The Correlation Between Physical Appearance and Shyness - Research Paper Example

Running Head: physical appearance and shyness The correlation between physical appearance and shyness? of the of the The correlation between physical appearance and shyness? Researchers have hypothesized that behavioral problems are accentuated when a shy person wants to be with other people. This need to be with others--sociability--is a related but separate construct from shyness. The purpose of the present study was to examine the correlation between physical appearance and shyness. Shyness is often a considerable handicap in social and occupational situations, and it can be inwardly distressing. Alfano, Joiner, and Perry (1994) pointed out that shyness is often associated with both depression and dysfunctional cognitions. How these dimensions relate to one another, however, is unclear. Does shyness have the properties of a personality trait and lead to depression, or does depression lead to shyness? Are both manifestations of some deeper personality trait? Where does attributional style, in the sense of dysfunctional cognitions, fit in? In their 1994 study, Alfano et al. hypothesized that attributional style is a mediating variable--"the differences in depression between shy and non-shy subjects would be due to differences in their attributional style" (p. 290). They subsequently measured shyness, attributional style (two measures: negative cognitive achievement and negative interpersonal achievement), and depression (two measures) with standardized instruments in a sample of 251 college students. They then applied analyses of variance and covariance to the data to test their hypothesis that attributional style mediates between shyness and depression (as opposed to depression mediating between shyness and attributional style). Having confirmed that their shy participants were more depressed and more negative than their non-shy counterparts, Alfano et al. (1994) tested their hypothesis in two stages, first by eliminating the effect of attributional style on depression and second by eliminating the effect of depression on attributional style. With the effect of attributional style statistically controlled, the differences between the shy and non-shy groups in depression became nonsignificant. However, with the effect of depression statistically controlled, the differences in attributional style between the shy and the non-shy groups still remained. On the basis of this evidence, Alfano et al. concluded that their hypothesis that attributional style mediates the effects of shyness on depression was supported, albeit weakly. In discussing their results, Alfano et al. (1994) drew attention to some of the limitations of their study. In particular, they reminded us that the temporality of relationships among variables cannot be determined with confidence from cross-sectional data; longitudinal data are needed. Accordingly, inferences about the appropriate ordering of their variables remained speculative. For instance, they pointed out that although their "data are consistent with the possibility that shyness is a vulnerability factor for the development of a negative attributional style which may then lead to depression . . . the converse, namely, that a negative attributional style leads to the behaviors and affects that constitute shyness is also possible" (p. 295), We agree with this second possibility--that shyness, rather than negative attributions, may give rise to depression directly. There is some empirical support for such an alternative sequencing (e.g., Anderson & Arnoult, 1985; Bruch & Pearl, 1995; Teglasi & Hoffman, 1982). Bruch and Pearl (1995), for example, found that the main attributional factor associated with shyness was their respondents perceived lack of controllability of events when interacting socially. Moreover, in considering the relationship between dysfunctional cognitions and maladaptive behavior, Leary (1990) argued that "people weigh the potential rewards and costs involved with certain behavioral options involved and estimate the likelihood that these options will occur. The potential costs of interaction include not only social failure and rejection, but their subjective consequences such as anxiety" (p. 41). Adding the word depression to anxiety, provides an explanation of what can happen to people who believe that others have a low opinion of them and are likely to ignore or reject them. These individuals can experience both social anxiety, which manifests itself as shyness, and depression. A more radical approach to the problem is to dispense with causal conceptions of this kind altogether, in favor of assessing the dimensionality of the Alfano et al. (1994) data by means of factor analysis. It is quite possible that all the variables measured are manifestations, in varying degrees, of a single general factor identified with negative affect and closely tied to self-esteem (Eysenck, 1970; Hallworth, 1965; Rosenberg, 1965). Another, more conservative, alternative to the Alfano et al. (1994) conception, which nonetheless stays true to their preferred sequencing (in a "path modeling" sense; Duncan, 1975), is to view shyness (the exogenous variable) as a personality trait (Plomin & Daniels, 1986) that exerts an influence simultaneously on both attributional style and depression (the endogenous variables). Moreover, because the Alfano et al. sample was composed of college students, it seems plausible to assume that the students would be concerned about their academic achievement and that disappointing grades could precipitate negative attributions about their performance. Once they had doubts about their own competence, the students may have started to believe that they were being negatively evaluated by their peers in social situations. As a consequence, negative cognitive achievement (attributional style) can be postulated to have fostered negative interpersonal achievement (attributional style), which itself fostered depression. A more parsimonious version of such a theoretical structure is one that combines the measures of attributional style and the measures of depression, respectively, into single composite variables or factors. This latter model also has the merit of using all of the Alfano et al. (1994) data. In the present study, we used the LISREL program Joreskog & Sorbom, 1989) for structural equation modeling to model the relationships among the measures of attributional style, shyness, and depression used by Alfano et al. (1994). Only one of the measures of depression--the Beck Depression Inventory (BDI; Beck, Rush, Shaw, & Emery, 1979)--was included in all but one of the analyses. Our goal in this study was to test the goodness of fit of the original Alfano et al. model and four alternative models to their data: Model 1: The preferred model of Alfano et al., in which both measures of attributional style mediate the effects of shyness on depression. Model 2: The suggested alternative model of Alfano et al., in which shyness mediates the effects of both measures of attributional style on depression. Model 3: A single-factor model accounting for the covariation among the measures of Alfano et al. Model 4: A modified version of the preferred sequencing of Alfano et al., in which shyness affects, in varying degrees, both attributional style and depression. Model 5: A variant of Model 4, in which a single latent variable (factor) replaces both attributional style variables and a single latent variable replaces both depression variables. Method We have noted that Alfano et al. (1994) acknowledged that, in trying to infer temporal sequencing from cross-sectional data, they had weakened their study. In our approach, we did not try to circumvent such difficulties by statistical techniques but, instead, used modeling methods to evaluate competing alternative structures to account for the pattern of relationships in the observed data. The basic requirement is that such structures have strong theoretical foundations. That requirement is in line with Hoyle and Smiths (1994) argument that "directionality can be inferred from data generated by cross-sectional studies in which a clear, logical, theoretical or empirically based cause-effect sequence can be proposed" (p. 329). Factor analysis, traditionally the most common approach to modeling the correlations among a set of variables of the kind under consideration here, is such a structure. To evaluate the theoretical interpretations of the results of the analysis in Alfano et al., we used the alternative path modeling approach, in which some variables are postulated to precede others in functional sequences. The structural equation modeling program used here (LISREL) has the advantage of enabling us to evaluate the fit of both kinds of models (factors and paths), separately and in combination, to the observed data. The program provides both parameter estimates (factor loadings, path coefficients, variances, and covariances) for each model and a number of indices to assess goodness of fit. The indices of fit include (a) chi-square, derived from a likelihood ratio test of the difference between the correlations (or covariances) implied by the model and the observed correlations; (b) a goodness-of-fit index (AGFI), ranging between 0 and 1, expressing the proportion of the observed variance and covariance in the data that can be accounted for by the model and adjusted for degrees of freedom; and (c) the root mean square residual (RMSR), which is the square root of the mean squared difference between the correlation coefficients in the observed and the implied matrices, with 0 indicating a perfect fit. Finally, we can evaluate the models in terms of their explanatory value in accounting for the depression scores--the main outcome variable in each of them; this value is given by the percentage of variance in the depression scores explained by the variables preceding depression in the model. Evidence of a model with a good fit is a necessary, if not sufficient, condition for establishing the validity of a given theoretical structure. A poorly fitting model, however, immediately rules out the structure to which it refers. Thus, by evaluating the fit of the different models, we limit theoretical possibilities and gain a purchase on the ones that are appropriate for the data. (For a fuller exposition of this methodology see Romney, 1994; Romney & Bynner, 1992.) Data In the Alfano et al. (1994) study, shyness was measured by a single scale (Cheek & Buss, 1981); negative attributional style concerning achievement events and interpersonal events was scored separately, with the Extended Attributional Style Questionnaire (EASQ; Metalsky, Halberstadt, & Abramson, 1987); and depression was assessed by two scales--the Beck Depression Inventory (BDI; Beck et al., 1979) and the Depressive Symptom Index (DSI; Metalsky, Joiner, Klocek, & Potthoff, 1992). The correlation table in the original study provided the data entered as input for the LISREL 7 program and is reproduced here as Table 1. However, because the DSI was developed merely as a "supplement to the BDI" and is very highly correlated with it (r = .84), we decided to exclude the DSI as a variable from all but one of the subsequent analyses, especially because similar findings were reported for both inventories, making one of them redundant. Results The first model we tested (Model 1) was the model initially proposed by Alfano et al. (1994), in which shyness is the independent variable, negative attributional style is the mediating variable (with pathways running from shyness simultaneously to negative attributional style for cognitive achievement events and negative attributional style for interpersonal events), and depression is the dependent variable (with pathways running to it simultaneously from achievement events and interpersonal events). Because the two measures of attributional style intercorrelated quite highly (r = .72), their residuals were allowed to intercorrelate. However, despite our relaxation of this constraint, the model did not fit the data very well, and only 13% of the variance in depression was explained by the model. (See Table 2 for the goodness-of-fit indices and the percentages of variance in depression explained for the five models that were tested.) In Model 1, the path coefficient from shyness to attributional style for cognitive achievement (.35) hardly differed from the path coefficient from shyness to attributional style for interpersonal events (.40). This finding, incidentally, is consistent with that of Alfano et al. (1994), that "shy subjects did not differ in their attributional style for negative interpersonal (ASQ-IP) versus achievement (ASQ-ACH) events" (p. 293). However, the paths from the two attributional style measures to depression were quite different,-.33 for interpersonal events and .04 for cognitive achievement events. These scores point to the need to place the two measures of attributional style in sequence in the model (i.e., cognitive achievement events should precede interpersonal events). In Model 2, attributional style precedes shyness and depression. As in Model 1, ASQ-IP and ASQ-ACH were set at the same point in time. In this case, the model fit the data much more poorly than did Model 1 (see Table 2), accounting for a mere 7% of the variance in depression. The single general factor model constitutes Model 3. This model fit the data better than Model 2 (see Table 2) but showed no improvement over Model 1. The highest factor loadings were on the attributional style variables, especially negative attributional style for interpersonal events. Model 4 resembles Model 1; in both models, attributional style mediates between shyness and depression, although in Model 4, in light of the consideration of the path coefficients estimated for Model 1, ASQ-ACH was located before ASQ-IP. This model fit the data almost perfectly: chi-square and RMSR both approached 0, and AGFI approached unity. The percentage of variance in depression explained by the model also increased over that in Model 1 to 15% (see Figure 1). The final model, Model 5, is in analogue of the previous model but uses latent variables (factors) in place of the attributional style measures and both of the depression measures (BDI and DSI) used by Alfano et al. (1994); thus, measurement error in these variables is taken into account. This model did not improve on the fit of Model 4 (see Table 2). However, the percentage of variance in depression accounted for increased to 18% (see Figure 2). Discussion The results of fitting the different models to the Alfano et al. (1994) data rule out certain theoretical accounts of the relationships and support others. For example, the alternative model, which postulates an intermediate position for shyness between attributional style and depression--a possibility suggested by Alfano et al.--is excluded. A single-factor model of the kind favored by personality factor theorists such as Hallworth (1965) similarly gains relatively little support from the model fitting. This finding leaves us with a choice between Model 1 (Alfano et al., 1994) and Models 4 and 5 (our variants). On grounds of fit, Models 4 and 5 are superior to Model 1, with Model 4 marginally the best, although when the dangers of capitalizing on chance with LISREL models are taken into account, all three may be considered to meet the criterion of a "good" fit. With respect to explanatory value, as assessed through the percentage of variance in depression explained, the choice between the models is easier to make. Model 4 is clearly superior to Model 1, and Model 5 is clearly superior to Model 4. The improved explanatory value of Model 5 resides in its correction for attenuation, which is achieved by taking measurement error in the attributional style and the depression variables into account. Conceptually, the two models are very similar, although pooling the measures of attributional style into a single latent variable throws away the undoubted explanatory interest in one attributional style variable operating as a function of the other. Having shyness as the independent or exogenous variable, which precedes negative attributional style and depression temporally, makes good theoretical (cf. Plomin & Daniels, 1986) and empirical sense. There is evidence of a "genetic component for shyness in young children" as well as "mothers self-reports of shyness, low sociability, and introversion, thereby suggesting the influence of a shared home environment" (Rothbart & Mauro, 1990, p. 152). Shyness, therefore, is evident at a very early age, before the cognitive capacity for making attributions and acquiring an attributional style has had time to develop. On another front, Eysencks (1970) two factors, introversion and neuroticism, which, in combination, characterize the individual who is socially anxious and withdrawn, also appear to be partially inherited. Moreover, the shy individual is designated by Eysenck as being "dysthymic" and, consequently, predisposed toward depression. Hence, there is strong justification for putting shyness first in the developmental sequence. In the model we selected (Model 4), negative attributional style for interpersonal events is preceded by negative attributional style for achievement events. It is not difficult to see that negative attributions about academic events can lead to negative attributions about evaluations from others. The link between attributional style for these two types of events was illuminated by Leary (1990,p. 41), who discussed a parallel topic, the relationship between test anxiety and social anxiety: "People worry about their test performance because they are concerned about how others will evaluate them (My parents will be mad and my friends will think Im stupid if I [ail this test)." The crucial question, however, is how these negative attributions about academic events occur in the first place. The answer may lie in low academic achievement: Students who perform poorly in college develop negative but realistic attributions about achievement and consequently start thinking that other people have a low opinion of them in general. They may then withdraw from the company of others and become depressed. Another explanation is not that academic achievement is too low but rather that expectations are too high. In other words, there may be individuals with high aspirations who are disappointed when they fail to live up to their own academic standards. These individuals may then develop (unrealistic) negative attributions about their achievement, which may, as before, develop into negative attributions about interpersonal events, leading again to depression. Before either of these two explanations can be given credence, they need to be verified empirically. To test the latter hypothesis, one has to compare the grades these individuals expect to obtain with their actual grades. Those students whose expected grades are much higher than their actual grades should prove to be the most vulnerable individuals. Finally, we need to return to the cautions we expressed earlier, about the interpretation of cross-sectional data. The patterns in correlations reflected in our structural model go only a short way toward validating the theory implied by any one of them. Applying structural equation modeling to cross-sectional data can rule out some of the models, but to determine which is the most appropriate with respect to temporal order, researchers need to have access to longitudinal data in which there are repeated measures of each of the critical variables. In sum, more research needs to be done to cross-validate the model, not only on another sample but also through a repeated measures longitudinal design. Also, much more needs to be done to substantiate the models theoretical basis. Meanwhile, a good working model exists that has obvious practical and heuristic implications. Student counseling services should be made aware of the risks, either real or imagined, for depression among poor achievers and be prepared to take on these cases for early intervention. Cognitive therapy designed to alter attributional style and enhance self-esteem seems to be the treatment of choice. Tables TABLE 1 Correlations Among Measures of Shyness, Attributional Style, and Depression Measure 1 2 3 4 5 1. Shyness -- 2. ASQ-IP .40 -- 3. ASQ-ACH .35 .72 -- 4. BDI .27 .36 .28 -- 5. DSI .26 .34 .29 .84 -- Note, ASQ-IP = Attributional Style--Generally for Interpersonal Events; ASQ-ACH = Attributional Style--Generally for Achievement Events; BDI = Beck Depression Inventory; DSI = Depressive Symptom Inventory. From "Attributional Style: A Mediator of the Shyness-Depression Relationship" by M.S. Alfano, T.E. Joiner, Jr., and M. Perry, 1994, Journal of Research in Personality, 28, p. 292. Copyright 1994 by Academic Press. Reprinted with permission. TABLE 2 Goodness-of-Fit Indices for LISREL Models (N = 251) Model Chi2 df p AGFI RMSR 1 5.31 1 .02 .896 .039 2 21.35 2 .00 .803 .099 3 4.11 2 .13 .960 .031 4 0.08 1 .78 .998 .004 5 1.34 4 .85 .992 .006 Note. AGFI = adjusted goodness-of-fit index; RMSR = root mean square residual. FIGURE 1. Structural equation model showing the relationship between shyness, negative attributional style, and depression (Model 4). FIGURE 2. Structural equation model (with the endogenous variablescorrected for attenuation) showing the relationship between shyness, negative attributional style, and depression (Model 5). References Alfano, M. S., Joiner, T. E., Jr., & Perry, M. (1994). Attributional style: A mediator of the shyness-depression relationship. Journal of Research in Personality, 28, 287-300. Anderson, C. A., & Arnoult, L. H. (1985). Attributional style and everyday problems in living: Depression, loneliness, and shyness. Social Cognition, 3(10), 16-35. Beck, A. T., Rush, A. J., Shaw, B. E, & Emery, G. (1979). Cognitive therapy of depression. New York: Harper and Row. Bruch, M. A., & Pearl, L. (1995). Attributional style and symptoms of shyness in a heterosexual interaction. Cognitive Therapy and Research, 19, 91-107. Cheek, J. M., & Buss, A. H. (1981). Shyness and sociability. Journal of Personality, and Social Psychology, 41, 330-339. Duncan, O. D. (1975). Introduction to structural equation models. San Diego: Academic Press. Eysenck, H. J. (1970). The structure of human personality (3rd ed.). London: Methuen. Hallworth, H. J. (1965). Dimensions of personality and meaning. British Journal of Social and Clinical Psychology, 4, 161-168. Hoyle, R. H., & Smith, G. T. (1994). Formulating clinical research hypotheses as structural equation models: A conceptual overview. Journal of Clinical and Consulting Psychology, 62, 429-440. Joreskog, K. G., & Sorbom, D. (1989). LISREL 7: A guide to the program and applications (2nd ed.). Chicago: SPSS Inc. Leary, M. A. (1990). Anxiety, cognition, and behavior: In search of a broader perspective. Journal of Social Behavior and Personality, 5, 39-44. Metalsky, G. I., Halberstadt, L. J., & Abramson, L. Y. (1987). Vulnerability to depressive mood reactions: Toward a more powerful test of the diathesis-stress and causal mediation components of the reformulated learned helplessness theory. Journal of Personality and Social Psychology, 52, 386-393. Metalsky, GM., Joiner, T. E., Jr., Klocek, J. W., & Potthoff, J. G. (1992). Vulnerability to depressive symptomatology: A prospective test of the diathesis-stress and causal mediation components of the hopelessness theory of depression. Journal of Personality and Social Psychology, 63, 667-675. Plomin, R., & Daniels, D. (1986). Genetics and shyness. In W. H. Jones, J. M. Cheek, & S. R. Briggs (Eds.), Shyness (pp. 63-80). New York: Plenum. Romney, D. M. (1994). Cross-validating a causal model relating attributional style, self-esteem, and depression: An heuristic study. Psychological Reports, 74, 203-207. Romney, D. M., & Bynner, J. M. (1992). The structure of personal characteristics. Westport, CT: Praeger. Rosenberg, M. (1965). Society and the adolescent self-image. Princeton, NJ: Princeton University Press. Rothbart, M. K., & Mauro, J. A. (1990). Temperament, behavioral inhibition, and shyness in childhood. In H. Leitneberg (Ed.), Handbook of social and evaluation anxiety (pp. 139-160). New York: Plenum. Teglasi, H., & Hoffman, M. A. (1982). Causal attributions of shy subjects. Journal of Research in Personality, 16, 376-385. Read More