Complete Research Paper from attached feedback

RESEARCH ARTICLE

WHEN FILLING THE WAIT MAKES IT FEEL LONGER: A PARADIGM SHIFT PERSPECTIVE FOR

MANAGING ONLINE DELAY1

Weiyin Hong Department of ISOM, Hong Kong University of Science and Technology, Clear Water Bay, Kowloon, HONG KONG

and Department of MIS, University of Nevada, Las Vegas,

Las Vegas, NV 89120 U.S.A. {[email protected]}

Traci J. Hess Isenberg School of Management, University of Massachusetts, Amherst,

Amhest, MA 01003 U.S.A. {[email protected]}

Andrew Hardin Department of MIS, University of Nevada, Las Vegas,

Las Vegas, NV 89120 U.S.A. {[email protected]}

As one of the most commonly experienced problems on the Internet, download delay is a significant impediment to the success of e-commerce websites. While some research has examined how such delays can be reduced and how much delay online users will tolerate, little research has taken a theoretically grounded approach to managing perceptions of the wait. Based on time perception theories, we develop a research model of the effects of actual wait time, amount of information, and direction of attention on perceptions of the wait. Two empirical studies were conducted using an experimental travel website to test the proposed hypotheses. The results show that with shorter waits, providing additional visual content, such as a travel picture, may make the wait feel longer. However, with longer waits, additional visual content that distracts the user from the passage of time makes the wait feel shorter and reduces users’ negative affect toward the wait. Further, the benefits of providing visual content in longer waits are enhanced as more content is provided. Visual content should also be chosen to distract the user from time and temporal processing, as reminding users of the passage of time can encourage temporal processing and make the wait feel longer, especially in longer waits or when the amount of temporal visual content is high. Our findings extend time perception theories and contribute to the literature by identifying a potential paradigm shift, from the retrospective to the prospective paradigm, when waiting times are prolonged. Post hoc study results confirm the practical contribution of our research, demonstrating that several key findings are counter-intuitive to professional web designers.

1

Keywords: Online waiting, time perception theory, perceptions of wait, amount of information, visual content, direction of attention, download delay

1Mike Morris was the accepting senior editor for this paper. Andrew Burton-Jones served as the associate editor.

The appendix for this paper is located in the “Online Supplements” section of the MIS Quarterly’s website (http://www.misq.org).

MIS Quarterly Vol. 37 No. 2, pp. 383-406/June 2013 383

Hong et al./A Paradigm Shift Perspective for Managing Online Delay

Introduction

Waiting on the Internet has been recognized as a commonly experienced problem (Ceaparu et al. 2004; Rose et al. 2005) that negatively impacts users’ information search and evalua- tion of websites (Dabholkar and Sheng 2008; Galletta et al. 2006; Rose and Straub 2001). The cost of this online waiting is billions in lost revenue to e-commerce vendors annually (Keynote Systems 2005; Zona Research Study 2001). While server-side technologies and data-transmission speeds have improved (e.g., Curran and Duffy 2005; Datta et al. 2004), waiting on the Internet is likely to continue due to the tremen- dous growth in Internet users, an increase in multimedia content, and the density of Internet traffic (Dennis and Taylor 2006; Galletta et al. 2006; Nah 2002).

In response to these unavoidable delays, studies of online waiting have been conducted, with many of them focused on identifying tolerable waiting times. There is some consensus that users will wait 8 to 15 seconds for a basic web page to load (Galletta et al. 2004; Hoxmeier and DiCesare 2000; Nah 2002), but reasonable wait times are expected to vary with the nature of the online task (Ryan and Valverde 2006). While these findings provide guidance to website designers, the actual waiting time experienced by users is jointly determined by the size and complexity of the Web application, the con- figuration of Web servers and client computers, network bandwidth, and the infrastructure of the Internet itself (Rose and Straub 2001; Ryan and Valverde 2003). Thus, it is important to manage users’ online waiting experiences, which explains why many websites provide feedback to users during online waits. For example, many websites use indicator bars to show that a search is in progress (see Figure A1 in Appen- dix A), while others display a variety of visual content including pictures (see Figures A2–A4 in Appendix A), a flash of cartoon characters (see Figure A5 in Appendix A), or a clock (see Figure A6 in Appendix A), most often in addition to indicator bars. While research suggests that providing feedback and visual content can increase users’ tolerable waiting time (Nah 2004), theoretical guidance on the type of visual content that should be provided is limited (Ryan and Valverde 2003).

Time perception theories can inform research on managing online waits, as these theories explain how humans estimate the passing of time and suggest how perceptions of a wait might be improved. Models of memory (Block 1989; Orn- stein 1969) and attention (Brown 1997; Hicks et al. 1977) describe how the amount of information conveyed during a wait can fill an individual’s memory, making the wait seem longer, or distract an individual from the passing of time, making the wait seem shorter. The resource-allocation model (Zakay 1989) reconciles these two theoretical perspectives by

describing how humans estimate time in retrospect, after the wait, and prospectively, when they are alerted that they will be asked to judge the duration of the wait. In this study, these theories are applied to the context of online waiting to manage user perceptions of shorter and longer waits. We propose that a change of perspective occurs when waiting is prolonged based on the unique characteristics of the online waiting context, and test this extension to time perception theories in this research.

Specifically, we examine how online users’ perceptions of shorter or longer waits can be managed by providing different visual content. The marketing literature has shown that im- proving customers’ perceptions of the wait can be as effective as reducing the actual length of the wait in traditional waiting environments (Antonides et al. 2002; Katz et al. 1991; Pruyn and Smidts 1998), and that an assessment of wait time per- ceptions is critical for understanding customers’ evaluations of the service (Davis and Heineke 1998; Durrande-Moreau 1999; Katz et al. 1991; Pruyn and Smidts 1998). While most studies of online waiting have focused on the detrimental effects of waiting, there is growing research interest in how the online wait experience can be managed (Buell and Norton 2011; Lee et al. 2012). Based on time perception theories, we address the following research questions in this study: How does (1) the amount of visual content, and (2) the temporal nature of the visual content, affect online users’ perceptions of a wait? We further consider: (3) Does the length of the online wait change the effect that visual content has on per- ceptions of the wait? Two experimental studies are conducted to answer these questions.

The rest of this paper is organized as follows. We first describe time perception theories and our proposed theoretical extension in more detail, and then develop our research model based on a review of the relevant wait time literature in IS and marketing. Next, we describe two experimental studies, each with hypotheses, research design, data collection and analysis, and a discussion of the results. Finally, we conclude the paper with theoretical and empirical contributions, and sug- gest directions for future research.

Time Perception Theories

Time is a highly complex notion, as it lacks an obvious physical presence and there is no human organ that is respon- sible for counting time (Fraisse 1984). Numerous studies have been conducted by psychology researchers to understand how humans estimate time intervals (for reviews, see Block et al. 2010; Block and Zakay 1996, 1997). There are two main theoretical streams of time perception models that

384 MIS Quarterly Vol. 37 No. 2/June 2013

Hong et al./A Paradigm Shift Perspective for Managing Online Delay

explain how humans estimate time: memory-based and attention-based.

Memory-Based Models

Memory-based models hypothesize that the perceived dura- tion of an interval is based on the amount of the memory cues associated with the judged interval (Ornstein 1969; for reviews, see Block 1989, 2003). The more information cues that one remembers seeing during a time interval, the longer his or her estimation of the time. There are a number of fac- tors that can influence the amount of memory cues, including the amount and the complexity of the information presented (Ornstein 1969), the amount of change in cognitive contexts (Block and Reed 1978), and the degree to which an interval is segmented (Poynter 1989). These factors positively affect the amount of memory cues, which leads to higher duration estimates. First introduced by Ornstein (1969), memory- based models have been the most frequently cited explanation of duration judgment, and have received wide empirical support from many studies (e.g., Block 2003; Hicks 1992; Macar and Jackson 1992; Staddon 2005).

Attention-Based Models

While memory-based models have received significant empirical support, contradictory findings have also been reported such that a negative relationship between information complexity and duration estimates was found (e.g., McClain 1983; Zakay 1993). The conflicting results can be explained by attention-based models, which hypothesize a positive relationship between the amount of attention allocated to the passage of time and duration estimates (Brown 1997; Casini and Macar 1997; Hicks et al. 1977; Lejeune 1998). Attention- based models propose that there is a cognitive timer which people use to keep track of the passage of time. This timer demands attention resources for its operation. When addi- tional, nontemporal information needs to be processed during a time interval, an individual’s limited attention resources are divided between the processing of the nontemporal informa- tion and the cognitive timer, reducing the individual’s ability to track time. Thus, nontemporal information distracts the individual, directing attention away from the cognitive timer and reducing duration estimates.

Resource-Allocation Model (RAM)

While both memory-based models and attention-based models have received empirical support, these models make contra-

dictory predictions, with memory-based models predicting longer time estimates when more information is presented, and attention-based models predicting that more information should reduce time estimates. In an effort to reconcile the two streams of models, Zakay (1989) proposed the resource- allocation model (RAM). The basic assumption of RAM is a that human’s central attention has limited capacity and attention resources are allocated among different tasks at all times (Kahneman 1973). In the case of time estimation, atten- tion resources are divided between temporal and nontemporal information processing, which are denoted by P(t) and P(i), respectively. As central attention capacity is limited (Kahneman 1973), when more resources are allocated to non- temporal information processing, P(i), less attention is avail- able for temporal information processing, P(t), and vice-versa.

In addition, RAM differentiates between prospective and retrospective time estimations. In the prospective paradigm, “participants know in advance that they will be asked to judge the duration of a time period” and therefore would direct resources to estimate the passing of time (Block and Zakay 1997, p. 184). In the retrospective paradigm, “participants do not know until after a time period that they are being asked to judge its duration” and thus are only able to estimate time in retrospect, based on the retrieval of memory cues that were processed during the time interval (Block and Zakay 1997, p. 184). RAM proposes that in the prospective paradigm, P(t) dominates as subjects know that they will be asked to estimate the time interval later and will thus allocate more attention to processing temporal information. Thus, if subjects are given additional, nontemporal information to process in the prospec- tive setting, less attention resources are available for P(t), resulting in shorter time estimates because the timer is only active while attention is paid to P(t). This is consistent with the prediction made by attention-based models. On the other hand, in the retrospective paradigm, resources are not directed toward P(t) during the time interval, and when subjects are asked to estimate time in retrospect, they will rely on P(i) as the primary basis for time estimation. Thus, if subjects are given complex information to process in a retrospective setting, more attention is directed to P(i), resulting in longer time estimation. This is consistent with the prediction made by memory-based models.

In short, RAM provides a useful framework to reconcile two seemingly contradicting models of time estimation and provides guidance for managing online waiting perceptions through information load and content. Results of recent meta- analysis studies support the propositions of RAM, with attention-based models explaining prospective time estima- tion, and memory-based models explaining retrospective time estimation (Block et al. 2010; Block and Zakay 1997).

MIS Quarterly Vol. 37 No. 2/June 2013 385

Hong et al./A Paradigm Shift Perspective for Managing Online Delay

Paradigm Shift

When applying time perception theories to an online waiting context, the retrospective paradigm is particularly relevant because users typically will not be thinking about or esti- mating the wait time when they begin an online search. However, as the waiting continues, users may start wondering about the amount of time they have been waiting, as they consider whether the website is functioning or whether they should continue to wait. This focus on the passage of time falls within the prospective paradigm, suggesting that a paradigm shift may occur in an online waiting context. Most psychology research on time estimation has been conducted in either a pure prospective or retrospective setting through strong experimental manipulations, so it is unclear whether a paradigm shift (from retrospective to prospective) is possible during longer waits. We believe that the unique character- istics of the online waiting context and anecdotal evidence in the psychology literature provide support for a paradigm shift.

First, a key assumption of the retrospective paradigm is that “subjects have no advance knowledge that they will subse- quently be asked to make a duration judgment so they presumably do not deliberately attend to temporal informa- tion” (Block 1992, p. 142). However, psychology researchers have also noted that if a person has little information to process, or feels bored, a temporal motive may arise (Doob 1971). Some even suggest the possibility that “when one is waiting, time draws one’s attention and one becomes engaged in prospective timing” (Zakay and Block 1997, p. 16). In a meta-analytic review, Block and Zakay defined temporal motive as “the extent to which boring or repetitive conditions may have led participants in the retrospective paradigm to attend to time during the target duration (therefore making it more like the prospective condition)” (1997, p. 190). How- ever, the meta-analysis showed that “although temporal motive was a significant moderator [i.e., the effect on duration judgments marginally differed between the two paradigms], it was not an important one” (Block and Zakay 1997, p. 191), and duration judgments with or without temporal motive were similar within both paradigms. In short, prior research sug- gests that temporal motive might be a factor in duration judgment, but no research has explicitly proposed or empiri- cally tested for a paradigm shift.

Second, a potential paradigm shift is particularly relevant to an online waiting context, as no tasks are assigned to users while they are waiting. In most psychology experiments on time estimation, concurrent, active tasks are assigned to sub- jects while waiting, such as learning and recalling a list of terms (McClain 1983; Predebon 1996) or conducting a visual search (Brown 1997). While some type of feedback is

typically provided on a web page during an online wait, users are not required to perform any task or process the informa- tion provided. Thus, the temporal motive may be stronger in an online context, leading to a paradigm shift. In addition, the type of information displayed while online users are waiting is similar to the so-called “passive” stimuli referenced in psychology experiments, where subjects passively view different numbers or stimuli and then are asked to estimate time (Block et al. 2010; Hicks et al. 1976). Psychology researchers believe that the cognitive processes involved when viewing passive, rather than active, stimuli during a wait are more complex, and a recent meta-analytic review explicitly excluded studies with passive stimuli in order to control the cognitive load experienced by subjects (Block et al. 2010). We believe that a paradigm shift is more likely to occur for subjects facing passive stimuli, as they are more likely to think about the wait.

Third, a paradigm shift is more likely to occur when the infor- mation provided during a wait arouses a temporal motive. For example, an apology for the wait or a picture of a clock may inadvertently remind users of the fact that they are waiting, and arouse a temporal motive. To our knowledge, this possi- bility has not been examined in psychology research, probably because subjects were typically given nontemporal tasks to perform while waiting. But for website designers, there are relatively few guidelines on how to select feedback information and visual content displayed during a wait, and as a result, the selection may remind users of the passing time. Thus, the website design may arouse a temporal motive, leading to a paradigm shift.

Research Model

We reviewed the IS and marketing literature on how infor- mation provided during a wait influences waiting time perceptions in both online and traditional service environ- ments, as summarized in Table B1 (in Appendix B). The literature is further categorized by the valence of the findings, depending on whether providing additional information or stimuli during the wait has no effect, positive effect, negative effect, or mixed effect on users’ perceptions of the wait.

The literature suggests that there are two key wait time perceptions: (1) the perceived quickness or duration of the wait, and (2) affective response to the wait (Baker and Cameron 1996; Houston et al. 1998; Hui et al. 1998; Hui and Tse 1996; McGuire et al. 2010). The perceived quickness or duration of the wait is the cognitive aspect of the waiting experience which captures customers’ loss of a very valuable

386 MIS Quarterly Vol. 37 No. 2/June 2013

Hong et al./A Paradigm Shift Perspective for Managing Online Delay

asset: time. The affective response to the wait is the emo- tional aspect of the waiting experience, capturing the stress customers experience during the wait.

In traditional service environments (e.g., waiting on the tele- phone or in a bank), empirical studies have proposed that providing temporal information, such as waiting duration information (Chebat et al. 2010; Groth and Gilliland 2006) and queuing information (e.g., Antonides et al. 2002; Whiting and Donthu 2006), and nontemporal information, such as music (Cameron et al. 2003; Hui et al. 1997), TV (Pruyn and Smidts 1998), and an electronic newsboard (Katz et al. 1991), are effective tools for managing consumer wait perceptions. In an online environment, a moving indicator bar (e.g., Geel- hoed et al. 1995; Myers 1985; Nah 2004) is commonly provided along with textual or graphical content, often referred to as “fillers” (Taylor 1995), to manage online wait perceptions. Research on indicator bars suggests that such interface elements provide valuable feedback to the user (Nah 2004), but only recently has the impact of other visual content, such as images and text, been examined. For example, Lee et al. (2012) found that providing filler infor- mation increased user attention to the web page, which in turn reduced the perceived wait time. However, they did not differentiate between the prospective and retrospective paradigm, or shorter and longer waits, and their investigation was limited to text and static images. Hence, a more theoretically grounded examination of different fillers (e.g., flash and video, temporal versus nontemporal) and varied wait times would be beneficial.

Finally, Table B1 (in Appendix B) shows that findings related to the provision of information during the wait are sur- prisingly inconsistent across different studies or even within a single study. Providing waiting related information was found to have no effect (Groth and Gilliland 2006; Whiting and Donthu 2006), a positive effect (Dellaert and Kahn 1999; Lee et al. 2012; Myers 1985), a negative effect (Chebat et al. 2010), or mixed effects (Antonides et al. 2002; Harrison et al. 2007; Hui and Tse 1996; Hui and Zhou 1996; Katz et al. 1991) on users’ waiting experience. These inconsistent results can be attributed to a number of factors. One factor is the limitations of RAM, as RAM does not theorize that the retrospective paradigm may change as the wait lengthens, or how RAM applies in waiting contexts with passive stimuli. Another factor is that these studies did not necessarily consider the amount of information conveyed or the temporal nature of the information. Finally, many of these studies did not consider a theoretically grounded model, such as RAM, for support. There is also evidence suggesting that these inconsistent findings might be due to the length of the actual

wait time (Antonides et al. 2002; Hui and Tse 1996; Katz et al. 1991). In other words, users may react to feedback infor- mation differently when facing waits of different lengths. Hence, more research is needed to examine how providing different information during different wait times affects online users’ perceptions of the wait.

According to resource-allocation theory, the (1) amount of information and the (2) direction of attention are the key factors that can influence perceptions of the wait, and the effects of these factors can change over time (Block et al. 2010; Zakay 1989; Zakay and Block 1997). Based on time perception theories and our literature review, we propose the research model depicted in Figure 1. Specifically, we identify amount of information (i.e., visual content), direction of attention (i.e., toward and away from time), and actual wait time (shorter and longer) as the key independent variables; and perceived quickness of the wait (PQW) and negative affect toward the wait (NAW) as the two key dependent variables. Table 1 summarizes the two experimental studies conducted to test the proposed model. Study 1 examines the effect of adding visual content that either directs attention away from or toward time on online users’ perceptions of the wait during shorter and longer waits. Study 2 focuses on longer waits, and examines the effects of increasing the amount of information when attention is directed away from or toward time.

Study 1

Study 1 examines how providing visual content directing attention away from or toward time affects online users’ perceptions of the wait with shorter and longer waits.

Study 1: Hypotheses Development

There is a general consensus that the actual waiting time significantly impacts the perceived wait duration (Dabholkar and Sheng 2008; Hornik 1981). As wait time increases, users will find the website to be slower. Hence, we hypothesize that actual waiting time reduces PQW (H1a). Further, waiting can be detrimental to users not only because of economic costs from time loss but also because of psychological costs due to stress and frustration (Osuna 1985). As a result, waiting is often described as frustrating, annoying, and aggravating (Dubé et al. 1991; Gardner 1985; Katz et al. 1991). Thus, we hypothesize that an increase in actual waiting time leads to NAW (H1b).

MIS Quarterly Vol. 37 No. 2/June 2013 387

Hong et al./A Paradigm Shift Perspective for Managing Online Delay

Figure 1. Research Model

Table 1. Summary of Experimental Studies

Study 1 Study 2

Experiment Design 2 × 3 Between-subject 2 × 2 Between-subject

Focus Examined the effects of adding visual content that either directs attention away from or toward time on online users’ perceptions of the wait during shorter and longer waits.

Examined the effects of increasing the amount of information when attention is directed away from or toward time on online users’ perceptions of the wait during longer waits.

Independent Variables

• Wait time (short/long) • Amount of information (no picture/with

picture) • Direction of attention (travel/clock picture)

• Amount of information (a static picture/a flash series of pictures)

• Direction of attention (travel/clock pictures)

Dependent Variables

• Perceived quickness of the wait • Negative affective toward the wait

• Perceived quickness of the wait • Negative affective toward the wait

Control Variables • Impatience • Attribution

• Impatience • Attribution

Analysis Method MANOVA MANOVA

Number of Subjects 207 139

H1a: Actual waiting time reduces the perceived quickness of the wait.

H1b: Actual waiting time increases negative affect toward the wait.

In a typical online waiting environment, users’ time estima- tion begins in the retrospective paradigm, as they are focused on their online task instead of estimating the wait time. According to RAM (Zakay 1989), memory-based models dominate in the retrospective paradigm and predict that providing visual content during the wait will give users more memory cues, leading to longer estimates of the wait (Ornstein 1969). Hence, providing visual content, such as a picture, to fill the wait may have the unintended consequence of making the wait seem longer. We propose that during shorter waits, providing visual content reduces PQW (H2a).

The provision of visual content also influences affect toward the wait. On one hand, while time perception theories do not specifically address affect, a longer wait is likely to result in negative feeling among users (Houston et al. 1998; Katz et al. 1991). On the other hand, prior research in traditional waiting environments has shown that providing something for users to look at while they are waiting helps to reduce boredom, lower stress, and make the waiting experience more inter- esting (Katz et al. 1991; Whiting and Donthu 2006). Thus, providing a picture for users to look at while waiting online (assuming that the picture was not distressing or disturbing) should improve their general affective state. These two potential effects are in opposite directions and could counter- act one another (i.e., a filler can either make the wait feel longer, which increases negative affect, or make the wait more interesting and reduce negative affect). In reviewing the prior literature, we found that the majority of findings showed

388 MIS Quarterly Vol. 37 No. 2/June 2013

Hong et al./A Paradigm Shift Perspective for Managing Online Delay

a positive effect from providing a filler, such that it improved users affective response toward the wait regardless of whether the filler makes the wait feel longer (Hui et al. 1997), shorter (Cameron et al. 2003), or neither longer nor shorter (Hui and Zhou 1996). Hence, we hypothesize that during shorter waits, providing visual content (e.g., a picture) reduces NAW (H2b).

H2a: During shorter waits, providing visual content (e.g., a picture) reduces the perceived quickness of the wait.

H2b: During shorter waits, providing visual content (e.g., a picture) reduces the negative affect toward the wait.

When waiting is prolonged, we expect users to begin focusing on time, in line with a prospective approach. According to RAM (Zakay 1989), attention-based models dominate in the prospective paradigm, which predicts that providing addi- tional content may distract the user from processing time, reducing wait time perceptions. Thus, we propose that during longer waits, providing visual content (e.g., a picture) in- creases PQW (H3a).

The effect of providing visual content on users’ affective response to the wait is more straightforward in longer waits. First, if providing visual content increases PQW as hypothe- sized, users should feel more positive toward the wait. Meanwhile, providing viewing content while waiting also helps make the waiting experience more interesting, espe- cially in longer waits when users are more likely to get bored. Hence, we propose that during longer waits, providing visual content (e.g., a picture) reduces NAW (H3b).

H3a: During longer waits, providing visual content (e.g., a picture) increases the perceived quickness of the wait.

H3b: During longer waits, providing visual content (e.g., a picture) reduces negative affect toward the wait.

Furthermore, the nature of the visual content may have an influence on users’ perceptions of the wait, depending on whether it directs users’ attention away from or toward time. RAM proposes that attention resources are divided between temporal and nontemporal information processing, which are denoted by P(t) and P(i) respectively. While the dominance of P(t) or P(i) in a particular waiting scenario is primarily determined by the paradigm of estimation (prospective versus retrospective), temporal visual content provided during the wait may inadvertently direct users’ attention to time, or P(t). For example, images of clocks or clock components, such as

the hands of a clock (e.g., DiClemente and Hantula 2003), may cause users to think about how long they have waited and therefore direct more attention to P(t). This inadvertent focus on P(t) may help explain the mixed results reported in the wait time literature, where temporal cues (e.g., duration information, queuing, and actual clocks) have been provided to reduce users’ anxiety and frustration as they wait.

In shorter waits, users’ time estimation begins in the retro- spective paradigm, with dominance of P(i). However, the provision of temporal information may direct more attention to P(t). The increased attention resources allocated to P(t) are likely to increase duration estimates. Hence, we hypothesize that during shorter waits, directing attention away from time results in higher PQW than directing attention toward time (H4a). From the affective perspective, a filler that directs attention to the passage of time is more likely to annoy users than a filler that directs attention away from time. Hence, we hypothesize that during shorter waits, directing attention away from time results in lower NAW than directing attention toward time (H4b).

H4a: During shorter waits, directing attention away from time results in higher perceived quickness of the wait than directing attention toward time.

H4b: During shorter waits, directing attention away from time results in lower negative affect toward the wait than directing attention toward time.

In longer waits, the dominance of P(t) and attention-based models predicts that PQW can be improved if the visual content can distract users from the passage of time, which is unlikely to occur when the content is temporal in nature. Hence, as with shorter waits, we propose that during longer waits, directing attention away from time results in higher PQW than directing attention toward time (H5a). From the affective perspective, when users are more likely to become bored and think about the passage of time, providing visual content that directs users’ attention away from time should help to reduce stress and make the waiting experience more interesting. Hence, as with shorter waits, we hypothesize that during longer waits, directing attention away from time results in lower NAW than directing attention toward time (H5b).

H5a: During longer waits, directing attention away from time results in higher perceived quickness of the wait than directing attention toward time.

H5b: During longer waits, directing attention away from time results in lower negative affect toward the wait than directing attention toward time.

MIS Quarterly Vol. 37 No. 2/June 2013 389

Hong et al./A Paradigm Shift Perspective for Managing Online Delay

Study 1: Experiment Procedure

We conducted a lab experiment with a 2 (actual wait time: short/long) by 3 (visual content: none, away from time, toward time) full-factorial between-subject design. Subjects were undergraduate business majors recruited from a large public university in the United States. Similar to other online waiting research (e.g., Galletta et al. 2004, 2006; Nah 2004), we consider the use of students subjects to be appropriate because students frequently use websites, and because we are examining a basic psychological phenomenon (i.e., web delay) to which people should react similarly. Subjects received course credit for participating. Those choosing not to participate were provided the opportunity to earn course credits by completing an alternative assignment. In addition, subjects were told that if they “thoughtfully” completed the study they would be eligible to receive a $10 gift card. Over 20 gift cards were distributed after the experiment.

Experimental Website and Task. An experimental website was developed for the study. We chose a travel website because travel is of general interest to our subjects and wait time on a travel website may vary depending on the com- plexity of the search task (Nah 2004; Ramsay et al. 1998; Teal and Rudnicky 1992), with longer waits expected for more complex online tasks (Ryan and Valverde 2006). For example, Lee et al. (2012) found an average wait time of 16.5 seconds for a basic flight search, while Figure A4 in Appendix A shows a travel website with a disclosed wait time of up to 90 seconds.

We also carefully balanced the need for ecological validity in the experiment and the need to compare relatively longer and shorter waits (as a paradigm shift is only expected during longer waits). We provided subjects with a cover story that they were evaluating a new prototype travel website, which utilized advanced data mining techniques to recommend less well-known, exotic travel destinations based on users’ travel and personal interests. As compared to basic flight searches, it is reasonable to expect that searches using advanced data mining technologies that algorithmically match travel destinations with one’s interests and requirements will take longer to complete. Moreover, offering a relatively new type of service reduces any possible anchoring bias to prior experiences with other travel sites. In summary, this design provides the high internal validity required for applying and testing theory, while preserving a reasonable level of ecological validity (Calder et al. 1981).

Pilot Study. We conducted a pilot study with 121 subjects to identify appropriate waiting times for the experimental web- site. We preselected eight delay intervals from 5 to 75

seconds based on a review of the literature. We employed a within-subject design using a similar cover story, and asked subjects to evaluate the speed of eight servers used by the website. The eight servers performed the same search func- tions but offered different search speeds and provided different fillers. Each subject was therefore exposed to eight delay intervals and eight types of feedback, randomized using an eight by eight Latin Square Design (Kirk 1995). Based on subjects’ responses, we selected two intervals to be used in the experiment (i.e., 10 seconds for the short wait, and 45 seconds for the long but still acceptable wait).

Variables Manipulated. Actual waiting time was manipulated as short and long (10 and 45 seconds). On the waiting page, a simple line of text (i.e., “Please wait while SmartTravel.com finds travel packages for you!”) was displayed (see Figure 2a) along with a fading-dots indicator bar at the bottom of the screen. The bar did not provide any indication of the progress of the download, and only showed that the website was still working on the user’s request. The indicator bar was included across all treatments to maintain realism (ecological validity) and to provide essential feedback to the users that the search was in process. In the control condition, only the line of text and the indicator bar were included (see Figure 2a). In terms of visual content, a number of travel and clock pictures were reviewed by the three researchers and a few MS MIS students with web design experience. A travel picture of a plane in flight (see Figure 2b) and an animated clock picture with moving hands (but otherwise static, see Figure 2c) were selected as it was believed that a plane in flight would remind users of travel and therefore direct attention away from time, while a “ticking” clock would direct attention toward time.

Variables Measured. All dependent variables were measured using existing scales from published studies whenever possible (see Appendix C). Specifically, PQW was measured using an established three-item semantic differential scale (Gorn et al. 2004).2 NAW was measured by a four-item, seven-point Likert-type scale taken from existing instruments (Hui et al. 1997; Hui and Tse 1996). Two control variables, impatience and attribution, were also measured. Impatience is identified as one of the most important time-related individual difference factors that may impact time estimation

2PQW, rather than a quantitative estimate of the number of seconds that subjects waited, was chosen for two reasons. First, PQW has managerially relevant consequences (such as attitudes and intentions about the website), and more substantive implications than an estimate of the number of seconds elapsed (Gorn et al. 2004). Second, a specific quantitative estimate may mean different things to different subjects. For example, a perceived wait of 10 seconds may seem short for one subject, but long for another. So, what ultimately matters is not the quantitative estimate, but how quick users perceive the download to be.

390 MIS Quarterly Vol. 37 No. 2/June 2013

Hong et al./A Paradigm Shift Perspective for Managing Online Delay

(a) Study 1: Control Condition (b) Studies 1 & 2: A Static Travel Picture

(c) Study 1: A Clock Picture with Moving Hands (d) Study 2: A Static Clock Picture

(e) Study 2: A Flash of Static Travel Pictures (f) Study 2: A Flash of Static Clock Pictures

Figure 2. Waiting Page Displays

MIS Quarterly Vol. 37 No. 2/June 2013 391

Hong et al./A Paradigm Shift Perspective for Managing Online Delay

(Francis-Smythe and Robertson 1999), and was measured with three items taken from an existing instrument (Spence et al. 1987). Attribution of the website delay (i.e., whether the subjects attribute the delay to the website features) has also been shown to influence online service evaluations (Houston et al. 1998; Rose et al. 2005; Taylor 1994), and was measured using two items (Houston et al. 1998; Rose et al. 2005). Finally, we asked subjects about the amount of visual content they saw on the waiting page using four self-developed items (which were pretested using a small pool of subjects) as a manipulation check for visual content.

Measurement Validation. Appendix D reports the results of an exploratory factor analysis that contains combined data from both studies for cleaner presentation. The variables demonstrated adequate reliability, with Cronbach alphas of 0.98 for PQW, 0.95 for NAW, 0.71 for impatience, 0.76 for attribution, and 0.98 for visual content, all above the recom- mended level of 0.70 for field research (Nunnally 1978). Correlations among the indicators intended to measure the same variable were all higher than their correlations with indicators measuring the other variables, providing evidence of both convergent and discriminant validity (Campbell and Fiske 1959). A principle component analysis with oblimin rotation showed that all indicators loaded higher on the vari- ables they were designed to measure than on other variables. All factor loadings exceeded the recommended threshold of 0.50 with most greater than 0.70 and differences of 0.30 or greater between loadings and cross-loadings (Hair et al. 2006). In summary, the scales demonstrate adequate psycho- metric properties.

Subjects and Procedure. A total of 207 subjects completed the experiment, with 65.2 percent being male, and with an average age of 21.97. On average, the subjects had 11.39 years of Internet experience, and spent 19.21 hours online per week. Subjects were randomly assigned to one of the six treatment conditions (see Table 2). A series of MANOVA tests showed that age (F = 0.239, p = 0.787), gender (F = .881, p = 0.416), years of Internet experience (F = 0.030, p = 0.970), or hours spent on the Internet per week (F = 0.732, p = 0.482), did not affect the dependent variables.

Subjects were provided with a task sheet that contained the cover story. They were then directed to the experimental website, where they were first asked to answer questions about their preferences for travel destinations and vacation activities (see Figures E1 and E2 in Appendix E for input pages). After clicking the search button, participants were taken to the waiting page. Upon expiration of the treatment wait time, the subjects were redirected to an online survey that measured PQW, NAW, attribution, and visual content.

Note that subjects’ perceptions of time were not confounded by the display of the travel recommendations, as they were asked to assess the wait before any recommendations were presented. Upon completion of the questionnaire, subjects were presented with travel recommendations provided by the experimental website. After reading the recommendations and clicking the continue button, subjects were redirected to a second survey where they were asked to respond to ques- tions measuring general impatience and demographic ques- tions such as gender, age, and Internet experience. Subjects were then debriefed and let go.

Study 1: Data Analysis Results

Manipulation and Control Checks. A one-way ANOVA showed that subjects considered the treatments with pictures as having significantly more visual content than the control treatments (F = 5.169, p = 0.024). Unexpectedly, users con- sidered the travel picture to have significantly more visual content than the clock picture (F = 39.019, p = 0.000). We address the implications of this finding in light of the Study 1 results in the discussion section.

A manipulation check for the direction of attention was per- formed using a separate study to avoid subjects guessing the purpose of the study. We collected data from 37 subjects using a between-subject design, and confirmed that a clock picture directs attention toward time (F = 36.499; p = 0.000) while a travel picture directs attention away from time (F = 15.242, p = 0.000).3 In addition, NAW was relatively low (between 2.95 ~ 4.37 on a scale of 1 to 7 across treatment conditions in longer waits), indicating that a 45 second wait was reasonable.

Impatience was not significantly related to PQW (F = 0.037, p = 0.848), but was significantly related to NAW (F = 5.852, p = 0.016). Attribution was significantly related to both perceived quickness (F = 63.241, p = 0.000), and negative affect (F = 52.990, p = 0.000), indicating that subjects attri- bute the delay to the provision of graphic information during the wait. Impatience and attribution were the only control or demographic variables that influenced wait time perceptions and thus were included in the MANOVA used for hypotheses testing in Study 1.

3A static clock picture was compared against a static travel picture (see Figures 2b and 2d), which is a more conservative test than comparing the clock picture with moving hands with a travel picture. The measures (on a five point scale) were “Do you think this picture made you think more or less about (1) the fact that you were waiting, (2) the travel recommendations?”

392 MIS Quarterly Vol. 37 No. 2/June 2013

Hong et al./A Paradigm Shift Perspective for Managing Online Delay

Table 2. The Assignment of Subjects and Descriptive Statistics

Study 1

Actual Wait Time

Short Long

Amount of Information (i.e., Visual Content)

No Picture

Number of Subjects 36 36

Perceived Quickness of the Wait 5.111 (1.135)* 2.269 (1.146)

Negative Affect toward Wait 1.750 (1.163) 4.368 (1.051)

With Picture

Away from Time (Travel)

Number of Subjects 38 35

Perceived Quickness of the Wait 4.508 (1.373) 2.848 (1.232)

Negative Affect toward Wait 1.868 (1.199) 3.357 (1.367)

Toward Time (Clock)

Number of Subjects 33 29

Perceived Quickness of the Wait 4.394 (1.526) 3.092 (1.241)

Negative Affect toward Wait 1.909 (1.531) 2.948 (1.378)

Study 2

Actual Wait Time = Long

Amount of Information

A Static Picture

A Flash of Pictures

Direction of Attention (i.e., toward and away from time)

Away from Time (Travel)

Number of Subjects 35 36

Perceived Quickness of the Wait 2.848 (1.232) 3.111 (1.163)

Negative Affect toward Wait 3.357 (1.367) 2.604 (1.341)

Toward Time (Clock)

Number of Subjects 38 30

Perceived Quickness of the Wait 2.649 (1.062) 1.889 (0.760)

Negative Affect toward Wait 3.322 (1.517) 4.175 (1.344)

*Mean (Standard Deviation). Study 2 results are included here and discussed later for ease of comparison with Study 1.

Hypotheses Testing. Table 2 reports the descriptive statistics of the two wait perception measures by treatment. A 2 (actual wait time: short, long) by 2 (amount of information: control, picture treatments)4 MANOVA revealed significant results for the main effects of actual waiting time (F = 52.339, p = 0.000), visual content (F = 5.010, p = 0.008), and the inter- action effect between wait time and visual content (F = 6.998, p = 0.001). Following Brambor et al. (2006), we used follow- up ANOVAs to test the effects on each dependent variable. Table 3 presents the results of the MANOVA and subsequent

ANOVA tests,5 while Figures 3a-b depict the interaction effects.

Consistent with H1a-b, actual waiting time significantly reduced perceived quickness (F = 72.527, p = 0.000) while increasing negative affect (F = 56.171 p = 0.000). The results generally support the moderating effect of wait time on the relationship between visual content and perceptions of the wait. In shorter waits, providing pictures significantly re- duced PQW (F = 5.688, p = 0.019), supporting H2a. How- ever, H2b was rejected as providing pictures did not reduce the NAW (F = 0.269, p = 0.605). In longer waits, providing pictures increased PQW (F = 7.586, p = 0.007) and reduced NAW (F = 20.442, p = 0.000). Hence, H3a–b were supported.

4A 2 (actual wait time: short vs. long) by 3 (amount of information: control versus travel picture versus clock picture) MANOVA analysis generated very similar results, with significant main effects of actual wait time (F = 45.735, p = 0.000), visual content (F = 2.549, p = 0.039), and the interaction effect between actual wait time and visual content (F = 3.395, p = 0.010). We chose to report the 2 by 2 MANOVA analysis results in the main paper because it allows us to test H1–H3 directly.

5Given the potential for family-wise (Type 1) error, we considered how correcting for multiple dependent variables (PQW and NAW) and multiple tests in Study 1 would influence our results. In Study 1, we divided alpha by 4 (.05/4 = .0125), given two dependent variables and two tests as shown in Table 4 (2 × 2), and found that all results remained significant (F1,201 > 6.35).

MIS Quarterly Vol. 37 No. 2/June 2013 393

Hong et al./A Paradigm Shift Perspective for Managing Online Delay

Table 3. Overall MANOVA and Individual ANOVAs Results

Factors

Overall MANOVA

Individual ANOVAs

PQW NAW

F Sig. η2 F Sig. η2 F Sig. η2

Study 1 Main Results Adjusted R2 = .532 Adjusted R2 = .500 Actual Waiting Time 52.339 0.000*** 0.344 72.527 0.000*** 0.265 56.171 0.000*** 0.218

Amount of Information 5.010 0.008** 0.048 0.029 0.864 0.000 9.290 0.003** 0.044

Actual Wait Time × Amount of Information

6.997 0.001*** 0.065 8.642 0.004** 0.041 8.606 0.004** 0.041

Study 1 Additional Results Adjusted R2 = 0.472 Adjusted R2 = 0.403 Actual Waiting Time 16.732 0.000*** 0.207 23.887 0.000*** 0.156 15.799 0.000*** 0.109

Direction of Attention 0.264 0.768 0.004 0.511 0.476 0.004 0.000 0.990 0.000

Actual Wait Time × Direction of Attention

0.048 0.953 0.001 0.011 0.916 0.000 0.095 0.758 0.001

Study 2 Adjusted R2 = 0.216 Adjusted R2 = 0.444 Amount of Information 0.869 0.422 0.013 1.749 0.188 0.013 0.116 0.734 0.001

Direction of Attention 11.274 0.000*** 0.146 14.270 0.000*** 0.097 13.565 0.000*** 0.093

Amount of Information × Direction of Attention

4.511 0.013* 0.064 4.567 0.034* 0.033 6.521 0.012* 0.047

Note: *p < 0.05; **p < 0.01; ***p < 0.001. Study 2 results are included here for ease of comparison with Study 1.

We performed additional analyses to examine how the temporal nature of the picture may affect the above results. In shorter waits, providing either a travel or a clock picture reduced the PQW (F = 4.205, p = 0.044 for the travel picture; F = 4.959, p = 0.029 for the clock picture), but did not influ- ence NAW (F = 0.186, p = 0.668 for the travel picture; F = 0.239, p = 0.627 for the clock picture), as compared to the control condition. In longer waits, providing either a travel or a clock picture increased the PQW (F = 4.208, p = 0.044 for the travel picture; F = 7.705, p = 0.007 for the clock picture), and reduced NAW (F = 12.242, p = 0.001 for the travel picture; F = 22.205, p = 0.000 for the clock picture), as compared to the control condition. Hence, the results were consistent with the overall results reported above.

As the effects of providing visual content are largely con- firmed, we proceeded with a detailed comparison between the effects of providing travel versus clock pictures in shorter and longer waits (see Table 3 and Figures 3c–d). In shorter waits, providing a travel or a clock picture resulted in similar levels of PQW (mTravel = 4.51, mClock = 4.39; F= 0 .111, p = 0.740, η

2

= 0.002) and NAW (mTravel = 1.87, mClock = 1.91; F = 0.016, p = 0.901, η2 = 0.000), rejecting H4a-b. In longer waits, similar results were observed such that directing attention away from time or towards time resulted in similar levels of PQW (mTravel = 2.85, mClock = 3.09; F = 0.620, p = 0.434, η

2 = 0.010) and NAW (mTravel = 3.36, mClock = 2.95; F = 1.408, p = 0.240, η

2 =

0.022). Thus, H5a–b were also rejected.6

Study 1: Discussion

The results of study 1 confirm the detrimental effects of longer waits, consistent with findings in the extant literature. More importantly, the Study 1 findings (H2a, H3a, and the additional analysis in particular) provide support for the proposed paradigm shift. Providing additional visual content (e.g., a picture) in shorter waits made users feel that the website was slower, which is consistent with the retrospective paradigm or memory-based models (where pictures serve as memory cues). This result is particularly interesting con- sidering that the amount of information contained in our pictures was rather limited. Nevertheless, by increasing the amount of memory cues, the pictures made users feel that the

6The effect sizes (η2, partial eta squared) shown in Table 3 and the analysis above suggest that the effect sizes related to the unsupported hypotheses, H4a–b and H5a–b, are small (with partial eta squared, small effects = 0.01, medium effects = 0.06, and large effects = 0.14). While an increase in sample size might bring some of these non-significant results closer to significance, the effect size would remain small, limiting the practical significance of such effects.

394 MIS Quarterly Vol. 37 No. 2/June 2013

Hong et al./A Paradigm Shift Perspective for Managing Online Delay

(a) Study 1: Actual Wait Time × Amount of Information on PQW (b) Study 1: Actual Wait Time × Amount of Information on NAW

(c) Study 1: Actual Wait Time × Direction of Attention on PQW (d) Study 1: Actual Wait Time × Direction of Attention on NAW

Figure 3. Interaction Effects of Wait Time

wait was longer. On the other hand, providing the same pic- tures in longer waits made users feel that the website was faster, which is consistent with the prospective paradigm or attention-based models (where pictures serve as a distracter from time). The benefit of providing visual content during longer waits is further confirmed by the finding of reduced negative affect toward the wait.

The expected benefit of visual content reducing negative affect in shorter waits was not confirmed, most likely because of mitigating effects from the increased content, making the wait feel longer. A closer examination of the manipulation of fillers in the prior literature shows that studies that found a positive influence on affect typically used more engaging, complex treatments, such as music (Cameron et al. 2003; Hui et al. 1997), TV (Pruyn and Smidts 1998), or an electronic newsboard (Katz et al. 1991). Our manipulation of simple pictures is probably less engaging, and thus not strong enough

to mitigate the negative affect resulting from the perception of a prolonged wait.

A surprising finding is that there were no differences in wait perceptions for direction of attention (away from/toward time), either in shorter or longer waits. The direction of atten- tion manipulation was significant, as previously discussed, confirming that a clock picture directs attention toward time and thus should make the wait feel longer than a travel picture. A possible explanation is that the insignificant result was due to the unexpected difference in the amount of visual content between the travel picture and the clock picture. In order to test this possibility, we collected additional data from 37 subjects using a new treatment (a 10 second wait with attention directed toward time) that provided greater visual content. We presented a series of static clock pictures during a 10 second wait and found that the perceived content (amount of information) was then similar to the treatment

MIS Quarterly Vol. 37 No. 2/June 2013 395

Hong et al./A Paradigm Shift Perspective for Managing Online Delay

presenting a static travel picture (mTravel = 3.39, mClock = 3.82; F = 2.032, p = 0.158, η2 = 0.028). The results showed that the new clock treatment and the existing travel picture treatment were not significantly different in PQW (F = 0.000, p = 0.998, η2 = 0.000) or NAW (F = 0.365, p = 0.548, η2 = 0.005), confirming the insignificant result of H4a–b. In short, while the temporal nature of the visual content may have directed more attention to time, the increased attention to P(t) did not offset the dominance of P(i) in shorter waits.

In longer waits, if the paradigm shift occurs and the attention model or P(t) dominates, in theory, the amount of information should have no impact, as memory cues only affect P(i). Given that our manipulation check confirmed that the clock picture attracts more attention to time as compared to the travel picture, we speculate that the insignificance of H5 is due to the diminishing effect of visual content over time. Considering that both the travel and clock picture were simple, users may complete their processing of the content fairly quickly. As the wait continues with no additional content provided after the initial picture, users may simply revert to focusing on P(t) after processing the content. In other words, if users are already focused on P(t) and if the distracting effect of a picture weakens over time (assuming a prospective paradigm), the difference between a travel picture and a clock picture is minimized in longer waits. Hence, it would be interesting to see if increasing the amount of temporal and nontemporal information over time strengthens the effects of direction of attention, and leads to the expected results. We designed a second study to examine these possibilities.

Study 2

In Study 2, we focused on longer waiting conditions only, as this is when the paradigm shift from a memory-based to an attention-based model occurs. The amount of visual content was increased to investigate whether insignificant differences for direction of attention could be due to the lower ratio of visual content per unit of time as the wait time lengthens. In other words, the ratio of visual content per unit of time is lower when displaying a single picture for 45 seconds as compared to displaying multiple, different pictures during the same wait.

Study 2: Hypotheses Development

The effects of directing attention away from time may be influenced by the cognitive load associated with processing visual content. Cognitive load is defined as “the amount of

information-processing (especially attentional or working- memory) demands during a specified time period” (Block et al. 2010, p. 331) and can be varied on a number of stimulus attributes, including complexity, difficulty, quantity, or the different contexts presented during a time interval. With simple stimuli (i.e., lower cognitive load), there is less dis- tracting information per unit of time, and thus users have more attention for temporal processing, resulting in waits that seem longer. However, increasing the number of nontem- poral stimuli shown during a time period decreases the per- ceived wait time (Predebon 1996). Thus, we hypothesize that as the amount of visual content directing users’ attention away from time increases (e.g., more distracting pictures), the wait will seem shorter (H6a). Perceptions of a shorter wait should make users feel more positive about the waiting experience. Further, having more distracting information to look at while waiting is more likely to keep users from getting bored, resulting in more positive affect toward the wait. Hence, we hypothesize that increasing visual content that directs users’ attention away from time should decrease NAW (H6b).

H6a: During longer waits, greater visual content directed away from time (e.g., a series of travel pictures) increases the perceived quickness of the wait as compared to less visual content (e.g., a travel pic- ture) directed away from time.

H6b: During longer waits, greater visual content directed away from time (e.g., a series of travel pictures) reduces negative affect toward the wait as compared to less visual content (e.g., a travel picture) directed away from time.

On the other hand, if the visual content directs users’ attention toward time, we should expect the opposite to occur. As more information is provided that reminds users of the passage of time, P(t) becomes more dominant, and longer perceptions of the wait should result. In other words, the amplifying effect of greater visual content will make the download seem slower when this visual content directs the user toward temporal processing. Hence, we hypothesize that when visual content directs users’ attention toward time (e.g., more time-related pictures), increasing the amount of visual content decreases the PQW (H7a). Similarly, perceptions of a longer wait should make users feel more negative about the waiting experience. In addition, providing more time-related information is likely to remind users of the wait, increasing the possibility that they will be annoyed by the wait. Hence, we hypothesize that when visual content directs users’ atten- tion toward time, increasing the amount of visual content leads to more NAW (H7b).

396 MIS Quarterly Vol. 37 No. 2/June 2013

Hong et al./A Paradigm Shift Perspective for Managing Online Delay

H7a: During longer waits, greater visual content directed toward time (e.g., a series of clock pictures) de- creases the perceived quickness of the wait as com- pared to less visual content (e.g., clock picture) directed toward time.

H7b: During longer waits, greater visual content directed toward time (e.g., a series of clock pictures) in- creases negative affect toward the wait as compared to less visual content (e.g., clock picture) directed toward time.

With H6a–b addressing the effects of greater visual content directed away from time, and H7a–b addressing the effects of greater visual content directed toward time, we can now combine them to make a direct comparison between directing attention away from or toward time. This effort also replicates H5 in Study 1, but differentiates between less visual content in H8 (e.g., a picture) and greater visual content in H9 (e.g., a series of pictures). Following the findings in Study 1, we hypothesize that less visual content directed away from time results in similar PQW (H8a) and similar NAW (H8b) as less visual content directed toward time.

H8a: During longer waits, less visual content (e.g., a pic- ture) directed away from time results in similar perceived quickness of the wait as less visual content directed toward time.

H8b: During longer waits, less visual content (e.g., a pic- ture) directed away from time results in similar negative affect toward the wait as less visual content directed toward time.

When the amount of visual content increases, however, we expect to see greater differences in wait perceptions for direction of attention (away from or toward time), following H6a–b and H7a–b. Hence, we hypothesize that greater visual content directed away from time results in higher PQW (H9a) and lower NAW (H9b) than greater visual content directed toward time.

H9a: During longer waits, greater visual content (e.g., a series of pictures) directed away from time results in higher perceived quickness of the wait than greater visual content directed toward time.

H9b: During longer waits, greater visual content (e.g., a series of pictures) directed away from time results in lower negative affect toward the wait than greater visual content directed toward time.

It is worth noting that H6–H9 together describe the interaction effects between direction of attention and amount of visual content in longer waits. We presented separate hypotheses rather than proposing overall interaction effects as our theories enabled us to make more detailed predictions on both the direction and the specific pattern of interaction effects.

Study 2: Experiment Procedure

We conducted a lab experiment with a 2 (direction of atten- tion: toward/away from time) by 2 (visual content: low/high) full-factorial between-subjects design. Subjects were re- cruited from the same pool as those that participated in Study 1. Direction of attention was manipulated by providing either clock or travel-related pictures. In the low visual con- tent treatment, a static picture of either a plane or a clock (see Figures 2b and 2d) was provided. In the high visual content treatments, a series of static clock or static travel-related pictures (including the ones used in the low visual content treatment conditions) was provided using Macromedia Flash (see Figures 2e–f for snapshots of some of the pictures used7). The same set of dependent and control variables were mea- sured following the same experimental procedure as in Study 1.

A total of 139 subjects completed the experiment (of which 35 subjects were taken from the longer wait, travel picture condition in Study 18), with 58.3 percent being males and an average age of 21.21. On average, the subjects had 11.14 years of Internet experience, and spent 19.33 hours online per week. Subjects were randomly assigned to one of the four treatment conditions (see Table 2). A series of MANOVA tests showed that age (F = 1.017, p = 0.439), gender (F = 0.024, p = 0.976), years of Internet experience (F = 1.381, p = 0.109), or hours spent on the Internet per week (F = 1.092, p = 0.328) did not affect the dependent variables.

7As the background color of web pages may affect users’ perception of time (Gorn et al. 2004), we kept the background color white across all treatment conditions, and made efforts to balance out the use of color in the travel and the clock pictures. For example, in the series of clock pictures, we used clocks of different colors, including blue and yellow.

8We combined the 35 subjects from Study 1 with the 104 subjects recruited for Study 2, as all subjects were recruited from the same subject pool (under- graduate business students at a public university) over a seven-month period, and each subject was randomly assigned to an experimental treatment.

MIS Quarterly Vol. 37 No. 2/June 2013 397

Hong et al./A Paradigm Shift Perspective for Managing Online Delay

Study 2: Data Analysis Results

Manipulation and Control Checks. A one-way ANOVA showed that the flash series of pictures was perceived as con- taining significantly more visual content than static pictures (F = 97.431, p = 0.000). Hence, our manipulation of the increased visual content per time unit was successful.

Similar to Study 1, impatience was not significantly related to PQW (F = 0.137, p = 0.712), but impatient users were more likely to perceive the waiting experience as negative (F = 14.163, p = 0.000). Also consistent with Study 1, attribution related to both PQW (F = 17.380, p = 0.000) and NAW (F = 63.860, p = 0.000), indicated that subjects attribute the delay to the provision of visual content. Impatience and attribution were the only control variables that influenced wait time perceptions in Study 2, and thus were included in the MANOVA described below.

Hypotheses Testing. Table 2 reports on the descriptive statistics. A 2 by 2 MANOVA generated significant results for the main effect of direction of attention (F = 11.274, p = 0.000), and the interaction effect between direction of attention and amount of visual content (F = 4.511, p = 0.013). Table 3 presents the results of the MANOVA and ANOVA tests,9 with Figures 4a–b depicting the interaction effects. Specifically, the interaction effects between direction of atten- tion and amount of visual content, are significant for both PQW (F = 4.567, p = 0.034) and NAW (F = 6.521, p = 0.012). In order to directly test the hypotheses, more detailed analyses of simple effects are needed.

First, when visual content directs users’ attention away from time, a series of travel pictures did not increase PQW as compared to a static travel picture (F = 0.859, p = 0.357), so H6a was rejected. However, a flash of travel pictures did reduce NAW as compared with a static travel picture (F = 5.490, p = 0.022), supporting H6b. Second, when visual content directs users’ attention toward time, a series of clock pictures significantly decreased the perceived quickness (F = 10.930, p = 0.002) and led to more negative affect (F = 5.850, p = 0.018), as compared to a static clock picture. Thus, H7a-b were supported. Third, a direct comparison between a static travel picture and a static clock picture showed no significant difference for PQW (F = 0.546, p = 0.463, η2 = 0.008) or

NAW (F = 0.011, p = 0.919, η2 = 0.000), supporting H8a–b.10

Finally, a series of travel pictures resulted in higher PQW (F = 24.416, p = 0.000) and lower NAW (F = 22.409, p = 0.000) than a series of clock pictures. Hence, H9a-b were also supported. Table 4 summarizes the hypothesis testing results for the two studies.

Study 2: Discussion

The findings of Study 2 show that in longer waiting condi- tions, directing attention away from time makes the wait feel shorter and more positive than directing attention toward time, but only when the amount of visual content is high. When the amount of visual content is low, the temporal nature of the content does not make a significant difference. This result is consistent with that of Study 1. More importantly, findings in Study 2 help to identify the boundary conditions under which direction of attention has a significant effect on wait perceptions.

The only unsupported hypothesis was H6a. Results revealed that increasing the amount of nontemporal visual content did not make the wait feel shorter, when compared to less visual content. If PQW is solely determined by attention-based models, then increasing the amount of nontemporal informa- tion should further distract attention from P(t), leading to shorter waits. And according to RAM, the opposite would be true if the memory model dominates. The finding that in- creasing nontemporal information does not affect PQW could be due to the nature of the online waiting experience. Online waiting scenarios are very different from the experimental set- tings in which time perception theories are normally tested, as users are not obligated to process the visual content provided during the wait. Thus, this insignificant effect has two pos- sible explanations. First, it is possible that users were not motivated to process the increased amount of nontemporal information and, therefore, there was no increased distraction from P(t). This explanation is unlikely, however, because the significant effects of providing increased temporal informa- tion on wait perceptions shows evidence of active processing of visual content during longer waits. Second, while findings from the first study suggest that there is a paradigm shift from retrospective to prospective in longer waits, it is unclear if the shift is complete. In other words, it is possible that in longer

9In order to control family-wise (Type 1) error, we divided alpha by 2 (.05/2 = .025), given two dependent variables and one test, and found that the effect of the interaction on PQW became marginally significant (F1,133 > 5.14). Since the pattern of overall results was unchanged, we concluded that family- wise error did not distort our results.

10Since 8a–b hypothesized nonsignificant differences, the supportive results should be interpreted with caution. However, the effect sizes suggest that the related effects are small (i.e., 0.008 and 0.000). While an increase in sample size might bring some of these nonsignificant results closer to significance, the effect size would remain small, thus limiting the practical significance of such effects.

398 MIS Quarterly Vol. 37 No. 2/June 2013

Hong et al./A Paradigm Shift Perspective for Managing Online Delay

(a) Study 2: Amount of Information × Direction of Attention on PQW

(b) Study 2: Amount of Information × Direction of Attention on NAW

Figure 4. Interaction Effects of Information × Direction

waits, although P(t) dominates, P(i) may still play a role. So when the amount of nontemporal information increases, while it primarily serves as a distracter from P(t), it also increases memory-based cues and therefore increases time estimation through P(i). The two effects may have balanced out, resulting in a nonsignificant difference between the high and low visual content directing attention away from time in longer waits.

Despite the lack of support for H6a, providing a higher level of nontemporal visual content did reduce NAW as compared to the lower visual content treatment (H6b). This result could be attributed to the distraction from thinking about time, and by giving users something fun to look at during the longer waits. If the users were not motivated to process the in- creased nontemporal information, then we should not have seen a reduction in their NAW. Hence, the result from the affective perspective also helps to exclude the first possible explanation above.

In addition, our results provide strong support for the effects of direction of attention when we increased the visual content. A waiting page filled with a series of static clock pictures was not only perceived as having a longer wait time than a static clock picture, but also longer than a series of travel pictures. This result confirms our speculation in Study 1, that in- creasing the amount of visual content per unit of time strengthens the effects of direction of attention, and brings out the expected differences between temporal and nontemporal visual content.

Post Hoc Studies

We conducted two post hoc studies to further validate our key findings and confirm the practical implications. First, we collected additional data from 37 subjects, who viewed either a static clock or a static travel picture while waiting 45 seconds for search results. Subjects then responded to three open-ended questions asking them to describe what they were thinking about at 15, 30, and 45 seconds, respectively. The open-ended responses were coded as relating to (1) travel topics, (2) time/length of the wait, or (3) other topics. Three of the authors coded the comments independently with 0.91 inter-rater reliability and an average Cohen’s Kappa of .81, which suggests very good agreement. Preliminary analysis of the open-ended responses supports the notion of a paradigm shift, as a greater percentage of the comments were directed toward time as the waiting was prolonged. After 15 seconds, an average of 70.3 percent of the participants provided travel- related responses while 15.3 percent provided time/wait related responses. After 30 seconds, these percentages were 20.7 percent for travel and 50.5 percent for time/wait. After 45 seconds, only 1.8 percent were directed toward travel while 84.7 percent were directed toward time.

Second, the two experimental studies contribute to practice through findings that are driven by theory rather than by intuition. Some of our findings are not intuitive, and it is unlikely that web designers would have anticipated our findings. To test this assumption, we designed a web based survey asking designers to predict the direction of the effects

MIS Quarterly Vol. 37 No. 2/June 2013 399

Hong et al./A Paradigm Shift Perspective for Managing Online Delay

Table 4. Hypotheses Testing Results

IVs DVs* Hypotheses Results

S tu

d y 1

• Actual wait time (Short/Long Wait)

PQW H1a: Actual waiting time reduces the perceived quickness of the wait. Supported

NAW H1b: Actual waiting time increases negative affect toward the wait. Supported

• Actual wait time (Short/Long Wait)

• Amount of information (Visual Content)

PQW H2a: During shorter waits, providing visual content (e.g., a picture) reduces the

perceived quickness of the wait. Supported

NAW H2b: During shorter waits, providing visual content (e.g., a picture) reduces the negative

affect toward the wait Rejected

PQW H3a: During longer waits, providing visual content (e.g., a picture) increases the

perceived quickness of the wait. Supported

NAW H3b: During longer waits, providing visual content (e.g., a picture) reduces the negative

affect toward the wait. Supported

• Actual wait time (Short/Long Wait)

• Direction of Attention (Away from/toward time)

PQW H4a: During shorter waits, directing attention away from time results in higher perceived

quickness of the wait than directing attention toward time. Rejected

NAW H4b: During shorter waits, directing attention away from time results in lower negative

affect toward the wait than directing attention toward time. Rejected

PQW H5a: During longer waits, directing attention away from time results in higher perceived

quickness of the wait than directing attention toward time. Rejected

NAW H5b: During longer waits, directing attention away from time results in lower negative

affect toward the wait than directing attention toward time. Rejected

S tu

d y 2

• Longer Wait Time • Amount of

information (Less/Greater)

• Direction of Attention (Away from/toward time)

PQW H6a: During longer waits, greater visual content directed away from time (e.g., a

series of travel pictures) increases the perceived quickness of the wait as compared to less visual content (e.g., a travel picture) directed away from time.

Rejected

NAW H6b: During longer waits, greater visual content directed away from time (e.g., a

series of travel pictures) reduces negative affect toward the wait as compared to less visual content (e.g., a travel picture) directed away from time.

Supported

PQW H7a: During longer waits, greater visual content directed toward time (e.g., a series of

clock pictures) decreases the perceived quickness of the wait as compared to less visual content (e.g., clock picture) directed toward time.

Supported

NAW H7b: During longer waits, greater visual content directed toward time (e.g., a series of

clock pictures) increases negative affect toward the wait as compared to less visual content (e.g., clock picture) directed toward time.

Supported

• Longer Wait Time • Direction of

Attention (Away from/toward time)

• Amount of information (Less/Greater)

PQW H8a: During longer waits, less visual content (e.g., a picture) directed away from time

results in similar perceived quickness of the wait as less visual content directed toward time.

Supported

NAW H8b: During longer waits, less visual content (e.g., a picture) directed away from time

results in similar negative affect toward the wait as less visual content directed toward time.

Supported

PQW H9a: During longer waits, greater visual content (e.g., a series of pictures) directed

away from time results in higher perceived quickness of the wait than greater visual content directed toward time.

Supported

NAW H9b: During longer waits, greater visual content (e.g., a series of pictures) directed

away from time results in lower negative affect toward the wait than greater visual content directed toward time.

Supported

*PQW: Perceived quickness of the wait; NAW: Negative affect toward the wait.

associated with three of our most counterintuitive findings. The director of website systems at a large online retailer was contacted, and 25 professional web designers agreed to complete our survey. No compensation was provided to the participants; however, we agreed to share the survey results. Survey questions were framed in practitioner terms and

worded so that the direction of the results would not be readily apparent. For example, the survey question for H2 was worded as: “During shorter waits, providing a picture will make the wait seem _____,” with response choices of shorter, longer, similar, and don’t know. Results revealed that only 24 percent of the web designers were able to accurately

400 MIS Quarterly Vol. 37 No. 2/June 2013

Hong et al./A Paradigm Shift Perspective for Managing Online Delay

predict the direction of H2a. For H6a, only 16 percent were able to predict the direction of the results. Finally, only 12 percent of the designers were able to correctly predict the results for H8a. These results help to reinforce our contribu- tion to practice by demonstrating how theory driven research can often provide insights that appear counterintuitive to practitioners.

Implications, Limitations, and Conclusion

Theoretical Implications

This paper makes a number of theoretical contributions. First, it highlights the significance of understanding users’ percep- tions of the wait. As Ryan and Valverde (2006, p. 185) observed in their review of the online waiting literature, “research on online waiting behavior is still in its infancy.” Existing IS research on online waiting tends to focus on objective (or tolerable) wait time. Until recently, little atten- tion has been given to the subjective perceptions of the wait, even though the importance of this concept has been estab- lished for over 20 years in the marketing literature. By showing that the visual content provided on a waiting page significantly impacts perceived quickness and negative affect, our findings confirm the importance of understanding online users’ wait perceptions. This paper is, therefore, one of the first attempts by IS researchers to direct research attention toward the management of online waiting experiences.

Second, our study effectively applies time perception theories to the online waiting context and extends time perception theories by demonstrating that a paradigm shift occurs in longer waits. In general, our results show support for RAM and the distinction between the prospective and retrospective paradigms. We identify conditions under which memory- based models and attention-based models would apply. Specifically, in shorter waiting conditions, a retrospective paradigm dominates and memory-based models provide a better explanation of the results; whereas in longer waiting conditions, there is a paradigm shift from retrospective to prospective in estimating time, and attention-based models dominate. Our findings advance time perception theories by showing that a paradigm shift is likely to occur in longer waits when temporal motives are strong, most likely because (1)-users are not given specific tasks to complete while waiting; (2)-only passive stimuli are presented to users; or (3) because information provided during the wait inadver- tently arouses temporal motive. As a result, our research

contributes to the duration judgment literature in terms of the more complex and less well-understood retrospective para- digm, by showing conditions under which a paradigm shift may occur.

Further, we found that while a paradigm shift occurs in longer waits, the shift may not be complete. Results in Study 2 show that in longer waits, while P(t) dominates, P(i) plays a smaller but influential role. Such findings reveal the uniqueness of online waiting conditions, in the sense that users do not expect to be asked how long they have waited, but neverthe- less need to keep an eye on the wait time so that they can abandon the website when the waiting is too long (Dabholkar and Sheng 2008). Nor are users instructed to process any of the information presented on the waiting page, but they tend to look at it anyway, probably because they are bored or the presented information is appealing. Therefore, our study advances time perception theories by revealing how they could be applied in the unique online waiting context.

Third, this paper examines various forms of feedback that are unique to the online waiting context. We identified and systematically varied the amount of information and direction of attention under different wait times. Our findings suggest that the amount of visual content provided during the wait (i.e., the amount of content per unit of time in longer waits), as well as the temporal nature of the content have significant effects on online users’ waiting perceptions, and such effects vary with wait time.

Finally, our study provides a strong theoretical framework that may explain some of the prior findings in Table B1, where including waiting-related information was found to have either no effect, positive effect, negative effect, or mixed effects on wait perceptions. Our results suggest that the inconsistent findings may be due to the length of the wait, as users react to information differently under different wait times. For example, Whiting and Donthu (2006) found no effect of waiting duration, queuing information, or music on PQW in a field study of telephone waiting experience. As the field study used the actual wait time on the phone, which varied greatly, the insignificance could be the result of off- setting opposite effects under different wait times.

Practical Implications

Waiting on the Internet is likely to continue to exist, given that download speeds can be affected by a number of factors and that technical solutions tend to be very costly (Weinberg 2000). Waiting for a basic webpage to load is just one short

MIS Quarterly Vol. 37 No. 2/June 2013 401

Hong et al./A Paradigm Shift Perspective for Managing Online Delay

form of online waiting, and various, longer forms of waiting have been identified, including complex searches and queries, file downloads, and waiting for information or confirmation (Ryan and Valverde 2005). This study explores alternative, cost-effective solutions for short to moderate online waits. The results of our study provide useful guidelines on how website designers can successfully manage users’ waiting ex- periences by providing appropriate feedback and visual con- tent based on the nature of the task and the expected wait.11

First, in shorter waiting conditions, providing additional con- tent (regardless of the nature of the content) may make the wait feel longer. The result of our post hoc study shows that website designers may not be aware of this phenomenon, which is probably why we see increased visual content on waiting pages, including promotional and recreational content (see Figures A2–A4). Thus, it is beneficial for website designers to understand the potential negative consequence of doing so when the wait time is relatively short. Fortunately, the negative effect of providing such information from a cognitive perspective (i.e., PQW), is somewhat alleviated from an affective perspective, as users do not seem to feel more NAW as a result. They probably appreciate having something to look at while waiting, even if the visual content makes the wait feel longer.

Second, in longer waiting conditions, the effects of providing additional visual content are more complicated. While it is critical for website designers to provide something for users to look at during longer waits, they need to be careful about the temporal nature and intensity of such content. Visual content that does not remind users of the passage of time is safe to include on the waiting page, and such content only makes users feel more positive about the wait. On the other hand, website designers should avoid content that may remind users of the passage of time. While a low amount of time- related visual content may minimally affect wait perceptions, a large amount of time-related visual content may direct users’ attention to the waiting itself, making them feel that they have waited longer and making them feel more annoyed.

Third, the results of our study show that users may attribute some delay to the design of the waiting pages (i.e., the visual content included). Therefore, website designers need to con- sider the tradeoff between providing greater visual content (e.g., video) to improve perceptions of the wait and the risk

that any download delay may be attributed to this content, generating NAW. Fortunately, our results indicate that even after taking into account the negative attribution effects, providing visual content, especially if unrelated to time, is beneficial in longer waits.

Overall, it is important for website designers to understand the value of managing online users’ perceptions of the wait. While the actual waiting time may not be under the control of Web designers, they can effectively manage users’ waiting experience by providing appropriate feedback. Our results reinforce the predictions of RAM, and show that providing appropriate content makes the waiting experience more posi- tive, while providing inappropriate content can worsen the waiting experience. Figure 5 graphically depicts the main findings of this study. As shown, during shorter waits, the inclusion of visual content has a negative effect on wait time perceptions regardless of specific content (i.e., temporal nature of the cue). However, with longer waits and more visual content, wait time perceptions can vary greatly depending upon the nature of that content (i.e., distracts or directs the user from/toward time).

Limitations

There are a few limitations to this study. First, we used a simulated website and a simulated waiting scenario instead of an actual website and waiting context for the experiment. Although carefully designed and pretested, a simulated web- site may not be as complete or complex as real websites, and a simulated waiting scenario may not be perceived as “real” by our subjects. However, designing an experimental, simulated website allowed us to control the wait time and visual content in the same waiting context, which is not feasible when utilizing existing websites. Thus, the use of a simulated website is appropriate in this study as it strengthens the internal validity of the experiment design.

Second, online users may switch to other websites if a web- page takes too long to load. The subjects in our study were evaluating a new travel website without concurrent task demands or high opportunity costs. Perceptions of the wait might differ if online users feel that the wait is costly or attempt to multitask while waiting. Nevertheless, our findings provide insight into users’ wait perceptions under the so- called “low-cost waiting conditions” (Cameron et al. 2003). We also took steps to ensure that subjects would not want to leave the website while waiting by carefully selecting waiting times that were acceptable based on pilot study results. Subjects’ affective responses to the wait were generally posi- tive across both studies (mean = 2.90 on a 1 to 7 scale, with

11While wait time is difficult to control, designers can reasonably predict whether short or long waits are expected based on the online waiting context (e.g., complex searches, purchase processing, and file downloads). In the future, Internet technologies may be able to detect the connection/execution speed of each client computer, and provide feedback pages accordingly.

402 MIS Quarterly Vol. 37 No. 2/June 2013

Hong et al./A Paradigm Shift Perspective for Managing Online Delay

Figure 5. Summary of Main Findings

smaller numbers indicating less NAW). Further, we included an indicator bar in all treatments to inform subjects that the website was processing their request. Hence, a controlled experimental setting provided internal validity while pre- serving a reasonable level of reality.

Epilogue

Managing online users’ waiting experiences, instead of the waiting length per se, has received increased attention from IS researchers. This study makes a unique contribution by documenting the paradigm shift that may occur as a wait lengthens, exposing the dramatically different effects of visual content in waiting scenarios of different lengths. Taking a theoretically grounded approach, this study reveals results that are somewhat contradictory to commonly held beliefs. We found that providing extra content during shorter waits, regardless of its substance, makes the wait feel longer. On the other hand, when wait times are prolonged it is beneficial to provide additional visual content to users, but only if the content does not remind users of the passage of time. These findings theoretically advance research on online users’ waiting experiences and provide useful guidelines for website designers, while also stimulating new research.

As research on online waiting is still in its infancy, there are ample avenues for future research. First, our study provides initial evidence that users react differently to different forms

of visual content over different wait periods. Future empirical studies are needed to examine other forms of content, such as music, videos, and text-based references to waiting. Even subtle waiting messages, such as “this may take up to xx seconds,” may have unintended negative effects. Second, prior research suggests that individual factors, such as age, gender, and Internet experience, may play an important role in affecting users’ perceptions of time and web resources (Jacko et al. 2000; Kellaris and Mantel 1994). Although we only found impatience to be a significant individual factor in longer waiting conditions, future research may examine a larger set of individual characteristics. Third, while most empirical research on this topic has employed a quantitative approach with experiments, qualitative research using real online waiting scenarios, in which users share their thoughts and feelings, would be extremely valuable. Fourth, we examined an online search task, but waiting with other online tasks should be investigated, as it is likely that users may react differently to different waiting situations, particularly those in which different opportunity costs exist. Finally, research on “mind-wandering” (Smallwood and Schooler 2006) may provide new theoretical insight as we consider how to design web pages that curb (or encourage) mind-wandering, de- pending on the context of the online wait. If we can harness the wandering of the human mind through the design of fillers, we may be able to further enhance tolerance for delay.

Our study makes a pioneering effort by identifying several key factors that affect users’ perceptions of the wait. These findings not only confirm that waiting perceptions can be

MIS Quarterly Vol. 37 No. 2/June 2013 403

Hong et al./A Paradigm Shift Perspective for Managing Online Delay

managed, but also support a paradigm shift, offering exciting new insights into how filler information should be used during shorter and longer waits. Future research should extend these findings by examining other types of fillers, waiting scen- arios, and tasks.

References

Antonides, G., Verhoef, P. C., and van Alas, M. 2002. “Consumer Perception and Evaluation of Waiting Time: A Field Experi- ment,” Journal of Consumer Psychology (12:3), pp. 193-202.

Baker, J., and Cameron, M. 1996. “The Effects of the Service Environment on Affect and Consumer Perception of Waiting Time,” Journal of the Academy of Marketing Science (24:4), pp. 338-349.

Block, R. A. 1989. “Experiencing and Remembering Time: Affor- dances, Context, and Cognition,” in Time and Human Cognition: A Life-Span Perspective, I. Levin and D. Zakay (eds.), Amster- dam: North-Holland, pp. 333-363.

Block, R. A. 1992. “Prospective and Retrospective Duration Judg- ment: The Role of Information Processing and Memory,” in Time, Action and Cognition: Towards Bridging the Gap, F. Macar, F. Pouthas, and W. J. Friedman (eds.), Dordrecht: Kluwer, pp. 141-152.

Block, R. A. 2003. “Psychological Timing Without a Timer: The Role of Attention and Memory,” in Time and Mind II: Infor- mation Processing Perspectives, H. Helfrich (ed.), Göttingen, Germany: Hogrefe & Huber, pp. 41-59.

Block, R. A., Hancock, P. A., and Zakay, D. 2010. “How Cognitive Load Affects Duration Judgments: A Meta-Analytic Review,” Acta Psychologica (134), pp. 330-343.

Block, R. A., and Reed, M. A. 1978. “Remembered Duration: Evidence for a Contextual Change Hypothesis “ Journal of Experimental Psychology: Human Learning and Memory (4), pp. 656-665.

Block, R. A., and Zakay, D. 1996. “Models of Psychological Time Revisited,” in Time and Mind, H. Helfrich (ed.), Kirkland, WA: Hogrefe & Huber, pp. 171-195.

Block, R. A., and Zakay, D. 1997. “Prospective and Retrospective Duration Judgments: A Meta-Analytic Review,” Psychonomic Bulletin & Review (4:2), pp. 184-197.

Brambor, T., Clark, W. R., and Golder, M. 2006. “Understanding Interaction Models: Improving Empirical Analyses,” Political Analysis (14:1), pp. 63-82.

Brown, S. W. 1997. “Attentional Resources in Timing: Inter- ference Effects in Concurrent Temporal and Non-Temporal Working Memory Tasks,” Perception & Psychophysics (59:7), pp. 1118-1140.

Buell, R. W., and Norton, M. I. 2011. “Think Customers Hate Waiting? Not So Fast...,” Harvard Business Review (89:5), pp. 34-34.

Calder, B. J., Phillips, L. W., and Tybout, A. M. 1981. “Designing Research for Application,” Journal of Consumer Research (8:2), pp. 197-207.

Cameron, M. A., Baker, J., Peterson, M., and Braunsberger, K. 2003. “The Effects of Music, Wait-Length Evaluation, and

Mood on a Low-Cost Wait Experience,” Journal of Business Research (56), pp. 421-430.

Campbell, D. T., and Fiske, D. W. 1959. “Convergent and Discriminant Validation by the Multitrait-Multimethod Matrix,” Psychological Bulletin (56), pp. 81-105.

Casini, L., and Macar, F. 1997. “Effects of Attention Manipulation on Judgments of Duration and of Intensity in the Visual Modality,” Memory & Cognition (25:6), pp. 812-818.

Ceaparu, I., Lazar, J., Bessiere, K., Robinson, J., and Shneider- man, B. 2004. “Determining Causes and Severity of End-User Frustration,” International Journal of Human–Computer Inter- action (17:3), pp. 333-356.

Chebat, J.-C., Salem, N. H., Poirier, J.-F., and Gëlinas-Chebat, C. 2010. “Reactions to Waiting Online by Men and Women,” Psychological Reports (106:3), pp. 851-869.

Curran, K., and Duffy, C. 2005. “Understanding and Reducing Web Delays,” International Journal of Network Management (15:2), pp. 89-102.

Dabholkar, P. A., and Sheng, X. 2008. “Perceptions of Download Delays: Relation to Actual Waits, Web Site Abandoning, and Stage of Delay,” The Service Industries Journal (28:10), pp. 1415-1429.

Datta, A., Dutta, K., Thomas, H., Vandermeer, D., and Ramam- ritham, K. 2004. “Proxy-Based Acceleration of Dynamically Generated Content on the World Wide Web: An Approach and Implementation,” ACM Transactions on Database Systems (29:2), pp. 403-443.

Davis, M. M., and Heineke, J. 1998. “How Disconfirmation, Perception and Actual Waiting Times Impact Customer Satisfac- tion,” International Journal of Service Industry Management (9:1), pp. 64-73.

Dellaert, B. G. C., and Kahn, B. E. 1999. “How Tolerable is Delay? Consumers’ Evaluation of Internet Web Sites after Waiting,” Journal of Interactive Marketing (13:1), pp. 41-54.

Dennis, A. R., and Taylor, N. J. 2006. “Information Foraging on the Web: The Effects of “Acceptable” Internet Delays on Multi- Page Information Search Behavior,” Dec Support Systems (42:2), pp. 810-824.

DiClemente, D. F., and Hantula, D. A. 2003. “Optimal Foraging Online: Increasing Sensitivity to Delay,” Psychology & Mar- keting (20:9), pp. 785-809.

Doob, L. W. 1971. Patterning of Time, New Haven, CT: Yale University Press.

Dubé, L., Schmitt, B. H., and Leclerc, F. 1991. “Consumers’ Affective Response to Delays at Different Phases of a Service Delivery,” Journal of Applied Social Psychology (21:10), pp. 810-820.

Durrande-Moreau, A. 1999. “Waiting for Service: Ten Years of Empirical Research,” International Journal of Service Industry Management (10:2), pp. 171-189.

Fraisse, P. 1984. “Perception and Estimation of Time,” Annual Review of Psychology (35), pp. 1-36.

Francis-Smythe, J. A., and Robertson, I. T. 1999. “Time Related Individual Differences,” Time and Society (8), pp. 273-292.

Galletta, D. F., Henry, R., McCoy, S., and Polak, P. 2004. “Web Site Delays: How Tolerant Are Users?,” Journal of the Associat- ion for Information Systems (5:1), pp. 1-28.

404 MIS Quarterly Vol. 37 No. 2/June 2013

Hong et al./A Paradigm Shift Perspective for Managing Online Delay

Galletta, D. F., Henry, R., McCoy, S., and Polak, P. 2006. “When the Wait Isn’t So Bad: The Interacting Effects of Website Delay, Familiarity, and Breadth,” Information Systems Research (17:1), pp. 20-37.

Gardner, M. P. 1985. “Mood States and Consumer Behavior: A Critical Review,” Journal of Consumer Research (12), pp. 281-300.

Geelhoed, E., Toft, P., Roberts, S., and Hyland, P. 1995. “To Influence Time Perception,” in Proceedings of the CHI’95 Conference on Human Factors in Computing Systems, New York: ACM Press, pp. 272-273.

Gorn, G. J., Chattopadhyay, A., Sengupta, J., and Tripathi, S. 2004. “Waiting for the Web: How Screen Color Affects Time Perception,” Journal of Marketing Research (41:2), pp. 215-225.

Groth, M., and Gilliland, S. W. 2006. “Having to Wait for Service: Customer Reactions to Delays in Service Delivery,” Applied Psychology (55:1), pp. 107-129.

Hair, J., Black, B., Babin, B., Anderson, R., and Tatham, R. 2006. Multivariate Data Analysis, (6th ed.) Upper Saddle River, NJ: Prentice-Hall.

Harrison, C., Amento, B., Kuznetsov, S., and Bell, R. 2007. “Rethinking the Progress Bar,” in Proceedings of the 20th Annual ACM Symposium on User Interface Software and Technology, New York: ACM Press, pp. 115-118.

Hicks, R. E. 1992. “Prospective and Retrospective Judgments of Time: A Neurobehavioral Analysis,” in Time, Action and Cogni- tion: Towards Bridging the Gap, F. Macar, F. Pouthas and W. J. Friedman (eds.), Dordrecht: Kluwer, pp. 97-108.

Hicks, R. E., Miller, G. W., Gaes, G., and Bierman, K. 1977. “Concurrent Processing Demands and the Experience of Time- in-Passing,” American Journal of Psychology (90:3), pp. 431-446.

Hicks, R. E., Miller, G. W., and Kinsbourne, M. 1976. “Prospec- tive and Retrospective Judgments of Time as a Function of Amount of Information Processed,” American Journal of Psychology (89), pp. 719-730.

Hornik, J. 1981. “Time Cue and Time Perception Effect on Response to Mail Survey,” Journal of Marketing Research (18:2), pp. 243-248.

Houston, M. B., Bettencourt, L. A., and Wenger, S. 1998. “The Relationship between Waiting in a Service Queue and Evaluations of Service Quality: A Field Theory Perspective,” Psychology and Marketing (15:8), pp. 735-753.

Hoxmeier, J. A., and DiCesare, C. 2000. “System Response Time and User Satisfaction: An Experimental Study of Browser-Based Applications,” in Proceedings of the 6th Americas Conference on Information Systems, Long Beach, CA, pp. 10-13.

Hui, M. K., Dubé, L., and Chebat, J.-C. 1997. “The Impact of Music on Consumers’ Reactions to Waiting for Services,” Journal of Retailing (73:1), pp. 87-104.

Hui, M. K., Thakor, M. V., and Gill, R. 1998. “The Effect of Delay Type and Service Stage on Consumers’ Reactions to Waiting,” Journal of Consumer Research (24:4), pp. 469-479.

Hui, M. K., and Tse, A. C. 1996. “What to Tell Consumers in Waits of Different Lengths: An Integrative Model of Service Evaluation,” Journal of Marketing (60), pp. 81-90.

Hui, M. K., and Zhou, L. 1996. “How Does Waiting Duration Information Influence Customers’ Reactions to Waiting for

Services?,” Journal of Applied Social Psychology (26), pp. 1702-1717.

Jacko, J. A., Sears, A., and Borella, M. S. 2000. “The Effect of Network Delay and Media on User Perceptions of Web Resources,” Behavior and Information Technology (19:6), pp. 427-439.

Kahneman, D. 1973. Attention and Effort, Englewood Cliffs, NJ: Prentice-Hall.

Katz, K. L., Larson, B. M., and Larson, R. C. 1991. “Prescription for the Waiting-in-Line Blues: Entertain, Enlighten, and Engage,” Sloan Management Review (32:2), pp. 44-53.

Kellaris, J. J., and Mantel, S. P. 1994. “The Influence of Mood and Gender on Customers’ Time Perceptions,” Advances in Customer Research (21), pp. 514-518.

Keynote Systems, 2005. “Serving Your Customers an Outstanding Online Experience,” White Paper, Keynote Systems, Inc., San Mateo, CA (http://www.keynote.com/docs/whitepapers/ Serving CustomersOutstandOnlineExp.pdf).

Kirk, R. E. 1995. Experimental Design: Procedures for Beha- vioral Science (3rd ed.), Pacific Grove, CA: Brooks/Cole.

Lee, Y., Chen, A. N. K., and Ilie, V. 2012. “Can Online Wait Be Managed? The Effect of Filler Interfaces and Presentation Models on Perceived Waiting Time Online,” MIS Quarterly (36:2), pp. 365-394.

Lejeune, H. 1998. “Switching or Gating? The Attentional Chal- lenge in Cognitive Models of Psychological Time,” Behavioral Processes (44:2), pp. 127-145.

Macar, F., and Jackson, J. 1992. “Towards an Understanding of Subjective Judgments of Time,” in Time, Action and Cognition: Towards Bridging the Gap, F. Macar, V. Pouthas, and W. J. Friedman (eds.), Dordrecht: Kluwer, pp. 141-152.

McClain, L. 1983. “Interval Estimation: Effect of Processing Demands on Prospection and Retrospection Reports,” Perception & Psychophysics (34), pp. 185-189.

McGuire, K. A., Kimes, S. E., Lynn, M., Pullman, M. E., and Lloyd, R. C. 2010. “A Framework for Evaluating the Customer Wait Experience,” Journal of Service Management (21:3), pp. 269-290.

Myers, B. A. 1985. “The Importance of Percent-Done Progress Indicators for Computer-Human Interfaces,” in Proceedings of the SIGCHI Conference on Human Factors in Computing Systems, New York: ACM Press, pp. 11-17.

Nah, F. F. H. 2002. “A Study of Web Users’ Waiting Time,” in Intelligent Support Systems Technology: Knowledge Manage- ment, V. Sugumaran (ed.), Hershey, PA: IGI Global, pp. 145-152.

Nah, F. F. H. 2004. “A Study of Tolerable Waiting Time: How Long Are Web Users Willing to Wait?,” Behavior & Information Technology (23:3), pp. 153-163.

Nunnally, J. C. 1978. Psychometric Theory (2nd ed.), New York: McGraw-Hill.

Ornstein, R. E. 1969. On the Experience of Time, Middlesex, England: Penguin.

Osuna, E. E. 1985. “The Psychological Cost of Waiting,” Journal of Mathematical Psychology (29), pp. 82-105.

Poynter, D. 1989. “Judging the Duration of Time Intervals: A Process of Remembering Segments of Experience,” in:Time and Human Cognition: A Life Span Perspective, I. Levin and D. Zakay (eds.), Amsterdam: North-Holland, pp. 305-331.

MIS Quarterly Vol. 37 No. 2/June 2013 405

Hong et al./A Paradigm Shift Perspective for Managing Online Delay

Predebon, J. 1996. “The Relationship between the Number of Presented Stimuli and Prospective Duration Estimates: The Effect of Concurrent Task Activity,” Psychonomic Bulletin and Review (3:3), pp. 376-379.

Pruyn, A., and Smidts, A. 1998. “Effects of Waiting on the Satis- faction with the Service: Beyond Objective Time Measures,” International Journal of Research in Marketing (15), pp. 321-334.

Ramsay, J., Barbesi, A., and Preece, J. 1998. “A Psychological Investigation of Long Retrieval Times on the World Wide Web,” Interaction with Computers (10), pp. 77-86.

Rose, G., Meuter, M., and Curran, J. 2005. “On-Line Waiting: The Role of Download Time and Other Important Predictors on Attitude toward E-Retailers,” Psychology and Marketing (22:2), pp. 127-151.

Rose, G., and Straub, D. W. 2001. “The Effect of Download Time on Consumer Attitude Towards the E-Service Retailer,” e-Service Journal (1:1), pp. 55-76.

Ryan, G., and Valverde, M. 2003. “Waiting Online: A Review and Research Agenda,” Internet Research: Electronic Networking Applications and Policy (13:3), pp. 195-205.

Ryan, G., and Valverde, M. 2005. “Waiting for Service on the Internet,” Internet Research (15:2), pp. 220-240.

Ryan, G., and Valverde, M. 2006. “Waiting in Line for Online Services: A Qualitative Study of the User’s Perspective,” Infor- mation Systems Journal (16), pp. 181-211.

Smallwood, J., and Schooler, J. W. 2006. “The Restless Mind,” Psychological Bulletin (132:6), pp. 946-958.

Spence, J. T., Helmreich, R. L., and Pred, R. S. 1987. “Impatience Versus Achievement Strivings on the Type a Pattern: Differ- ential Effects on Students' Health and Academic Achievement,” Journal of Applied Psychology (72:4), pp. 522-528.

Staddon, J. E. R. 2005. “Interval Timing: Memory, Not a Clock,” Trends in Cognitive Sciences (9:7), pp. 312-314.

Taylor, S. 1994. “Waiting for Service: The Relationship between Delays and Evaluations of Service,” Journal of Marketing (58), pp. 56-69.

Taylor, S. 1995. “The Effects of Filled Waiting Time and Service Provider Control over the Delay on Evaluations of Service,” Journal of the Academy of Marketing Science (23:1), pp. 38-48.

Teal, S. L., and Rudnicky, A. 1992. “A Performance Model of System Delay and User Selection Strategy,” in Proceedings of the SIGCHI Conference on Human Factors, New York: ACM Press, pp. 295-305.

Weinberg, B. D. 2000. “Don’t Keep Your Internet Customers Waiting Too Long at the (Virtual) Front Door,” Journal of Interactive Marketing (14:1), pp. 30-39.

Whiting, A., and Donthu, N. 2006. “Managing Voice-to-Voices Encounters: Reducing the Agony of Being Put on Hold,” Journal of Service Research (8:3), pp. 234-244.

Zakay, D. 1989. “Subjective Time and Attentional Resource Allocation: An Integrated Model of Time Estimation,” in In Time and Human Cognition, I. Levin and D. Zakay (eds.), Amsterdam: North-Holland, pp. 365-397.

Zakay, D. 1993. “Time Estimation Methods: Do They Influence Prospective Duration Estimates?,” Perception (22:1), pp. 91-101.

Zakay, D., and Block, R. A. 1997. “Temporal Cognition,” Current Directions in Psychological Science (6:1), pp. 12-15.

Zona Research Study. 2001. “The Need for Speed II,” Zona Market Bulletin, Issue 5, April (http://www.keynote.com/downloads/ Zona_Need_For_Speed.pdf).

About the Authors

Weiyin Hong is an Emeritus Associate Professor in the Department of Management Information Systems at the University of Nevada, Las Vegas. She is also an Adjunct Associate Professor in the Department of ISOM at the Hong Kong University of Science and Technology. She received her Ph.D. in Information Systems from the Hong Kong University of Science and Technology and her B.Sc. in MIS from Fudan University, China. Her research interests include human–computer interaction, user acceptance of emerging technologies, and user privacy concern. Her work has appeared in MIS Quarterly, Information Systems Research, Journal of Manage- ment Information Systems, International Journal of Human– Computer Studies, Communications of the ACM, Information & Management, Journal of the American Society for Information Science and Technology, and Journal of Database Management.

Traci J. Hess is an associate professor in the Isenberg School of Management at the University of Massachusetts Amherst. She received her Ph.D. and M.A. degrees in Information Systems/ Technology from Virginia Tech and a B.S. in Accounting from the University of Virginia. Her research interests include human– computer interaction, decision support systems, and user acceptance of information systems. Her work has appeared in journals such as MIS Quarterly, Journal of Management Information Systems, Jour- nal of the Association for Information Systems, Decision Sciences, Decision Support Systems, Journal of Strategic Information Systems, AIS Transactions on Human–Computer Interaction, Journal of Organizational Computing and Electronic Commerce, and The DATA BASE for Advances in Information Systems. Traci serves as an associate editor for Decision Sciences, a senior editor for AIS Transactions on Human–Computer Interaction, and on the editorial board for the Journal of the Association for Information Systems.

Andrew Hardin is the director of the Center for Entrepreneurship and an associate professor in the Lee Business School at the University of Nevada, Las Vegas. Andrew’s research is focused on organizational collaboration and virtual work, financial decision support systems, and research methodologies. His work has been published or is forthcoming in journals such as Management Science, MIS Quarterly, Organizational Behavior and Human Deci- sion Processes, Journal of Management Information Systems, European Journal of Information Systems, Journal of the Asso- ciation for Information Systems, The DATA BASE for Advances in Information Systems, Group Decision and Negotiations, Small Group Research, and Educational and Psychological Measurement. He currently serves as a senior editor for Information Systems Journal and The DATA BASE for Advances in Information Systems, as an associate editor for the European Journal of Information Systems, and as a guest associate editor for MIS Quarterly.

406 MIS Quarterly Vol. 37 No. 2/June 2013

RESEARCH ARTICLE

WHEN FILLING THE WAIT MAKES IT FEEL LONGER: A PARADIGM SHIFT PERSPECTIVE FOR

MANAGING ONLINE DELAY

Weiyin Hong Department of ISOM, Hong Kong University of Science and Technology, Clear Water Bay, Kowloon, HONG KONG

and Department of MIS, University of Nevada, Las Vegas,

Las Vegas, NV 89120 U.S.A. {[email protected]}

Traci J. Hess Isenberg School of Management, University of Massachusetts, Amherst,

Amhest, MA 01003 U.S.A. {[email protected]}

Andrew Hardin Department of MIS, University of Nevada, Las Vegas,

Las Vegas, NV 89120 U.S.A. {[email protected]}

Appendix A

Screenshots of Online Waiting Pages

Figure A1. Expedia.com Figure A2. Travelocity.com

MIS Quarterly Vol. 37 No. 2—Appendices/June 2013 A1

Hong et al./A Paradigm Shift Perspective for Managing Online Delay

Figure A3. United.com Figure A4. Alaskaair.com

Figure A5. Disneyland.com Figure A6. Vistaprint.com

A2 MIS Quarterly Vol. 37 No. 2—Appendices/June 2013

Hong et al./A Paradigm Shift Perspective for Managing Online Delay

MIS Quarterly Vol. 37 No. 2—Appendices/June 2013 A3

Hong et al./A Paradigm Shift Perspective for Managing Online Delay

A4 MIS Quarterly Vol. 37 No. 2—Appendices/June 2013

Hong et al./A Paradigm Shift Perspective for Managing Online Delay

MIS Quarterly Vol. 37 No. 2—Appendices/June 2013 A5

Hong et al./A Paradigm Shift Perspective for Managing Online Delay

References

Antonides, G., Verhoef, P. C., and van Alas, M. 2002. “Consumer Perception and Evaluation of Waiting Time: A Field Experiment,” Journal of Consumer Psychology (12:3), pp. 193-202.

Areni, C., and Grantham, N. 2009. “(Waiting) Time Flies When the Tune Flows: Music Influences Affective Responses to Waiting by Chnging the Subjective Experience of Passing Time,” Advances in Consumer Research (36), pp. 449-455.

Bhatti, N. 2000. “Integrating User-Perceived Quality into Web Server Design,” Computer Networks: The International Journal of Computer and Telecommunication Networking (33:1-6), pp. 1-16.

Cameron, M. A., Baker, J., Peterson, M., and Braunsberger, K. 2003. “The Effects of Music, Wait-Length Evaluation, and Mood on a Low-Cost Wait Experience,” Journal of Business Research (56), pp. 421-430.

Chebat, J.-C., Salem, N. H., Poirier, J.-F., and Gëlinas-Chebat, C. 2010. “Reactions to Waiting Online by Men and Women,” Psychological Reports (106:3), pp. 851-869.

Dellaert, B. G. C., and Kahn, B. E. 1999. “How Tolerable is Delay? Consumers’ Evaluation of Internet Web Sites after Waiting,” Journal of Interactive Marketing (13:1), pp. 41-54.

DiClemente, D. F., and Hantula, D. A. 2003. “Optimal Foraging Online: Increasing Sensitivity to Delay,” Psychology & Marketing (20:9), pp. 785-809.

Geelhoed, E., Toft, P., Roberts, S., and Hyland, P. 1995. “To Influence Time Perception,” in Proceedings of the CHI’95 Conference on Human Factors in Computing Systems, New York: ACM Press, pp. 272-273.

Groth, M., and Gilliland, S. W. 2006. “Having to Wait for Service: Customer Reactions to Delays in Service Delivery,” Applied Psychology (55:1), pp. 107-129.

Harrison, C., Amento, B., Kuznetsov, S., and Bell, R. 2007. “Rethinking the Progress Bar,” in Proceedings of the 20th Annual ACM Symposium on User Interface Software and Technology, New York: ACM Press, pp. 115-118.

Hui, M. K., Dubé, L., and Chebat, J.-C. 1997. “The Impact of Music on Consumers’ Reactions to Waiting for Services,” Journal of Retailing (73:1), pp. 87-104.

Hui, M. K., and Tse, A. C. 1996. “What to Tell Consumers in Waits of Different Lengths: An Integrative Model of Service Evaluation,” Journal of Marketing (60), pp. 81-90.

Hui, M. K., and Zhou, L. 1996. “How Does Waiting Duration Information Influence Customers’ Reactions to Waiting for Services?,” Journal of Applied Social Psychology (26), pp. 1702-1717.

Katz, K. L., Larson, B. M., and Larson, R. C. 1991. “Prescription for the Waiting-in-Line Blues: Entertain, Enlighten, and Engage,” Sloan Management Review (32:2), pp. 44-53.

Kellaris, J. J., and Kent, R. J. 1992. “The Influence of Music on Consumers’ Temporal Perceptions: Does Time Fly When You’ve Having Fun?,” Journal of Consumer Psychology (1:4), pp. 365-376.

Lee, Y., Chen, A. N. K., and Ilie, V. 2012. “Can Online Wait Be Managed? The Effect of Filler Interfaces and Presentation Models on Perceived Waiting Time Online,” MIS Quarterly (36:2), pp. 365-394.

Myers, B. A. 1985. “The Importance of Percent-Done Progress Indicators for Computer-Human Interfaces,” in Proceedings of the SIGCHI Conference on Human Factors in Computing Systems, New York: ACM Press, pp. 11-17.

Nah, F. F. H. 2004. “A Study of Tolerable Waiting Time: How Long Are Web Users Willing to Wait?,” Behavior & Information Technology (23:3), pp. 153-163.

Oakes, S., and North, A. C. 2008. “Using Music to Influence Cognitive and Affective Responses in Queues of Low and High Crowd Density,” Journal of Marketing Management (24:5-6), pp. 589-602.

Pruyn, A., and Smidts, A. 1998. “Effects of Waiting on the Satisfaction with the Service: Beyond Objective Time Measures,” International Journal of Research in Marketing (15), pp. 321-334.

Taylor, S. 1995. “The Effects of Filled Waiting Time and Service Provider Control over the Delay on Evaluations of Service,” Journal of the Academy of Marketing Science (23:1), pp. 38-48.

Tom, G., Burns, M., and Zeng, Y. 1997. “Your Life on Hold: The Effect of Telephone Waiting Time on Customer Perception,” Journal of Direct Marketing (11:3), pp. 25-31.

Weinberg, B. D. 2000. “Don’t Keep Your Internet Customers Waiting Too Long at the (Virtual) Front Door,” Journal of Interactive Marketing (14:1), pp. 30-39.

Whiting, A., and Donthu, N. 2006. “Managing Voice-to-Voices Encounters: Reducing the Agony of Being Put on Hold,” Journal of Service Research (8:3), pp. 234-244.

A6 MIS Quarterly Vol. 37 No. 2—Appendices/June 2013

Hong et al./A Paradigm Shift Perspective for Managing Online Delay

Appendix C

Summary of Items

Perceived quickness of the wait (1–7 semantic scale) Questions 1 through 3 relate to the speed of the search. What do you think of the speed of the search?

1) Slow … Fast 2) Not speedy … Speedy 3) Not quick … Quick

Negative affect toward the wait How much did the waiting make you feel ______________?

1) Irritated (1 not at all … 4 neutral … 7 very much) 2) Annoyed (1 not at all … 4 neutral … 7 very much) 3) Frustrated (1 not at all … 4 neutral … 7 very much) 4) Unsatisfied (1 not at all … 4 neutral … 7 very much)

Impatience Please read each statement below carefully. For each statement, circle the response which best represents your opinion. There are no right or wrong answers.

1) Typically, how easily do you get irritated? (1 not at all easily … 7 extremely easily) 2) How is your “temper” these days? (1 I seldom get angry … 7 very hard to control) 3) When you have to wait in line such as at a restaurant, the movies, or the post office, how do you usually feel? (1 accept calmly …

7 feel very impatient and refuse to stay long)

Attribution Please indicate your degree of agreement with the following statements, with 1 indicating “strongly disagree” and 7 indicates “strongly agree.”

1) There is a lot the website could have done to avoid or shorten the delay (1 strongly disagree … 7 strongly agree) 2) The delay was mostly caused by the design of the website (1 strongly disagree … 7 strongly agree)

Visual content The following questions relate to the web page that you saw while waiting for your travel recommendations. Please assess the visual content of the web page that you saw while waiting by responding to the questions below.

1) The web page that I saw while waiting provided (1 low visual content … 7 high visual content) 2) While waiting, I saw a web page that contained (1 very little visual content … 7 a lot of visual content) 3) The amount of visual content that I saw while waiting was (1 very low … 7 very high) 4) The web page that I saw while waiting provided (1 not much visual content at all … 7 quite a lot of visual content)

MIS Quarterly Vol. 37 No. 2—Appendices/June 2013 A7

Hong et al./A Paradigm Shift Perspective for Managing Online Delay

Appendix D

Exploratory Factor Analysis Results

Inter-item Correlation Analysis*

PQW1 PQW2 PQW3 NAW1 NAW2 NAW3 NAW4 IMP1 IMP2 IMP3 ATT1 ATT2 VC1 VC2 VC3 VC4

PQW1 1.00

PQW2 .923 1.00

PQW3 .911 .951 1.00

NAW1 -.548 -.532 -.534 1.00

NAW2 -.572 -.549 -.547 .901 1.00

NAW3 -.533 -.508 -.505 .858 .850 1.00

NAW4 -.645 -.615 -.619 .781 .775 .788 1.00

IMP1 -.021 -.044 -.039 .097 .093 .052 .069 1.00

IMP2 .045 .044 .053 .048 .061 .063 .052 .552 1.00

IMP3 .016 .014 .009 .158 .175 .156 .174 .388 .400 1.00

ATT1 -.596 -.592 -.585 .540 .528 .533 .574 .047 -.004 .008 1.00

ATT2 -.455 -.447 -.434 .443 .424 .437 .450 .041 -.093 -.008 .617 1.00

VC1 .065 .060 .057 -.044 -.048 -.073 -.091 -.087 -.076 .036 .033 .026 1.00

VC2 .063 .054 .054 -.055 -.068 -.087 -.089 -.023 -.051 .050 .008 .003 .923 1.00

VC3 .085 .072 .081 -.053 -.068 -.087 -.100 -.033 -.014 .092 -.009 .012 .893 .924 1.00

VC4 .075 .057 .059 -.066 -.074 -.110 -.107 -.051 -.032 .059 -.012 -.014 .883 .920 .938 1.00

*PQW: Perceived quickness of the wait; NAW: Negative affect toward the wait; IMP: Impatience; ATT: Attribution; VC: Visual Content

Factor Analysis (Principle Component Analysis with Oblimin Rotation)*

PQW NAW IMP ATT VC

PQW1 .976†

PQW2 .976

PQW3 .927

NAW1 .913

NAW2 .905

NAW3 .894

NAW4 .721

IMP1 .873

IMP2 .840

IMP3 .312 .646

ATT1 .930

ATT2 .652

VC1 .973

VC2 .970

VC3 .968

VC4 .955

*PQW: Perceived quickness of the wait; NAW: Negative affect toward the wait; IMP: Impatience; ATT: Attribution; VC: Visual Content †Factor loadings smaller than 0.30 were omitted for a clearer presentation.

A8 MIS Quarterly Vol. 37 No. 2—Appendices/June 2013

Hong et al./A Paradigm Shift Perspective for Managing Online Delay

Appendix E

Screenshots of Input Pages on the Experimental Website

Figure E1. Studies 1 and 2: Input Page 1

Figure E2. Studies 1 and 2: Input Page 2

MIS Quarterly Vol. 37 No. 2—Appendices/June 2013 A9

Copyright of MIS Quarterly is the property of MIS Quarterly & The Society for Information Management and

its content may not be copied or emailed to multiple sites or posted to a listserv without the copyright holder's

express written permission. However, users may print, download, or email articles for individual use.