Conflict Over Emotional Expression: Psychological

Copyright 1990 by the American Psychological Association, Inc. 0022-3514/90/S00.75

Journal of Personality and Social Psychology 1990, Vol. 58, No. 5, 864-877

Conflict Over Emotional Expression: Psychological and Physical Correlates Laura A. King

Robert A. Emmons

Michigan State University

University of California, Davis

This study addresses the construct of conflict or ambivalence over emotional expression. Ambivalence is seen as an important mediator in the link between emotional styles and psychological and physical well-being. Using the "personal striving" framework, a questionnaire measure of ambivalent emotional strivings (AEQ) was designed. In Study 1,292 Ss completed this measure along with questionnaire measures of expressiveness, social desirability, and intense ambivalence. Women scored significantly higher than men on both the AEQ and expressiveness. In Study 2, scores on the AEQ were found to be negatively correlated with self-reported and peer-rated expressiveness. In Study 3, 48 Ss participated in a 21-day study of mood and health. Expressiveness was positively correlated with some measures of well-being and with daily negative affect. Ambivalence was positively correlated with several indices of psychological distress. Although the AEQ correlated with questionnaire measures of physical symptomatology, neither the AEQ nor the expressiveness measures correlated with daily symptom reports. Results support the contention that conflict over emotional expressiveness is a variable worthy of study in its own right, having implications for research on personality and health.

Shields has commented that "as a culture we are ambivalent about emotion" (1987, p. 231). This cultural ambivalence is reflected in many common expressions that subtly denigrate emotion and emotional display. Expression of emotion is commonly termed a "giving in" to passion. It is said that disaster is avoided when "cooler heads prevail." Alongside this disdain for emotion is the assumption that emotions must be "let out"— that the healthy end to an emotional response is emotional expression. Indeed, Baumeister and Tice (1987) have stated that a primary component of emotional response is a motivation to express. From Freud (e.g, 1917/197 7) to the present (e.g., Fridlund, Newman, & Gibson, 1984; Pelletier, 1985), "bottled up emotions" have been blamed for symptomatology, both psychological and physical. Thus, expression is viewed as uncivilized and yet beneficial. Individuals' ambivalence over emotional expression is suggested here as a factor in the relation between emotional expression and well-being. Before describing the development of a measure to tap individual differences in ambivalence over expressing emotion, it may be helpful to consider some of the literature concerning this relation. Styles of emotional expression have been brought up in relation to several physical illnesses, including cancer (Cox & McCay, 1982; Derogatis, Abeloff, & Melisaratos, 1979; Greer & Morris, 1975; Jensen, 1987), coronary heart disease (CHD; Friedman & Booth-Kewley, 1987; Friedman, Hall, & Harris, 1985; Gentry, 1985; Goldstein, Edelberg, Meier, & Davis, 1988), and other diseases (Beutler, Engle, Oro-Beutler, Daldrup, &

This article is based on portions of a thesis submitted in partial fulfillment of the requirements for the degree of master of arts by Laura A. King under the supervision of Robert A. Emmons at Michigan State University. Correspondence concerning this article should be addressed to Laura A. King, who is now at Department of Psychology, University of California, Davis, California 95616. 864

Meredith, 1986; Pelletier, 1985; Udelman & Udelman, 1981). Cox and McCay (1982) concluded that the strongest psychosocial predictor of cancer is an antiemotional attitude, particularly an inability to express negative emotion. In the CHD literature, evidence has been equivocal as to the beneficial effects of expressiveness. Whereas nonverbal expressions of anger are a criterion for the Type A behavior pattern (Friedman & Booth-Kewley, 1987), studies have also supported the idea that unexpressed hostility is associated with CHD (Gildea, 1949, and Cady et al, 1961, cited in Friedman, Harris, & Hall, 1984). Friedman et al. (1985) acknowledged that there are individuals who are expressive of their emotions (including anger) who may be diagnosed as Type A but are not at risk for CHD ("natural charismatics"). Friedman and Booth-Kewley (1987) also argued that a subset of Type B persons may be in conflict over expressing emotion and are at greater risk for CHD than relaxed, "true" Type Bs. The fact that individuals who express hostility as well as those who "bottle it up" may be at risk for CHD demonstrates the importance of the processes underlying expressive behaviors in the development of pathology. Expressiveness itself cannot discriminate between groups whose emotional styles place them at risk and those whose styles do not (cf. Hall, Friedman, & Harris, 1986). One purpose of the ambivalence construct is to distinguish between persons whose expressive styles are similar but whose underlying ambivalence differs (e.g, relaxed, quiet people vs. repressed, tense inhibitors). The lack of utility of expressiveness alone as a means of distinguishing healthy and unhealthy styles is also evident in the processes thought to underlie the connection between emotional expression and ill-being. Many of the assumptions about the advantages of emotional expression are based on the well-documented inverse relation between emotional expression and autonomic reactivity (Buck, 1984; Hokanson & Burgess, 1962; Hokanson & Shetler, 1961; Jones, 1935). Typically, the inhibition of emotional expression has been associated with an increase in autonomic activity

CONFLICT OVER EXPRESSION (Pennebaker, 1985). Chronic autonomic arousal, in turn, is associated with development of psychosomatic disease. The inverse relation between autonomic reactivity and expression would seem to indicate that expression of emotion is healthy and lack of expression unhealthy. However, this uncomplicated view has proven inadequate as understanding of emotional expression and inhibition has grown. For instance, although there is evidence that facial display attenuates arousal (Buck, 1984), other evidence (e.g, Lanzetta, Cartwright-Smith, & Kleck, 1976) suggests that it augments arousal. Research has demonstrated that autonomic arousal can be related variously to nonexpression and expression as a function of situational factors, individual differences in characteristic styles of expression (Roth & Cohen, 1986; Sackheim, 1983), and the quality of the emotion itself (Murray, 1985; Tavris, 1984). For example, researchers (Murray, 1985; Tavris, 1984) have pointed out that negative emotions such as aggression, anger, and hostility present special problems for individuals because the expression of these emotions may hold important ramifications for the social group. Furthermore, inexpressive repressors have been found to report fewer health problems than expressive sensitizers (Bell & Byrne, 1978; Weinberger, in press). Perhaps the keyflawin the inverse conceptualization of the relation between expression and arousal is that it fails to make the previously mentioned distinction between comfortable inexpressiveness and inhibition. The assumption that emotion must or should be expressed carries with it the corollary that those who do not express emotion are inhibiting expression. Such an assumption may be erroneous. An appropriate point of transition from the simplistic inverse relation between expression and well-being to a more sophisticated, goal-based model of this relation is provided by Pennebaker (1985) in his suggestion that lack of expression per se may not be pathogenic. Rather, according to Pennebaker, lack of emotional expression coupled with the desire to express emotion is the (literally) fatal combination. Pennebaker and his colleagues (Pennebaker & Hoover, 1986; Pennebaker, Hughes, & O'Heeron, 1987) have found that individuals who inhibit their desire to confide in others about traumatic life events are at an increased risk for the development of later health problems. Pennebaker has coined the phrase "active inhibition" to refer to the process of willfully preventing oneself from a desired action. Active inhibition is thought to be related to obsessive thoughts about the inhibited action. Along similar lines, Emmons and King (1988) found that those goals that were in conflict with other goals or about which individuals felt ambivalent were the same goals about which individuals spent time thinking. According to Pennebaker, active inhibition leads to chronic autonomic arousal and eventual physical breakdown. Pennebaker's focus on the intent or desire to express emotion indicates that it is appropriate to look at individuals' goals with regard to emotional expressiveness as important factors in determining the potentially pathogenic quality of certain characteristic styles of emotion regulation. On the basis of the research of Pennebaker and Emmons and King (1988), we may conclude that when these goals are in conflict with social norms or with other goals, an individual may experience the detrimental effects of psychosomatic reaction, whether they are expressive or inexpressive.

865

The present investigations addressed individuals' goals about expression and the conflict that may exist between and within these goals. In particular, individuals' "personal strivings" about expression were used to construct a questionnaire addressing the ambivalence issue. Emmons (1986) has described the role of personal strivings in personality and subjective wellbeing. A personal striving refers to "what a person is characteristically trying to do" (Emmons, 1986, p. 1059), or the characteristic types of goals the person strives for. A formal elaboration of the personal striving construct is given in Emmons (1989). Subjects have generated strivings in a number of studies and, invariably, they spontaneously enumerate strivings dealing with expression, management, and regulation of emotion (Emmons, 1989). Personal strivings that pertain to emotion illustrate the dilemma presented by emotional expression on a microcosmic level. Examples of emotion-related personal strivings collected in a study by Emmons and King (1988) include "control my temper," "always appear cool," "not let my emotions take over," "always wear a smile on my face," "be honest and open about my feelings," and "let my anger out before it all builds up inside me." These strivings provide a pointed demonstration of the translation of cultural ambivalence into individual lives, revealing the belief that emotion should be honestly expressed but also that expression implies vulnerability. An emphasis on goals in a study of emotional expression seems justified, given the assertion by Baumeister and Tice (1987) that every emotion carries with it a motivation to express. Emmons (1986) found that ambivalence about one's strivings is associated with low subjective well-being, especially high negative affect. In addition, Emmons and King (1988) have found that conflict between personal strivings is related to physical symptomatology, both concurrently and prospectivery. In that study, it appeared that many of the subjects' ambivalent personal strivings (i.e, those strivings whose achievement would cause unhappiness as well as happiness) concerned emotional expression. Personal strivings are particularly suitable for a study of ambivalence over expression because what a person "tries to do" may be different from what a person actually does (Emmons, 1986). Within this framework, we can examine the importance of ambivalence about expression to well-being apart from the basic issue of expression itself. One premise of the current studies is that individuals may be either expressive or inexpressive and still be conflicted over their style. However, a more specific association between ambivalence and inhibition has recently been suggested by Emmons and King (1988). They found that individuals are unlikely to act on strivings about which they are ambivalent. Thus, ambivalent individuals may be likely to be inexpressive. The relation between ambivalence over expression and inhibition that is implied here is important because it indicates that ambivalent and inexpressive individuals are likely to be confused by a model that stresses the importance of expression per se. Inhibition here refers to both voluntary (as per Pennebaker's "active inhibition") and involuntary "attenuation of affective expression" (cf. Buck, 1984). In the context of the present investigations, inexpressiveness is not predicted to be pathogenic. Rather, goal conflict that may underlie inexpressiveness (i.e, inhibition) is predicted to be associated with ill-being. In order to examine the importance of ambivalence over ex-

866

LAURA A. KING AND ROBERT A. EMMONS

pression and expressiveness to well-being, two questionnaire measures were constructed—one to measure each of these constructs. The development and application of these questionnaires are described in the three studies reported here. In the first study, the scales were created to measure ambivalence over emotional expression and emotional expressiveness and were administered to a large group of students at Michigan State University. These data were used in factor analyses of the scales and a comparison of sex differences in self-reported expressiveness and ambivalence over expression. In a second study, the questionnaire measure of ambivalence over expression was compared with self- and peer-reported measures of expressiveness to determine whether, in fact, ambivalence is associated with lack of expression. In a third study, expressiveness and ambivalence over expression were used to predict the psychological and physical well-being of subjects participating in a daily mood study. Thus, the studies we report here culminate in an examination of the contributions of ambivalence over expression and expressiveness to psychological and physical wellbeing. The central thesis of this research is that expression is, in fact, of little importance to well-being. One can be expressive or inexpressive of emotion and be healthy. Expressive behaviors themselves do not predict well-being. Rather, from the current perspective, it is ambivalence about one's expressiveness style that fosters ill-being. Study 1 Development of the Questionnaire A list of 600 personal strivings that had been collected from subjects in an earlier study (Emmons, 1986) were examined, and those strivings that dealt with emotion were compiled into a list. This list was used as a basis for the generation of items in the Ambivalence Over Emotional Expressiveness Questionnaire (AEQ). The strivings were reworded in order to address their potentially ambivalent character. For instance, the striving "I typically strive to express my emotions honestly" became "I want to express my emotions honestly but I am afraid that it may cause me embarrassment or hurt." Ambivalence over expressing emotion can take on different forms, from wanting to express but not being able to, to expressing but not necessarily wanting to, to expressing and later regretting it. Ambivalence, then, encompasses both inhibition and rumination. Items were written so as to represent all of these possibilities. Instructions for the AEQ emphasized that the subject should answer each item with a view to its overall meaning. Thus, if a statement consisted of two thoughts, subjects were encouraged to give the item a high rating only if both thoughts applied to them. The rating scale for the questionnaire ranged from 1 to 5, with 1 indicating that the respondent never feels what the statement suggests and 5 indicating that the respondent frequently feels that way. It may be noted that no items on the AEQ are negatively worded. Response bias has been an enduring concern in psychological research. However, because of the complex nature of the items, reverse wording was deemed infeasible. The potential contamination of scores by acquiescence is not likely to be great in this case, as such bias is most likely to influence dichot-

omous judgments, not rating scales (Schriesheim & Hill, 1981). In addition, Schriesheim and Hill reported that negative items may impair response accuracy and reliability. In the case of a questionnaire with items that are already compound in nature, the disadvantages of negatively worded items were believed to outweigh the advantages. A second scale, the Emotional Expressiveness Questionnaire (EEQ), was also constructed rationally, focusing on actual expressive behaviors. Items were written tapping the expression of both positive and negative emotions. Items are responded to on a 1 to 7 rating scale, with 1 indicating the subject does not agree with the item at all and 7 indicating that the subject strongly agrees with the item. An example of an item from the EEQ is "When I am angry people around me usually know." High scores on the EEQ indicate a tendency to express emotion. A new measure was deemed necessary, as the most widely used measure of expressiveness, the Affective Communication Test (ACT; Friedman, Prince, Riggio, & DiMatteo, 1980), taps nonverbal expressiveness or charisma more than general expressiveness and may overlap with extraversion (Friedman et al, 1984). Predictions Social Desirability and Convergent Validity We predicted that neither the AEQ nor the EEQ would be positively correlated with social desirability. We further predicted that the AEQ would be significantly positively correlated with a trait measure of ambivalence. Such a correlation can be taken as evidence of the validity of the AEQ. In addition, we predicted that the AEQ and the EEQ would be slightly negatively correlated. This correlation would lend support to the contention that conflicted individuals may inhibit expression. Sex Differences We predicted that, overall, women would be significantly more expressive and more ambivalent over expression than men. Although, generally, emotional expressiveness has been associated more with women than with men (Shields, 1987), Hochschild (1983) and Shields (1987) have suggested that women may be less expressive of anger than are men. Thus, the overall difference in expressiveness should be due to differences on positive and intimacy emotions rather than anger. Method Subjects A total of 299 Michigan State University students (117 men and 182 women) completed a packet of questionnaires for extra credit in various undergraduate psychology courses. Male subjects ranged in age from 18 to 26, with a mean age of 19.8 years. Female subjects ranged in age from 18 to 32, with a mean age of 19.1. Subjects were informed in class that their packets would be checked for completeness and carefulness, after which their names would be removed from the packet, ensuring anonymity. Completion of the packet was entirely voluntary.

CONFLICT OVER EXPRESSION

Materials The surveys that were completed by this sample of subjects included the AEQ; the EEQ; the Raulin Intense Ambivalence scale (Raulin, 1984), which measures pathological ambivalence; and the MarloweCrowne Social Desirability Questionnaire {Marlowe & Crowne, 1964).

Results and Discussion Reliability and Factor Analysis of the AEQ Several AEQ items were deleted because they seemed to pertain more to self-assertion than emotional expression. Additional items were deleted on the basis of low hem-total correlations. The deletion of items left a total of 28 items in the revised AEQ, with a mean interitem correlation of .23, a minimum interitem correlation of .02, and a maximum interitem correlation of .64. Item characteristics for the AEQ are shown in Table 1. The alpha reliability coefficient of the AEQ was .89. The mean for scores on the scale was 2.9 on a 5-point scale, and the scale standard deviation was .58. A separate sample of 50 students completed the AEQ on two occasions, with a 6-week interval separating testings. The test-retest correlation was .78. An exploratory factor analysis was conducted on the interitem correlation matrix for the AEQ for men and women separately to determine whether there were important differences between the sexes in the factor structure of the scale. For both male and female subjects, two large factors emerged—one for positive emotion and one for negative emotion. Because there were only slight differences in the factor solutions, data were combined across sexes in further analyses. Principal components extracted two factors with eigenvalues greater than 1. A varimax rotation was used because it was judged most likely to yield an interpretable solution (Hunter, 1984). Sixteen items loaded highest on the first factor (eigenvalue = 6.70), accounting for 17% of the variance. Items and loadings can be seen in Table 2. Items loading on this factor concerned love and affection, as well as a fear of expression of emotions that might lead to vulnerability. Examples of the types of items that loaded on the first factor include "I try to show people I love them although at times I am afraid that it may make me appear weak" and "I want to tell someone when I love them, but it is difficult to find the right words." This factor can be termed ambivalence over the expression of positive emotion. Twelve items loaded on the second factor (eigenvalue = 1,35), which accounted for 12% of the variance. Items loading on this factor pertained to ambivalence over expressing primarily negative emotions (e.g, "After I express anger at someone it bothers me for a long time.") However, one item dealing with pride also loaded on the second factor, as did an item pertaining to jealousy. Because items loading on the second factor pertained not only to anger but to feelings of pride and jealousy, the term entitlement (as per Shields, 1987) was selected to encompass all of these. The second factor was considered to tap ambivalence over expression of emotions of entitlement. A subsequent confirmatory factor analysis was performed on the results of the exploratory factor analysis. The confirmatory factor analysis allows for a check of the internal consistency of the clusters. If the analysis employs communalities in the diago-

867

nal, the items' loadings on the clusters represent an estimate of their correlation with the true score for the trait underlying the cluster. If the items all load highest on the cluster in which they have been placed by the exploratory analysis, the clusters can be concluded to be internally consistent (Hunter, 1984). Items loading on each factor were entered into a cluster analysis using PACKAGE (Hunter, 1988). Both clusters had high reliabilities & = .87 for the ambivalence over positive emotion cluster and a = .77 for the ambivalence over expressions of entitlement cluster). All of the items did load highest on the cluster into which the exploratory analysis placed them. Generally, however, items were highly correlated with both of the clusters. In addition, the two clusters were themselves highly correlated (r = .71). This high correlation, as well as the high correlations of items with both clusters, suggested that the AEQ might be best characterized as tapping a single general construct—ambivalence over emotional expression. A second-order confirmatory factor analysis was performed on the two clusters. A single general factor emerged with an eigenvalue of 1.39, on which both clusters loaded highly. Although these results support the contention that the AEQ is unidimensional, we retained the subclusters that emerged in these analyses in order to investigate the proposed sex differences for the different specific types of emotions, after which the subscales were disregarded.

Reliability and Factor Analysis of the EEQ Initially, 50 items were constructed to assess emotional expressiveness. Items were subsequently deleted from the EEQ because they were judged to tap positive self-assertion rather than emotional expression. Also, items were deleted on the basis of low item-total correlations. The deletion of items left a total of 16 items in the EEQ, with a mean interitem correlation of. 18, a minimum interitem correlation of .06, and a maximum interitem correlation of .59. Item characteristics for the EEQ are shown in Table 3. The mean for scores on the scale was 4.6 (on a 7-point scale), and the scale standard deviation was .76. The alpha reliability coefficient of the EEQ was .78. Initially, factor analyses for the EEQ were conducted for men and women separately; however, because the solutions were virtually identical, the group was used as a whole in all further analyses. The interitem correlations were entered into an exploratory factor analysis, using communalities to correctforattenuation due to error of measurement Principal components extracted three factors with eigenvalues greater than 1, which were rotated using varimax. The eigenvalue for the first factor was 3.41; the second factor, 1.30; and the third factor, 1.10. Items and their loadings appear in Table 4. Seven items loaded on the first factor. All of these items refer to the expression of positive emotions, such as laughter, liking, and affection. Thus, this factor was labeled expression of positive emotion. It accounted for 14% of the variance. Five items loaded highest on the second factor. AH of these items pertain to liking, love, gratitude, and apologizing. Because these emotions center around relational concerns, this factor was retained as different from the first factor and was termed expression ofintimacy. The second factor accounted for 12% of the variance.

868


Table 1 Items in Final Version of the Ambivalence Over Emotional Expressiveness Questionnaire Item

Item-total r

M

SD

24. It is hard tofindtherightwords to indicate to others what I am really feeling 25. I worry that if I express negative emotions such as fear and anger, other people will not approve of me. 1. I want to express my emotions honestly but I am afraid that it may cause me embarrassment or hurt. 27. I often cannot bring myself to express what I am really feeling. 11. I'd like to talk about my problems with others, but at times I just can't. 18. I want to tell someone when I love them, but it is difficult to find the right words. 19. 1 would like to express my disappointment when things don't go as well as planned, but I don't want to appear vulnerable. 8. Often I'd like to show others how I feel, but something seems to be holding me back. 21. I try to hide my negative feelings around others, even though I am not being fair to those close to me. 17. Often Ifindthat I am not able to tell others how much they really mean to me. 10. I try to keep my deepest fears and feelings hidden, but at times I'd like to open up to others. 22. I would like to be more spontaneous in my emotional reactions but I just can't seem to do it. 20. I canrecalla time when I wish that I had told someone how much I really cared about them. 26. I feel guilty after I have expressed anger to someone. 6. I would like to express my affection more physically but I am afraid others will get the wrong impression. 23. I try to suppress my anger, but I would like other people to know how I feel. 15. I try to apologize when I have done something wrong but I worry that I will be perceived as incompetent. 28. After 1 express anger at someone, it bothers me for a long time. 14. I try to show people I love them, although at times I am afraid that it may make me appear weak or too sensitive. 9. I strive to keep a smile on my face in order to convince others I am happier than I really am. 12. When someone bothers me, I try to appear indifferent even though I'd like to tell them how I feel. 4. I try to avoid sulking even when I feel like it. 5. When I am really proud of something I accomplish I want to tell someone, but I fear I will be thought of as conceited. 13. I try to refrain from getting angry at my parents even though I want to at times. 7. I try not to worry others, even though sometimes they should know the truth. 2. I try to control my jealousy concerning my boyfriend/girlfriend even though I want to let them know I'm hurting. 16. I think about acting when I am angry but I try not to. 3. ! make an effort to control my temper at all times even though I'd like to act on these feelings at times.

.58

3.0

1.1

.56

2.3

I.I

.56 .56 .55

3.1 3.0 3.2

1,1 1.2 1.1

.55

2.9

1.3

.55

2.7

1.2

.52

3.2

l.l

,52 .51

2.6 3.1

1.1 1.1

.48

3.2

1.2

.46

2.8

1.1

.45 .44

3.5 2.9

1.2 1.1

.42 .42

2.7 3.2

1.2 1.0

.41 .41

2.4 2.7

1.1 1.2

.40

2.5

1,2

,40

2.6

1.2

.38 .36

3.4 2.9

1.1 1.1

,36 .36 .36

3.1 2.9 3,1

1.2 1.1 1.0

.34 .33

3.1 2.9

1.3 1.1

.25

3.5

1.1

Note, n = 299.

Four items loaded highest on the third factor. All but one of these items concern the expression of negative emotions (e.g., anger and disappointment). Thus, this factor was termed expression of negative emotion. It accounted for 10% of the variance. A subsequent confirmatory factor analysis (using PACKAGE; Hunter, 1988) was conducted using the clusters that emerged in the exploratory analysis. Each cluster had a substantial reliability coefficient considering the small number of items in each; for the expression of positive emotion cluster, a = .74; for the expression of negative emotion cluster, a — 67; and for the expression of intimacy cluster, a = .63. All of the clusters were positively correlated; however, the expression of negative emotion cluster was not correlated as strongly with either the ex-

pression of positive emotion cluster (r= .29) or the expression of intimacy cluster (r = .37) as those two clusters were correlated with each other (r = .58). Social Desirability and Ambivalence Correlations were computed between scores on the AEQ, the EEQ, their subscales, and the Marlowe-Crowne Social Desirability scale (MCSD) to ensure that the new scales were not contaminated by social desirability. The MCSD correlated slightly negatively with both scales (for the AEQ, r = -.20, p < .001; for the EEQ, r = - . 11, p < .03.) As a test of convergent validity, correlations between scores on the AEQ and the Raulin Intense Ambivalence scale were

869


Table 2 Loadings for Two-Factor Solution for Revised Ambivalence Over Emotional Expressiveness Questionnaire Item 10. 6. 8. 5. 11. 3. 4. 1. 15. 7. 9. 22. 14. 9. 20. 7. 28. 26. 23. 25. 16. 9. 3. 4. 13. 15. 2. 5.

Often I find that I am not able to tell others how much they really mean to me, I want to tell someone when I love them, but it is difficult to find the right words. Often I'd like to show others how I feel, but something seems to be holding be back. I'd like to talk about my problems with others, but at times I just can't. I try to keep my deepest fears and feelings hidden, but at times I'd like to open up to others. I want to express my emotions honestly but I am afraid that it may cause me embarrassment or hurt. I often cannot bring myself to express what I am really feeling. It is hard to find the right words to indicate to others what I am really feeling. I would like to express my affection more physically but I am afraid others will get the wrong impression. I would like to express my disappointment when things don't go as well as planned, but I don't want to appear vulnerable. I try to hide my negative feelings around others, even though I am not being fair to those close to me. I would like to be more spontaneous in my emotional reactions but I just can't seem to do it. I try to show people I love them, although at times I am afraid that it may make me appear weak or too sensitive. I strive to keep a smile on my face in order to convince others I am happier than I really am. I can recall a time when I wish that I had told someone how much I really cared about them. I try not to worry others even though sometimes they should know the truth. After I express anger at someone, it bothers me for a long time. I feel guilty after I have expressed anger to someone. I try to suppress my anger, but I would like other people to know how I feel. I worry that if I express negative emotions such as fear and anger, other people will not approve of me. I think about acting when I am angry but I try not to. When someone bothers me, I try to appear indifferent even though I'd like to tell them how I feel. I make an effort to control my temper at all times even though I'd like to act on these feelings at times. I try to avoid sulking even when I feel like it. I try to refrain from getting angry at my parents even though I want to at times. I try to apologize when I have done something wrong but I worry that I will be perceived as incompetent. I try to control my jealousy concerning my boyfriend/girlfriend even though I want to let them know I'm hurting. When I am really proud of something I accomplish I want to tell someone, but I fear I will be thought of as conceited.

computed. As expected, the AEQ was significantly positively correlated with the Raulin scale (r =.35, p < .001). We predicted that the AEQ and EEQ would be slightly negatively correlated, and this prediction was borne out (r = —.24, p < .01).

Sex Differences Because sex differences are likely to be included in cumulative studies and the correlation coefficient lends itself to metaanalysis (Hunter, Schmidt, & Jackson, 1982), Pearson correlation coefficients (rather than t tests) were computed between sex of subject and scores on the EEQ, AEQ, and the respective cluster scores for these two scales. Because female was coded as 1 and male as 0, positive correlations indicate that women scored higher on the scale in question. With regard to the AEQ, predictions were supported. Women scored significantly higher than men on the AEQ (r =. 10, p < ,05). The means on the AEQ were 2.99 for women and 2.88 for men. Women scored higher than men on the ambivalence over expression of positive emotion subscale (r = . 11, p< .03), whereas there was no sex difference on the ambivalence over emotions of entitlement subscale (r = .03, ns). As was predicted, women scored significantly higher on the EEQ than did men (r =. 15, p < .005). The means on the EEQ were 4.72 for women and 4.47 for men. Also as

Factor 1 (Positive)

Factor 2 (Entitlement)

68 66 61 59 57 55 54 54 51

.04 .12 .13 .20 .12 .27 .30 .34 .09

46

.35

45 44 41

.33 .24 .15

39 38 38 10 16 15 36 09 20 03 17 19 29 18

.21 .31 .15 .63 .59 .51 .49 .49 .42 .38 .38 .37 35 .34

.24

.32

expected, women reported themselves as more expressive of positive emotion and intimacy (r =. 14, p < .009, and r =. 17, p < .002, respectively). Contrary to predictions, men were not more expressive of negative emotions than women, although the correlation was in the predicted direction (r = —.06, ns). Several conclusions can be drawn from the results of Study 1. First, whereas the AEQ is best characterized as unidimensional, the EEQ taps three different subsets of expressiveness: expression of positive emotions, expression of negative emotions, and expressions of intimacy. The factor structure of the EEQ is not surprising, given that the experience of positive and negative emotion has been conceptualized as independent (Diener & Emmons, 1984). Whether or not the individuals experienced differing degrees of positive or negative emotions, however, cannot be addressed by Study 1. The AEQ was significantly positively correlated with the Raulin I-A scale, suggesting that it also taps ambivalence. It is important to note, however, that the AEQ is not simply equivalent to the Raulin scale. The two scales were correlated only .35. Whereas the Raulin scale taps intense ambivalence in general, the AEQ specifically taps ambivalence over emotional expression. The AEQ and the EEQ were negatively correlated, which suggests that individuals who are ambivalent about expression

870

LAURA A. KING AND ROBERT A. EMMONS received 3 hr of course credit. Among the women, the age range was 19 to 47, with a mean age of 23.3 years. Among the men, ages ranged from 19 to 27, with a mean age of 21.6 years.

Table 3 Items in the Final Version of the Emotional Expressiveness Questionnaire Item 1. I often tell people that I love them. 16. I show that I like someone by hugging or touching that person. 3. I often touch friends during conversations. 9. Watching television or reading a book can make me laugh out loud. 4. I laugh a lot, 2. When I am angry people around me usually know. 5. People can tell from my facial expressions how I am feeling. 6. Whenever people do nice things for me, I feel "put on the spot" and have trouble expressing my gratitude. (-) 7. When I really like someone they know it. 11. I often laugh so hard that my eyes water or my sides ache. 13. When I am alone, I can make myself laugh by remembering something from the past. 15. My laugh is soft and subdued. (-) 12. If a friend surprised me with a gift, I wouldn't know how to react. (-) 8. I apologize when I have done something wrong. 10. If someone makes me angry in a public place, I wiil "cause a scene." 14. I always express disappointment when things don't go as I'd like them to.

Itemtotal r

M

SD

Procedure

.54

3.7

2.0

.49

3.4

1.9

.49

3.4

1.9

.49 .42

5.5 5.7

1.5 1.3

.40

4.6

1.5

.37

4.6

1.6

Expressiveness measures. Subjects completed the following questionnaires: the AEQ; the EEQ; the Affective Communication Test (ACT; Friedman et al, 1980), a measure of nonverbal expressiveness; the Family Expressiveness Questionnaire (Halberstadt, 1986), a 40item scale measuring verbal and nonverbal emotional expression in the home; and the Affective Intensity Measure (AIM; Larsen & Diener, 1987), another 40-item questionnaire measuring the intensity of emotional response. Family expressiveness has been related to selfexpressiveness (Halberstadt, 1986); however, it has not been considered in relation to ambivalence over expression. Peer reports. Subjects were asked to provide the names and ad-

.37 .37

5.0 5.1

1.6 1.5

.34

5.0

1.8

31 30

5.0 5.4

1.6 1.5

29

4.9

1.5

Item

24

5.6

1.2

19

2.2

1.5

18

3.8

1.5

9. Watching television or reading a book can make me laugh out loud. 13. When I am alone, I can make myself laugh by remembering something from the past. 11. I often laugh so hard that my eyes water or my sides ache. 4. I laugh a lot. 16. I show that I like someone by hugging or touching that person. 3. I often touch friends during conversations. 15. My laugh is soft and subdued.

Note, n = 299.

tend to be inexpressive. This correlation may explain why inexpressiveness has been associated with distress in past research (e.g., Cox & McCay, 1982; Gentry, 1985). It may be that individuals who seem to suffer the adverse consequences traditionally associated with inexpressiveness are actually suffering from the consequences of ambivalence. This statement supports the contention made earlier that ambivalence may help to separate those who are healthily inexpressive from those who are actually inhibited. In the second study, the AEQ was compared with other measures of expressiveness and emotional intensity to more fully examine this possibility as well as to compare the AEQ with other scales and constructs. Study 2 Method Subjects A sample of 48 Michigan State University undergraduates (35 women and 13 men) participated in a 10-week-long course and research project titled "Research on Goals, Mood, and Health." Subjects were from various academic levels and various courses of study. They were recruited through announcements posted in the psychology department. Participation was open to virtually anyone, and participants

Table 4 Three-Factor Solution for Revised Emotional Expressiveness Questionnaire Loading

H 6. Whenever people do nice things for me, I feel "put on the spot" and have trouble expressing my gratitude. (—) 12. If a friend surprised me with a gift, I wouldn't know how to react. 1. I often tell people that I love them. 7. When I realty like someone they know it. 8. I apologize when I have done something wrong. 2. When I am angry people around me usually know. 5. People can tell from my facial expressions how 1 am feeling. 14. I always express disappointment when things don't go as I'd like them to. 10. If someone makes me angry in a public place, I will "cause a scene."

Factor 1 Factor 2 Factor 3 (Positive) (Intimacy) (Negative)

.72

.10

.07

.67

-.01

-.05

.58 .50

.06 .24

-.02 .02

.49

.35

.29

.42

.39

.19

.30

.05

.20

.07

.74

-.05

.02

.67

-.08

.20

.49

.41

.24

.47

.33

.19

.24

.13

.07

.12

.67

.12

.15

.51

.00

-.06

.50

.03

.01

.50

871


Table 5 Correlations for Ambivalence Over Emotional Expressiveness Questionnaire (AEQ) and Expressiveness Measures: Study 2 Measure 1. 2. 3. 4. 5. 6.

AEQ EEQ Peer rating ACT Family expressiveness AIM

1

2

3

4

5

-.25 -.3! -.12 -.33 -.00

— .43 .19 .46 .50

— .23 .40 .29

.0? -.05

— .33

Note. For all but the peer ratings of expressiveness, n = 48. For n = 48, r a .24 is significant, p < .05, and r a .33 is significant, p < .01. For peer ratings, n - 207, r a . 10 is significant, p < .05 and r a . 16 is significant, p < .01. EEQ = Emotional Expressiveness Questionnaire; AIM = Affect Intensity Measure; ACT = Affective Communication Test.

dresses of 10 individuals who knew them well. These individuals were contacted by mail and were asked to complete a 6-item expressiveness scale (fx = .59) about the subjects. These peer ratings of expressiveness provided an additional measure ofexpressiveness. The total number of peers responding was 207. The number of peers per subject returning the packet ranged from 0 to 10. The mean number of peers per subject was 4, with a standard deviation of 1.9. Of the individuals who completed and returned the peer questionnaire, 58% were friends, 14% were parents, 13% were other relatives, 4% were boyfriends or girlfriends, and 11% shared some other relationship with the subject. Only those subjects for whom at least 2 peers responded to the mailer were used in peerratinganalyses. Three subjects wereeliminated from analyses involving peer data because of lack of peer respondents.

viewed previously presents equivocal results pertaining to the relation between expression and well-toeing. In this study, scores on the AEQ, the EEQ, peer ratings, and the ACT will be used to predict psychological and physical well-being as measured by self-report questionnaires, daily reports of mood and physical well-being, and objective measures of physical symptomatology. Predictions AEQ, Expressiveness, and Mood Measures Emotional consequences of conflict include tension, uncertainty, confusion, and vacillation (Miller, 1959; Van Hook & Higgins, 1988). Therefore, we predicted that the AEQ would be positively associated with negative mood. We did not predict that scores on the EEQ, the ACT, and peer ratings ofexpressiveness would be related to daily mood. Questionnaire Measures of Psychological Well-Being We predicted that AEQ scores would be positively correlated with measures of distress and negatively correlated with measures of positive well-being. Scores on the three measures of expressiveness should not be correlated with measures of negative affect. These predictions are in keeping with the rationale that expressiveness itself is no insulation against the adverse psychological consequences that have been associated with inexpressiveness. Measures of Physical

Results and Discussion Pearson correlations for scores on the AEQ and the measures of expressiveness were computed and are presented in Table 5. The means for the AEQ and the EEQ for Study 2 were comparable to those for Study 1—2.84 for the AEQ and 4.33 for the EEQ. In Study 2, the AEQ had a standard deviation of .60 and the EEQ a standard deviation of .63. As expected, the AEQ was significantly negatively correlated with both the EEQ and peer ratings of expressiveness. However, the ACT was not significantly correlated with the AEQ. Scores on the AEQ were negatively related to family expressiveness, and they were not significantly correlated with affect intensity as measured by the AIM. The latter finding suggests that ambivalent individuals are not less likely to experience strong emotions. These results lend further support to the idea that ambivalent individuals may be likely to inhibit emotional expression, but this inhibition may not necessarily affect emotional experience. The results indicate that measures ofexpressiveness alone may confuse individuals who are relaxed and inexpressive with those who are tense and inhibited. The third study addressed this problem further by comparing the predictive ability of these measures of expressiveness and ambivalence over expression with regard to psychological and physical well-being. Study 3 The purpose of the third study was to determine the extent to which the AEQ and the EEQ predict well-being. Literature re-

Symptomatology

With regard to physical symptoms, we predicted that scores on the AEQ would be positively correlated with dairy symptom scores, scores on questionnaire measures of symptoms, health center visits, and illnesses. We predicted that expressiveness scores would show no relation to these symptom indices. Method Subjects in the third study were the same as in Study 2.

Procedure Mood reports. Daily mood reports were completed by the subjects for 21 consecutive days. Two forms were completed each day, one at the "middle" of the subject's day (usually in the mid- to late afternoon) and one before the subject went to sleep. These mood reports consisted of a list of adjectives describing four positive and five negative emotions (happy, joyful, pleased, enjoyment/fun, unhappy, angry, anxious, depressed, and frustrated). The adjectives listed on these reports were chosen on the basis of earlier factor-analytic work by Diener and Emmons (1984). Subjects rated the extent to which they had experienced the emotion 1 isted during the part of the day prior to completion of the form on a 6-point scale ranging from not at all (I) to extremely much (6). The ratings on the positive affect adjectives were summed over the 21 days to produce a composite positive affect score, and the ratings on the negative affect adjective were summed to produce a composite negative affect score. These composite scales have been used in other studies (see Diener, 1984, for a review), and their temporal reliability and internal consistency coefficients approach .90. Psychological well-being measures. During the first weeks of the

872


term-long study (prior to the daily mood assessment), subjects completed several questionnaires measuring psychological well-being, including the Brief Symptom Inventory (BS1; Derogatis & Spencer, 1982), which measures psychological symptoms on several dimensions, including depression, guilt, anxiety, paranoia, phobias, and obsessive-compulsive tendencies. Subjects also completed the WellBeing scale from the Multidimensional Personality Questionnaire (MPQ; Tellegen, 1982), the Satisfaction With Life Scale (SWLS; Diener, Emmons, Larsen, & Griffin, 1985), Rosenberg's Self-Esteem Scale (Rosenberg, 1963), and a revised version of the Bradburn Affect Balance scale (Bradburn, 1969; revised by Warr, Barter, & Brownbridge, 1983), which assesses the extent to which the respondent has recently experienced positive and negative affect. Subjects also completed the Beck Depression Inventory (Beck & Beamesderfer, 1974) as a further measure of psychological distress. Finally, subjects completed the neuroticism scale from the Eysenck Personality Questionnaire (EPQ; Eysenck & Eysenck, 1975), which is a measure of chronic negative affect. Physical symptomatology. Five measures of physical symptoms were obtained from these subjects. First, they completed the Pennebaker Inventory of Limbic Languidness (PILL; Pennebaker, 1982), a questionnaire measure of common physiological symptoms. Second, they completed the Somatization scale from the BSI. Third, the daily mood reports included a symptom checklist; subjects checked the symptoms that they had experienced on the given day. The symptoms listed on the daily report included headaches, stomach ache/pain, chest pain, runny/congested nose, coughing/sore throat, faintness/ dizziness, out of breath, acne/pimples, and stiff/sore muscles ipt = .78). The nine categories were chosen on the basis of factor-analytic work on a large list of physical symptoms (Pennebacker, 1982). Friedman and Booth-Kewley (1987) stated that stomach ache/pain, chest/heart pain, out of breath, and sore/stiff muscles can be considered precursors to serious illnesses that have also been related to emotion: ulcers, heart disease, asthma, and arthritis. These symptom reports were summed over all days to create a global symptom score for each subject. Subjects were asked to sign a consent form giving the experimenters permission to access their health records at the university health center. They were assured that their records would be kept completely confidential. Given these records, a rater, blind to the subjects' AEQ and EEQ scores, counted the number of visits made to the health center as well as the number of different illnesses diagnosed in the past year. The fourth and fifth measures of symptomatology were provided by these two counts (i.e, health center visits and different illnesses). The number of visits ranged from 0 to 18, with a mean of 2,89 and a standard deviation of 4.35. The number of illnesses ranged from 0 to 10, with a mean and standard deviation of 1.81 and 2.19, respectively.

Results and Discussion Ambivalence Over Expression, Expressiveness, and Psychological Well-Being Pearson correlation coefficients were computed for measures of mood, psychological well-being, the AEQ, and the measures of expressiveness. The results are shown in Table 6. Generally, the AEQ was negatively associated with measures of psychological well-being and positively associated with measures of psychological ill-being, as predicted. The AEQ correlated significantly negatively with both life satisfaction and selfesteem. Positive correlations were found for the AEQ and daily negative affect, obsessive/compulsive tendencies, depression, paranoid ideation, and phobic anxiety. Ambivalence was also posi-

tively associated with depression, as measured by the Beck Depression Inventory, and negative affect, as measured by the Bradburn Affect Balance Scale. In sum, the AEQ behaved largely as predicted, correlating positively with measures of poor functioning and negatively with measures of healthy functioning. To demonstrate that ambivalence, and not expressiveness, is the critical variable in the relation between emotion and wellbeing, partial correlation coefficients were computed between the AEQ and well-being measures, controlling for the EEQ. None of the partial correlations, shown in Table 6, decreased. Finally, in order to demonstrate the discriminant validity of ambivalence over expression from neuroticism, partial correlations were calculated for the AEQ scores and the well-being measures, controlling for neuroticism. Several of these partial correlations were lower than the zero-order correlations, however, and, importantly, a few remained significant, including the correlations between the AEQ and the BDI and the AEQ and several of the BSI scales. The correlation between the AEQ and the Neuroticism scale was .35. The EEQ correlated significantly positively with the MPQ measure of well-being and the Bradburn Positive Affect Scale. Peer ratings of expressiveness correlated significantly positively with the MPQ as well. In terms of distress, the EEQ was positively associated with daily reported negative mood, and anxiety and guilt from the BSI. Peer ratings of expressiveness did not correlate with measures of distress. The ACT did not correlate significantly with any of the measures of psychological functioning. To summarize, the EEQ tended to correlate with some measures of both positive and negative functioning. Those correlations that existed between expression and distress were positive, not negative as would be predicted by a perspective emphasizing expression per se as healthy. Ambivalence Over Expression, Expressiveness, and Physical Well-Being Pearson correlation coefficients were computed for ambivalence over expression, the expressiveness measures, and the five measures of physical complaints. Results are shown in Table 7. Although the questionnaire measure of symptomatology, the Somatization scale from the BSI, was marginally correlated with the AEQ (r = .20, p .16 is significant, p < .01. For partials, controlling for neuroticism, n = 36, r a .26 is significant, p < .05, and r a .36 is significant, p .36 is significant, p < .01.

874


questionnaire measures of symptoms, as predicted. However, contrary to predictions, ambivalence over expression showed no relation to health center visits or number of different illnesses. As predicted, expressiveness was generally not associated with measures of health status, supporting further the contention that expression of emotion is not the key to healthy emotion management. Indeed, expressiveness as measured by the EEQ was marginally correlated with health center visits. General Discussion It may be helpful to begin this discussion with a recap of the major findings from both studies. The first important finding from Study 1 was the unidimensionality of the AEQ. The positive correlation between the AEQ and the Raulin scale, as well as the negative correlation between the AEQ and the EEQ, lend support to the idea that the AEQ taps ambivalence over emotional expression. Also, in Study 1, we predicted that women would be more expressive and more ambivalent over expression than men. This prediction was supported, with the added consideration that women were more expressive of and ambivalent over primarily positive emotions. The results of Study 2 lend support to the notion that ambivalent individuals may inhibit expression. This contention would explain the traditional view of inexpressiveness as pathogenic. Among the majorfindingsof Study 3 were the relations between the AEQ and questionnaire measures of psychological ill-being and physical symptoms and the absence of a relation between expressiveness and these measures of ill-being. Also of importance in Study 3 was the lack of relation among health center visits, number of illnesses, and expressiveness or ambivalence over expression. The first of these findings to be addressed here is the sex differences found in Study 1. Although these differences were statistically significant, they were not high in magnitude. Women were more expressive and more ambivalent over expression. These results suggest that, as postulated earlier, although men are less expressive, they may be comfortable within their inexpressiveness. It should be noted here that there was no sex difference in scores on the Marlowe-Crowne. Thus, the differences in expressiveness and ambivalence over expression between male and female subjects is not likely to be due to a male reluctance to self-disclose or to admit to negative affect. Predicted sex differences did not emerge with regard to the expression of anger. Although researchers in the area of sex differences in emotional expression have focused on anger as a stumbling block for women (e.g., Hochschild, 1975, 1983; Shields, 1987), the current results indicate that it was positive emotions about which women were more ambivalent. Hochschild (1975) has suggested that women are placed in a conflictful situation routinely as a result of cultural expectations of nurturance and situational demands for neutrality or emotional detachment. The present results support her contention. It is noteworthy that the subjects participating in the first study were all college women—those who can be said to be trying, at least to some extent, to balance a traditional female role with individual ambitions. It may be that for these young women, it is not negative emotions that cause problems but positive emotions—those that may seem inappropriate to their career concerns but are part and parcel of traditional femininity.

Predictions about the other analyses in Study 1 were confirmed for the most part, although there were some exceptions. The AEQ and the EEQ were negatively correlated with the Marlowe-Crowne scale. Because high scorers on the MCSD tend to deny negative affect (Weinberger, in press), this negative correlation indicates that for an individual to score highly on the AEQ, he or she must be willing or able to admit to negative affect. This conclusion reveals one weakness of the AEQ measure: It does not tap into ambivalence that may be felt by individuals who do not or cannot admit to experiencing negative affect. This conclusion would indicate that the scale cannot uncover unconscious conflict over expression. This shortcoming may not be fatal, however, because Emmons and King (1988) have demonstrated the implications of conscious conflict for psychological and physical well-being. The fact that the AEQ was positively correlated with daily negative affect is all the more impressive in this light. Because conflicted individuals are likely to be less expressive, the presentfindingshave probably underestimated the degree to which ambivalent individuals are distressed. Reporting one's emotional state is a type of expressive behavior. The construct of ambivalence over expressing emotion bears resemblance to the well-established repressive-coping style, or repressive defensiveness (Jensen, 1987; Weinberger, in press). Persons in conflict over expressing emotion may very well share similar characteristics with repressive individuals, linked through common inhibitory processes. However, we view ambivalence as a more general construct, encompassing both positive and negative emotions as well as both inhibition and rumination, whereas repressiveness refers to the lack of recognition and expression of negative emotion only. Because individuals must admit to experiencing emotion before they can be in conflict over expressing it, it is unlikely that repressors would score high on the AEQ. The relations between these constructs need to be examined in future research. The results of Study 3 provide surprising evidence about the roles of expression and ambivalence over expression in wellbeing. Although expressiveness was associated with positive affect, it was not negatively correlated with measures of distress. This result contradicts many of our assumptions about expression and health, underscoring the need for more sophisticated views of emotional expression. The AEQ, however, was positively correlated with a number of measures of poor psychological functioning. Although the AEQ correlated with measures of psychological distress and a questionnaire measure of symptoms, it did not correlate with the number of illnesses suffered or with daily reported symptoms. Several explanations seem possible. The first is, of course, that there is no relation between ambivalence over expression and physical distress. This explanation would hold that ambivalence over expression is a purely psychological phenomenon that does not affect physical well-being. Although the existence of any such phenomenon would seem highly unlikely given the previous literature on the relation between conflict and health, it is still possible that this particular form of emotional turmoil is not physiologically pathogenic. The second explanation is that ambivalence over expression is an emotional experience that requires time to affect physical health. Longitudinal data on the individuals who were highly

875


ambivalent over expression would demonstrate whether, over time, this ambivalence affects physical health. A third possibility is that the symptom measures used were insufficient to detect the relation between ambivalence over expression and physical illness. EEQ scores were marginally correlated with health center visits. There are two possible explanations for this unexpected result. First, if symptom reporting is thought of as an expressive behavior, then the correlation between symptom reports and the EEQ is understandable. However, such an explanation would also predict that EEQ scores would be positively associated with daily symptoms, and this was not the case. A second explanation is provided by research in health psychology. The concept of "readiness to use services" or "orientation to care" has been used to describe the seeking out of treatment for symptoms (Rosenstock & Kirscht, 1979). This readiness involves the belief that one's health is in one's control and that health professionals and services are a means of maintaining one's good health (Rosenstock & Kirscht, 1979). In this sense, seeking out treatment may be viewed as a sign of positive psychological functioning. Thus, emotional expressiveness may be positively associated not with physical symptoms, but with "orientation to care." Several problems are apparent in the present studies and deserve to be addressed here. One issue that is brought to the forefront by this study is the question of to what extent selfreports of symptomatology are in themselves expressive behaviors. It has already been suggested that the correlation between health center visits and expressiveness may be due to differences between expressive and inexpressive individuals in symptom reporting. We join with others (e.g, Costa & McCrae, 1987; Watson & Pennebaker, 1989) in calling for more objective health status measures. Such a methodology might involve ambulatory heart and blood pressure monitoring and tests of immune functioning. Also, galvanic skin response monitoring might be necessary, as skin conductance is generally accepted as being a sensitive index of inhibition (Fowles, 1980; Pennebaker, 1985). One potential problem with the measures of expression used in this investigation is that they ignore the social context of expression. In none of these scales was the individual to whom an emotion was to be expressed mentioned clearly. This disregard for social context presents an important problem, for it may be that individuals are more likely to confide in or express emotions to some individuals in their lives (e.g., spouse, therapist, parent, or friend) than others (e.g, boss). Balswick (1988) provided a means by which social context of expression can be included and examined. He had subjects fill out expressiveness questionnaires several times—for different individuals in their social environment. For instance, subjects would fill out the questionnaire once with regard to their spouse, again with regard to a friend, and so forth. Such a methodology would offer several benefits to researchers in the area of emotional expression. Not only would knowledge of the importance of expressiveness increase, but knowledge of the adaptiveness of individual styles of expression would increase as well. The current investigations suggest several avenues for future research, in terms of both improving the methodology in work on the ambivalence over expression construct and defining the

conceptual framework of emotion research. It is necessary that research begin to focus on the individual goals that underlie expression. In addition, the use of measures of expressiveness alone in health psychology research will not accomplish the task of differentiating the healthy from the potentially unhealthy. A final point that isfittinglyrepeated here as a summary of the findings of this investigation is that expression in and of itself is not the key to well-being. Rather, it may be how one feels about one's expressive behaviors that determines the potentially harmful character of lack of expression. There may be other components in the relation between emotion and health. Researchers must continue to look for those factors that separate inexpressiveness from inhibition—to distinguish between individuals who are expressive and healthy and those who are expressive and at risk for serious illness. This is a continuing challenge for those conducting studies on emotional expression and health psychology. The construct of ambivalence over expression and one measure of it, the Ambivalence Over Emotional Expressiveness Questionnaire, have been presented as a step in meeting that challenge.

References Balswick, J. (1988). The inexpressive male. Lexington, MA: Lexington Books. Baumeister, R., & Tice, C. (1987). Emotion and self-presentation. In R. Hogan & W H. Jones (Eds.), Perspectives in personality (Vol. 2, pp. 181-200). Greenwich, CT: JAI Press. Beck, A. T., & Beamesderfer, A. (1974). Assessment of depression: The depression inventory. In R Pichot(Ed.), Psychological measurement in psychopharmacology: Modern problems in pharmacopsychiatry (Vol. 7, pp. 151-169). Basel, Switzerland: Karger. Bell, P A., & Byrne, D. (1978). Repression-sensitization. In H. London & J. E. Exner (Eds.), Dimensions ofpersonality (pp. 449-485). New York: Wiley. Beutler, L. E, Engle, D, Oro-Beutler, M. E, Daldrup, R., & Meredith, K. (1986). Inability to express intense affect: A common link between depression and pain. Journal of Consulting and Clinical Psychology, 54, 752-759. Bradburn, M. N. (1969). The structure ofpsychological well-being. Chicago: Aldine. Buck, R. W (1984). The communication of emotion. New York: Guilford. Costa, R X, & McCrae, R. R. (1987). Neuroticism, somatic complaints, and disease: Is the bark worse than the bite? Journal of Personality, 55,299-316. Cox, X, & McCay, C. (1982). Psychosocial factors and psychophysiological mechanisms in the aetiology and development of cancers. Social Science and Medicine, 16, 381-396. Derogatis, L. R, Abeloff, M. D., & Melisaratos, N. (1979). Psychological coping mechanisms and survival time in metastatic breast cancer. Journal of the American Medical Association, 242, 1504-1508. Derogatis, L. R., & Spencer, R M. (1982). The BriefSymptom Inventory (BSI). Baltimore: Clinical Psychometric Research. Diener, E. (1984). Subjective well-being. Psychological Bulletin, 95, 542-575. Diener, E., & Emmons, R. A. (1984). The independence of positive and negative affect. Journal of Personality and Social Psychology, 47, 1105-1117.

876


Diener, E., Emmons, R. A, Larsen, R. J., & Griffin, G. (1985). The satisfaction with life scale. Journal of Personality Assessment, 49, 71-75. Emmons, R. A. (1986). Personal strivings: An approach to personality and subjective well-being. Journal ofPersonality and Social Psychology, 51, 1058-1068. Emmons, R. A. (1989). The personal striving approach to personality. In L. Perrin (Ed.), Goal concepts in personality and social psychology (pp. 87-126). Hillsdale, NJ: Erlbaum. Emmons, R. A., & King, L. A. (1988). Personal striving conflict: Immediate and long-term implications for psychological and physical well-being. Journal of Personality and Social Psychology, 54, 1040-1048. Eysenck, H. 1, & Eysenck, S. B. G. (1975). Manual of the Eysenck Personality Questionnaire. San Diego: Educational Testing Service. Fowles, D. C. (1980). The 3 arousal model: Implications of Gray's 2 factor theory for heart rate, electrodermal activity and psychopathy. Psychophysiology, 17, 87-104. Freud, S. (1977). Introductory lectures on psychoanalysis (J. Strachey, Trans.). New York: W W Norton & Co. (Original work published 1917) Fridlund, A. J., Newman, J. B., & Gibson, E. L. (1984). Putting emotion in behavioral medicine: Discrete-emotion psychophysiology and its relevance for research and therapy. In C. Van Dyke, L. Temoshok, & L. S. Zegans (Eds.), Emotions in health and illness (pp. 117-134). New York: Grune & Stratton. Friedman, H. S., & Booth-Kewley, S. (1987). Personality, Type A behavior, and coronary heart disease: The role of emotional expression. Journal of Personality and Social Psychology, 53, 783-792. Friedman, H. S., Hall, J. A., & Harris, M. J. (1985). Type A behavior, nonverbal expressive style, and health. Journal of Personality and Social Psychology, 48,1299-1315. Friedman, H. S., Harris, M. J., & Hall, J. A. (1984). Nonverbal expression of emotion: Healthy charisma or coronary-prone behavior? In C. Van Dyke, L. Temoshok, & L. S. Zegans (Eds.), Emotions in health and illness (pp. 151-165). New York: Grune & Stratton. Friedman, H. S., Prince, L. M., Riggio, R. E., & DiMatteo, M. R. (1980). Understanding and assessing nonverbal communication of emotion: The Affective Communication Test. Journal ofPersonality and Social Psychology, 39, 333-351. Gentry, W D. (1985). Relationship of anger-coping styles and blood pressure among Black Americans. In M. A. Chesney & R. H. Rosenman (Eds.), Anger and hostility in cardiovascular and behavioral disorders (pp. 139-148). New York: Hemisphere Publishing Corporation. Goldstein, H. S., Edelberg, R., Meier, C. F., & Davis, L. (1988). Relationship of resting blood pressure and heart rate to expressed anger. Psychomatic Medicine, 50, 321-329. Greer, S., & Morris, T. (1975). Psychological attributes of women who develop breast cancer: A controlled study. Journal ofPsychosomatic Research, 79,147-154. Halberstadt, A. (1986). Family socialization of emotional expression and nonverbal communication of styles and skills. Journal ofPersonality and Social Psychology, 51, 827-836. Hall, J. A, Friedman, H. S., & Harris, M. J. (1986). Nonverbal cues, the Type A pattern, and coronary heart disease. In R Blanck, R. Buck, & R. Rosenthal (Eds.), Nonverbal communication in the clinical context (pp. 144-168). University Park: Pennsylvania State University Press. Hochschild, A. R. (1975). The sociology of feeling and emotion. In M. Millman & R. Kanter (Eds.), Another voice (pp. 280-307). Garden City, NY: Anchor Books. Hochschild, A. R. (1983). The managed heart. Berkeley: University of California Press. Hokanson, J. E., & Burgess, M. (1962). The effects of 3 types of aggres-

sion on vascular processes. Journal ofAbnormal and Social Psychology, 64, 446-449. Hokanson, J. E., & Shetler, S. (1961). The effect of overt aggression on physiological arousal level. Journal ofAbnormal and Social Psychology, 63, 446-448. Hunter, J. (1984). Factor analysis. In E Menge & J. Cappella (Eds.), Multivariate techniques in human communication research (pp. 229-257). New York: Academic Press. Hunter, J. (1988). PACKAGE: A manual for the IBM mainframe. Unpublished manuscript, Michigan State University. Hunter, J., Schmidt, E L., & Jackson, G. B. (1982). Meta-analysis: Cumulating researchfindingsacross studies. Beverly Hills, CA: Sage. Jensen, M. R. (1987). Psychobiological factors predicting the course of breast cancer. Journal of Personality, 55, 317-342. Jones, H. E. (1935). The galvanic skin response as related to overt emotional expression. American Journal of Psychology, 47, 241-251. Lanzetta, J. T, Cartwright-Smith, J., & Kleck, R. E. (1976). Effects of nonverbal dissimulation on emotional experience and autonomic arousal. Journal of Personality and Social Psychology, 33, 354-370. Larsen, R. J, & Diener, E. (1987). Affect intensity as an individual difference characteristic: A review. Journal of Research in Personality, 21,1-39. Marlowe, D, & Crowne, D. D. (1964). The approval motive: Studies in evaluative dependence. New York: Wiley. Miller, N. E. (1959). Liberalization of basic S-R concepts: Extensions to conflict behavior, motivation, and social learning. In S. Koch (Ed.), Psychology: A study of a science (Vol. 2). General systematic formulations, learning, and special processes (pp. 196-292). New York: McGraw-Hill. Morris, T, Greer, S., Pettingale, K. W, & Watson, M. (1981). Patterns of expression of anger and their psychological correlates in women with breast cancer. Journal ofPsychosomatic Research, 25,111 -117. Murray, E. J. (1985). Coping and anger. In T. Fields, R McCabe, & N. Schneiderman (Eds.), Stress and cop ing(pp. 243-261). Hillsdale, NJ: Erlbaum. Pelletier, K. R. (1985). Mindashealer, mindasslayer. New York: Delacorte Press. Pennebaker, J. W (1982). The psychology of physical symptoms. New York: Springer-Verlag. Pennebaker, J. W (1985). Traumatic experience and psychosomatic disease: Exploring the roles of behavioral inhibition, obsession, and confiding. Canadian Psychology, 26, 82-95. Pennebaker, J. W, & Hoover, C. W (1986). Inhibition and cognition: Toward an understanding of trauma and disease. In R. J. Davidson, G. E. Schwartz, & D. Shapiro (Eds.), Consciousness and selfregulation (Vol. 4, pp. 107-136). New York: Plenum Press. Pennebaker, J. W, Hughes, C. E, & O'Heeron, R. C. (1987). The psychophysiology of confession: Linking inhibitory and psychosomatic processes. Journal of Personality and Social Psychology, 52, 781-793. Raulin, M. L. (1984). Development of a scale to measure intense ambivalence. Journal ofConsulting and Clinical Psychology, 52, 63-72. Rosenberg, M. (1963). The association between self-esteem and anxiety. Journal of Psychiatric Research, 1, 135-152. Rosenstock, I. M., & Kirscht, J. R (1979). Why people seek healthcare. In G. C. Stone, N. E Cohen, & N. E. Adler(Eds.), Health psychology —A handbook (pp. 161-188). San Francisco: Jossey-Bass. Roth, S., & Cohen, L. J. (1986). Approach, avoidance, and coping with stress. American Psychologist, 41, 813-819. Sackheim, H. A. (1983). Self-deception, self-esteem, and depression: The adaptive value of lying to oneself. In J. Masling (Ed.), Empirical studies ofpsychoanalytic theories (Vol. 1, pp. 101 -157). Hillsdale, NJ: Erlbaum.

877

CONFLICT OVER EXPRESSION Schriesheim, C. A., & Hill, K. D. (1981). Controlling acquiescence response bias by item reversals: The effect on questionnaire validity. Educational and Psychological Measurement, 41,1101-1115. Shields, S. A. (1987). Women, men, and the dilemma of emotion. In R Shaver &C.Hendrick(Eds.), Review of personality and social psychology (Vol. 7, pp. 229-250). Beverly Hills, CA: Sage. Tavris, C. (1984). On the wisdom of counting to ten: Personal and social dangers of anger expression. In P. Shaver (Ed.), Review of personality and social psychology (Vol. 5, pp. 170-191). Beverly Hills, CA: Sage. Tellegen, A. (1982). Briefmanual for the Multidimensional Personality Questionnaire. Unpublished manuscript, University of Minnesota, Minneapolis. Udelman, H. D., & Udelman, D. L. (1981). Emotions and rheumatologic disorders. American Journal of Psychotherapy, 35, 576-587. Van Hook, E., & Higgins, E. T. (1988). Self-related problems beyond the

self-concept: Motivational consequences of self-guides. Journal of Personality and Social Psychology, 55, 625-633. Warr, R, Barter, J, & Brownbridge, G. (1983). On the independence of positive and negative affect. Journal of Personality and Social Psychology, 44, 644-651. Watson, D., & Pennebaker, J. W (1989). Health complaints, stress, and distress: Exploring the central role of negative affectivity. Psychological Review, 96, 234-251. Weinberger, D. A. (in press). The construct validity of the repressive coping style. In J. Singer (Ed.), Repression and dissociation: Defense mechanisms and personality styles. Chicago: University of Chicago Press. Received February 21,1989 Revision received July 20,1989 Accepted October 3,1989 •

American Psychological Association Subscription Claims Information

Today's Date_

This form is provided to assist members, institutions, and nonmember individuals with any subscription problems. With the appropriate information provided, a resolution can begin. If you use the services of an agent, please do NOT duplicate claims through them and directly to us. PLEASE PRINT CLEARLY AND IN INK IF POSSIBLE.

PRIOT RILL NAMB OR K E Y N A M E S INSTITUTION

MHMB8R OR CUSTOMER NUMBER ( MAY BB POUND ON ANY PAST ISSUB LABEL)

ADDRESS

DATE YOUR ORDER WAS MAILED ( OR PHONED) P. O. NUMBER

CITY

STATE/COUNTRY

ZIP

PREPAID

OHBCK W _CHAI»OB.

CHECK/CARD CLBARSD D A T B _ _

(If possible, send a copy,frontand back, of your cancelled check to help us in

YOUR NAME AND PHONE NUMBER

our research of your claim.) ISSUBS: _ _ M B S W O

Tmn(s)

VOLVYR.

ISSUERS)

NO./MONTH

Thank you. Once a claim is received and resolved, delivery of replacement issues routinely takes 4-6 weeks. M I BM mm mm «

^

HH • • m B«(TO BE PILED OUT BY APA STAHOB— ^

DATE RECEIVED.

DATBOFACTION_

ACTION T A K B N _

INV. N O . & DATB_

STAFF NAMB

LABEL #, DATE

^

^m mm ^

^

^

^

SEND THIS FORM TO: APA Subscription Claims. 1400 N. Uhle Street, Arlington, VA 22201 PLEASE DO NOT REMOVE, A PHOTOCOPY MAY BE USED.

_JCXMAOBD