The perpetuation of entrained behavior during computer-mediated communication

Abstract

The present investigation examined the development and persistence of mutual entrainment during dyadic computer-mediated interactions. Entrainment refers to the synchronization and/or modification of endogenous temporal rhythms by external forces (Ancona & Chong, 1996; Kelly, 1988; McGrath, Kelly, & Machatka, 1984). Our findings indicate that entrainment readily occurs during computer-mediated communication, as our participants altered their rates of communication to match those of a confederate who was responding at a consistently slow pace. Subsequently, when participants were paired during a second computer-mediated interaction session, their initial rates of entrainment transferred to these new groups and served to mutually influence each other's communication rates.

Keywords:

• Entrainment
• Computer-mediated communication
• Behavioral synchrony

Interpersonal coordination is an underlying social process that is present in the majority of human social interaction. Bernieri and colleagues (Bernieri, Davis, Rosenthal, & Knee, 1994) theorize that interactional synchrony (coordinating the pace and rhythm of an interaction) and behavioral matching (the unconscious mimicking of physical behaviors) are the primary components of this interpersonal coordination. It has been argued that our ability to automatically synchronize and coordinate communication patterns and behaviors serves important evolutionary and social functions (Lakin, Jefferis, Cheng, & Chartrand, 2003). Entrainment falls under the interactional synchrony component of interpersonal coordination—humans have a number of endogenous temporal rhythms that can become “captured” or mutually entrained with dominant external agents (e.g., sleep cycle, menstruation cycle, mood, language, and so on). For example, one form of entrainment is exhibited when conversants mutually synchronize aspects of their individual speech patterns and behaviors (e.g., taking turns, frequency and length of utterances and pauses, etc.) with those of a partner (Giles & Ogay, 2006; Kelly, 1988; McGarva & Warner, 2003). Interpersonal coordination is a fairly robust phenomenon that has been demonstrated across a variety of situations and behaviors including verbal communication (Giles & Ogay, 2006; McGarva & Warner, 2003), nonverbal actions (Chartrand & Bargh, 1999; van Baaren, Holland, Kawakami, & van Knippenberg, 2004), virtual interactions (Bailenson & Yee, 2005), and mood effects (Kelly & Barsade, 2001). Researchers suggest that interaction synchrony between group members serves an adaptive function that may lead to increased positive intra-group affect, rapport, and enhanced task performance and decision making in task and social groups (Blount & Janicik, 2002; McGrath & Kelly, 1986; van Baaren, Horgan, Chartrand, & Dijkmans, 2004). Conversely, an inability to socially synchronize can result in frustration, misunderstandings, lack of cohesion, performance deficits, and even life-threatening consequences (Blount & Janicik, 2002; McGrath & Kelly, 1986).

The present study aimed to extend the theory of interaction synchrony by examining the presence and influence of entrainment during computer-mediated interactions. In particular we investigated whether the pace of communication in computer-mediated communication (CMC) would become mutually entrained during electronic interactions, whether these entrained behaviors would then persist when paired with similarly or dissimilarly entrained partners, and the impact of entrainment on interpersonal liking.

Social Influence and Computer-Mediated Communication

One commonly held tenet of social behavior is that interpersonal influence increases with proximity and social presence (Latané & Wolf, 1981). That is to say, the closer in proximity an individual, the more potential influence they have over one's thoughts and actions; however, there have been mixed findings regarding the overall impact of social presence on interpersonal influence during CMC. The cues-filtered-out model asserts that the reduction of social presence and physical proximity in CMC attenuates many of the social influence factors that regulate behavior in traditional face-to-face (FTF) groups (e.g., dominance, synchronization, conformity, etc.), leading to increased levels of disinhibited, hostile, and antinormative behaviors (Baltes, Dickson, Sherman, Bauer, & LaGanke, 2002; Kiesler, Seigel, & McGuire, 1984; Kiesler & Sproull, 1992). Based on the cues-filtered-out paradigm, it is possible that interaction synchrony may not develop during CMC. The decrease in proximity, social context cues, and social presence may prevent group members from mutually influencing each other's rates of electronic communication, resulting in deregulated communication patterns for discussants.

Alternatively, Walther's social information processing model (SIP) suggests that the lack of positive socio-emotional content and mutual influence in CMC may be the indirect result of increased temporal demands experienced by electronic groups, rather than the medium (Walther, 1992). When temporal demands are alleviated by providing groups with sufficient time to interact, electronic teams develop similar levels of positive and normative relational communication as traditional FTF groups (Hobman, Brodia, Irmer, & Chang, 2002; Walther & Burgoon, 1992).

Similarly, the social identity model of deindividuation effects (SIDE) asserts that increased anonymity and reduced social presence in internet communication serve to enhance social identification with groups and strengthen the impact of social forces on group behavior, rather than reducing social influence (Reicher, Spears, & Postmes, 1995; Walther, 1995). That is to say, instead of disinhibition, typical or dominant behaviors will be quickly adopted as normative standards during computer-mediated interactions as a direct result of the anonymity and reduced social cues that characterize the medium (Postmes, Spears, Sakhel, & De Groot, 2001). Providing support for these normative models, Kalman, Ravid, Raban, and Rafaeli (2006) found evidence for the development of normative chronemic patterns during asynchronous discourse (i.e., email), which has even less social presence and impact than synchronous CMC.

Therefore it is our contention that interaction synchrony (e.g., entrainment) will readily develop during synchronous computer-mediated discussions as communicators adapt to a normative pace established by a visually anonymous confederate. We predict that individuals will alter their communication pace to match that of a confederate who responds to their electronic messages at a consistently slow or fast pace. Although previous research has investigated interpersonal coordination in computer-mediated environments (Bailenson & Yee, 2005; Bailenson, Yee, Patel, & Beall, 2008), these studies were focused on the influence of behavioral mimicry as opposed to entrainment or interaction synchronization, as is the case in the present study.

The Perpetuation of Entrainment

Although interaction synchrony has been demonstrated to occur under a wide range of settings, it is not clear whether these entrained behaviors will persist over time, and in the absence of a dominant external pacer (Blount & Janick, 2002). Prior entrainment studies have only examined face-to-face groups that maintained stable compositions throughout their life spans (Kelly, Futoran, & McGrath, 1990). For example, Kelly et al. (1990) found that FTF groups would maintain their initially entrained work rates (i.e., working at a fast or slow pace to solve anagrams), even when working on subsequent tasks that required differing amounts of cognitive effort. While the researchers found evidence of “carry-over effects” in entrained participants, their research paradigm held the groups’ compositions constant throughout their task interactions. It also stands to reason that if one's behavior becomes mutually entrained with one set of teammates (e.g., working at a slow pace), then this entrained behavior may become internalized and persist even when interacting with a new set of team members (Jacobs & Campbell, 1961; Neilsen & Miller, 1997).

In the present research we examined not only whether individuals’ behaviors become entrained with one another during computer-mediated interactions, but also whether these behaviors will then perpetuate and mutually influence their new group members, even ones that were entrained at different rates than themselves. We expected participants to alter the speed of their electronic communication to match that of a confederate who responds in a consistently slow or fast manner. These entrained behaviors eventually would become normative and persist throughout subsequent electronic interactions. Participants who subsequently interact with similarly entrained partners (either fast or slow) will reinforce each other's response styles leading to the perpetuation of this entrained behavior, demonstrating mutual entrainment maintenance. However, when participants interact with dissimilarly entrained partners their incompatible communication patterns will result in desynchronized and moderated response rates compared to the similarly entrained dyads. These desynchronized partners will initially communicate at different rates until they mutually establish a new averaged norm that falls between the similarly entrained fast and slow dyads’ response rates.

Consistent with the similarity-attraction hypothesis (Newcomb, 1963), we also anticipated that participants who were initially entrained at similar rates would form more positive impressions of each other following their electronic interactions than those who were entrained at differing rates of communication (Blount & Janicik, 2002; McGrath & Kelly, 1986; Newcomb, 1963). Research has consistently found that the more attitudinally and behaviorally similar individuals are, the more liking and attraction they will have for each another (Newcomb, 1963), even when we are not explicitly aware of their similarities (Chartrand & Bargh, 1999).

Overview and Hypotheses

In order to study behavioral entrainment and its persistence, we divided each experimental session into two phases: an initial interaction with a confederate and a second interaction with another participant. Participants in the first phase interacted via a computerized chat program with a confederate who communicated using a consistently fast (immediate) or slow (lagged) response style. Participants in an external control condition did not interact with a confederate responder. Instead they were paired with another participant in order to gather a baseline rate of responding for comparison purposes. We expected participants to match the confederate's speed and that these rates would differ from the baseline control. We propose the following three hypotheses for the confederate interaction phase of the experiment.

• Hypothesis 1a: Participants who are paired with a consistently fast-responding confederate will have faster average response times compared to participants who are paired with a slow confederate.

• Hypothesis 1b: Participants who are paired with a consistently fast-responding confederate will have faster average response times compared to control participants.

• Hypothesis 1c: Participants who are paired with a slow-responding confederate will have slower average responses compared to control participants.

Participants who had previously interacted with confederates were paired with each other for the second interaction. Half of these dyads were comprised of “matched” participants (i.e., both participants had initially interacted with a fast or slow responding confederate) and half were comprised of “unmatched” participants (i.e., one participant initially interacted with a slow-responding confederate and the other communicated with a fast-responding confederate). After this interaction, participants privately rated how much they liked their dyadic partner. We propose the following four hypotheses for the participant interaction phase of the experiment.

• Hypothesis 2: Matched dyads, composed of participants who were entrained at a fast rate of responding, will have faster average response times than matched dyads, composed of participants who were entrained at a slower rate of responding.

• Hypothesis 3a: Unmatched dyads will have response rates that are slower than the matched dyads composed of participants who were entrained at a fast rate.

• Hypothesis 3b: Unmatched dyads will have response rates that are faster than the matched dyads composed of participants who were entrained at a slow rate.

• Hypothesis 4: Participants in matched dyads will form more positive impressions of each other than participants in unmatched dyads.

Method

Participants and design

A total of 73 undergraduate students participated in the study in exchange for extra credit in their introductory psychology courses. Two participants were recruited for each experimental session and randomly assigned to either the immediate (N = 25) or lagged (N = 31) condition. We later recruited 17 additional participants from the same population for an external control condition.

Procedure

Upon arrival, each participant was escorted by an experimenter to an isolated research cubicle containing a personal computer and two photographs that were lying face down on a table. Similar to Chartrand and Bargh (1999), participants were told the purpose of the study was to develop a revised version of the Thematic Apperception Test (TAT) that could be delivered and assessed in an online environment. They were informed that the TAT was a projective test that was frequently used by clinical psychologists in personality assessments and that they would be serving as pilot participants during the initial development of this instrument. It was stressed that their participation was needed to ascertain the feasibility of this delivery method for the TAT, and that the research was not interested in assessing their individual personalities.

Participants were told that they would be describing two pictures to another participant using the America Online instant messaging program (AIM). AIM is a synchronous chat program where users type messages into their instant messenger window and then hit “enter” to send the message to their partner. The AIM program keeps a running log of the entered conversation that is visible to both communicators. Using the chat program, their task was to describe the pictures, one sentence at a time, to a partner who would be doing the same but with a different set of photographs. Participants were free to talk about any aspects of the pictures or how the photographs made them feel during the interaction session. They were instructed to alternate their descriptive sentences with their partner as if they were having an actual online conversation (i.e., person 1 types a message, then person 2, then person 1, and so on). Students were never given any identifying information about the person with whom they were communicating.

When they were finished describing the first photograph students were instructed to type “done” in the chat window. If their partners typed “done” first and they still had comments to make about the photograph, they could continue to post messages without waiting for their partners to reply. Participants were told not to move on to the second photograph until both partners had typed “done.” When they were finished describing both photographs they were again instructed to type “done” and then notify the researcher. Lastly, all participants in the experimental conditions were told that they were randomly selected to send the first message and they could do so whenever they were ready. In actuality there was no random selection; participants always sent the first message to ensure that the confederates’ responses rates were consistent over the course of the interaction.

Entrainment manipulation

Confederates served as chat partners for each of the participants in the experimental conditions during the initial interaction sessions. Each of the confederates had five prepared photograph descriptions stored on their computers that they cut and pasted into the chat window. Each of the pre-scripted messages was a simple one-sentence description of an aspect of the photograph, which was purposefully neutral in tone (e.g., There is a close-up of a little boy who is around the age of 3 or 4). Following each of the participants’ messages, they would enter and send their scripted responses to their partners. Confederates were randomly instructed to respond to each of the participants’ messages in either a consistently Immediate (every 15 seconds) or Lagged (every 90 seconds) fashion throughout the interaction. Based on confederate and participant feedback during pilot testing, these response rates were viewed as sufficient yet realistic times for each of the two experimental conditions.1 After the confederate posted their fifth message they would immediately type “done.” If the student participant finished describing their photograph before the confederate's fifth message, the confederates kept responding in either an Immediate or Lagged fashion until all five of their messages were posted. Participants in the control condition were paired with each other in order to establish baseline rates of communicating during the entrainment phase of the study.

Following the computer-mediated interaction, participants were given an eight-item rating form to complete. The instrument contained five filler items about the photographs and computer set-up (e.g., How difficult was it to see the text on the screen?) mixed with three questions assessing their impressions of their chat partner and the interaction (How likeable was the person that you just interacted with? How smoothly would you say your interaction went with the other participant? Would you like to work with the person you just communicated with in the future?). All questions were formatted using a seven-point Likert-type scale with higher scores indicating stronger agreement with the statements. Responses to the three impression questions were combined to create a single liking index (α =.83).

Participant conversations

After completing the rating forms, participants who had previously communicated with a confederate were moved into new research cubicles containing two additional photographs and another personal computer running AIM. For the second interaction the two participants would now be communicating with each other in another computerized chat session. Again, participants were given no identifying information about their partners. As in the first interaction they were instructed to describe two pictures to their partners using the same alternating conversation-like format. Participants were randomly assigned to communicate with a partner who had been entrained at the same rate as they had (Matched Immediate or Matched Lagged) or with a partner who had been entrained at a different rate (Unmatched). Once participants were finished with the second interaction, they were given the same rating scale as before, thanked and debriefed. Again the participants’ responses to the three impression items were combined to create a liking index (α =.83).

Results

Confederate interactions

Entrainment

To assess whether the participants became entrained to the pace of the confederates’ communications, we conducted a one-way analysis of variance (ANOVA) comparing participants’ average response rates (in seconds) for the Immediate, Lagged, and control groups, over the course of their first interactions. Because Levene's test indicated a violation of the homogeneity of variance assumption, F(2, 70) = 4.92, p =.01, we used the Welch statistic to test Hypotheses 1a–c. The rates of participant responding significantly differed across the three groups F(2, 37.03) = 5.96, p <.01, ω²=.12. In support of Hypotheses 1a and 1c, Games-Howell post hoc comparisons indicated that the Lagged participants responded significantly slower (M = 42.00, SD = 16.54) than those who were paired with a fast-responding confederate (M = 30.07, SD = 8.79), p <.05 and marginally slower than control participants (M = 31.36, SD = 15.63), p =.08. Inconsistent with Hypothesis 1b, there were no significant differences in response rates between participants in the control and Immediate conditions, p >.05.

These data suggest that social entrainment may develop in computer-mediated contexts, as participants in the Lagged condition altered their pace of communication to match that of the slow-responding confederates. Although participants in the Immediate condition were significantly faster at responding than their Lagged counterparts, they were not statistically different from the control groups. However, it is likely that these findings represent a floor effect, in that the nature of the task may have precluded these participants from further increasing their rates of communication to match that of the fast-responding confederates. This notion seems even more plausible considering the relatively small standard deviation scores of the Immediate groups (SD = 8.79) compared to the control (SD = 15.63) and Lagged (SD = 16.54) conditions. It is also possible that our fast-responding confederates were responding at relatively normative rates for the task.

Impressions

Although we did not make a priori predictions regarding the impressions of the confederate responders, participants formed more positive impressions of the confederate in the Immediate (M = 17.00, SD = 2.52) than the Lagged condition (M = 13.26, SD = 3.61), F(1, 57) = 20.89, p <.01, η =.27. These results further suggest that the fast-responding confederate may have been communicating at a rate that was closer to “normal” than the slow responders.

Participant interactions

Entrainment persistence

In order to test the perpetuation of entrained behavior during the second interactions, a one-way ANOVA was conducted comparing the participants’ average response times (in seconds) in the Unmatched, Matched Immediate, and Matched Lagged conditions. The data revealed that the dyads’ average response times were significantly affected by their initial rates of entrainment, F(2, 26) = 3.55, p <.05, η =.21. Supporting Hypothesis 2, the Tukey's HSD procedure indicated that participants in the Matched Immediate condition responded at a significantly faster rate (M = 20.91, SD = 6.21) than those in the Matched Lagged condition (M = 29.40, SD = 9.70), p <.05. Participants’ response times in the Unmatched condition (M = 23.17, SD = 4.55) did not significantly differ from either of the Matched conditions, although their response rates did fall predictably between the Matched Immediate and Matched Lagged times, consistent with Hypotheses 3a and 3b.

Impressions

A planned comparison was conducted to assess whether participants who were paired with a similarly entrained partner would form more positive impression of each other compared to the Unmatched pairs. Contrary to Hypothesis 4, impression ratings for participants in the Matched conditions (M = 15.81, SD = 2.66 for Immediate groups and M = 17.20, SD = 2.98 for Lagged groups) did not differ from those in the Unmatched groups (M = 16.25, SD = 3.49), t(57) = 1.00, p >.05.

Content analyses

To rule out the possibility that the differences in the content of the participants’ discussions may have influenced their respective communication rates across the conditions, we conducted a content analysis on a subset of the computerized transcripts. Two independent coders, who were blind to the study's hypotheses, counted occurrences of procedural remarks (e.g., Who goes first?), photograph descriptions (e.g., the dog is furry), inferences (e.g., I think they are at a party), emotional reactions (e.g., This photograph makes me sad), and off-topic comments (e.g., I’m going to be late for my next class), and rated emotional tone of the discussions (using a five-point Likert scale with higher scores indicating more positivity) in randomly chosen transcripts from the Unmatched (N = 6), Matched Immediate (N = 6), and Matched Lagged (N = 5) groups. Coders also tallied the length of the discussions by counting the total number of comments made by participants in each of the transcripts. Using the Spearman-Brown formula, inter-rater reliability estimates for the averaged coder ratings were.92 for procedural remarks,.98 for descriptions,.83 for inferences,.91 for emotional reactions, 1.00 for off-topic comments,.91 for emotional tone, and.98 for discussion length.

We conducted a series of one-way ANOVAs comparing the content of the Matched and Unmatched groups from the second phase of the study using the average of the two coders’ ratings. The analysis revealed no significant differences in the content or length of the discussion logs between these groups (ps >.05). Because there were no differences in the discussion content or length between the Matched and Unmatched groups, it does not appear that the substance of their conversations influenced the pace differences between these groups.

Discussion

This study was the first to test the development and influence of entrainment during computer-mediated interactions. Our findings revealed that participants will readily change and alter the pace of their electronic communication to match that of a confederate who is responding at a consistently slow pace. These findings are intriguing because synchronous chat is not subject to the production-blocking effects seen in traditional face-to-face communication, where waiting for one's speaking turn blocks one's own thoughts and communication. During CMC, communicators are able to type messages at the same time without blocking their partner's communication attempts. Although we told the students not to “enter” their messages until their partner had responded to their previous post, participants were able type their thoughts into the message window while waiting for their partner to respond. Even though the participants in the lagged condition had more “lag time” in between messages to view the pictures and type out their responses, they still waited 12 seconds longer on average to send messages to their partners than those in the immediate condition, who had significantly less time to type.

We did not find initial evidence of entrainment for participants communicating with fast-responding confederates. Although entrainment during CMC could be a unidirectional phenomenon, it is more likely that these participants were suffering from a floor effect when describing their photographs to the confederates. Due to the constraints of the task, they may have been physically unable to continuously generate photograph descriptions at a fast speed, regardless of the confederates’ accelerated pace. It is also plausible that the confederates in our immediate condition were in fact responding at a relatively normative pace for the task, because the average response rates of the participants in this condition did not significantly differ from those in the control group. Future researchers may want to use an alternate task or further increase the speed of confederate responding in order to determine whether entrainment is bidirectional.

Even without direct evidence for entrainment in the Immediate condition, when matched with a similarly entrained partner participants maintained their initial pace of electronic messaging during subsequent interactions. Participants who were paired with a similarly entrained partner (immediate or lagged) maintained their initial rates of entrainment during subsequent interactions, demonstrating mutual entrainment maintenance. These results are consistent with classic research on norm transmission in FTF groups (Jacobs & Campbell, 1961; Nielsen & Miller, 1997), as well as modern social influence theories of computer-mediated interaction. Both Walther's hyperpersonal model and the SIDE model posit that there are nonverbal social context cues present during CMC that serve to establish, maintain, and regulate normative behavior during electronic interactions (Reicher et al., 1995; Walther 1995). Consistent with these models, our findings suggest that electronic groups will readily establish normative standards regarding communication pace, which will then persist even in the absence of the original dominant pacer.

Organizational researchers continue to stress the importance of examining the cognitive and behavioral consequences of our escalating reliance on online interactions for both social and professional purposes (Haythornthwaite & Nielson, 2006). In light of the increasing use of synchronous chat in organizational settings (Pew Internet & American Life Project, 2004) there are numerous practical implications for these findings in the virtual workplace. For example, supervisors may want to consider the composition of their digital workgroups, not just in traditional areas of knowledge and experience but also in terms of their normative rates of communication. Task groups may perform suboptimally if their mutually influenced rates of communication are disrupted or unsuited for the task (Kelly et al., 1990), which could then lead to an increase in negative intra-group affect (Blount & Janicik, 2002; McGrath & Kelly, 1986).

Although we did not examine performance per se, participants did form more positive impressions of confederates who responded immediately to their messages than those communicating with Lagged confederates. These findings may be partially attributed to the fact that the 15-second response time of our Immediate confederates was more similar to the “normal” participant response times for the task than that of the Lagged confederates, as evidenced by the similar response rates of our control and immediate participants. Research has demonstrated that groups that feel more “in synch” have more positive intra-group affect than those that feel out of synch (Blount & Janicik, 2002). It is possible that participants in our Immediate condition were more easily able to synchronize with their confederate partners’ communication rates and this resulted in more positive intra-group affect.

Furthermore, we anticipated that the similarly entrained groups in our second interactions would also experience increased positive intra-group affect. Because Matched participants were entrained at relatively the same rate, it stands to reason that these dyads would be more “in synch” than the Unmatched groups, and consequently develop more favorable impressions of each other. However, our data failed to support these claims. One explanation may be that our three-item impression scale did not adequately measure the breadth of the participants’ emotional responses to being in or out of synch with their partners. Also, Blount and Janick (2002) found that participants’ subjective feelings of being in or out of synch, and not their entrained behaviors, influenced their affective reactions. Our research did not specifically assess whether the Matched participants felt more “in synch” than the Unmatched dyads. Participants in our Matched groups may have felt somewhat out of synch with their new partners because their average response times were different from those of the original confederates (roughly 30 seconds vs 15 for Matched Immediate groups and 42 seconds vs 90 seconds for Matched Lagged groups). Future research may need to employ a more sensitive impression measure that examines participants’ subjective feelings of being in or out of synch with one another.

Limitations and Future Research

Time, pace, and synchronization are crucial but often overlooked issues in social and organizational behavior, especially in terms of our computer-mediated interactions (Campbell & Stasser, 2006). The current study examined the development and perpetuation of entrainment during CMC. The results indicate that interaction synchrony does readily occur in computer-mediated interactions and will persist through subsequent interactions, even when interacting with new group members. It should be noted that the “one-shot” dyads in our study were working on an artificial task, which required very little cognitive effort. It would be interesting to examine groups working on more difficult and meaningful tasks that may serve to enhance or interfere with these entrainment effects during computer-mediated interactions. Also one could investigate the development of entrainment during more natural CMC interactions (e.g., pre-existing friends instant messaging over the Internet or text messaging via cellular phones).

While our research demonstrated that entrainment is perpetuated in computer-mediated groups, we do not know about the longevity of these effects. We only found evidence of entrainment transmission over a single generation, whereas previous studies have found similar types of norm transmission in FTF groups for up to six generations (Jacobs & Campbell, 1961). We speculate that entrainment during CMC will persist over more than one generation as well. This notion is supported by Walther's (1992) SIP model, which predicts that electronic groups will become increasingly cohesive and normative over time as the team members increase their exchange and sampling of information. While these “generational” studies, including our own, generally utilized the same types of tasks across generations, it could also be argued that these entrainment effects would persist even when groups are exposed to different types of tasks (Harrison, Mohammed, McGrath, Florey, & Vanderstoep, 2003). Future research should look at the perpetuation of these entrainment effects in subsequent generations, and the factors that mitigate their transmission.

Notes

¹ Initially we attempted to use faster response time for confederates in the Immediate condition; however, 15 seconds was the fastest that confederates could reliably cut and paste their pre-formatted responses into the IM chat window.