Seeing Red Running head: MAJORITY INFLUENCE ON A COLOR RECALL TASK

1

Why did they all see red? Contrasting the effects of normative vs. informational influence on a color recall task Bryan Kennedy (15855492) bryank@berkeley.edu Dang Dao, Diana Deguzman, Takuma Psychology 101 Professor Wickens Section Fri 9-1;1Sullivan

Seeing Red Abstract The effects of conformity were contrasted in an anonymous verses a group environment. Thirty-five participants completed a color recognition task, which was refined during a

2

series of pretests to achieve a better than chance accuracy rate. The subjects were divided into 3 conditions, which differed only by the manner in which they responded: a blank sheet of paper, a paper with responses already filled in, and a paper written in a group setting. While the effects of conformity among the three conditions did not differ significantly, some interesting observations can be noted from the findings.

Seeing Red Introduction The purpose of this experiment is to better understand conformity by examining

3

the relationship between normative and informational influence found in group behavior. Forty years ago, Milgram et al (1963) established the importance of any study on conformity with his findings that a majority of individuals would obey an authoritarian leader to no end. In his groundbreaking study, 60% of subjects ended up administering what they believed to be fatal electric shocks to others. Solomon Asch (1951) organized a study that looked specifically at majority influence, finding that 75% of people would conform to an obviously incorrect majority at least once. Participants were ushered into a room with seven confederates of the study. They were seated such that they were always the second-to-last in the row, so they would know the answers of six of their peers before they themselves answered. Over eighteen trials, they were asked to view a slide depicting four lines or varying length. It was their job to determine which of the three lines on the right were the same as the line on the left. The task varied in ambiguity, but in twelve of the eighteen trials, the confederates were instructed to give a consistent and incorrect answer. That is, if Line 1 was the same as Line 2 in the particular slide, they would all answer Line 3. All answers were given aloud, so that subjects were keenly aware of the answers given by their peers. In the control experiment, where no confederates were present, only one percent of trials resulted in an error. In the non-control condition, however, an amazing 75% of students made at least one error. This suggests that these students were susceptible to conforming at least some of the time when faced with an overwhelming majority. Leading researchers in the field have since distinguished between two forms of conformity in groups: normative and informative. In a study by Deutsch and Gerard (1955), subjects were given a similar task to that employed by Asch, but were separated

into three conditions. The first condition involved a face-to-face interaction, whereby subjects gave their answers in a situation where they could see (and be seen) by the confederates. The second condition added on a group goal, so that the subject felt they were working as part of a team with the confederates to win a prize for the team with the best score. A third condition removed the visual interaction of the first group, but allowing subjects to answer anonymously. In the group goal condition, subjects produced the greatest number of errors, suggesting that people will conform most when faced with a present and attentive majority (48%). The conformity was least in the anonymous activity (23%), suggesting that group influence is weakest when the group is obscured. The truly interesting finding, however, was that subjects who responded in the anonymous condition still demonstrated a significant amount of influence in their answers over the control group. It seemed then that there were two powers at play here rather than one. First, there was obviously a strong influence caused by the face-to-face interaction. Second, there also appeared to be some influence caused by the answers themselves. These findings led Deutsch and Gerard to propose two new terms to describe group conformity. In normative influence, an individual will avoid taking a minority position because they fear being isolated and scorned by the group. In informational influence, however, this individual joins the majority with the thought that “truth lies in numbers”. They see the answers of the majority as a fact that supports their answer, which may or may not outweigh the other factors they use to arrive at their answer. In their study on conformity in a word-recognition task, Schneider and Watkins (1996) found that subjects were likely to conform to responses given by a single confederate, despite the correctness of their initial response. In a follow-up to this study, Wright et al (2000) found that this finding extends to the recognition of images seen

previously, and does not depend on the contributions of a confederate. Two subjects were given individual photobooks filled with different cars, and were asked to work together on a recognition task. Unbeknownst to the subjects, their photobooks were not the same, so there were many instances where their recognition of a particular car would not have been seen before by the other subject. However, during discussion, it was found that those that spoke first were most likely to be successful in “converting” the opinions of the other, despite the fact that they may have not seen that particular car. This finding suggests that primacy of response can have an effect on answers. A face-to-face interaction is not required to induce conformity. In a wordrecognition experiment conducted by Reysen (in press), subjects were convinced that their responses entered into a computer would be seen by a fellow participant, and they would likewise be able to view their answers. The setup was illusionary, however, since the answers being fed back to the terminal were computed using a simple algorithm. However, it was found that subjects still conformed to the answers provided by this “virtual confederate”. Walther et al. (2002) took this a step further by placing four subjects in a room with computers. Each participant was given the impression that they would be seeing the answers of everyone else in the group; however, this was illusionary, as all participants acted independently and saw the answers provided by a computer program. Subjects were then presented with a variety of slides, and their task was to recall these slides afterward. The slides were varied in salience, in that some pictures were of items from the same lexical class (example: hammer, wrench, saw, etc), while others were completely disparate (example: cat, house, mountain, etc). This resulted in the more distinctive slides being recalled with greater accuracy. It was also discovered that subjects demonstrated

far less conformity for incorrect answers on slides that were most salient; that is, the surer they were of the answer, the less they were likely to conform. For the present study, the experimenters wished to distinguish between normative and informational influence using a less salient task than that employed in experiments by Asch, and Deutsch and Gerard. It was hypothesized that such a method would increase conformity over these original studies. To this end, we developed a color identification and recall task, whereby subjects were shown a series of colored slides, and then asked to recall the color of the first slide. According to theories on the serial position effect of memory, this task of recalling the first slide should be easier than remembering the other slides (Murdock 1962). To distinguish between the two different types of influence, a 3group design was formed. In the control group C, subjects were individually presented with the task, and were asked to write their answers on a sheet of paper. In group A, the conditions were the same except that the answer sheet already had bogus answers from “past participants” indicating the same incorrect answer. In group B, subjects were placed in a room with four experimental confederates, and wrote down their answer on a sheet containing the confederates’ incorrect responses. It was proposed that such a design would result in a visible increase in conformity, with group C being the lowest, followed by B, and then C with the highest demonstrated conformity. Method Participants Thirty-five volunteers were selected from the UC Berkeley campus. Group C had 13 participants, Group B had 11 participants, and Group A had 12 participants. For Groups C and A, two experimenters approached individual patrons of Pat Brown’s café, soliciting subjects by describing the general procedure and time involved. Participation in Groups C and A required approximately 5 minutes and no rewards were offered.

For Group B, one experimenter approached loiterers in the Dwinelle Hall quad, describing the general procedure and offering them an incentive for their participation. Participation for Group B required approximately 10 minutes and subjects were rewarded with a candy bar if requested. The data from one subject in Group B had to be disregarded due to the fact that he arrived when one of the confederates was out of the room, and thus sat in the incorrect spot. Apparatus The slides were presented using a Dell Latitude D600 Pentium-M 1.8Ghz laptop running Windows XP SP2, with 1GB of RAM, and an ATI Radeon 9000 Mobility 32MB video card set at 32-bit color. The built-in SXGA+ (1400x1050) LCD screen was set to the highest brightness setting, which was judged during our pretests to be sufficiently visible in our two test environments. Microsoft Powerpoint 2003 was used to generate and display the slides. Colors were selected using the FOCOLTONE® Solid Matte color wheel provided by Adobe Systems. Four bright, primary colors were chosen for the slides, all were judged during the pretest to have enough visual saliency to be distinguished and named consistently: Blue (FOCOLTONE 1076), Red (FOCOLTONE 2237), Yellow (FOCOLTONE 3396), Green (FOCOLTONE 7004). Instructions were presented as white text on a black background. The answer sheets in all three groups were identical with the exception of the differing treatments used in Group A/B. Procedure Subjects in Groups C (control) and A were selected during two mid-day (11am – 1pm) sessions at Pat Browns café on the north side of the UC Berkeley campus. Subjects were chosen non-systematically from the patrons inside the café. Though gender was not overtly recorded in this study, an attempt was made to recruit an equal number of males and females for all groups.

One experimenter approached a prospective subject to request their participation, and, if they accepted, another experimenter brought over the laptop system and answer sheet and set them up on the table the subject was sitting at. The first experimenter then advised the subject to read the instructions on the screen carefully, and then both experimenters faced away from the subject to minimize potential confounds. To assure consistency across subjects in Groups C and A, all instructions were presented on the laptop screen for fifteen seconds prior the test. The test itself consisted of twelve slides, three each of the four colors. Each slide was shown on the screen for two seconds, without any additional delay between them. The order of the slides was chosen non-systematically and was the same across all subjects and groups, though no two of the same color slides ever appeared back-to-back. The first color was blue, followed by red, yellow, and green. At the conclusion of the slides, the subjects were instructed to locate the answer sheet and write down their answer to the question prompt that appeared on that sheet. For all subjects and groups, the prompt was: “What was the color of the first slide?”. In Group C, this answer sheet was blank (Figure 1), in Group A, it contained the responses of four fake previous participants (Figure 2), each having “answered” red. These answer sheets were shuffled, so the experimenters did not know the subject’s group assignment until after they returned the sheet. Subsequent to subjects indicating they were done, the experimenters retrieved the laptop and answer sheet, and debriefed them if requested. Subjects in Group B were selected during one mid-day session (10am to 2pm) session in Dwinelle Hall. They were drawn non-systematically from those in or around the Dwinelle quad. One experimenter approached potential subjects with the guise that

she needed one more subject for her experiment, to fill the place of a no-show. Each was promised a candy reward for their participation. The participants were led by the experimenter to a small room in Dwinelle Hall, where four confederates were already waiting. The chair closest to the door was left empty for the subject to sit in. The experimenter then shut off the lights and read the instructions from the laptop screen aloud to the five now seated around the table. Following the slide display, a blank answer sheet was passed around from left to right, with each confederate answering red. That way, the subject was the final person to write down their answer. At the conclusion of their writing down their answer, the subjects were debriefed. Most subjects in Group B were debriefed due to the involved nature of the activity; subjects in the other two conditions were less likely to ask for a debriefing, so they were simply thanked at the conclusion of their participation. Those who asked to be debriefed were told that we were studying group conformity for a Psychology 101 Research Methods class. Analyses The data were analyzed using a chi square (χ2), with the level of significance set to 0.05. Results Our null hypothesis was that our treatment would have no effect on the data in each of the three conditions (μC=μA=μB). Our alternative hypothesis was that our conformity treatment would have some effect on the mean number of correct subject answers among the three groups (μC≠μA≠μB). In order to reject the null hypothesis, at the level of significance stated above, and using our degrees of freedom (df) of 2, the χ2 critical value is a minimum of 5.99.

For our expected and observed frequencies, we compared the number of subjects who responded “red”, the incorrect answer used for our conformity treatment. A significant value in this statistic could indicate that our subjects conformed to the information presented to them in groups A and B. However, the χ2 value obtained was 0.7448, well below the critical value of 5.99. In addition, we compared the number of subjects who responded blue, the correct answer. A significant value in this statistic would show that the mean of each group differs significantly, and could indicate that subjects in each group differed in their level of confidence in their answers. Here again, however, our χ2 value of 5.484 was below the critical value of 5.99 required for rejecting the null hypothesis. According to these results we have failed to reject our null hypothesis and must assume that the treatments we employed to induce conformity did not have a significant effect on the answers given by our subjects. Discussion While we ere unable to reject our null hypothesis, this was very likely the result of having too few participants in our three groups, rather than an inherent flaw in the design of the study. This was a limitation of the resources available to us, as students. Though it did not reach the .05 level, we did note an unexpected finding in that more subjects indicated the correct answer in Group B than in Group C. There are two possible explanations for this finding. First, the conditions inside Dwinelle Hall were far more ideal than Pat Brown’s café. Not only was the room darker, allowing for a better view of the screen, but the overall environment was less distracting. It is entirely possible that students in Groups C and A were distracted by the commotion in the busy café environment, and thus did not attend to the slides as well as those in Group B. Though also not a significant finding, this would explain why Group C exhibited a higher overall

conformity to answering “Red”; Deutsch and Gerard’s study clearly demonstrates that those who paid less attention to the stimulus were more likely to conform to the responses of confederates. Additionally, it is possible that our Group B environment was too informal and relaxed. Fairey (1986) examined the relationship between self-doubt and attention to the stimulus when responding in group situations. It was found that, in a high-group-pressure condition, those with high self-doubt would pay less attention to the stimulus than those with low self-doubt. The reverse was true for subjects in the low-group-pressure condition, where high-self-doubters paid more attention to the stimulus than those with low self-doubt. In the present study, Group B could definitely be considered a lowpressure group because of the lack of any punishments for an incorrect answer. Subjects in this group may have also been fairly apprehensive about the experience, considering that they were given few instructions and entered into a room already filled with other “subjects”. It is likely that the low-stress environment, ideal viewing conditions, and the relative nervousness of the students contributed to the relatively low exhibited conformity. We were keenly aware that holding Groups C and A in a different location than Group B would have a detrimental effect on our results. Ideally, we would have preferred to conduct all our experiments in Dwinelle Hall, but the difficulties in securing a room for any significant amount of time proved troublesome. To control for the many confounds caused by this discrepancy, future experimenters with more time and resources should conduct all three groups under similar conditions. While in Group B we were attempting to study the effects of conformity in a group, it is possible that our subjects viewed themselves less as a part of a group and more as an individual against a group. For normative influence to take place, Turner

(1978) suggests that individuals must identify with a group through some means. Most often experimentally, this takes shape as a shared group reward of performance, though group identity can be established through many other social factors. However, the scope of the current experiment did not allow us to adequately foster such a group identity, and thus, it is likely that subjects in Group B exhibited less cohesion because they simply did not trust the other group members, nor feel it important to filter their answer to align with the group consensus. As a contrasting example, Asch established group trust by running the same subjects and experimental cohorts across many different trials, allowing the cohorts to build trust and demonstrate expertise in the minds of the subjects. A more elaborate version of the current study might increase the number of trials in all groups to duplicate this cohesive effect. Lastly, it is also possible that our convenience sample of UC Berkeley students is not representative of the population at large. A study conducted by Pasupathi (1999), for example, found that older individuals generally showed less social influence on a task involving judgments of geometric shapes and emotional faces, than their younger counterparts. Bond and Smith (1996) also found that the tendency toward social conformity has declined in the US since the 1950’s (when Asch’s groundbreaking study was published), due to the increase in pressures of be individualistic. Further confining the potential findings of our study, is their other finding that conformity differs across nations; individuals raised in more collectivist nations exhibit a greater predisposition for conforming. Though many significant confounds prevented us from arriving at any significant results, this experimental design is a worthwhile and effective exploration of the effects of conformity within groups. As one subject asked of the experimenter before being debriefed, “why did they all see red?”

References Asch, S.E. (1951). Effects of group pressure upon the modification and distortion of judgments, in Brown, Rupert (2000) Group Processes. Malden, MA: Blackwell Press. Bond, R. and Smith, P.B. (1996) Culture and Conformity: A meta-analysis of studies using Asch’s line judgment task. Psychological Bulletin, 119, 111-37. Deutsch, M. and Gerard, H.B. (1955) A study of normative and informational social influence upon individual judgement. Journal of Abnormal and Social Psychology, 51, 629-36. Fairey, Patricia J. (1986). Conformity and attention to the stimulus: Some temporal and contextual dynamics. Journal of Personality and Social Psychology, 51, 315-24. Milgram, S. (1963) Behavioral study of obedience. Journal of Abnormal Psychology, 67, 371-8. Murdock, B.B. (1962) The serial position effect of free recall. Journal of Experimental Psychology, 64, 482-8. Pasupathi, Monisha (1999) Age differences in response to conformity pressure for emotional and non-emotional material. Psychology and Aging, 14, 170-4. Reysen, M.B. (in press) The effects of conformity on recognition judgements. Memory. Schneider, D.M., and Watkins, M.J. (1996) Response conformity in recognition testing. Psychonomic Bulletin & Review, 3 (4), 481-5. Turner, J.C. (1978) Social categorization and social discrimination in the minimal group paradigm, in Brown, Rupert (2000) Group Processes. Malden, MA: Blackwell Press. Walther, Eva et al. (2002) Conformity effects in memory as a function of group size, dissenters, and uncertainty. Applied Cognitive Psychology, 17, 793-810. Wright, Daniel B. et al. (2000) Memory conformity: Exploring misinformation effects when presented by another person. British Journal of Psychology, 91, 189-202.