Backup of thesis5

Document Sample
Backup of thesis5 Powered By Docstoc
					Emotional Primes and the Stability of Self-Knowledge Daniel Kauwe Brown University

1

Introduction
The question of self-awareness has constantly resonated throughout human history. Socrates gave us his famous, “Know thyself” quote, and the Chinese philosopher Lao Tzu stated, “To understand others is to have knowledge; To understand oneself is to be illumined.’ Clearly, the meaning and nature of self-awareness has long occupied the collective attentions of great philosophers, but today issues of self-awareness are no longer the purview of philosophy. Psychologists also wrestle with self-awareness; whether from a cognitive, social or evolutionary perspective, we search to better understand the origins and implications of human self-awareness. In the Alcibiades, Socrates asserts that self-knowledge is wisdom, and draws a connection between self-awareness and self-knowledge, “if we have no self-knowledge and no wisdom, can we ever know our own good and evil?” Socrates’ quote suggests that our awareness, particularly in regards to abstract thinking, depends upon selfknowledge. If such is true, decades of research on self- and group-knowledge should uniquely position modern psychology to answer questions regarding self-awareness (Clement & Krueger, 2000, 2002; Craik et al., 1999; Gaertner, Sedikides, & Graetz, 1999; Hogg & Turner, 1987; J. Krueger, 2000; J. I. Krueger, 2003; Markus, 1977; Markus & Kunda, 1986). Modern psychology often splits human awareness into self versus group categories (self-perceptions vs. group-perceptions, self-judgments vs. group-judgments, and selfconcept versus group-concept). Of these categories, we will concern ourselves with selfknowledge (what we know about ourselves) and group-knowledge (what we know about others). Examples of the former theory include self-schema theory and social projection (Clement & Krueger, 2000; J. Krueger, 2000; J. I. Krueger, 2003; Markus, 1977; Markus & Kunda, 1986), while examples of the latter include self-stereotyping theory and social categorization theory (Biernat, Vescio, & Green, 1996; Hogg & Turner, 1987; Lorenzi Cioldi, 1991; Onorato & Turner, 2004; Pickett, Bonner, & Coleman, 2002). One branch of social cognition argues that relative to group-knowledge, self-knowledge is more

2

constant and concrete (Gaertner et al., 1999; Gramzow, Gaertner, & Sedikides, 2001); and the other branch maintains that self-knowledge is more contextual and malleable (Hogg & Turner, 1987; Onorato & Turner, 2004). Given that self- and group-knowledge are important components to human perception and judgment (Kunda, 1999), ascertaining the relative stabilities of self- and group-knowledge is critical to understanding our social cognitive processes. Many theories of social cognition, like social perception or social judgment, depend upon assumptions regarding self- and group-knowledge. However, these assumptions are anything but uniform, with psychological research divided into those who regard selfknowledge as relatively stable and those who regard self-knowledge as relative malleable. The hypothesis and conclusions of both branches have important implications for social cognition, but given their polar separation, they both cannot be true. Hogg and Turner initially argued that salience of social categories determines selfdefinition (Hogg & Turner, 1985, 1987), thereby creating the theory of selfcategorization. By their arguments, self-perceptions and self-judgments are determined by contextual forces, and thus self-concept and self-knowledge are fundamentally influenced by social context. According to Hogg and Turner, social categories contain individuals who “define, describe and evaluate themselves in terms of the group/category label” (1987, p. 325). In this respect, self-concept is ancillary to external, groupconcepts. If individuals “define, describe and evaluate” themselves according to group labels, then we must conclude that knowledge of the group determines self-knowledge. Furthermore, according to Hogg and Turner, the primary mechanism for this process is the transformation of external category labels into internal self-definition through the self-categorization processes. Brewer’s theory of optimal distinctiveness (Brewer, 1991) has often been cited as an explanatory model for the motivation and process of self-categorization (Pickett et al., 2002). According to Brewer’s model of optimal distinctiveness, individuals are constantly fluctuating between assimilation to the group and differentiation away from the group. Thus, optimal distinctiveness is a theory of opposing mental processes that characterize social cognition. As noted earlier, self-categorization relies upon the assumption that individuals utilize external category labels when defining themselves.

3

Optimal distinctiveness further characterizes and explains this process of self-definition through group attributes. In particular, optimal distinctiveness argues that people are motivated by need arousal to either relate to or individuate from the group. This need arousal is circular in its origin; individuals who feel highly differentiated from the group will be motivated to assimilate, and individuals who feel depersonalize will be motivated to individuate from the group. Therefore, self-categorization (i.e. application of groupknowledge) depends upon the individual’s emotional state, and individuals are constantly cycling through assimilation and differentiation need and fulfillment. Ultimately, self-categorization is a theory of cognition shaped by social context and interpreted from social perceptions (Turner, Oakes, Haslam, & McGarty, 1994). In marked contrast lie the theories of stable self-concept. Of these theories, Markus’ theory is most seminal and most often cited (Markus, 1977). Markus’ is typically cited as the originator of self-schema theory (Kunda, 1999), and thus an important contributor to theories of stable self-concept and self-knowledge. Her initial research points to the existence self-schemata that are a) integral to the processing of self-information, b) the basis of self-prediction, c) filter against counter-schematic information (Markus, 1977). She eloquently describes self-schemata as “implicit theories that used by individuals to make sense of their own behavior and to direct future courses of behavior.” (p. 78). Markus followed this research with more work on self-schemata, and later revised some aspects of her theory to include the theory of working self-concept (Markus & Kunda, 1986). Within to this theory, some elements of the self-concept are stable and other elements are unstable. The unstable elements alter depending on social context (similar to self-categorization theory), but unlike self-categorization, Markus and Kunda argue that certain elements of the self are always stable. These elements, called core elements, are integral to conceptions of similarity and difference with other people. Markus and Kunda found core elements to be resistant to change, though working selfconcepts may change according to external stimuli. Self-schemata theory postulates that social perceptions are determined by self-concepts, while self-categorization theory postulates that self-concepts are determined by social perceptions. According to selfschemata theory, self-knowledge determines self-schemas, which in turn determine selfconcepts, which ultimately determine perceptions of others (Markus, Smith, & Moreland,

4

1985). The Present Quandary The present research intends to further clarify this contradiction by investigating the relative malleability of self- versus group-knowledge. To this end we seek to address the following question: Is self-knowledge stable or not? To answer this question, we will operationally assess malleability of selfknowledge through the measure of self- and group-ratings. Regarding the measure of self- and group-ratings, our research corresponds to this main hypothesis: If self-knowledge is more stable than group-knowledge, then self-ratings should also be more stable than group ratings. Thus following our experimental manipulation we expect self-ratings to fluctuate less than group-ratings.

Purpose: To determine the relative stabilities of self- and group-knowledge we have focused on self- and group-ratings as indicators of self- and group-knowledge. Furthermore, the associative strength between self-and group-ratings provides an index of relative stabilities when an experimental manipulation occurs between the self- and group-ratings. Simple logic suggests that the changes in associative strength between the self- and group-ratings will also indicate changes in the production of self- or groupratings. By counterbalancing self- and group-ratings and examining patterns of associative strength, we can examine which type of ratings is most susceptible to manipulation. Finally, the type of rating(s) most affected by the experimental manipulation should reflect the stability of the knowledge upon which the ratings are based. In the current set of experiments, an emotional prime to the group was chosen a s the experimental manipulation (Table 1 Main Experimental Design).

5

Group 1 Group 2 Group 3 Group 4

Self ratings Self ratings Group ratings Group ratings

Positive priming Negative priming Positive priming Negative priming

Group ratings Group ratings Self ratings Self ratings

Desirability ratings Desirability ratings Desirability ratings Desirability ratings

American Ideals American Ideals American Ideals American Ideals

Table 1 Main Experimental Design

The group prime was selected with the intention that differential primes to the target group, should affect how people make ratings for themselves and the target group, thus producing a pattern of associative strengths between self- and group-ratings. For this experiment, associative strength was measured through strength of correlation between self- and group-ratings. The manipulation of participants’ emotions to affect their perceptions and judgments has commonly been used in experiments that measure the relationship between self- and group-judgments (Agostinelli, Sherman, Presson, & Chassin, 1992; Campbell, 1986; Clement & Krueger, 2002; Crocker, Alloy, & Kayne, 1988; Sherman, Chassin, Presson, & Agostinelli, 1984). For this experiment, we reason that the positive prime should produce strong association between self- and group-ratings, while the negative prime should produce weak association between self-and groupratings. Self-categorization theory (and other theories of fluid self-knowledge) suggests that emotional primes to the group should lead to differential patterns of self-ratings. It has been argued that that individuals define themselves not as unique and separate but as members social categories (Turner et al., 1994). Turner et al. (1994) declared that if group membership is made salient, members will accentuate similarities to other members while enhancing differences from non-members. It has been argued that in situations where the social identity is salient, the self is defined in terms of group membership and individuals respond in agreement with the beliefs of group members (Haslam et al., 1996). Therefore, if theories of instable self-knowledge are true, emotionally primed self-ratings should show the most change. Thus self-categorization theory would suggest that though self-ratings will fluctuate with changes in group-

6

context, the individual’s sense of group-ratings should not fluctuate with changes in group-context. Predictions Prediction One: Positive priming to the group will produce strong correlations between selfratings and group-ratings, and this effect will be greatest when self-ratings precede group-ratings (Group 1). Given that we predict group-ratings to be less stable than selfratings, we expect a higher mean score for Group 1 relative to Groups 2, 3, 4. This effect is illustrated in both Figure 1 and Figure 2. Prediction Two Negative priming to the group will decrease correlations between self-ratings and group-ratings, but self-ratings will be more stable than group ratings thus this reduction will be more pronounced for Group 2 than Group 4 (Figure 1). If the opposite pattern is found and Group 4 is lower than Group 2, then the data will suggest that self-ratings are more malleable than group-ratings (Figure 2).

2 3 4 Group Projection scores
Figure 1 Predicted Projection Scores

1

2 3 4 Group Projection s core s
Figure 2 Predicted Projection Scores

1

7

Methods and Materials
Participants:

Preliminary Study Participants were undergraduate students, male (n = 30, mean age = 19.40) and female (n = 16, mean age = 19.50), derived from the Brown Psychology Department’s Subject Pool. Participants were presented with a questionnaire, and given course credit in exchange for their voluntary participation.

Final Study Participants were Internet users eighteen years of age or older, male (n = 132, mean age = 21.71) and female (n = 196, mean age = 21.57). Advertisements for this study were made to public Internet forums to solicit voluntary participation in the Final Study phase of this experiment. All advertisements for the study contained a link to the actual study. No compensation was offered or given. Procedure:

Preliminary Study The study took place within a Brown Psychology lab. The main portion of the questionnaire was administered via computer, and participants’ responses were automatically encoded and stored in a database with no individual identifying information. Table 1 demonstrates the main experimental design. Participants performed an emotional priming activity (independent variable) interspersed between sets of self- and group-ratings (dependent variable). The priming activity asked participants to write about five negative or five positive America contributions to world history. As shown in Table

8

1, the order of these self- and group-ratings was counterbalanced for two of the four treatment groups. Finally, to control for social desirability, participants complete desirability ratings for all projection items. A manipulation check was also administered to assess degree of identification with American culture. Over all, the experimental design consisted of a social judgment measure (either self or group), followed by the emotional priming activity (positive or negative), the social judgment measure that complemented the first (self or group, depending on the first projection measure), and finally the social desirability ratings and manipulation check. All aspects of the survey were computerized and conducted on an Apple computer. Excel was used to create the survey and corresponding database. Participants were randomly assigned to one of four treatment groups by the survey program, and all responses were stored in a database for later analysis. Following the experimental measures, participants were debriefed and then dismissed. Final Study The Final Study Procedure was very similar to The Preliminary Study. The main alteration was participant recruitment and the presentation of the survey. Participants were recruited through advertisements posted to various public Internet forums. No compensation was offered for participation, and the advertisement text appealed to the general curiosity of participants. Presentation of the survey was done via the web, thus the survey was completely available during the entire run-time. An initial consent form was used to access the survey, thus potential participants were warned that only those 18 years or older could participate. Participants indicated their ability and willingness to participate by clicking on an “Agree” button. Participants who clicked on the “Agree” button could then access the site. The survey was presented similar to The Preliminary Study. Thus, the survey consisted of a social judgment measure (either self or group), followed by the emotional priming activity (positive or negative), the social judgment measure that complemented the first (self or group, depending on the first projection measure), and finally the social desirability ratings and manipulation check. 9

Given that the survey was hosted over the Internet, participants presumably accessed the survey through private or public computers, using their available web browsers. The survey itself was created as web document and hosted from a computer in the Brown Psychology Department. Materials: A simple questionnaire format was used to create the survey, and four sets of questions composed the bulk of the survey. Two sets of questions were the self- and group-ratings. Questions were fundamentally identical for both the self- and groupratings, that is the base questions consisted of ten items taken from the Minnesota MultiPhasic Personality Inventory (MMPI):
I certainly feel useless at times. My hardest battles are with myself. I like poetry. It does not bother me that I am not better looking. I am neither gaining nor losing weight. I think most people would like to get ahead. I enjoy detective or mystery stories. I often think, “I wish I were a child again.” I do not worry about catching disease. I seldom worry about my health.

These items have been used in past research that measures the relationship between self- and group-ratings (Clement & Krueger, 2002). Participants were asked to personally respond to these items and give an endorsement of agree or disagree (selfratings), and participants were also asked to respond and give endorsements for the percentage of Americans that would agree with the item in question (group-ratings). The same MMPI items were later presented within a third set of questions, the social desirability ratings. For each of the ten items, participants were asked to indicate the desirability of the item. The emotional prime consisted of a prompt and a short writing period. The negative prompt was as follows, Please think of America's role across history. In your opinion, how has America negatively impacted the world, past or present? Please think about and generate a list of five examples that illustrate America's negative impact. Any kind of example is fine, there is no right or wrong answer.

10

The positive prompt was identical, except “positively” and “positive” were substituted for “negatively” and “negative.” Finally participants completed a manipulation check that assessed degree of American identification. The manipulation check consisted of a set of questions composing a scale designed to measure degree of identification with modern American ideals (Phinney, DuPont, Espinosa, Revill, & Sanders, 1994). The scale begins with the following instructions:
To what extent do the following phrases accurately describe what “America” or “being American” means for you personally? Please rate on a 4-point scale from with 1 = strongly agree and 4 = strongly disagree.

The questions consisted of:
A democratic country in which the laws protect my interests. A society in which I can say what I want. A land of economic opportunity for me. A country whose public media has kept me well-informed about world problems. A society that is accepting and tolerant of my cultural background or ethnic group A society that guarantees my basic rights. A society where I will be able to acquire personal wealth. A land that allows me to practice my own religion in freedom. A society that is concerned about the welfare of my cultural group. A land in which an education is available to me if I want it. A society that supports freedom and justice for me. A country in which hard work will pay off in success for me.

11

Results
Results Preliminary Study Unpartialed Data Raw self and group-ratings were correlated ideographically across subject, and the resultant scores were then transformed using the Fisher transformation for correlation scores. Individual Fisher transformed scores were then summed within conditions (means and standard errors are graphed and presented in Figure 3).

Mean Scores Preliminary Study
0.70 0.60 0.50

Fisher Z Score

0.40 0.30 0.20 0.10 0.00 Group 1: Positive Group 2: Negative Group 3: Positive Group 4: Negative

r = 0.48

r = 0.27

r = 0.50

r = 0.42

Order: Self-Group

Order: Group-Self
Figure 3

In line with our predictions, the graphed data indicates that the negative emotional prime did produce a reduction in Groups 2 and 4, relative to Groups 1 and 3, suggesting that the negative prime substantially reduced the relationship between self-and groupratings. However, 2 X 2 ANOVA (Emotional Prime by Order of Ratings) found no main effect by order. Thus, the order of self-ratings before or after group-ratings was not significant across conditions. A trend towards significance was found for the main effect 12

of the emotional prime (F (3, 45) = 2.795, p = .101). No significant interaction effect was found for the emotional prime and the order of ratings. Based on our prior predictions, we were particularly interested in the differential effects of the positive treatments (Group 1, Self-Group; Group 3, Self-Group) and the negative treatments (Group 2, Self-Group, Group 4, Self-Group). For the positive treatment, we predicted that Group 1 (Emotion: Positive; Order: Self-Group) would show the strongest associative strength between self- and group-ratings. For the negative treatment, we predicted that Group 2 (Emotion: Negative; Order: Self-Group) would show the weakest associative strength between self- and group-ratings. These predictions each yield pairs of complementary contrasts (Figure 4). For the positive treatment, we expect Contrast A (Group 1: strongest correlation between selfand group-ratings) to be more significant than Contrast B (Group 3: strongest correlation between self- and group-ratings). Likewise, for the negative treatment, we expect Contrast C (Group 2: weakest correlation between self- and group-ratings) to be more significant than Contrast D (Group 4: weakest correlation between self- and groupratings).

1

Contrast A

2

3

4

1

Contrast B
Group 1 2 3 4

2

3

4

1

Contrast C

2

3

4

1

Contrast D

2

3

4

Key Treatment Positive Negative Positive Negative

Order Self/Group Self/Group Group/Self Group/Self

Figure 4 Contrast Analysis

Contrast analysis revealed a clear lack of significance for Contrasts A, B, and C. (Table 2). Thus suggesting that the positive treatment did not differentially affect self- or group-ratings (Contrast A and Contrast B), nor did the negative treatment significantly 13

self-ratings relative to group-ratings (Group 4, Contrast D). However, Contrast C though not significant, did demonstrated a strong trend towards significance (t (45) = 1.80, p = .078). This trend suggests that group-ratings, relative to self-ratings, are more susceptible to negative stimuli.

Contrast A B C D

1 3 -1 1 1

Group 2 3 -1 -1 -1 3 -3 1 1 1

4 -1 -1 1 -3

t 0.78 1.27 1.81 0.02

df 45 45 45 45

Significance (2-tailed)

0.438 0.213 0.078 0.985

Table 2 Unpartialed Data Preliminary Study

Partialed Data Given that the social desirability of an item may lead to higher endorsements simply because that item is desirable, we also controlled for the effects of social desirability by partialing the social desirability ratings from the correlation scores between self and group-ratings (this procedure was preformed ideographically by individual). These individual partialed scores were once again transformed via the Fisher transform and summed and average by treatment condition. The graphical form of the data is presented in Figure 5.

14

Mean Partialed Scores Preliminary Study
0.70 0.60 0.50

Fisher Z Score

0.40 0.30 0.20 0.10 0.00 Group 1: Positive Group 2: Negative Group 3: Positive Group 4: Negative r = 0.46 r = 0.19 r = 0.48 r = 0.30

Order: Self-Group
Figure 5

Order: Group-Self

The partialed data follows the patterns and trends discussed with the unpartialed data, with one major difference – partialing the correlation scores improved the statistical significance of the findings. As before, the ordering of self-ratings before or after groupratings was not significant across conditions (F (3, 45) =.40, p = .53). However, the main effect of emotional prime reached strong significance (F (3, 45) = 5.65, p = .022). Thus there was a statistically significant deprecation in relationship between the negatively primed groups and the positively primed groups. Again, there was no significant interaction effect. Contrast analysis was run based on the contrasts outlined for the unpartialed data (Figure 4), and results are reported in Table 3. Contrast analysis did not yield statistical support for predictions for the positive treatment (Contrast A versus Contrast B). Thus, there was no statistical evidence that the positive treatment differentially affected the associative strength between self- and group-ratings. Also, there was no evidence that the negative prime had a significant impact upon the self-ratings following group-ratings (Group 4, Contrast D). However, Contrast C though not significant, did demonstrate a strong trend towards significance (t (45) = 1.80, p = .078). This suggests that the negative prime strongly influences group-ratings when following self-ratings (Group 2, Contrast C).

15

Contrast A B C D

1 3 -1 1 1

Group 2 3 -1 -1 3 -1 -3 1 1 1

4 -1 -1 1 -3

t 1.26 1.63 1.87 0.74

df 45 45 45 45

Significance (2-tailed)

0.213 0.109 0.067 0.465

Table 3 Contrast Analysis Partialed Data Preliminary Study

Results Final Study The Final Study is based upon the data from the Preliminary Study and the Internet survey. Given that the studies were identical in nature, the data from the Preliminary Study were pooled with the Internet survey data, resulting in the Final Study data.

Unpartialed Data As with the data from The Preliminary Study, raw self and group-ratings were correlated ideographically across subject, and the resultant scores were then transformed using the Fisher transformation for correlation scores. Individual Fisher transform scores were then summed within conditions. The means and standard deviations are reported in graphically in Figure 6.

16

Mean Scores Final Study
0.50

0.40

Fisher Z Score

0.30

r = 0.34

r = 0.32

r = 0.41

r = 0.41

0.20

0.10

0.00 Group 1: Positive Group 2: Negative Group 3: Positive Group 4: Negative

Order: Self-Group
Figure 6

Order: Group-Self

Contrary to the Preliminary Study, the data did not indicate a main effect by emotional prime (F (3, 368) = .170, p = .680), but there was an indication of a main effect by order of ratings (F (3, 368) = 3.23, p = .073. No significant interactions effect was found between the emotional prime and the order of ratings. As with the Preliminary Study, contrast analysis was conduced according to the patterns presented in Figure 4. No contrasts demonstrated statistical significance. However, Contrast B appears to be most descriptive.

Contrast A B C D

1 3 -1 1 1

Group 2 3 -1 -1 -1 3 -3 1 1 1

4 -1 -1 1 -3

t -0.82 1.13 1.16 -0.84

df 368 368 368 368

Significance (2-tailed)

0.412 0.190 0.248 0.400

Table 4 Contrast Analysis Unpartialed Data Final Study

Partialed Data

17

As in The Preliminary Study, we also controlled for the effects of social desirability by partialing the social desirability ratings from the correlation scores between self and group-ratings. These individually partialed scores were again transformed via the Fisher transform and summed and averaged. The graphical form of the data is presented in Figure 7.

Mean Partialed Scores Time 2
0.50

0.40

Fisher Z Score

0.30

0.20

r = 0.30

r = 0.26

r = 0.40

r = 0.37

0.10

0.00 Group 1: Positive Group 2: Negative Group 3: Positive Group 4: Negative

Order: Self-Group
Figure 7

Order: Group-Self

Compared to the unpartialed data, the main effect for order of ratings increased in significance (F (3, 368) = 6.66, p = .010). As with the unpartialed data, there was no main effect for emotional prime (F (3, 368) = .97, p = .326), nor was there an interaction effect. The partialed data was also analyzed according to our main contrasts of interest (Figure 4 Contrast Analysis). It was found that for the positive prime, Contrast B was far more significant than Contrast A. Contrast B postulates a significant change in positively primed group-ratings (Group 1) relative to the positively primed self-ratings, and the negatively primed self- and group-ratings, but this pattern was not supported by the data. On the other hand, the opposite pattern proved significant for positively primed self-

18

ratings. Thus, the data suggests that the positive prime significantly affected self-ratings for Group 3, relative to the group-ratings of Group 1 (also positively primed) and Groups 2 and 4. Contrast analysis also found that negatively primed group-ratings were strongly affected relative to the negatively primed self-ratings and the positively primed self/group-ratings. Although Contrast D was not significant, (F (3, 368) = -1.02, p = .306), Contrast C showed a strong indication of approaching significance (F (1, 368) = 1.91, p = .056. Contrast C postulated that Group 2 (group-ratings last) would show the weakest relation between self- and group-ratings. Though not as significant as Contrast B, Contrast C does warrant consideration.

Contrast A B C D

1 3 -1 1 1

Group 2 3 -1 -1 -1 3 -3 1 1 1

4 -1 -1 1 -3

t -0.90 2.07 1.91 -1.02

df 368 368 368 368

Significance (2-tailed)

0.369 0.039 0.056 0.306

Table 5 Contrast Analysis Partialed Data Final Study

Analysis By Identification The weak association between self- and group-ratings for Group 1 (positive prime) was highly unusual given our a priori predictions. Given that a manipulation check was also included in the form of a group identification measure (American Ideals Measure, AIM (Phinney et al., 1994)) the decision was made to analyze the data according to the identification measure. Simple inspection of the means indicates that there was little fluctuation in AIM scores (Table 6), and this is confirmed by an insignificant One-way ANOVA (F (3, 368) = .69, p = .563).

19

Group
1 2 3 4

N
97 69 98 108

Mean
1.9501 2.0350 1.9400 1.9608

SD
0.6951 0.6429 0.6705 0.6502

Table 6 American Ideals Measure Final Study

However, a median split by AIM scores revealed a more complicated scenario. The Final data (partialed) was subdivided into High/Low groups consisting of those who scored high on the AIM and those who scored low on the AIM. As Table 7 indicates there is a significant differential pattern based on identification with the target group. Group 1 demonstrates a significant difference in mean correlation scores when split between High/Low AIM scores (t = 2.994, p = .004). Groups 2 and 3 also showed an internal difference based on AIM scores, but this difference between High/Low groups was not significant. These data reveal that the high identifiers show depressed scores following the negative prime, but low identifiers show strong scores (t (90) = 2.994, p = .004). This trend was mirrored in Groups 2 and 3, but the trend did not reach standard significance. Figure 8 presents a view of the Final data (partialed), with Group 1 divided into High/Low groups.

Group
1 High 1 Low 2 High 2 Low 3 High 3 Low 4 High 4 Low

N
44 48 34 30 48 45 52 50

Fisher Scores Group Mean
0.1839 0.4587 0.1875 0.3760 0.3715 0.5022 0.3880 0.3779

Fisher Scores Group SD
0.4188 0.4581 0.3730 0.5385 0.3519 0.4360 0.3710 0.3793

Mean Group Difference
0.2748 -0.1885 0.1307 -0.0101

Significance (Two tailed)
t = 2.994 p = 0.004 t = -1.607 p = .113 t = 1.596 p = .114 t = -.126 p = .900

Table 7 Median split on American Ideals Measure, Final Study

20

0.50

0.40 p = 0.004

Fisher Z Score

0.30

0.20

0.10

r = 0.34

r = 0.32

r = 0.41

r = 0.41

r = 0.41

0.00 Group 1 High Group 1 Low Group 2 Group 3 Group 4

Figure 8

21

Discussion This study was begun with the intent of examining the relative stabilities of selfknowledge versus group-knowledge. To accomplish this goal we operationally measured self- and group-knowledge through self- and group-ratings, respectively. We hypothesized that following an emotional prime, self-ratings would be more less influenced by the prime than group-ratings. We also reasoned that this change would be reflective in the relationship between self- and group-ratings. Self- and group-ratings were measured before and after an emotional prime (positive or negative), with the order of self- and group-ratings, counterbalanced across treatment groups. Finally, we correlated self- and group-ratings to produce an index of associative strength. A Preliminary Study was first run to explore our hypotheses. Given our experimental paradigm (Table 1), we hypothesized that negatively or positively primed group-ratings would exhibit more change than negatively or positively primed selfratings. To this end, we predicted a specific pattern of results; we predicted that the relationship between self- and group-ratings would be disrupted by the negative prime when group-ratings follow the prime but not when group-ratings precede the prime. Likewise, we expected the positive prime to enhance the relationship between self- and group-ratings, but only when group-ratings follow the positive prime.

Group 1 Group 2 Group 3 Group 4

Self ratings Self ratings Group ratings Group ratings

Positive priming Negative priming Positive priming Negative priming

Group ratings Group ratings Self ratings Self ratings

Desirability ratings Desirability ratings Desirability ratings Desirability ratings

American Ideals American Ideals American Ideals American Ideals

Table 1 Main Experimental Design (Repeated)

Based on our experimental design, our predictions yielded a specific set of contrasts, presented in Figure 4. Of these contrasts, a tendency towards significance was found for Contrast C, but not for the other contrasts (for significance values, see Table 2 and Table 3). Contrast C stipulated that Group 3, and only Group 3, would show a 22

reduction in the relationship between self- and group ratings, due to the negative prime. Based on our predictions, negative primed group-ratings should lead to reduced association between self- and group-ratings, but negatively primed self-ratings should not lead to the same reduction. Contrast C’s tendency towards significance, suggests that the data is best described by a pattern in which there is marked reduction in associative strength for self- and group ratings due to a negatively primed group-ratings, but no reduction in associative strength for any other condition, regardless of negative or positive prime. In other words, group-ratings that follow a negative prime are less stable than self-ratings that follow the same prime. We also predicted that positively primed group-ratings would exhibit increased associative strength relative to positively primed self-ratings (i.e. Contrast A). However, this contrast did not prove significant, nor was there an indication of a trend towards significance. Similarly, the opposite contrast (Contrast B) did not exhibit significance or a trend towards significance. This lack of significance for Contrast A and Contrast B suggests that positively priming was did not differentially affect the relationship between self- and group-ratings, depending on order or presentation. The statistical patterns observed in the Preliminary encouraged us to pursue a larger study. The Preliminary Study lacked the necessary power to achieve full significance, and thus we believed that a larger study would provide greater statistical power. However, the results of the Final Study were not so clear. We saw confirmation and disconfirmation of the statistical patterns from the Preliminary Study. Likewise the results of the Final Study also confirmed one a priori prediction and contradicted another. First, the data from the Preliminary Study suggested that the positive prime has no differential affect whether it precedes self- or group-ratings. Thus, the Preliminary Study data did not indicate that self-ratings are more or less affected by the positive prime, relative to group-ratings. In the Final Study, contrast analysis strongly indicated that the positive prime affected self-ratings more than group-ratings, as reflected in the stronger relationship between self- and group ratings for Group 3 relative to Groups 1, 2, and 4 ( Contrast B). The opposing contrast (Contrast A) did not achieve significance, suggesting that the group-ratings of Group 1 were less affected by the positive prime than the selfratings of Group 3. The significance of Contrast B over Contrast A, contradicts our first

23

a priori prediction, and suggests that our hypothesis was partially incorrect or the experiment was inadequate for the intended purposes. Like the Preliminary Study, the Final Study strongly indicated that negative priming does differentially affect self-ratings relative to group-ratings. We saw a strong reduction in the associative strength between self- and group ratings for Group 2 (negatively primed self-ratings) relative to Group 4 (negatively primed self-ratings). Although standard statistical significance was not achieved, a trend towards significance was noticed for Contrast C relative to Contrast D. Contrast C stipulates a reduced relationship between self- and group-ratings for Group 2 relative to all other group, while Contrast D stipulates a reduced relationship for Group 4 relative to all other groups. Given that Contrast C displays a greater tendency towards significance than Group 4, we interpret this disparity as evidence that negatively primed group-ratings are more affected than negatively primed self-ratings. At first glace, our overall data support two divergent hypotheses. On one hand, part of the data supports theories of stability in self-knowledge. On the other hand, the data also supports theories of instability in self-knowledge. Prediction One was supported by the data of the Preliminary Study, but not the Final Study. Thus, though the Preliminary Study indicated that following a positive prime, self-ratings are more stable than group-ratings, the Final Study found support for the opposite pattern. Yet, support also exists for Prediction Two. Both the Preliminary Study and the Final Study consistently and strongly suggest that the negative prime differentially affects groupratings relative to self-ratings. How can we explain these results? If these data reveal self-ratings to be simultaneously stable and unstable, then further explanation is required. One possibility is that more a complex theoretical framework is required than the simple dichotomy of stable self-knowledge versus unstable self-knowledge. A second possibility is that unforeseen complications or confounds arose that make interpreting the data problematic. Of these two possibilities, this current research is best suited to addressing the latter. If we return to discussing the results according to the identification data, several interesting points emerge. First, Group 1 shows a significant break between high and low identifiers, such that high AIM scores lead to low associative strength between self- and

24

group ratings. In contrast, low AIM scores lead to high associative strength between selfand group-ratings. This significant pattern suggests that the more individuals identify with the target group (Americans), the less relationship between their self-ratings and subsequent group-ratings. Given that this pattern was only observed for the positive treatment and only for the positively primed group-ratings, we must ask whether or not strength of identification with the target group was a confound. Similarly, it is possible that the selected target group (American) was inadequate for this experiment. Choosing “American” as the target group may have introduced a target too large for the desired manipulation, thus leading to weak and undesired affects of priming. This is reflected in the significant relationship between AIM scores and associative strength between self- and group-ratings for Group 1. Past research does not support the inverse relationship between AIM scores and the associative strength of self-group ratings, in particular self-categorization theory and optimal distinctiveness theory have great difficult accounting for this pattern. Optimal distinctiveness theory stipulates that excessive individuation is always undesirable (Brewer, 1991), thus it is difficult to understand why positively primed group-ratings would show such a marked reduction in relationship with the preceding self-ratings.In this specific example, it is not obvious why participants would be motivated to produce divergent self- and group-ratings. Also, given that general self-categorization theory hypothesizes that individuals extract group attributes and then apply those attributes to themselves, the positively primed groupratings should have been highly related to self-ratings. It is difficult to justify why individuals who show strong identification with the target group “American,” and who have been positively primed to the target group, then demonstrate reduced relationship between self-ratings and ratings of the target group. It is especially difficult to explain why the positively primed group-ratings would show a similar reduction in associative strength to the negatively primed group-ratings. These findings are also problematic for the research of Bierat et al. (1996) who declared that individuals would reject negative stereotypes, but embrace positive stereotypes. For the sake of argument, the data were re-examined with the Group 1 High AIM scores removed (Figure 9). In doing this, we were addressing a simple question: Would different patterns of significance emerge if self- and group-ratings for Group 1 were not

25

confounded by the High/Low AIM scores? For Group 1, individuals high on the AIM scale showed exceptionally low relationship between self- and group-ratings. Therefore, had all respondents in Group 1 tended towards the responses of the high identifiers, there would be no significant changes in statistical patterns. However, if the data for Group 1 had tended in the direction of the low identifiers, than significant changes would occur in the data. As presented in Table 8, the significance of Contrasts A-D would dramatically change if the variation in Group 1 was reduced. First, were would be no statistical support for the difference in positive primes. Therefore, while there would be so support for Prediction One, neither would the data contradict the prediction. Second, there would be strong statistical support for Prediction Two. Thus the data would indicate that a statistically significant reduction in associative strength for Group 2, and Group 2 alone, and so indicating that group-ratings are statistically unstable relative to self-ratings.

Mean Scores Adjusted Data
0.50

0.40

Fisher Z Score

0.30

0.20

r = 0.40

r = 0.26

r = 0.40

r = 0.37

0.10

0.00 Group 1: Positive Group 2: Negative Group 3: Positive Group 4: Negative

Order: Self-Group

Order: Group-Self

Figure 9 Adjusted Final Data (Partialed Fisher Scores)

26

Contrast A B C D

1 3 -1 1 1

Group 2 3 -1 -1 -1 3 -3 1 1 1

4 -1 -1 1 -3

t 0.98 1.32 2.50 -0.28

df 324 324 324 324

Significance (2-tailed)

0.328 0.118 0.013 0.779

Table 8 Adjusted Final Data (Partialed Fisher Scores)

27

Conclusion
Based on the actual data and the adjusted data, it appears that further investigation is needed. Given that the results of the Final Study simultaneously support and contradict theories of stable self-knowledge, and given that the data when adjusted according to identification, show only support for the theory of stable self-knowledge, we must conclude that further research is necessary to better clarify these results. The most accurate conclusion is that identification was an unforeseen confound that has influence the experiment in a problematic fashion. The most likely direction for future research is the selection of a target group that will allow for more specific and more powerful experimental manipulation.

28

29

References
Agostinelli, G., Sherman, S. J., Presson, C. C., & Chassin, L. (1992). Self-protection and self-enhancement biases in estimates of population prevalence. Personality and Social Psychology Bulletin, 18(5), 631-642. Biernat, M., Vescio, T. K., & Green, M. L. (1996). Selective self-stereotyping. Journal of Personality and Social Psychology, 71(6), 1194-1209. Brewer, M. B. (1991). The social self: On being the same and different at the same time. Personality and Social Psychology Bulletin, 17(5), 475-482. Campbell, J. D. (1986). Similarity and uniqueness: The effects of attribute type, relevance, and individual differences in self-esteem and depression. Journal of Personality and Social Psychology, 50(2), 281-294. Clement, R. W., & Krueger, J. (2000). The primacy of self-referent information in perceptions of social consensus. British Journal of Social Psychology, 39(2), 279299. Clement, R. W., & Krueger, J. (2002). Social categorization moderates social projection. Journal of Experimental Social Psychology, 38(3), 219-231. Craik, F. I. M., Moroz, T. M., Moscovitch, M., Stuss, D. T., Winocur, G., Tulving, E., et al. (1999). In search of the self: A positron emission tomography study. Psychological Science, 10(1), 26-34. Crocker, J., Alloy, L. B., & Kayne, N. T. (1988). Attributional style, depression, and perceptions of consensus for events. Journal of Personality and Social Psychology, 54(5), 840-846. Gaertner, L., Sedikides, C., & Graetz, K. (1999). In search of self-definition: Motivational primacy of the individual self, motivational primacy of the

30

collective self, or contextual primacy? Journal of Personality and Social Psychology, 76(1), 5-18. Gramzow, R. H., Gaertner, L., & Sedikides, C. (2001). Memory for in-group and outgroup information in a minimal group context: The self as an informational base. Journal of Personality and Social Psychology, 80(2), 188-205. Haslam, S., Oakes, P. J., McGarty, C., Turner, J. C., Reynolds, K. J., & Eggins, R. A. (1996). Stereotyping and social influence: The mediation of stereotype applicability and sharedness by the views of in-group and out-group members. British Journal of Social Psychology, 35(3), 369-397. Hogg, M. A., & Turner, J. C. (1985). Interpersonal attraction, social identification and psychological group formation. European Journal of Social Psychology, 15(1), 51-66. Hogg, M. A., & Turner, J. C. (1987). Intergroup behaviour, self-stereotyping and the salience of social categories. British Journal of Social Psychology, 26(4), 325340. Krueger, J. (2000). The projective perception of the social world: A building block of social comparison processes. In J. Suls & L. Wheeler (Eds.), Handbook of social comparison: Theory and research. The Plenum series in social/clinical psychology (pp. 323-351). Dordrecht, Netherlands: Kluwer Academic Publishers. Krueger, J. I. (2003). Return of the ego--Self-referent information as a filter for social prediction: Comment on Karniol (2003). Psychological Review, 110(3), 585-590. Kunda, Z. (1999). Social cognition: Making sense of people. Cambridge, MA: The MIT Press. Lorenzi Cioldi, F. (1991). Self-stereotyping and self-enhancement in gender groups. European Journal of Social Psychology, 21(5), 403-417.

31

Markus, H. (1977). Self-schemata and processing information about the self. Journal of Personality and Social Psychology, 35(2), 63-78. Markus, H., & Kunda, Z. (1986). Stability and malleability of the self-concept. Journal of Personality and Social Psychology, 51(4), 858-866. Markus, H., Smith, J., & Moreland, R. L. (1985). Role of the self-concept in the perception of others. Journal of Personality and Social Psychology, 49(6), 14941512. Onorato, R. S., & Turner, J. C. (2004). Fluidity in the self-concept: the shift from personal to social identity. European Journal of Social Psychology, 34, 257-278. Phinney, J. S., DuPont, S., Espinosa, C., Revill, J., & Sanders, K. (1994). Ethnic identity and American identification among ethnic minority youths. In A.-M. Bouvy & F. J. R. van de Vijver (Eds.), Journeys into cross cultural psychology (pp. 167-183). Bristol, PA: Swets & Zeitlinger Publishers. Pickett, C. L., Bonner, B. L., & Coleman, J. M. (2002). Motivated self-stereotyping: Heightened assimilation and differentiation needs result in increased levels of positive and negative self-stereotyping. Journal of Personality and Social Psychology, 82(4), 543-562. Sherman, S. J., Chassin, L., Presson, C. C., & Agostinelli, G. (1984). The role of the evaluation and similarity principles in the false consensus effect. Journal of Personality and Social Psychology, 47(6), 1244-1262. Turner, J. C., Oakes, P. J., Haslam, S., & McGarty, C. (1994). Self and collective: Cognition and social context. Personality and Social Psychology Bulletin, 20(5), 454-463.

32


				
DOCUMENT INFO
Shared By:
Categories:
Tags:
Stats:
views:12
posted:10/16/2009
language:English
pages:32