Journal of psr~nality and Social Psychology Copyright 1989 by the American Psychological Amociatkm, Inc. 1989, Vol. 56, No. 6, 950-957 0022-3514/89/$00.75

Private Self-Consciousness, Self-Awareness, and the Reliability of Self-Reports

William Nasby Pacific Graduate School of Psychology

Two studies found that individuals high in private self-consciousness provide self-reports of greater reliability across time than individuals low in private self-consciousness. In addition, Study 2 found that a successful manipulation of self-awareness did not affect test-retest reliability of self-reports among Ss either high or low in private self-consciousness. The hypothesis that individuals high in private self-consciousness have articulated self-schemata of greater temporal stability than individu- als low in private self-consciousness received support. The discussion considers (a) how private self- consciousness, but not self-awareness, influenced test-retest reliability of self-reports, (b) how private self-consciousness and self-awareness may both influence, albeit differentially, the eriterion- related validity of self-reports, and (c) the possibility that current models of self-consciousness-self- awareness require reformulation.

Private self-consciousness (e.g., Fenigstein, scheier, & Buss, 1975), a cognitive style, denotes habitual attentiveness to or awareness of the self. Since Fenigstein et al. (1975) originally proposed the construct, research has examined numerous cor- relates of private self-consciousness, which Buss (1980), Carver and Scheier (1981), and scheier and Carver (1983) have re- viewed.

Not surprisingly, several correlates pertain to the processing of self-referent information. For example, individuals high in private self-consciousness (a) write more detailed self-descrip- tions (Turner, 1978b), (b) decide more quickly whether trait ad- jectives describe the self (Mueller, 1982; Turner, 1978c), (c) re- call more self-referent material under conditions of incidental learning (Agatstein & Buchanan, 1984; Hull & Levy, 1979, Ex- periment 1; Turner, 1980), and (d) commit more false alarms to self-descriptive traits but fewer to non-self-descriptive traits during a recognition memory task (Nasby, 1985, 1989) than do individuals low in private self-consciousness.

The previous studies have established relations between pri- vate self-consciousness and either the accuracy or the efficiency (e.g., speed, productivity) of processing self-referent informa- tion. Cognitive tasks that assess either accuracy or efficiency provide two means of inferring the degree to which an articu- lated schema has guided performance (e.g., Fiske & Taylor, 1984; Kihlstrom & Nasby, 1981; Nasby & Kihlstrom, 1986). Consequently, from the convergent results of the previous stud- ies, several investigators (e.g., Agatstein & Buchanan, 1984; Nasby, 1985) have concluded that individuals high in private self-consciousness have articulated self-schemata (e.g., Kihl- strom & Cantor, 1983; Markus, 1980; Rogers, 1981) more ex- tensively than individuals low in private self-consciousness.

Cognitive tasks that assess response consistency provide a

Correspondence concerning this article should be addressed to Wil- liam Nasby, Pacific Graduate School of Psychology, 935 East Meadow Drive, Palo Alto, California 94303.

950

third means of inferring the degree to which an articulated schema has guided performance (e.g., Fiske & Taylor, 1984; Markus, 1977). Considering the relation between schematic ar- ticulation and response consistency, individuals high in private self-consciousness should also process self-referent information more consistently than do individuals low in private self-con- sciousness. Indeed, McFarland and Sparks (1985) have found that the self-reports of subjects high in private self-conscious- ness are characterized by greater internal consistency than are the self-reports of subjects low in private self-consciousness.

Whereas McFarland and Sparks (1985) related private self- consciousness to the consistency of processing self-referent in- formation within one test session, the following studies examine relations between private self-consciousness and the consis- tency of processing self-referent information across two test ses- sions, and hence across time. The extension actually provides a test of a new hypothesis--namely, that individuals high in pri- vate self-consciousness have articulated self-schemata of greater temporal stability than have individuals low in private self-con- sciousness. More specifically, the following studies examine whether subjects high in private self-consciousness provide more reliable self-reports across time than do subjects low in private self-consciousness.

Interestingly enough, the same hypothesis derives from some research that has assessed behavioral correlates of private self- consciousness. For example, studies have consistently revealed positive correlations between private self-consciousness and the validity of self-reports across several domains (e.g., Froming & Carver, 1981; Gibbons, 1983; scheier, Buss, & Buss, 1978; scheier & Carver, 1980; Turner, 1978a). Therefore, self-reports from individuals high in private self-consciousness predict and postdict more accurately, and hence correspond more closely to, behavioral (nontest) criteria than do self-reports from indi- viduals low in private self-consciousness. Because validity pre- supposes reliability, positive and significant correlations ought to occur between private self-consciousness and temporal sta- bility (test-retest reliability) of self-reports.

SELF-CONSCIOUSNESS AND SELF-REPORTS 951

A related literature (e.g., Gibbons, 1983; Pryor, 1980; Pryor, Gibbons, Wicklund, Fazio, & Hood, 1977; Wicldund, 1982) has indicated that experimentally manipulating self-awareness also enhances the criterion-related validity of self-reports. Moreover; private self-consciousness and self-awareness have related similarly to a host of other var iables--a convergence that has assumed considerable theoretical importance. (For an extensive review, see Carver & Scheiet; 1981.) If the trait of pri- vate self-consciousness relates positively to the test-retest reli- ability of self-reports, then one might expect that experimen- tally manipulating the state of self-awareness would produce similar effects.

However, the proposed relation between private self-con- sciousness and temporal stability of self-schemata actually leads to a different expectation. Indeed, enhancing the test-retest reli- ability of self-reports through an experimental manipulation of self-awareness could seriously damage the hypothesis. Should an experimental manipulation of self-awareness not affect the test-retest reliability of self-reports, then theoretical repercus- sions of some import would ensue.

Overv iew o f Studies

I conducted two studies, both of which examined whether private self-consciousness relates positively to the test-retest re- liability of self-reports about behavioral dispositions. Study 1 sought primarily to support the hypothesis that individuals high in private self-consciousness report more reliably about the self across time than do individuals low in private self-conscious- ness. Although a negative outcome would undermine the hy- pothesis, a positive outcome would still require additional re- search to examine an alternative explanation. Study 2 exam- ined the alternative explanation by adding an experimental manipulation of self-awareness to the basic design of Study 1.

S tudy 1

M e t h o d

S u b j e c t s

Prospective subjects included 76 undergraduate volunteers (31 men and 45 women) who (a) completed the Self-Conscionsness Scale (SCS; Fenigstein et al., 1975) and Form A of the Personality Research Form (PRF; Jackson, 1984), and (b) agreed to complete other measures when contacted later during the semester. Seven prospective subjects (5 men and 2 women), however, did not complete the additional measures, which included Form B of the PRF. The actual subjects, therefore, in- eluded the 69 undea~graduates (26 men and 43 women) from the pro- speetive sample who completed both Form A and Form B of the PRF, thereby providing the necessary data to assess test-retest reliability.

I n s t r u m e n t s

Self-Consciousness Scale. I used the SCS (Fenigstein et al., 1975) to assess private self-conscionsness. Of the 23 items contained in the SCS, 10 define a subscale that measures private self-consciousness. Research (e&, Carver-& Glass, 1976; Fenigstein et al., 1975; Turner, Scheier, Carver, & Ickes, 1978) has amply documented the reliability and valid- ity (construct and discriminant) of the subscale.

Among the sample of actual subjects, scores ranged from 15 to 37 (M = 26.25, SD = 4.67). Prospective subjects who did not complete

Form B of the PRF obtained comparable scores, ranging from 14 to 38 (M = 26.43, SD = 4.16). Private self-consciousness also did not differ between men and women who completed both administrations of the PRF(t < 1).

Personality Research Form. Form A and Form B of the PRF (Jack- son, 1984) provided suitable means to gauge the test-retest reliability of self-reports about behavioral dispositions. Each form includes 15 scales, each ofwhich comprises 20 items. The 15 scales of each form contain no items that overlap, and the parallel forms include no common items. Indeed, according to Jackson (1984), Form A and Form B of the PRF meet the psychometric criteria of strictly parallel tests (Lord, 1964). Of the 15 scales, 14 measure personality variables from a comprehensive theory proposed by Murray (1938). Consequently, the content scales cover a broad, if not exhaustive, range of human functioning. (The re- maining scale provides a "validity" index.) Each item consists of a self- referent statement to which the subject responds true or false. Subjects usually require 30-45 min to complete the 300 items of either form.

P r o c e d u r e

During the first week of the second semest~ the prospective subjects completed the SCS and the PRF (From A) according to standard in- structions. Half of the prospective subjects (15 men and 23 women) completed the SCS first and the PRF (Form A) second, whereas the other half (16 men and 22 women) did the reverse. Approximately 2 months later, the actual subjects completed Form B of the PRF. De- briefings indicated that subjects did not recognize the hypothesis under investigation.

R e s u l t s

Column 1 and Column 2 of Table 1 present the means (T scores) of the 15 scales from the first administration (Form A) and second administration (Form B) of the PRF, respectively. Prospective subjects who only completed Form A of the PRF did not obtain scores that differed from actual subjects; see Col- umn 1.

To assess test-retest reliability, I first calculated the squared difference (d 2) between the T scores of each scale for each sub- ject. Then, for each subject, I summed the squared differences, thereby providing an index of profile similarity (global reliabil- ity) according to/)2 (Cronbach & Gleser, 1953; Osgood & Suci, 1952). Lower values would indicate greater similarity between profiles, and hence greater test-retest reliability, of self-reports than would higher values. If individuals high in private self-con- sciousness do provide more reliable self-reports than individu- als low in private self-consciousness, then a significant but nega- tive correlation should occur between private self-conscious- ness and the values of/32. Results confirmed expectations: Private self-consciousness and the profile index correlated sig- nificantly and negatively, r(67) = - .37 , p < .01.

The specific values of d 2 provide alternative means to exam- ine the hypothesis, separately for each scale. Again, lower values would indicate greater test-retest reliability of seW-reports than would higher values. Column 3 of Table 1 presents the corre- lations between private self-consciousness and the specific in- dexes that assess test-retest reliability for each scale. The num- ber of negative and significant correlations (12) far exceeded the number that one could reasonably attribute to chance (z = 13.33, p < .001). (One must interpret the significance of the z score cautiously, because each subject contributed to all corre-

952 WILLIAM NASBY

Table 1 Summary of Results for the Personality Research Form (PRF) Scales

PRF Scales 1 2 3 4 5

Achievement 51.25 49.48 -.29** .12 .07 Affiliation 48.75 5 0 . 7 2 -.33*** -.05 .05 Aggression 50.91 50.04 -.40"** .17 - .00 Autonomy 51.46 5 1 . 4 6 -.39"** .13 - . 0 3 Dominance 49.26 51.22 -.26* -.11 .10 Endurance 49.12 49.57 -.31"** .15 -.19 Exhibition 48.12 48.68 -.29"* - . 17 .01 Harmavoidance 48.80 50.22 - . 19 -.04 .05 lmpulsivity 51.30 49.20 -.29** -.08 - . 10 Nurturance 49.42 48.32 -.31"** -.07 .19 Order 50.09 51.36 .19 .13 .01 Play 50.26 48.38 -.35*** .19 .13 Social

Recognition 51.96 5 0 . 2 8 -.37*** .07 .06 Understanding 50.12 49.84 -.41"*** .12 .17 Infrequency 51.09 48.83 .17 -.05 - . l0

Note. Columns 1 and 2 present the means (Tscores) from the adminis- tration of PRF Form A and PRF Form B, respectively. Column 3 pre- sents the correlations between private self-consciousness and the values ofd 2. Columns 4 and 5, respectively, present the correlations between private self-consciousness and T scores from the administration Forms A and B of the PRF. (For each correlation, df= 67.) *p < .05 (two-tailed test). **p < .02 (two-tailed test). ***p < .01 (two-tailed test). **** p < .001 (two-tailed test).

lations. Nevertheless, failure of the z score to attain significance would have seriously undermined the hypothesis.)

Content of the Self-Reports and Discriminant Validity Columns 4 and 5 of Table 1 list the correlations between pri-

vate self-consciousness and the 15 scales from, respectively, the first administration (Form A) and the second administration (Form B) of the PRE No correlation from either column reached significance, indicating that private self-consciousness did not affect the content of the self-reports. One cannot, there- fore, attribute the results of the primary analyses to content differences that vary according to private self-consciousness. Moreover, the discriminant validity of private self-conscious- ness receives further support from the nonsignificant corre- lations (of. Carver & Glass, 1976).

The data can provide additional evidence about discriminant validity, because the SCS also contains a subscale that measures public self-consciousness, which denotes habitual attentiveness to or awareness of the impression(s) one conveys to others. None of the correlations between public self-consciousness and the scales of the PRF from either administration reached sig- nificance. Of greater importance, considering the correlations between public self-consciousness and each index of reliability (profile and scale), no significant results occurred. (Study 2, similarly, provided no evidence that public self-consciousness and indexes of reliability covaried.) Because the subjects com- pleted the personality questionnaires anonymously, thereby minimizing the likelihood of impression management, public self-consciousness ought not to have moderated test-retest reli- ability of self-reports.

Discussion Although Study 1 provides considerable evidence that indi-

viduals high in private self-consciousness report more reliably about the self across time than do individuals low in private self- consciousness, a rival hypothesis warrants additional research. The rival hypothesis does not assume that individuals high in private self-consciousness have articulated self-schemata of greater temporal stability than have individuals low in private self-consciousness. Instead, the alternative hypothesis proposes that under appropriate circumstances individuals low in private self-consciousness can report about the self as reliably (consis- tently) over time as can individuals high in private self-con- sciousness.

A construct that actually pertains to private self-conscious- ness may pinpoint the appropriate circumstances and provide the basis for deriving the alternative hypothesis. Private self- consciousness, a trait, denotes a predisposition to experience a corresponding state of self-awareness. According to prevalent accounts (e.g., Buss, 1980; Carver & Seheier, 1981; Dural & Wicklund, 1972), an individual experiences self-awareness whenever, or as long as, some facet of the self occupies focal attention. (When attention shifts away from the self, the state presumably ceases.) Self-awareness may enhance the accessibil- ity of, and therefore facilitate the retrieval of, information from the self-schema. Once retrieved, the information may then guide how an individual responds to personality items (e.g., the statements of the PRF). If an individual cannot readily access and retrieve the relevant information from the self-schema (even though the information exists), then other (extraneous) factors will shape and otherwise contribute to responses (e.g., response sets, transient moods, etc.).

Individuals high in private self-consciousness ought to expe- rience states of self-awareness more often than individuals low in private self-consciousness. In Study 1, a greater proportion of subjects high in private self-consciousness may have experi- enced self-awareness while completing both self-reports than did subjects low in private self-consciousness. Across time and test sessions, then, the subjects high in private self-conseions- ness may have accessed and retrieved the same or similar infor- mation from the contents of the self-schema more consistently than did subjects low in private self-consciousness. Extraneous

In an earlier study that examined the same hypothesis, subjects twice completed the Adjective Check List (ACL; Gough & Heilbrun, 1983) with a test-retest interval of 3 months separating the two admin- istrations. The results supported the hypothesis; subjects high in private self-consciousness reported more reliably about the self across time than did subjects low in private self-consciousness. A feature that origi- nally recommended the ACL over the PRF concerns breadth of cover- age, because the ACL measures 37 personality variables, whereas the PRF measures only 14. Nevertheless, a potential weakness of the ACL eventually motivated the shift to the PRF. Specifically, the ACL, unlike the PRF, lacks alternate or parallel forms. Because subjects responded to the same items twice in the earlier study, the results could have impli- cated "contamination" from memory, because evidence indicates that subjects high in private self-consciousness remember self-referent infor- mation more than do subjects low in private self-consciousness (e.g., Agatstein & Buchanan, 1984; Hull & Levy, 1979, Experiment 1; Turner, 1980).

SELF-CONSCIOUSNESS AND SELF-REPORTS 953

factors, conversely, may have determined or influenced self-re- ports among the latter subjects during either or both assess- ments.

To examine the rival hypothesis, Study 2 both measured pri- vate self-consciousness and manipulated self-awareness while subjects completed the self-reports. If, through an experimental manipulation, subjects low in private self-consciousness but high in self-awareness provide self-reports of(a) greater reliabil- ity than subjects low in both private self-consciousness and self- awareness, and (b) comparable reliability to subjects high in private self-consciousness, then the hypothesized relation be- tween private self-consciousness and temporal stability of self- schemata would not receive support, whereas the alternative hy- pothesis would.

S tudy 2

Method

Subjects

The subjects included 80 undergraduate volunteers (40 men and 40 women) who completed first the PRF (Forms A and B) under conditions that either did or did not elevate self-awareness (20 men and 20 women per experimental group) and then the SCS. Prospective subjects had originally included 92 undergraduates (48 men and 44 women) who completed the first administration of the PRF. Of the prospective sub- jects, therefore, 12 (8 men and 4 women) did not complete the second administration of the PRF or the SCS.

Instruments

Self-Consciousness Scale. Again, I used the SCS (Fenigstein et ai., 1975) to assess private self-consciousness. Among the sample of actual subjects, scores ranged from 13 to40(M= 26.40,SD= 4.87).

Personality Research Form. I used Form A and Form B of the PRF to assess the test-retest reliability of self-reports.

Manipulation check. Davis and Brock (1975, Experiment 2) devised and validated a practical and straightforward technique to assess state levels of self-awareness. Basically, the technique assumes that verbal ref- erences to the self and level of self-awareness covary. The actual task presents a passage that contains 43 sentences, including 50 pronouns, from a foreign language (Swahili) that none of the subjects had pre- vionsly encountered. According to the instructions, the task measures sensitivity to foreign languages and the subject must guess the meaning of each pronoun. The number of first-person, singular pronouns that the subject chooses (e.g, I, me, my) provides the index of self-awareness.

Procedure

A female experimenter who did not know the purpose of the research tested each subject individually. All subjects completed Form A of the PRF first. Haifthe sample (20 men and 20 women) completed the PRF (Form A) under conditions that attempted to induce or heighten the state of self-awareness, whereas the remaining half did not. Specifically, the former subjects completed the PRF while seated before a large mir- ror (121.92 em× 106.68 cm) that reflected an image of the face, neck, and shoulders. The latter subjects completed the personality question- naire while seated before the nonreflective surface of the same mirror. Considerable research indicates that exposure to a mirror image of the face does increase self-awareness (e.g~, Buss, 1980; Carver & Scheier, 1981). To verify that the experimental manipulation succeeded, all sub- jects completed the state measure of self-awareness (Davis & Brock,

1975, Experiment 2). The order of administering the PRF and the ma- nipulation check varied. Within each group, half the subjects (10 men and 10 women) completed first the PRF and then the manipulation check. The remaining subjects from each group did the reverse.

Between 6 and 10 weeks after the initial session (Mdn = 8 weeks), subjects completed Form B of the PRF, the manipulation check to gauge self-awareness, and the SCS. Subjects completed both questionnaires and manipulation checks before either the reflective or the nonreflective surface of the mirror. Also, for each subject, the order of completing the PRF and the manipulation check did not vary between sessions. Finally, administration of the SCS occurred just before a debriefing that I con- dueted outside the laboratory setting. (I did not know the experimental status of any subject before administering the SCS. Nor did the room where debriefing occurred contain a mirror.)

Results

Preliminary Analyses

Private self-consciousness. A 2 (self-awareness: high vs. low) × 2 (sex: male vs. female) × 2 (order. PRF first and manip- ulation check second vs. manipulation check first and PRF sec- ond) analysis of variance (ANOVA) of the private self-conscious- ness scores revealed no significant effects, Fs( l , 72) < 1.48, ps > .20. Given random assignment to groups, one would not have expected to find significant effects of self-awareness and order (the manipulated variables). Also, that private self-conscious- ness did not differ systematically between men and women rep- resents a common and expected result.

State measure of self-awareness. To conduct the next analy- sis, I first blocked subjects according to level of private self-con- sciousness (high vs. low), using a median split, and then assessed the scores from the state measure of self-awareness (Davis & Brock, 1975, Experiment 2), using a 2 (private self-conscious- ness: high vs. low) × 2 (self-awareness: high vs. low) × 2 (sex: male vs. female) × 2 (order: manipulation check first and PRF second vs. PRF first and manipulation check second) × 2 (test session: first and second) mixed, unweighted-means ANOVA. The analysis produced a significant main effect of private self- consciousness, F ( I , 65) = 4.02, p < .05, and a significant main effect of self-awareness, F( 1,65 ) = 7.14, p < .01. The main effect of private self-consciousness indicated that subjects high in pri- vate self-consciousness chose more first-person, singular pro- nouns (M = 17.88, or 35.75%) than did subjects low in private self-consciousness (M = 14.73, or 29.45%). Similarly, the main effect of self-awareness indicated that subjects who responded before the reflective surface of the mirror chose more first-per- son, singular pronouns (M = 19.65, or 39.30%) than did sub- jects who responded before the nonreflective surface of the mir- ror (M = 12.95, or 25.90%). The first effect supports the conclu- sion that subjects high in private self-consciousness experienced more self-awareness overall than did subjects low in private self- consciousness; the second effect indicates that the experimental manipulation of self-awareness succeeded. Some of the remain- ing effects, none of which attained or approached significance, deserve comment. The nonsignificant main effect of test ses- sion, F( I , 65) = 1.32, p > .20, and the nonsignificant interaction between self-awareness and test session, F( I , 65) = 1.14, p > .20, support the inferences that the experimental manipulation of self-awareness succeeded across both sessions and to the

954 WILLIAM NASBY

same extent. In addition, from the nonsignificant interactions between private self-consciousness and self-awareness, F(I, 65) -- 1.63, p > .20, private self-consciousness and test session, F(I, 65) = 1.49, p > .20, and private self-consciousness, self- awareness, and test session (F < 1), one may reasonably con- clude that private self-conscionsness did not moderate or qual- ify the effects of experimentaUy manipulating self-awareness.

Correlational analyses supplemented the ANOVA. Among subjects who completed the PRF and the manipulation cheek before the reflective surface of the mirror, private self-con- sciousness and the state measure of self-awareness correlated positively although insignificantly during both the first and see- ond test session, r(38) = .12 and r(38) = .16, respectively. Among subjects who completed the PRF and the manipulation check before the nonreflective surface of the mirror, private self- consciousness and the state measure of self-awareness again correlated positively although insignificantly during both the first and second test session, r(38) = .21 and r(38) = . 19, respec- tively. (None of the correlations differed from each other.)

Further analyses did reveal significant relations between pri- vate self-consciousness and a combined index of self-focus, which consisted simply of the sum of each subject's scores on the state measure from the first and second test session. Specifi- cally, private self-conscionsness and the combined index of self- focus correlated significantly both among subjects who re- sponded before the reflective surface of the mirror, r(38) = .30, p < .05, and among subjects who responded before the nonre- flective surface of the mirror, r(38) ffi .35, p < .05. (The corre- lations did not differ from each other.) The significant corre- lations here would explain how the previous ANOVA produced a significant main effect of private self-consciousness, even though private self-consciousness and the state measure of self- awareness did not correlate significantly within each level of test session.

Primary Analyses

The primary analyses involve indexes of/)2 and d 2. The for- mer index provides the measure of profile similarity across test sessions and operationalizes global reliability, whereas the com- ponents of the global index, d 2, provide the specific indexes that assess reliability from each scale.

Considering first the global index, the results indicated that 192 among subjects who completed the PRF before the reflective surface of the mirror (M = 867.18) did not significantly differ from 192 among subjects who completed the PRF before the nonreflective surface (M = 879.30), t(78) = - 1.14, p > .20. (A test for difference between variances of two independent sam- ples did not attain or approach significance. Consequently, for the global index, one can assume homogeneity of independent variances among the two groups.)

In contrast, private self-consciousness and 192 did correlate significantly and negatively among both the former group, r(38) ffi - .42, p < .01, and the latter group, r(38) = - .39, p < .02. The correlations, however, did not differ between the groups (z = . 15). The state manipulation, therefore, did not pre- dict global reliability of self-reports, whereas the trait measure did. That private self-consciousness and D 2 correlated signifi- cantly and negatively replicates a major result of Study 1.

Table 2 Mean Values of d 2for Each Personality Research Form (PRF) Scale as a Function of Self-Awareness and Corresponding Values oft

Self-awareness

PRF Scales High Low t

Achievement 49.13 54.00 - 1.56 Affdiation 52.33 59.98 - 1.68 Aggression 59.95 55.93 1.49 Autonomy 64.03 61.90 1.26 Dominance 62.13 51.88 2.14" Endurance 60.98 57.95 1.45 Exhibition 53.98 53.03 0.88 Harmavoidance 53.00 54.03 -0.82 Impulsivity 58.85 59.18 -0.46 Nurturance 60.03 64.95 - 1.60 Order 49.93 61.03 -2.27* Play 61.95 64.08 - 1.09 Social Recognition 61.90 62.10 -0.23 Understanding 53.93 55.15 -1.17 In frequency 65.05 65.00 0.02

Note. For each t ratio, df= 78. * p < .05 (two-tailed test).

I then analyzed the specific components of the profile mea- sure. Table 2 lists the mean values of d 2 across each scale ac- cording to the experimental manipulation of self-awareness (high vs. low), and also presents the t ratios that assess whether corresponding pairs of group means differ. Two of the ratios indicated significant differences between the experimental groups. One test (Order scale) revealed greater reliability among subjects who experienced the heightened level of self-awareness; however, the other (Dominance scale) revealed the opposite. Sampling error alone can explain the number of significant effects; the opposite signs also underscore the conclusion that the state of self-awareness per se did not affect reliability. (Cor- responding tests for difference between variances of two inde- pendent samples did not attain significance. Consequently, for each scale, one can assume homogeneity of independent vari- ances among the two groups.)

Table 3 lists the correlations between private self-conscious- ness and the specific indexes that assess test-retest reliability across each scale according to whether subjects completed the PRF before the reflective surface of the mirror (high self-aware- ness) or before the nonreflective surface (low self-awareness). Negative and significant correlations dominate the table. More specifically, Columns 1 and 2 each contain 11 correlations that reached significance, and one cannot attribute the number to chance (z = 12.14, p < .001; again, one must interpret the z score cautiously).

Table 3 also presents the tests of difference between indepen- dent correlations (one per scale). Of the 15 comparisons, none approached, let alone exceeded, the critical value (z = 1.96). Therefore, akin to the global index, analyzing the data from scale to scale also produced results that largely replicated those of Study 1 and extended across two levels of self-awareness.

To examine the alternative hypothesis more directly, I per- formed a series of 2 (self-awareness: high vs. low) × 2 (private

SELF-CONSCIOUSNESS AND SELF-REPORTS 955

Table 3 Correlations Between Private Self-Consciousness and d e for Each Personality Research Form (PRF) Scale as a Function of Self-Awareness and Corresponding Tests of Difference Between Independent Correlations (z)

Self-awareness

PRF Scales High Low z

Achievement -.35" -.24 0.52 Atfdiation -.33* -.32* 0.04 Aggression -.37* -.35* 0.10 Autonomy -.43"** - .39** 0.21 Dominance - . 19 -.21 0.09 Endurance -.39** -.32* 0.34 Exhibition -.39** -.35* 0.20 Harmavoidance - . 17 .09 1.13 lmpulsivity -.32" -.38"* 0.29 Nurturance -.37* -.35* 0.10 Order -.09 -.33* 1.09 Play -.32* -.35* 0. i 4 Social Recognition -.36* -.32* 0.19 Understanding -.40"* -.43"** 0.15 lnfrequency .08 .19 0.48

Note. For each correlation, dr= 38. * p < .05 (two-tailed test). ** p < .02 (two-tailed test). *** p < .01 tailed test).

(two-

self-consciousness: high vs. low) unweighted-means ANOVAS. One ANOVA examined the global index of reliability and the remaining ANOVAS, one per scale, examined the specific in- dexes.

The alternative hypothesis would have received support if sig- nificant interactions between private self-consciousness and self-awareness had emerged and if the interactions had con- formed to a specific pattern. Specifically, the strongest evidence would have occurred from significant interactions indicating that the group low in private self-consciousness and high in self- awareness produced self-reports of (a) greater reliability than the group low in both private self-consciousness and self-aware- ness and (b) comparable reliability to both groups high in pri- vate self-consciousness. Significant interactions that conformed

"to a different pattern would have supported a weaker version of the alternative hypothesis; namely, a pattern that resembles the previous one, except that the group high in both private self- consciousness and self-awareness obtained the highest level of reliability.

Of the 16 ANOVAS, not one produced an interaction between self-awareness and private self-consciousness that attained or approached significance, Fs(l, 76) < 1.51, ps > .20. Conse- quently, neither the strong nor the weak version of the alterna- tive hypothesis received any support. The main effects, of course, simply indicated once again that private self-conscious- ness affected reliability, global and specific, whereas self-aware- ness did not. (A series of 16 Fm~ tests, one for each of the previ- ous ANOVAS, indicated that the data did not violate the assump- tion of homogeneity of independent variances.)

Content o f the Self-Reports and Discriminant Validity A series of 2 (self-awareness: high vs. low) X 2 (test session:

first and second, or Form A and Form B of the PRF) mixed

ANOVAS (one per scale) ofthe Tscores produced no effects that reached significance, Fs(1, 78) < 1.67, ps > .20. In addition, I computed correlations between private self-consciousness and T scores on each scale of the PRF within each level of self- awareness (high vs. low) and test session (first and second, or Form A and Form B of the PRF). Of the 60 correlations, only two attained significance. More specifically, private self-con- sciousness and T scores on the Aggression scale correlated sig- nificantly, r(38) = - .34, p < .05, among subjects who responded before the reflective surface of the mirror during the first ses- sion; private self-consciousness and Tscores on the Order scale also correlated significantly, r(38) = .32, p < .05, among sub- jects who responded before the nonreflective surface of the mir- ror during the first session. One can reasonably attribute both of the signilicant correlations to sampling error. Consequently, neither private self-consciousness nor self-awareness reliably affected the content of the self-reports. One cannot, therefore, attribute the results of the primary analyses to content differ- ences that vary according to private self-consciousness or self- awareness. Moreover, considering the correlational analyses, the discriminant validity of private self-consciousness again re- ceived support, and did so under two levels of self-awareness.

Discussion

To summarize, a standard means of manipulating self-aware- ness succeeded, and succeeded equally well across both ses- sions. In addition, subjects high in private self-conscionsness experienced more self-awareness overall than did subjects low in private self-consciousness; however, private self.conscious- ness did not moderate or qualify the effects of the experimental manipulation.

The primary analyses, therefore, provided a straightforward means of testing the alternative hypothesis. According to the alternative, manipulating self-awareness should have boosted test-retest reliability of self-reports among subjects low in pri- vate self-consciousness, which did not occur. Nevertheless, Study 2 replicated and extended Study 1: Private self-conscious- ness predicted test-retest reliability, global and specific, among both subjects who completed the self-reports before the nonre- flective surface of the mirror (a replication) and subjects who did the same before the reflective surface (an extension). Also, the correlations between private self-consciousness and the in- dexes that operationalized the test-retest reliability of self-re- ports did not differ between the experimental groups.

General Discussion

Beyond the support that the major hypothesis received from both studies, the result of perhaps the greatest interest was the nonsignificant effects of self-awareness. Why did the experi- mental manipulation of self-awareness in Study 2 not influence the test-retest reliabifity of self-reports either among subjects low in private self-consciousness or among subjects high in pri- vate self-consciousness?

Quite plausibly, a self-awareness manipulation would serve to activate or prime the self-schema, thereby enhancing the ac- cessibility and facilitating the retrieval of self-relevant informa- tion from memory. To the extent that the self-schemata ofindi-

956 WILLIAM NASBY

viduals low in private self-consciousness lack temporal stability, self-awareness manipulations at different points in time would activate or prime dissimilar information. At each point in time, their self-reports might be veridical, in the sense of actually re- flecting the information that the self-schema contains (cf. Gib- bons, 1983), but those reports would still differ when compared across time and hence would be unreliable. Among individuals low in both private self-consciousness and self-awareness, the reports when compared across time would not only lack reli- ability, but would also lack veridicality (Gibbons, 1983) at each time of assessment to the extent that factors other than the ac- tual content of self-schemata influenced responses.

This analysis of the effects that self-awareness manipulations produce would perhaps suggest that subjects high in both pri- vate self-consciousness and self-awareness ought in Study 2 to have provided self-reports of greater reliability than subjects high in private self-consciousness but low in self-awareness. In fact, however, this analysis suggests otherwise.

More specifically, studies of social judgment (e.g., Bargh & Tota, 1988) have indicated that frequently activated self-sche- mata can become chronically accessible over time. Frequently thinking about the self--a defining characteristic of high pri- vate self-consciousness--would correspond to frequently acti- vating information about the self. Therefore, among individuals high in private self-consciousness, self-schemata may not only be extensively articulated and temporally stable but may also be chronically accessible. Chronically accessible self-schemata would require less activation to access and retrieve from memory.

Referring specifically to Study 2, subjects high in private self- consciousness, having articulated easily accessible self-sche- mata, may not have required the "extra" activation from the self-awareness manipulation to report veridically about the self during either of the test sessions. Also, having articulated tem- porally stable self-schemata, individuals high in private self- consciousness would have reported about the self reliably across the test sessions, under conditions of either low self-awareness (in a relative, if not an absolute, sense) or high self-awareness.

Conceptualizing private self-consciousness and self-aware- ness from this vantage may address a second type of consistency, namely, consistency between a self-report about a behavioral disposition (e.g., a personality trait) and nontest or actual be- havior (criterion-related validity). Previous research (e.g., Froming & Carver, 1981; Gibbons, 1983; Scheier et al., 1978; Scheier & Carver, 1980; Turner, 1978a) has demonstrated greater correspondence between self-reports about behavioral dispositions and criterion measures of actual behavior among individuals high in private self-consciousness than among indi- viduals low in private self-consciousness.

Recall also that a related literature (e.g., Gibbons, 1983; Pryor, 1980; Pryor et al., 1977; Wicklund, 1982) has indicated that experimentally inducing self-awareness enhances the crite- rion-related validity of self-reports. Why should both the trait of private self-consciousness and the state of self-awareness moderate similarly the criterion-related validity of self-reports? Although explanations (e.g., Buss, 1980; Carver & Scheier, 1981; Gibbons, 1983; Pryor, 1980) differ somewhat, each attri- butes the greater correspondence between self-reports and crite- rion measures to certain effects that ensue from the state of self-

awareness. For example, Pryor (1980) asserted that self-aware individuals closely monitor their responses to a self-report in- ventory or questionnaire, which helps them to bring those re- sponses into closer alignment with their actual behavioral dis- positions or with their actual past behavior. To extend this ac- count, private self-consciousness would presumably moderate criterion-related validity of self-reports in the same way, be- cause at any given time individuals high in private self-con- sciousness would more likely experience the critical state of self- awareness than would individuals low in private self-conscious- ness. This account and others suggest that among individuals high in private self-consciousness (who happen to be low in self- awareness), one would not expect to observe high correspon- dence between their self-reports and criterion measures of their actual behavior.

Alternatively, the proposal under consideration here suggests that among individuals high in private self-consciousness (who have articulated easily accessible self-schemata), even low self- awareness (again, in a relative, if not an absolute, sense) may suffice to report veridically about the self. Moreover, among in- dividuals high in private self-consciousness, high self-awareness may not be a prerequisite for correspondence to occur between self-reports and criterion measures.

Conversely, among individuals low in private self-conscious- ness, elevating self-awareness through an experimental manip- ulation may enhance the accessibility and facilitate the retrieval of information that self-schemata contain at that time. If the criterion measure occurs in close temporal proximity to the self-report, then one could reasonably expect correspondence between the two. Results from Study 2, however, further suggest that if the criterion measure does not occur in close temporal proximity to the self-report, then the likelihood of correspon- dence would decline.

These differential predictions serve to distinguish this ac- count of self-awareness-self-consciousness from others that are currently available. In addition, these predictions derive from a theoretical account that can easily explain why private self- consciousness, but not self-awareness, moderates reliability of self-reports across time; other accounts would predict that self- awareness and private self-consciousness should both influence test-retest reliability in a similar way. Finally, this account can also explain a variety of relations that have been recently ob- tained between private self-consciousness and performance on cognitive tasks that require the processing of self-referent infor- mation. Given these points, it may be necessary to reexamine conceptualizations of self-consciousness and self-awareness that are currently available, most notably, their implicit as- sumptions that (a) private self-consciousness and self-awareness should affect outcome measures similarly, and (b) when private self-consciousness and self-awareness do affect outcome mea- sures similarly, they do so in the same way.

References

Agatstein, E C., & Buchanan, D, B. (1984). Public and private self- consciousness and the recall of self-relevant information. Personality and Social Psychology Bulletin, 10, 314-325.

Bargh, J. A., & Tota, M. E. (1988). Context-dependent automatic pro- cessing in depression: Accessibility of negative constructs with regard

SELF-CONSCIOUSNESS AND SELF-REPORTS 957

to self but not others. Journal of Personality and Social Psychology, 54, 925-939.

Buss, A. H. (1980). Self-consciousness and social anxiety. San Fran- cisco: Freeman.

Carver, C. S., & Glass, D. C. (1976). The self-consciousness scale: A discriminant validity study. Journal of Personality Assessment, 40, 169-172.

Carver, C. S., & Sebeier, M. E (1981). Attention and self-regulation: A control-theory approach to human behavior. New York: Springer-Vet- lag.

Cronbach, L. J., & Gleser, G. C. (1953). Assessing similarity between profiles. Psychological Bulletin, 5 O, 456--473.

Davis, D., & Brock, T. C. (1975). Use of first person pronouns as a function of increased objective self-awareness and performance feed- back, Journal of Experimental Social Psycholog3z, 11, 381-388.

Duval, S. & Wicklund, R. A. (1972). A theory of objective self-aware- ness. New York: Academic Press.

Fenigstein, A., Seheier, M. E ,& Buss, A. H. (1975). Public and private self-consciousness: Assessment and theory. Journal of Consulting and Clinical Psychology, 43, 522-527.

Fiske, S. T., & Taylor, S. E. (1984). Social cognition. Reading, MA: Ad- dison-Wesley.

Froming, W. J., & Carver, C. S. (1981~). Divergent influences of private and public self-consciousness in a compliance paradigm. Journal of Research in Personality, 15, 159-171.

Gibbons, E X. (1983). Self-attention and self-report: The 'veridicality' hypothesis. Journal of Personality, 51, 517-542.

Gough, H. G., & Heilbrun, A. B. (1983). The Adjective Check List man- ual. Palo Alto, CA: Consulting Psychologists Press.

Hull, J. G., & Levy, A. S. (1979). The organizational functions of the self: An alternative to the Duval and Wicldund model of self-aware- ness/Journal of Personality and Social Psychology,, 37, 756-768.

Jackson, D. N. (1984). Personality Research Form manual. Port Huron, MI: Research Psychologists Press.

Kihlstrom, J. E, & Cantor, N. (I 983). Mental representations of the self. In L. Berkowitz (Ed.), Advances in experimental social psychology (Vol. 17, pp. 1-47). New York: Academic Press.

Kihlstrom, J. E, & Nasby, W. 0981). Cognitive tasks in clinical assess- ment: An exercise in applied psychology. In P. C. Kendall & S. D. Hollon (Eds.), Assessment strategies for cognitive-behavioral inter- ventions (pp. 287-317). New York: Academic Press.

Lord, E M. (1964). Nominally and rigorously parallel test forms. Psy- ' chometrika, 29, 335-345.

Markus, H. (1977). Self-schemata and processing information about the self. Journal of Personality and Social Psychology,, 35. 63-78.

Markus, H. (1980). The self in thought and memory. In D. M. Wegner & R. R. Vallacher (Eds.), The self in social psychology (pp. 102-130). New York: Oxford University Press.

McFarland, S. G., & Sparks, C. M. (1985). Age, education, and the in- ternal consistency of personality scales. Journal of Personality and Social Psychology, 49, 1692-1702.

Mueller, J. H. (1982). Self-awareness and access to material rated as self- descriptive or nondescriptive. Bulletin of the Psychonomic Society, 19, 323-326.

Murray, H. A. (1938). Explorations in personality. Cambridge, MA: Harvard University Press.

Nasby, W. (1985). Private self-consciousness, articulation of the self- schema, and recognition memory of trait adjectives. Journal of Per- sonality and Social Psychology, 49, 704-709.

Nasby, W. (1989). Private and public self-consciousness and articulation of the self-scbema. Journal of Personality and Social Psychology, 56, 117-123.

Nasby, W., & Kihlstrom, J. E (1986). Cognitive assessment of personal- ity and psychopathology. In R. E. Ingrain (Ed.), Information process- ing approaches to psychopathology and clinical psychology (pp. 217- 239). New York: Academic Press.

Osgood, C. E., & Suci, G. J. (1952). A measure of relation determined by both mean differences and profile information. Psychological Bul- letin, 49, 251-262.

Pryor, J. B. (1980). Self-reports and behavior. In D. M. Wegner & R. R. Vallacher (Eds.), The self in social psychology (pp. 206-228). New York: Oxford University Press.

Pryor, J. B., Gibbons, E X., Wicklund, R. A., Fazio, R. A., & Hood, R. (1977). Self-focused attention and self-report validity. Journal of Personality, 45, 513-527.

Rogers, T. B. (1981). A model of the self as an aspect oftbe human information-processing system. In N. Cantor & J. E Kihlstrom (Eds. ), Personality, cognition, and social interaction (pp. 193-214). HiUsdale, N J: Eflbaum.

Seheie~; M. E, Buss, A. H., & Buss, D. M. (1978). Self-consciousness, self-report of aggressiveness, and aggression. Journal of Research in Personality, 12, 133-140.

Seheier, M. E, & Carver, C. S. (1980). Individual differences in self- concept and self-process. In D. M. Wegner & R. R. Vallacher (Eds.), The self in socialpsychology (pp. 229-251). New York: Oxford Uni- versity Press.

Sebeier, M. E, & Carver, C. S. (1983). Private and public aspects of the self. In L. Wheeler (Ed.), Review of personality and social psychology,, (Vol. 2, pp. 189-216). Beverly Hills, CA: Sage.

Turner, R. G. (1978a). Consistency, self-consciousness, and the predic- tive validity of typical and maximal personality measures. Journal of Research in Personality, 12, 117-132.

Turner, R. G. (1978b). Effects of differential request procedures and self-consciousness on trait attributions. Journal of Research in Per- sonality, 12, 431--438.

Turner, R. G. (1978c). Self-consciousness and speed of processing self- relevant information. Personality and Social Psychology Bulletin, 4, 456--460.

Turner, R. G. (1980). Self-consciousness and memory of trait terms. Personality and Social Psychology Bulletin, 6, 273-277.

Turner, R. G., Seheier, M. E, Carver, C. A., & Ickes, W. (1978). Corre- lates of self-consciousness. Journal of Personality Assessment, 42, 285-289.

Wieklund, R. A. (1982). Self-focused attention and the validity of self- reports. In M. P. Zanna, E. T. Higgins, & C. P. Herman (Eds.), Consis- tency in social behavior: The Ontario Symposium (pp. 149-172). Hillsdal¢, N J: Erlbaum.

Received August 27, 1988 Revision received November 23, 1988

Accepted November 28, 1988 •