INTRODUCTIONThere are several models of impression formation. Solomon Asch’s classic “warm-cold” study examined the processes used to form a unified impression of a target person. He concluded that rather than just a simple summation of traits an interactive process occurs between central and peripheral traits. The perceiver blends diverse features of the target person into a unified impression that includes not only discrete units of information about the person, but also both the meaning of the individual features and their relationships (Asch, 1946, as cited in Hock, 1995; Kunda & Thagard, 1996).Asch’s later work (1984) continued this Gestaltist theme of the unified nature of person-perception. He demonstrated that even when presented with incongruent pairs of dispositional traits (e.g., sociable-lonely), participants readily resolved the discordance by using the same lawful ordering and organizing processes. They formed impressions of the imagined target persons that contained not just a summation of the information presented about the persons, but also meaningful rationales that explained how two opposite dispositions could co-exist in one person. In line with gestalt theory, the whole was more, and different from, the sum of the parts. In addition to Asch’s early work on impression formation, N. H. Anderson proposed an elementaristic model of how impressions are formed. This process was assumed to occur by the perceiver’s assessing the meaning of each element of information about the target person separately and then combining them algebraically into a summary impression (Anderson, 1968, as cited in Kunda & Thagard, 1996). More recent models of impression formation are Brewer’s dual process model and Fiske and Newberg’s continuum model. Both are serial models that assume perceivers first use stereotype-based processes in forming impressions, and then, if strongly motivated to do so, or if the attempt to categorize the target person as belonging to any particular stereotype fails, attribute-based processes are used (Kunda & Thagard, 1996). Kunda and Thagard (1996) proposed a parallel-constraint-satisfaction model in which stereotypes, traits, and behaviors are all represented as interconnected nodes in a spreading activation network. The spread of activation between nodes is constrained by both positive and negative associations, and nodes not only activate but also deactivate each other. The theory assumes that information (behavior) that has been directly observed and so is known to be true of the target person constrains the impression formed of the person. Furthermore, impressions are formed holistically by the simultaneous activation and deactivation of associates of the observed information, and these processes jointly constrain the impression of the target person. In addition to these broad theories and models of impression formation, several studies have examined the elements of making an evaluation or a judgment about a person. Herringer (1991) found that participants exhibit a slight bias toward using stable-trait or mood-state words to describe themselves, whereas words used to describe others were more socially evaluative and metaphorical. Davis (1990) found that participants’ initial impression responses for target persons were classified into five perceptual categories: a) personal characteristics, 25.7 %; b) personal and social roles, 33.8%; c) appearance characteristics, 10.3%; d) context of situation, 10.5%; and e) extended references (e.g., responding with “Marlboro Man” to a slide of a cowboy on a horse), 19.7% of responses. The author concluded that category usage in initial impressions may be related to the participant’s inference about the social salience of the target person. Jussim (1987) found that participants’ self-esteem influenced not only their self-evaluations of performance on a task, but also their perceptions of evaluations given by others regarding their performance on the task. It was also found that self-esteem biases operated to allow participants with high self-esteem to interpret both positive and negative feedback more favorably regardless of whether they were told they had done a good or a poor job on the task. Koeck and Guthrie (1975) examined reciprocity by having participants rate confederates before performing a shared task with them and then again after the confederates had given either positive, neutral, or negative evaluations of the participants’ task performance. It was found that when participants with high self-esteem received neutral (i.e., rating of average performance) or negative evaluations from confederates, they reciprocated by giving negative evaluations of the confederates. This did not hold with participants who had low self-esteem. They rated confederates more negatively before performing a shared task with them, and the evaluations they gave confederates after the task were even more negative, regardless of whether they received a positive, neutral, or negative evaluation from the confederate. The authors suggested that this result with low self-esteem participants partially accords with cognitive dissonance theory. Cellar, Miller, and Halpert (1993) investigated whether the outcome of a decision about employment affected participants’ evaluation of the fairness of the decision and also of the personality of the decision-maker. They found that personality variables ascribed to the decision-maker were highly correlated with perceptions of the fairness outcome of the decision, which suggested a relationship between decision made and evaluation of the decision-maker’s personality. Our study investigated the relationship between college students’ expected rating on a task and their subsequent evaluation of the personality of the researcher who reports to them a rating on their performance of that task. We hypothesized that we would find a strong relationship between rating received and evaluation of the researcher’s personality. Specifically, we predicted that, using a 2 (hard versus easy task) X 2 (high versus low score) factorial design, participants in the hard task/high score group would reciprocate with a favorable evaluation of the researcher’s personality. We also predicted participants in the easy task/low score group would reciprocate with an unfavorable evaluation of the researcher’s personality. And finally, we predicted participants in both the easy task/high score group and in the hard task/low score group would give mixed evaluations of the researcher’s personality.
METHOD
PARTICIPANTS
The participants were 87 undergraduate students from Missouri Western State College who received extra credit in a general psychology course for participating in this study. The age range was not recorded. However, the ratio of men to women was 42:45, and the level of completed education of the participants ranged from 12 to 15 years. Additionally, participants were randomly assigned to one of four experimental conditions.APPARATUS/MATERIALS
The instrument used to collect data for this study was an 18-item adjective checklist identical to that used in the original impression formation studies conducted by Solomon Asch in 1946. Additionally, two versions of a sham intelligence (IQ) test were administered to all participants. These IQ test versions were similar, including format and number of items, but they varied in level of difficulty. PROCEDURE
Prior to administration of the impression formation scale the participants were randomly assigned to one of two conditions. Each participant completed one of two versions of the sham IQ test administered by the IQ researcher. One version was less difficult with the intent that participants should be able to complete it within the allotted time. The second version was quite difficult, and there was only a small chance that participants would be able to accurately finish the test within the time limit. The participants were told the IQ test was a valid measurement device and should be taken seriously. All participants had a total of five minutes to complete the test. After collecting the tests, the researcher assigned fake IQ scores in the following manner: 1) half of the hard test group was assigned scores within one standard deviation above the mean, and the other half was assigned scores within one standard deviation below the mean; and 2) half the easy test group was assigned scores within one standard deviation above the mean, and the other half was assigned scores within one standard deviation below the mean. During the subsequent class meeting, the IQ researcher gave participants their individual scores for the test and told them the (imaginary) class mean IQ score. Immediately following this, another researcher, collecting data for a seemingly unrelated project, was introduced to the participants. The second researcher asked participants to evaluate the first researcher who was collecting the IQ data. Participants were then given the impression formation adjective checklist and told that it was an evaluation to assess students’ perceptions of all psychology student researchers collecting data at the school. After participants completed the evaluation form it was collected, and participants were debriefed and given an opportunity to ask questions of both researchers.
RESULTS
Eighteen 2 (hard vs. easy IQ test) X 2 (high vs. low IQ score) between-groups analyses of variance (ANOVA) were performed using the impression formed by participants as the dependent variable. Table 1 lists individual results of each ANOVA. Significant findings occurred for participants’ perceptions of the sham IQ test researcher’s altruism, happiness, generosity, and seriousness.The ANOVA comparing hard vs. easy IQ test and high vs. low IQ score to participants’ perceptions of the researcher’s altruism found no main effect for score, F(1,83)=.192, p=.662. Therefore, by itself, score assigned to the IQ test did not affect perceptions of the researcher’s altruism. The main effect for test was significant, F(1,83)=7.717, p<.05. This indicates that difficulty level of the IQ test did affect participants’ perceptions of the researcher’s altruism. Additionally, there was an interaction between test score and test difficulty. The result of this ANOVA indicates participants who received a high score on a hard test were more likely to see the researcher as altruistic than were those individuals who received a low score on a hard test. It also indicates that participants who received a low score on an easy test were more likely to consider the researcher altruistic than were those individuals who received a high score on an easy test. The ANOVA comparing hard vs. easy IQ test and high vs. low IQ score to participants’ perceptions of the researcher’s happiness found no main effect for score, F(1,83)=1.543, p=.218, indicating that score received did not affect perceptions of the researcher’s happiness. The main effect for test was also not significant, F(1,83)=1.543, p=.218. This indicates that difficulty level of the IQ test did not affect perceptions of the researcher’s happiness. However, there was an interaction between test scores and test difficulty such that participants who received low scores on the more difficult test were less likely to perceive the researcher as happy. The ANOVA comparing hard vs. easy IQ test and high vs. low IQ score to participants’ perceptions of the researcher’s generosity found a main effect for score, F(1,83)=5.115, p<.05. This indicates that participants who received high scores were more likely to perceive the researcher as generous than were those individuals who received lower scores. The main effect for test difficulty was not significant, F(1,83)=1.279, p=.261, indicating that test difficulty did not affect participants’ perceptions of the researcher’s generosity. The ANOVA comparing hard vs. easy IQ test and high vs. low IQ score to participants’ perceptions of the researcher’s seriousness found no main effect for score, F(1,83)=.169, p=.682, indicating that score received did not affect perceptions of the researcher’s seriousness. The main effect for test, however, was significant, F(1,83)=5.412, p<.05. This indicates that test difficulty did affect perceptions of the researcher’s seriousness. In this case, participants who completed a hard test were more likely to perceive the researcher as serious regardless of score received.
DISCUSSION We hypothesized a strong relationship between participants’ score on a task and how they would evaluate the personality of the researcher reporting their score. Our three predictions were: a) participants in the hard task/high score group would reciprocate with a favorable evaluation of the researcher’s personality, b) participants in the easy task/low score group would reciprocate with an unfavorable evaluation of the researcher’s personality, c) and participants in both the easy task/high score group and in the hard task/low score group would give mixed evaluations of the researcher’s personality. The data we accumulated provided only weak support for our hypothesis. We found statistical significance when the following four traits were tested: a) altruistic, b) serious, c) happy, and d) generous. None of the remaining 14 ANOVAs showed any statistically significant results. Our strongest result was for altruistic. Participants in the hard test/high score group described the researcher as altruistic, which supports our first prediction. However, participants in the easy test/low score group also described the researcher as altruistic, contrary to our second prediction. The second strongest result was for the trait adjective serious. Participants who received the more difficult version of the sham IQ test described the researcher as serious regardless of whether they received a high or a low score. Therefore, this also does not fully support our hypothesis. The third strongest result was for generous. Participants who received high scores described the researcher as generous regardless of test difficulty. The fourth strongest result was for happy. There were no main effects for either score or test difficulty, but there was an interaction. Participants who received low scores on the more difficult version of the test described the researcher as happy less often than did the other participants. There are many theories and models of impression formation. Our results for the traits altruism and generous provide weak support for Jussim’s (1987) research into levels of self-esteem and evaluations of a target person. Additionally, our results for the trait happy support the findings of Koeck and Guthrie (1975) regarding reciprocity in impression formation. It is clear from the literature and the results presented here that impression formation is a complex theory and difficult to assess. There are several limitations to our study. Our participants were not necessarily representative of the general population. Additionally, although participants were randomly assigned to groups, many potential confounding variables were not controlled. There is a possibility that participants discussed the sham IQ test with other general psychology students in other classes before all classes were tested. Also, participants may have discussed the test and/or the researcher among themselves during the approximately 48 hours in between taking the sham IQ test and completing the trait adjective instrument. Therefore, both internal and external validity of this study are debatable. In future research in this area, control measures should be strengthened. In addition, a more representative sample of participants would improve generalizability.
REFERENCES
Asch, S. E. & Zulier, H. (1984). Thinking about persons. Journal of Personality and Social Psychology, 46, 1230-1240. Cellar, D. F., Miller, M. L., & Halpert, J. A. (1993). Effects of type of decision and organizational outcome on observers’ evaluations of the context for decision. Psychological Reports, 72, 335-344. Davis, L. L. (1990). Social salience: What we notice first about a person. Perceptual and Motor Skills, 71, 334. Herringer, L. G. & Haws, S. C. (1991). Perception of personality traits in oneself and others. The Journal of Psychology, 125, 33-43. Hock, R. R. (1995). Forty studies that changed psychology: Explorations into the history of psychological research (2nd Ed.). Englewood Cliffs, NJ: Simon and Schuster. Jussim, L., Coleman, L., & Nassau, S. (1987). The influence of self-esteem on perceptions of performance and feedback. Social Psychology Quarterly, 50, 95-99. Koeck, R. & Guthrie, G. M. (1975). Reciprocity in impression formation. Journal of Social Psychology, 95, 67-76. Kunda, Z. & Thagard, P. (1996). Forming impressions from stereotypes, traits, and behaviors: A parallel-constraint-satisfaction theory. Psychological Review, 103, 284-308.
APPENDIX A
Impression of Student InvestigatorsLast 6 numbers of Social Security Numer Sex: Male _____ Female______ Age: ______ You have just interacted with a student research investigator. Please rate that person according to the traits listed below. Circle the one word in each pair of words that most closely describes that person. You must circle only one word from each pair. Do not skip any pair. 1. Generous or… Ungenerous 2. Shrewd or… Wise 3. Unhappy or… Happy 4. Irritable or… Good-natured 5. Humorous or… Humorless 6. Sociable or… Unsociable 7. Popular or… Unpopular 8. Unreliable or… Reliable 9. Important or… Unimportant 10. Ruthless or… Humane 11. Good-looking or… Unattractive 12. --- Persistent or… Unstable 13. Frivolous or… Serious 14. Restrained or… Talkative 15. --- Self-centered --- or… Altruistic 16. Imaginative or… Hard-headed 17. Strong or… Weak 18. Dishonest or… Honest
TABLE 1
Table of ANOVA Results: Dependent Variable Degrees of Freedom Main Effect, Score Main Effect, Difficulty Difficulty X Score Altruist 1, 83 .192 7.717* 2.283 Good-Looking 1, 83 .108 .108 1.245 Good-Natured 1, 83 1.934 .220 .142 Generous 1, 83 5.115* 1.279 1.279 Happy 1, 83 1.543 1.543 4.603* Honest 1, 83 1.178 1.178 1.178 Humane 1, 83 2.979 .720 .579 Humorous 1, 83 1.934 .385 .008 Imaginative 1, 83 .000 .505 1.386 Important 1, 83 .003 .488 .488 Persistent 1, 83 .023 .023 1.898 Popular 1, 83 1.401 .667 1.034 Reliable 1, 83 .025 .003 .003 Serious 1, 83 .169 5.412* .918 Sociable 1, 83 .196 .310 .480 Strong 1, 83 .943 .034 1.586 Talkative 1, 83 1.320 .049 .880 Wise 1, 83 .082 .012 .012 | 
