Bystander Intervention in Video Chat 1 RUNNING HEAD: Bystander Intervention in Video Chat Bystander Intervention in Video Chat Troy Effner Carnegie Mellon University
Bystander Intervention in Video Chat 2 Abstract This paper explores the aspect of video conferencing and its effect on prosocial behavior. Modeling the study as closely to Darley and Latane (1968) study on bystander intervention we prove that the addition of facial cues increase helping behavior over a computer-mediated communication.
Bystander Intervention in Video Chat 3 Introduction Communication technologies have expanded and diversified the way that we interact with each other. Text messaging, email, video conference, public and private virtual spaces are technologies currently available to a significant number of people. The sender of communications over channels may have different intents and expectations than what the receiver of the communication may have. Additionally these communication channels are becoming more ubiquitous in our every day interactions. Mobile phones, Laptops, ipads are all now fully connecting us to each other via email, texting, text chat, and video chat to name a few. Because new ways to interact with each other are available, there is also opportunity to observe how bystander intervention plays a role in these various mediums. Darley s original paper (Darley & Latane, 1968) on bystander intervention utilized an audio intercom system as a communication channel to measure bystander intervention. The study measured the time a subject took to report that there was a potential problem with another person they were communicating with on an intercom system. The group size of this study varied resulting in different amounts of time it took to report a problem to the experimenter. Because the subjects in larger groups took longer to report that there was a problem, Darley posited this demonstrated that diffusion of responsibility increased as the group size increased. Studying new media, such as online chat rooms, also gives insight into how people behave in groups over this new communication medium. After a new story broke about how Larry Froistad confessed to the murder of his daughter to 200 people in a self-help chat room, P.M. Markey took the opportunity to study chat rooms to see what level of diffusion of responsibility actually appeared. Markey conducted a study (Markey, 2000) of Yahoo! Chat
Bystander Intervention in Video Chat 4 examining how many people would respond to the helping request Can anyone tell me how to look at someone s profile? The results showed that bystander intervention in Internet chat groups followed the same pattern as in Latane and Darley s theory (Darley & Latane, 1968) where the larger the group was the longer it took for people to respond. Additionally, it was shown that if specific people were asked for help, responses came faster. Email was also assessed for bystander intervention by Barron and Yechiam (Barron, 2002). The study was conducted by obtaining 240 university email addresses of graduate students, doctoral students, and faculty. In the three scenarios that were tested, emails requesting information were sent to either one person directly, to one person along with three other email addresses that have the same @university_name.edu, and finally email sent to person along with three other email addresses from random email services such as hotmail or gmail. The measure of this study was how many people responded along with the quality of information provided. Again, bystander intervention was demonstrated in that the people who were emailed directly responded more often than did the scenarios where other people were also on the TO: line. The quality of help was also measured and results indicated that in the scenarios with multiple people in the TO: line, better help was given when the recipient presumed that others were less capable of helping by virtue of the non-university email addresses. Lastly, video conferencing software has been tested within the context of framing of the other person and its impact on empathy (Nguyen & Canny, 2009). When two subjects are having a conversation over a video conferencing system, the framing of the body and face impacts how much empathy is demonstrated when the two meet for the first time. When subject A and subject B have a conversation over video conference the framing of the other person has significant implications in the empathy created from that conversation. In situations where subject B is
Bystander Intervention in Video Chat 5 framed so that the upper body can be seen subject A is more likely to help subject B in the second part of the study where Subject B drops a pencil holder. When the framing subject B is restricted to only the face subject A is less likely to help in the same pencil dropping scenario. Relevant to the video conferencing research is the study on Accurate Identification of Fear Facial Expressions Predicts Prosocial Behavior (Marsh, Kozak, & Ambday, 2007). In this study individuals who could recognize fear facial expressions were more likely to be prosocial. By adding a visual channel to the communication there is an increased chance that prosocial behavior. These studies have covered some of the computer-mediated communication landscape, but one area yet to be directly explored is emergency helping and bystander intervention in video conference sessions. In the original bystander intervention paper, the self reported, primary cause of inaction was that the subjects weren t sure what was happening (Darley et. al, 1968). Video conferencing is a medium where both audio and visual cues can be communicated and mistaking events are less likely to occur. Additionally, because there are visual cues, there is an expected increase in empathy due to the ability to see facial expressions (Marsh et al. 2007). Given this previous research I propose that a study that incorporates visual facial cues as a new independent variable conducted over video conferencing software. I hypothesize that the dependent measure of response times of subjects will decrease as compared to the Darley et. al. (1968) paper. Methods Sixty subjects for the study were recruited from the Carnegie Mellon student body and offered $15 for their time. Not only should this ensure randomization of the participants it would also give us subjects who would be most familiar with video conferencing software. The study
Bystander Intervention in Video Chat 6 will consist of a control group of thirty people and a test group of thirty people. Each of these will contain three conditions for testing bystander intervention with a independent variable of group size as well as presence of facial cues provided by the video conference. This study will attempt to model the Darly et. al (1968) study. The test group will be tested with just the subject and the presenter. The second group will consist of the subject, presenter, and ostensibly one other participant. In the final scenario the group will be the subject, presenter, and four other fictional participants. The reason for the fictional participants will be so that the web conference can actually be a recording and be replicated with the testing scenario. As the subject arrives to the testing lab they will be told that they are there to help assess video conferencing software for usability and its interface. The subject is then led into a room with a desk and a computer that is already set up with the video conference software up and running. The examiner assists the subject in typing in their login of participant 1 to sign into the system. As he logs in to the conferencing software he is presented with a video feed of a presenter who acknowledges his presence by saying I see participant 1 has joined the meeting has joined the meeting. The examiner says that the presenter will walk the subject through the rest of the study and that he has some paper work to do just outside the door. Before arriving at the study, participants were randomly assigned to one of three conditions. Condition 1 the presenter indicates to the target participant that he is the only one in this conference call. To represent a large scale conference call where there may be a hundred attendees his microphone has been muted so he won t be able communicate with anyone. He will be asked to do some tasks in the interface to validate that they are easy to accomplish. As the presenter is instructing on the participant on the features of the system the presenter begins to look pale and faints off his chair and out of the camera s view.
Bystander Intervention in Video Chat 7 Condition 2 modifies the first condition in the introduction when the presenter acknowledges that target participant and as well as one other participant who is also connected to the video conferencing system. This will be supported by having a list of attendees and the subject being able to see Participant 2 in the list. In the protocol the examiner will state that the other participant is there to help test the server load to ensure it can handle the work load. The same steps are followed where the presenter begins instructing the participants on how to use the system and then faints off screen. The third and final condition is with the presenter, the target participant, and then four other participants. As before the presenter goes through the list of participants acknowledging that they have joined the session and proceeds as the same with the prior two scenarios. Again the protocol for the examiner will be to state that there are more people on to test and see how more people impact server workload. The control group will be divided into the same group sizes and walked through the same steps as the test group without the benefit of the video images. The examiner will indicate that a part of the video conferencing service is down, but that won t affect the voice communication and the primary subject can still follow along. Again the examiner will log the subject in as participant 1. In this scenario the presenter will begin the presentation the same way indicating that participant 1 has joined and that they will not be able to directly communicate with each other. After beginning the process the presenter will articulate that something is wrong and begin making as if he were in distress suffering from a heart attack or seizure. The dependent variable we are testing is the time to help as a factor of how many people are in the group that is being presented to. When the presenter faints the examiner will begin
Bystander Intervention in Video Chat 8 timing to see how long it takes for the subject to notify the examiner of a problem. The examiner will be sitting at a desk just outside the doorway. If the subject has not come out of the room after six minutes, the examiner will enter the room to disclose the true nature of the experiment. Results The results of the time to respond measurements in the video will be put through a oneway ANOVA. From these results I expect that as group size increases the speed to which the subjects report a problem will increase. Concurrently the percent that report a problem will decrease as the size of the group increases. Table 2: Effects of group size and speed of response Control group without video chat present Group Size N % responding Time in Sec 2 (S & Presenter) 10 85 52 3(S, Presenter, & 1 other) 10 62 93 6(S, Presenter, & 4 others) 10 31 166 Table 1: Effects of group size and speed of response Test group with video chat present Group Size N % responding Time in Sec 2 (S & Presenter) 10 90 50 3(S, Presenter, & 1 other) 10 85 65 6(S, Presenter, & 4 others) 10 60 80
Bystander Intervention in Video Chat 9 After the test proves that they are valid variances I will perform a two way ANOVA on the groups to see if there is significant effect of adding in Video Conferencing as a component to this study. ANOVA SUMMARY Source SS df MS F P bg 28962 5 Video 4418 1 4418 4418 5.00E-07 GroupN 21693 2 10847 10847 <.0001 r x c 2851 2 1425.5 1425.5 <.0001 wg 12 12 1 These results indicate that both group size as well as presence of video conference has a definite impact on helping behavior. Discussion Given these results we can compare them to Darley s findings in his original Bystander Intervention paper. With the addition of visual cues and the increased helping behavior related to them one can infer that the video conferencing scenario people are more likely to help and help faster than reported in that paper Once the study has been completed we would need to assess if the participant was convinced that other people were present in the video chat and thought that they were capable of helping. Having sufficient understanding that other people are present and are capable of helping is a key factor in the bystander effect and to accurate compare that criteria would need to be satisfied. Listing the participants as they join the presentation may be enough, but one other potential further research idea would be to model the study closer to Darley s by creating an automatic switching feature where the participants can see each other, but not interact.
Bystander Intervention in Video Chat 10 Another potential research idea that stems from this is investigating how people organize each other in emergency helping situations over computer-mediated communication. This would look at if the participants were able to communicate with each other who would be the first to attempt to coordinate the others and what structure that would take. References Darley, J, and B. LatanÃ. (1968). Bystander intervention in emergencies: diffusion of responsibility..journal of personality and social psychology, 8 (4), pp. 377-383. Markey, P. (2000). Bystander intervention in computer-mediated communication. Computers in Human Behavior, 16 (2), pp. 183-188. Barron, G, and E. Yechiam. (2002). Private e-mail requests and the diffusion of responsibility. Computers in Human Behavior, 18 (5), pp. 507-520. Marsh, A, and M. Kozak, and N. Ambady. (2007). Accurate identification of fear facial expressions predicts prosocial behavior.. Emotion, 7 (2), pp. 239-251. Nguyen, D. T, and J. Canny. (2009). More than face-to-face: empathy effects of video framing. CHI '09 Proceedings of the 27th international conference on Human factors in computing systems, pp. 423-432.