver notification: Passive–assume bus will stop Active–move toward driver Active–pull stop cord Active–ask driver to stop • Decide to acknowledge driver or not (disembarking) • Disembark

 

The driver model also provides some insight into participant behavior. Most interesting to us were the actions (or workarounds) the drivers utilized in navigating the routes and responding to route events.

Driver Sequence Model

Activity	Intent	Abstract step
• Define the route	• Keep moving to next stop	• Driving the bus
• Stop bus	• Pick up/ drop off passengers	• Determine if you fill passenger requirements • Pause for opportunistic riders and stragglers • Determine stop needs
• Continue defining route	• Keep moving to next stop	Trigger: doors have been closed • Driving the bus • Conversations with passengers • Communicate location with radio
• Respond to route events	• Rectify problem events	• Avoid bunching with workaround

 

 

Cultural Model – Stinger Bus System

The cultural model is often the most difficult to illustrate. It requires the observer to understand and interpret relationships that are intangible. Our model focuses on the driver and the set of influences that surround and shape their work.

One noteworthy observation: the model demonstrates an indirect relationship between the riders and the drivers through the school and Argenbright (the transportation management company). Basically, riders complain to the school about the "bunching" of the buses, the school, in turn, complains to Argenbright, who tells the drivers to use their CB radios to distance the buses. This policy solution, however, did not work because the drivers (particularly on the Campus route) often ignored it. This has implications for any redesign—you cannot mandate the drivers use some technological solution without first securing their approval.

 

Physical Model – Stinger Bus System

The physical model represents a refinement of the activities we observed across the Stinger stops. We segmented this generic stop into zones where certain activities tended to occur. The loading and looking zones afford a wide field of vision and allow the passengers to see the bus arrival. The opportunistic zones provide similar affordances for non-waiting pedestrians. Finally, the waiting zones often constrain vision, but provide shelter and a place to rest.

We observed that the physical design of the student center stop often hindered the use of buses. We noticed that when the opportunistic zones were lacking, there was a decrease in opportunistic ridership. Pedestrians could not walk up and ride the bus because they did not have time to see the bus. Also, when the waiting zone has walls that occlude vision, riders must wait closer to the street to see incoming buses. The student athletic center stop is fairly open. Along the walking paths to the stop, a pedestrian can easily see incoming buses. At this stop we noted increased opportunistic ridership.

 

Design Implications

From our observations we noted four major issues in the Stinger Bus System that have design implications. First, riders endure long wait times at stops – sometimes as long as 20 minutes. Secondly, this wait time is exacerbated by the fact that there is very little information at stops about bus routes or arrival times. The third issue deals with the poor communication links between the drivers. Driver communication allows buses to be spaced evenly reducing wait time for passengers. Poor communication – either by choice or ineffective CB radios – adversely increases wait time. Finally, the physical construction of stops can hinder the use of buses.

A solution to combating the second problem (information failure among passengers) might include the installation of an information appliance (like the "BuzzBus" prototype) at the stop. This type of device could provide passengers data about the current status of the bus (location, time to arrival) and can assist in decision-making. This type of appliance, however, doesn’t solve the first issue of having to wait for the bus. Through our observations, we noticed a phenomenon we termed "bunching" – when buses would arrive at a stop within a few minutes of each other.

We know that the communication links between drivers are poor and this seems to cause bunching. However, the solution to the first problem is not simply solved by implementing a technical solution. We cannot go in and force drivers to use a new communication system through policy changes. Currently a CB radio is installed on every bus. Drivers on the campus route were told to use the system and ignored the mandate. Drivers on the MARTA route use the system and can effectively space themselves out on their routes. If a technological solution is proposed, drivers must be educated on the effects of their actions and motivated to use the system.

Finally, for the physical bus stop, we suggest optimizing the visual affordances of opportunistic zones. Walls, trees and bushes surround the student center. These effectively prevented the pedestrian from determining the buses’ location. Reducing these barriers might promote increased ridership.