Special Session Summary Waiting For Service: Affective Responses, Satisfaction and Decision-Making of Consumers Waiting in Queues

Dilip Soman, HKUST
Rongrong Zhou, HKUST
[ to cite ]:
Dilip Soman and Rongrong Zhou (2002) ,"Special Session Summary Waiting For Service: Affective Responses, Satisfaction and Decision-Making of Consumers Waiting in Queues", in NA - Advances in Consumer Research Volume 29, eds. Susan M. Broniarczyk and Kent Nakamoto, Valdosta, GA : Association for Consumer Research, Pages: 431-433.

Advances in Consumer Research Volume 29, 2002     Pages 431-433



Dilip Soman, HKUST

Rongrong Zhou, HKUST

Queues are ubiquitous. Every day, the average individual may have to wait in a queue to take a bus to work, to use an ATM or to speak to a telephone ticketing agent. The three presentations in this special session look at various aspects of consumer affective experiences, satisfaction and decision-making while waiting in a queue. In the first presentation, using a computerized experiment, Amit Pazgal, Sonja Rajdas and Ambar Rao studied the impact of providing various types of information on customers’ balking (decide not to join) and reneging (leave after having spend some time waiting) decisions in queues. They investigated both the quality of the decision-making as well as the customers’ satisfaction with the decisions, and compared these decisions to the queuing theory benchmarks. Their results suggested that for a given line length, information about the expected waiting time and actual time already waited increased the probability of joining and staying until served; of the two factors, information about the expected wait was by far more effective.

In the second presentation, Rongrong Zhou and Dilip Soman further looked at consumer experiences in, and decisions to leave a queue. While this should depend only on the size of the total expected wait, they showed that the number of people waiting behind a person largely influenced satisfaction, affect and reneging decisions. Specifically, with more people behind, the mood and satisfaction was better, and likelihood of reneging was lower. It was argued that the "number of people behind" effect was mainly due to social comparisons that people spontaneously make while waiting in queues. Results from five experiments and one field study were presented showing the robustness of this effect and testing the underlying social comparison explanation in three different ways.

In the third presentation, France LeClerc looked at situations in which consumers calling for service were in a "telephone queue". Prior literature suggests that in general the greater the number of distractions, the smaller is the perception of the time interval. Theories in time perception however suggest that whether this assumption holds is a function of whether consumers are indeed trying to evaluate the waiting time while they hold. More specifically, it was proposed that when people are told ahead of time to evaluate the length of a time period (prospective judgment), more "events" occurring during this period makes the period seem shorter. On the other hand, when people are asked about the length of the time period only after having experienced it (retrospective judgments), the more events, the longer the time period appears. She presented results from a laboratory experiment that confirmed these predictions.

At the conclusion of the three presentations, Donald Lehmann led the discussion. Besides summarizing the three papers and highlighting the common learning from them, he put forward interesting questions and issues and offered suggestions on directions of future research in this area.



Amit Pazgal, Washington University

Sonja Rajdas, Washington University

Ambar Rao, Washington University

You go to the campus ATM to get cash. You need to be back at your office for an appointment in a few minutes, but you think you will be able to get back in time. The line at the ATM is quite long. You can:

$ Join the line and wait until you get cash, even though you might be late.

$ BalkBnot join the line because you don’t want to be late, and return later.

$ RenegeBjoin the line but leave after a few minutes to avoid being late, and return later

For a service provider reneging and balking are both undesirable, because they may lead customers to switch to another service, (e.g., another bank.)

Queuing theory shows that in a single server queue with exponential inter-arrival times and service times (an M/M/1 queue) a customer with full information minimizes expected waiting time by balking immediately or joining the line and staying until served. Reneging is never optimal. However, customers typically do not have full information and considerations of psychological costs may lead them to different decisions than those prescribed by queuing theory.

In this paper, we study balking and reneging decisions in a computerized experiment. Our goals are to:

$ Determine the impact of various types of information on balking and reneging.

$ Assess participant satisfaction with decision-making.

$ Establish the rules that participants used in their decision-making and compare these rules to queuing theory benchmarks.

$ To provide guidance to managers of service systems.

We developed a computer simulation of an M/M/1 queue; participants had the option to join or to balk and return "tomorrow". If they joined, they still had the option of reneging and returning tomorrow. Those who balked or reneged faced another queue, the length of which was independent of the one they left. This time they had to wait until served. After two training sessions, each participant encountered ten queuing situations, where we randomly varied the length o the queue. We studied the impact of two informational factors on decision-making. The first factor, called the clock factor consisted of the presence or absence of data customers normally haveBnamely clock time and its variants, such as elapsed time. Participants would only be able to see and use this data after they joined the queue, but the practice sessions ensure that they know that this data will be available to them should they join the line. The second factor, called the information factor, consisted of the presence or absence of information about the expected time before service, provided before a participant joined the queue. In all conditions customers were asked to estimate their waiting time before receiving service. Those who balked or reneged were also asked to estimate the line length on the following day. Approximately 60% of participants systematically underestimated the expected waiting time for long lines and overestimated it in short lines, while 40% did the opposite. We captured this heterogeneity in a parameter bBas b increases, the overestimate of waiting time increasesBwhich was used as a covariate in the analysis of the data from the experiment. Participants in all experimental conditions were asked for their estimate of how long they had waited after they were finally served. We also asked participants to record their satisfaction with their decision-making (not the service).

We found that for a given line length, both information about the expected waiting time and clock data increased the probability of joining and staying until served; of the two factors, information on expected waiting time was by far more effective. The covariates, line length and b, had significant effects, both directly and via their interactions. The net effect of each of these covariates was to reduce the probability of staying. Considering only those who balked or reneged (i.e., all those who were not served on the first day), the information factor, by inducing participants to join, reduced balking but increases reneging. The clock factor had no impact in this setting. Satisfaction with decision-making was highest (on average) for those who stayed until served followed by balkers and renegers. Information about the expected wait increased the satisfaction of stayers. For balkers and renegers, when the actual number next day was greater than the expectation (a loss), satisfaction was reduced, while when the actual number was smaller than expectation (a gain) there was no significant impact. Over 90% of participants used a consistent threshold rule to determine when they would balk. We infer that this rule is based on two componentsBthe expected line length on the next day and the switching cost. Participants were remarkably accurate in their estimates of the expected queue length: However, consistent with prospect theory, participants overestimated the switching cost by more than a factor of 2, thus joining the line when they should balk (to minimize their expected waiting time). Reneging was infrequent, occurring on only about 4% of occasions.

We conclude that managers of service systems should offer estimates of expected waiting time to their customersBnot only does this information reduce balking, it also improves satisfaction with the decision making. However, efforts to switch customers, for example from peak load periods to off peak periods must provide significant incentives in order to overcome the loss aversion associated with switching costs.



Rongrong Zhou, HKUST

Dilip Soman, HKUST

Queues are ubiquitous (e.g. at bus stop, banks and even on the phone) and have been studied extensively in the operations literature. In the present research, we study consumer experiences in a queue and their decisions to leave (renege) or willingness to pay (WTP) to avoid further waiting. A rational evaluation of these decisions should depend on the time already spent waiting and the expectation of the remaining wait. However, we found that the number of people waiting behind a person also influences the value associated with the queue position, satisfaction, affect and decisions to leave the queue (reneging). Specifically, with more people behind, the mood and satisfaction is better, and likelihood of reneging is lower. In a series of studies, we test for the robustness of this effect of the number of people behind, and its underlying mechanisms.

Study 1 describes a waiting scenario taking place in the post office. Both number of people ahead (5 vs. 10) and behind (0 vs. 5 vs. 10) were manipulated. Not surprisingly, more people ahead led to more negative affect, greater willingness to renege and higher WTP to avoid further waiting. More interestingly, when there were more people behind, subjects reported more positive mood, were less likely to renege and indicated lower WTP. Similarly, Study 2 uses a scenario that involves hiring someone to wait in line for you and subsequently paying him for this waiting service. Results show that subjects perceived the service more valuable and were willing to pay much more for it when they found out there were more people behind the position that were held for them.

There exists a universal human tendency to learn about the self through comparison with others (Gilbet, Price and Allan, New Ideas in Psychology, 1995). We propose that the reason why consumers care about people waiting behind may be related to social comparison, more specifically, downward comparison. Waiting is usually an unpleasant experience which represents the loss of a valuable resourceCtime (Osuna, Journal of Mathematical Psychology, 1985). However, seeing people behind is somewhat a comfort since "there are people worse off than me", therefore makes the experience less painful. Three sets of factors (queue factors, personal factors, situational factors) might shape the extent of social comparisons, each of which is tested in a separate experiment.

First, characteristics of the queue may affect the salience of relative queue position and therefore affect the ease with which social comparisons can be made. In Study 3, we manipulated the type of queues (linear vs. take-a-number queue) as well as the number of people behind (0 vs. 3 vs. 6), leading to a 2 X 3 between-subjects design. In the linear queue condition, subjects were shown photos of people waiting in a straight (linear) queue to be served. In the take-a-number queue condition, subjects were shown photos of people sitting at randomly selected locations in a waiting area for their number to be called. Then they were asked to empathize themselves with a target person in the photo and fill out a series of measures. Results showed that the effect of number of people behind was weaker in the take-a-number queue condition when social comparisons were made more difficult.

Second, there are individual differences in the tendency to make social comparisons. In Study 4, we measured subjects’ general tendency to socially compare using the [INCOM] scale (Gibbons and Buunk, JPSP, 1999). Number of people behind was again varied between subjects. Results showed that this manipulation affected mood, satisfaction and reneging (as before) to a greater extent for subjects with a higher tendency to socially compare.

Third, we argue that situational factors (e.g. unusual events) determine the accessibility of counterfactual thoughts which, in turn, determine the degree of social comparisons made in a queue setting. For example, a situation in which the consumer could have been delayed prior to joining a queue could trigger counterfactual thoughts about their position relative to others (e.g. "if I had arrived 15 minutes late, I would be a long way behind the queue as compared t where I am now.") resulting in feelings of luck, positive affect and greater determination to stay in line. In Study 5, we manipulated the (situationally-driven) tendency to make social comparisons through the accessibility of counterfactuals. Results showed that the number behind effect was weaker when the counterfactuals were less accessible.

Our findings have theoretical implications for the perception/evaluation of time and waiting, as well as practical implications for the design of queues and waiting areas.



France Leclerc, University of Chicago

Please hold, all our agents are presently busy helping other customers! Every one of us has had to listen to this or a similar message while trying to get through to airlines, phone-ordering companies or utilities. In fact, it is common to hear such messages more than once in a single phone call since often the same message is repeated at regular intervals while the caller is waiting for the service provider to answer. Furthermore, there is frequently some music playing in the background that one can hear in-between messages.

Presumably, one of the objectives of the service provider for using such tactics is to affect how people experience the waiting time. Not surprisingly, the amount of time spent waiting for a service appears to be negatively correlated with customer satisfaction (Clemmer and Schneider 1989). Since reducing waiting time is not always possible (or profitable), managers may opt to rely on tactics that can affect subjective waiting time. As suggested by Maister (1985), if customers are kept busy, the wait should feel shorter. This view would suggest that the more messages heard during the waiting period, or if there is music played in the background, the shorter the wait will feel.

Theories of time perception ( see Zakay and Block 1997, for a review) suggest that whether this assumption holds is a function of whether consumers are indeed trying to evaluate the waiting time while they hold. More generally, it has been proposed that when people are told ahead of time to evaluate the length of a time period (prospective judgment), more "events" occurring during this period make the period seems shorter. On the other hand, when people are asked about the length of the time period only after having experienced it (retrospective judgments), the more events, the longer the time period appears. Said differently, when under a prospective mode, consumers appear to process time according to an attentional model. The attentional model holds that judged duration is a direct function of the amount of attention paid to the passage of time. The less attention spent focusing on waiting because of distractions the shorter the wait will feel. On the other hand, under a retrospective mode, since there is no memory trace of time, the judgments have to be reconstructed and duration estimations are based on encoded information available in memory. At the time the duration is judged, the person retrieves stimulus information and estimates duration based on the amount of information that was retrieved. Therefore, when a time interval is filled with more events, more information, more complex information, or when the interval is more segmented or altered in any other way that would increase the amount of information available in memory, the perceived duration is greater. As for music, given that it has been show to attract attention to non-temporal information, one would expect that its presence in the background would enhance the effect of number of messages on time estimation.

Predictions derived from these theories were tested in a laboratory study. Subjects were asked to evaluate a new "phone-ordering" system that was entirely computerized (no human interaction). They were told that they would have to order a good using the new system and would be asked afterwards to evaluate their interaction with the system on a number of dimensions. After having dialed the service, subjects were put on hold for 4 minutes. The temporal paradigm (i.e., whether or not they were explicitly told to attend to the wait per se) was manipulated as well as the number of messages ( 3 or 8) the subjects heard while waiting. For half of the subjects, light classical music was played in the background whereas the remaining half had no music. At the end of the wait, subjects were asked how long they had waited.

The pattern of results obtained supports the predictions. When subjects had music in the background, in the prospective condition, more messages lead to a shorter estimated waiting time whereas in the retrospective condition, the opposite was found. When subjects had no music in the background, the effect of number of messages did not reach significance.

This research suggests that the tactic of keeping someone busy while waiting will not necessarily lead to consumers remembering the waiting period as shorter. Clearly, a number of other factors have to be considered. This research highlights the challenge of defining optimal managerial practices in this domain.