You're like Me, No Matter What You Say: Self Projection in Self-other Comparisons

The present research explores self-projection as a determinant of judgments of another person, but departs from past research by allowing research participants to gain individuating information about the other person in a conversational format. 72 college students first completed a checklist of bad study habits. Next, in pairs, they discussed their study habits while being video-taped. Participants then rated their own study habits and their conversation partner's study habits. Participants' ratings of their own smdy habits robustly predicted their ratings of their partners'study habits. The number of bad study habits the partner mentioned during the conversation had no significant effect on participants' ratings of their partner. By seeking common conversational ground, discussion partners appear to have created a perception of greater similarity between themselves and the other person than that which objectively existed. Social psychology has a long history of identifying biases in our judgments of others. When we are asked to predict how the average person or the majority of our peers will respond to a request or attitude questionnaire, our answers often reflect self-projection: We think others will respond the same way we would (i.e., the false consensus effect-However, when we are asked to compare ourselves to others on evaluative dimensions or to assess the likelihood of certain life events happening to us as compared to others, we are not as willing to share o w responses for owselves with others. Instead, these judgments tend to reflect a self-enhancing bias

Social psychology has a long history of identifying biases in our judgments of others.When we are asked to predict how the average person or the majority of our peers will respond to a request or attitude questionnaire, our answers often reflect self-projection: We think others will respond the same way we would (i.e., the false consensus effect -Krueger & Clement, 1994; Ross, Greene, & House, 1977).However, when we are asked to compare ourselves to others on evaluative dimensions or to assess the likelihood of certain life events happening to us as compared to others, we are not as willing to share o w responses for owselves with others.Instead, these judgments tend to reflect a self-enhancing bias (Alicke, 1985;Weinstein, 1980), particularly when people are asked to compare themselves to generalized others (e.g., "other students at your university").
Outside of social psychology studies, judgments of other people often emerge from actual interactions with individuals (e.g., the other actor auditioning for the same part in a play or the other patient waiting in the doctor's office).Alicke, Klotz, Breitenbecher, Yurak and Vredenhurg (1995) demonstrated that self-enhancing biases are reduced when self-other comparisons involve an individualized (as opposed to a generalized) other.However, in Alicke et al.'s study, the individuating information provided about the other person was not directly relevant to the dimensions on which self-other evaluations were being made.In circumstances in which people receive such information, how will it affect their self-other comparisons?
In the current study, we first obtained self-report information from student participants about their bad study habits.We then asked them to discuss their bad study habits with another student while we videotaped them.Finally, we asked them to rate their own study habits and those of their conversation partner.If people self-enhance in this context, participants should rate their partner's study habits as worse than their own.If participants are basing their self-other ratings on individualized information, then the person in the interaction who admits to the most bad study habits during the conversation should be rated worst.However, if people are using themselves as a basis for making judgments about others, then participants' ratings of their partners should be similar to their ratings of themselves, regardless of what their partner says,

Participants
Seventy-two psychology students at the University of Oregon (52 females and 20 males) participated in the study in exchange for course credit.

Procedure
The participants were run two at a time.They were told that the study was about had study habits.First, the participants received a checklist of 25 had study habits (e.g., "I do homework for less than 7 hours a week" and "I study with the T. V. on") and were asked to indicate whether each statement on the checklist was true for them.Next, the participants were asked to discuss their bad study habits with each other while bemg videotaped.
After the discussion, participants were separated and asked to complete a comparison questionnaire.They were asked who they thought had the worst study habits, themselves or their partner, using a 12-point scale where high numbers indicated that their partner was worse.They were also asked for separate absolute evaluations of their own study habits and their partner's study habits on three items (e.g., "How had are yourlyour partner's study habits?') using 7-point scales where high numbers indicated worse study habits.
Before participants left, we asked them to rate how bad each of the study habits on the checklist was on a 7-point scale (high numbers indicated worse habits).

Coding of Videos
One of the researchers coded the videos to see which study habits the participants mentioned in their conversations.Another researcher coded a subset of the videos and the percent agreement between the two coders was 94%.We noted whether participants mentioned a had habit from the original list, as well as any additional bad study habits that were mentioned.

Results
On average, participants endorsed 9 out of the 25 bad study habits on the initial checklist (SD = 3.4, with a range from 3 to 17).In the course of the videotaped conversations, participants confessed to an average of 4.65 bad study habits (SD = 1.82, with a range of 1 to 9).Of these habits discussed on videotape, 2.65 on average were shared by the two participants (SD = 1.50, with a range of 0 to 6).The study habits on the checklist were clearly rated as bad by participants.No habit received an average rating of less than 4.0 (on a 7-point scale) except an item about eating while studying.
Participants' answers to the three items about their own study habits were highly intercorrelated (alpha = .77),and so we combined these questions into a mean rating of self study habits.Similarly, participants' ratings of the other person's study habits were also intercorrelated (alpha = .66),so these were also combined into a mean other rating.
Participants did not show a self-enhancing bias, and if anything, participants tended to give their partner more positive ratings than themselves.On the absolute scales that assessed the self and other separately, participants rated their own study habits as worse (M = 4.35, SD = 1.21) than they rated their partner's study habits (M = 3.87, SD = .96),t (71) = 3.50, p < .001.On the scale that asked participants to make a relative judgment of who had the worst study habits, participants' mean response was 6.08 (SD = 1.93), which was marginally below the 6.5 midpoint of the scale, t (71) = -1 .8 4 , ~ = .071,indicating that participants reported that their study habits were similar to their partner's, but slightly worse.
Consistent with Alicke et al.'s 1995 findings, individuating information about the other person might have prevented participants from showing a self-enhancing bias.However, this interpretation of our results is only speculative, as we had no baseline condition without individuating information to which we could compare our results.Given the realm in which the comparisons were made, our participants' evaluations of their study habits might have been characterized by pluralistic ignorance rather than self-enhancement.To the extent that students are ignorant of others' study habits (our results clearly indicate that participants did not share all their bad study habits with their partner during their conversations), they may overestimate the degree to which they deviate from the norm (Miller & McFarland, 1991).
Using regression analyses, we found that the more bad study habits participants checked off on the initial questionnaire, the worse they rated their own study habits, b = ,629, p < .OOll. (The overall r2 was ,395, F = 4 5 .7 3 , ~ < ,001.)Thus, participants' self-ratings appear to be based on individuating information.However, participants' ratings of their partner did not.Using simultaneous multiple regression, we regressed participants' ratings of their partners on three predictors: the number of bad study habits the partner mentioned during the conversation, the number of bad study habits that the partner checked off on the initial questionnaire (which participants did not have direct access to) and the participant's (not the partner's) own self ratings.The overall r2 for this equation was ,213, F = 6.12, p = ,001.The number of bad study habits that the partner mentioned during the conversation did not signiticantly predict participants' ratings of their partner's study habits, b = ,001, p = ,995, nor did the number of bad study habits that the partner checked off earlier on the list, b = ,155, p = .16.Far and away, the best predictor of participants' ratings of their partner's study habits was participants' ratings of their own study habits, b = ,434, p < ,001.
We considered the alternative possibility that the regression results were a measurement artifact, such that participants' self ratings were most predictive of partner ratings only because they were the most sensitive measure in the regression equation.However, even when partners' own ratings of their study habits (which were necessarily as sensitive as participants' self ratings) were included in the regression equation, participants' self ratings were still the only significant predictor of their ratings of their partners.
Another possibility why participants may have disregarded the number of bad study habits mentioned by the partner during the conversation is that participants knew that their own admission of bad habits during the conversation was non-representative of the number of bad habits that they had checked off (in private) on the questionnaire.However, the number of bad study habits a participant checked off on the initial questionnaire was a significant predictor of the number of bad study habits a participant mentioned in the conversation, b = ,248, p = ,035.
Unshared characteristics have been found to carry disproportionate weight in evaluative comparisons that do not involve the self and other (e.g., Hodges, 1997) and thus it is possible that it was not the total number of habits that the partners mentioned that determined participants' ratings of their partners, but instead only those habits mentioned by the partner that the participant did not share.In order to explore whether partners' unshared study habits carried disproportionate weight in determining participants' ratings, we divided the number of bad study habits mentioned only by the participant's partner during the conversation (i.e., habits that were not shared with the participant) by the total number of study habits mentioned by the partner in the video.Thus, the more unique bad study habits the partner mentioned, the greater this ratio.When this ratio was added to the regression equation, it was neither a significant predictor of participants' ratings of their partners, b = ,133, p = .22,nor did it significantly change the overall r2.

Final Thoughts
Strikingly, our results suggest that individuating information did not affect participants' ratings of their partners -even when participants heard information that was directly relevant to the dimension they were evaluating.Furthermore, participants showed no self-enhancing bias in their ratings of their partners.Instead, the best predictor of participants' evaluations of their partners was how participants rated themselves.No matter what the partners said, they were seen as being like the participants.
Although these results may merely be another example of self-projection (e.g., see Van Boven, Dunning & Loewenstein, 2000), we believe that it was the course of the conversation that drove the results.Our impressions of the interactions suggest that habits that participants shared were much more fertile conversational fodder than unshared habits, which were veritable conversation dead-ends.These conversational forces may not only explain the overwhelmingly high correspondence between self-other ratings, they also help explain why partners' reported habits during the conversation seemed to have had no impact on participants' ratings of their partners.Even a partner's unique habits may have been equated with self habits as the conversation (and the newly formed relationship) unfolded between participants and their partners.For example, talking on the phone while studying or e-mailing while studying could both be discussed as examples of letting contact with friends interrupt one's studies.A shared identity as students may have further contributed to participants' tendency to assimilate their judgments of themselves and the other (Mussweiler, 2001).In this "get to know you" setting, participants appeared to be searching for common ground with their partners, a process which inflated the similarity between self and other.