Journal of Personality and Social Psychology 1994. Vol. 66. No. 5,934-949 Copyright 1994 by the American Psychological Association, Inc 0022-3514/94/S3.00 Individual Differences in Dispositional Expressiveness: Development and Validation of the Emotional Expressivity Scale Ann M. Kring, David A. Smith, and John M. Neale Although emotional expressivityfiguresprominently in several theories of psychological and physi- cal functioning, limitations of currently available measurement techniques impede precise and eco- nomical testing of these theories. The 17-item Emotional Expressivity Scale (EES) was designed as a self-report measure of the extent to which people outwardly display their emotions. Reliability stud- ies showed the EES to be an internally consistent and stable individual-difference measure. Valida- tional studies established initial convergent and discriminant validities, a moderate relationship be- tween self-rated and other-rated expression, and correspondence between self-report and laboratory- measured expressiveness using both college student and community populations. The potential for the EES to promote and integratefindingsacross diverse areas of research is discussed. Other peoples' emotional expressions hold a certain fascina- tion for nearly everyone. News agencies always provide images of politicians' expressions on winning and losing elections. Re- ports of court cases routinely mention the defendant's emo- tional expressions during the reading of the verdict. Winning and losing locker-room photographs attempt to capture sports figures' expressive reactions. This level of fascination is proba- bly supported by the belief that something unique and interest- ing is communicated by emotional expressions���something that words may at times fail to express. As Fritz Perls (1969), the founder of Gestalt therapy, put it "What we say is mostly either lies or bullshit. But the voice is there, the gesture, the posture, the facial expression" (p. 54). People vary in the extent to which they outwardly exhibit emotions, and these differences have long posed unique and in- teresting challenges to psychologists. Indeed, emotional expres- siveness has captured the attention of researchers interested in areas as diverse as nonverbal communication, psychopathology, personality, social psychology, and health psychology. This arti- cle reports on the development of a new self-report measure capturing the general disposition to outwardly express emotion. At the outset, it is worth addressing several crucial questions. Can emotional expressiveness be defined operationally? Is emo- Ann M. Kring, Department of Psychology, Vanderbilt University David A. Smith, Departments of Psychology and Psychiatry, Ohio State University John M. Neale, Department of Psychology, State University of New \brk at Stony Brook. Portions of this research were presented at the 96th Annual Conven- tion of the American Psychological Association in Atlanta. This work was supported in part by a grant from the National Institute of Mental Health (MH4411602) awarded to John M. Neale and a grant from Sigma Xi, the Scientific Research Society, awarded to Ann M. Kring. We thank Joanne V. Wood, Arthur Stone, and Andrew J. Tomarken for their helpful comments on drafts of this article and Nadya A. Klinetob for her assistance with collecting data from Sample F. Correspondence concerning this article should be addressed to Ann M. Kring, Department of Psychology, Vanderbilt University, Nashville, Tennessee 37240. tional expressiveness meaningful and of theoretical interest? Can a self-report measure be constructed to adequately mea- sure emotional expressiveness? Defining Emotional Expressiveness The construct of emotional expressiveness recognizes indi- vidual differences in the extent to which people outwardly dis- play their emotions, and it differs in important ways from other modes of affective response. For example, emotionality is tradi- tionally conceptualized as the tendency to shift from a positive or neutral emotional state to a negative one (Buss & Plomin, 1975 Thurstone, 1951 Watson & Clark, 1984), or more gener- ally, as a disposition to experience positive or negative emotions (e.g., Tellegen, 1985 Tellegen et al., 1988). Emotional experi- ence is further encompassed by Larsen's (1984) conceptualiza- tion of affect intensity. Hedonic capacity, on the other hand, in- volves peoples' ability to experience pleasure (Chapman, Chap- man, & Raulin, 1976). Expressive self-control is captured by the construct of self-monitoring and refers to the ability to monitor and control one's own verbal and nonverbal behavior with re- spect to social cues (Snyder, 1974). For present purposes, emo- tional expressiveness refers simply to the outward display of emotion, regardless of valence (positive or negative) or channel (facial, vocal, or gestural). It is important to emphasize that our definition of emotional expressiveness does not include a priori assumptions about the type of emotion (e.g., happiness or sadness) expressed or the manner in which emotion is expressed (e.g., facially). That is, our conceptualization emphasizes a general disposition toward expressing different emotions across various channels. Of course, it may well be the case that important differences exist, for example, in the expression of positive versus negative emo- tions. However, these differences must be determined empiri- cally before being built into a measure of expressivity. The decision to define a construct in either a broad or narrow fashion is an issue that is at the forefront of current research on personality test construction. Negative Affect measures are a case in point. Briefly, the construct of Negative Affect refers to 934

EMOTIONAL EXPRESSIVITY SCALE 935 peoples' predispositions to experience distress and disengage- ment, including several negative mood states, such as anxiety, anger, contempt, disgust, guilt, and self-dissatisfaction. Watson and Clark (1984) convincingly demonstrated that several exist- ing scales all measured in varying degrees the same construct of Negative Affect. Several independent lines of research were conducted on presumably different and specific constructs (e.g., trait anxiety and neuroticism) however, these "specifics" were more meaningfully captured and represented by the general construct ofNegative Affect. Returning to the construct ofemo- tional expressivity, the premise on which this article is based is that the construct of generalized emotional expressiveness, including its relation to more specific expressive processes (e.g., inhibition of anger), must be empirically elucidated. Whether substantial gains in incremental validity might be achieved by creating several different "specialized" measures above and be- yond that which can be obtained with a general measure re- mains to be established. Theoretical and Empirical Interest in Emotional Expressiveness Althoughfiguringprominently in several major programs of research, expressiveness is most integral to the area of nonver- bal communication where it is studied in relation to autonomic arousal (e.g., Levenson, Ekman, & Friesen, 1990 Notarious & Levenson, 1979), the ability to perceive emotion in others (e.g., Zuckerman, Hall, DeFrank, & Rosenthal, 1976), the ability to produce affective displays on demand (e.g., Berenbaum & Rot- ter, 1992), the subjective experience of emotion (e.g., Ekman, Davidson, & Friesen, 1990 Ekman, Friesen, & Ancoli, 1980), social skills and communication (e.g., Buck, Losow, Murphy, & Costanzo, 1992 Riggio, 1986), and gender differences (e.g., Hall, 1985). To index emotional expressiveness, researchers in this area have relied heavily on judges' ratings of the accuracy of emotional displays. However, this method, although appro- priate from the standpoint of information exchange, attends only to the quality of expression (i.e., does the subject smile when looking at a happy stimulus?) and neglects the quantity and magnitude of expression (i.e., how often and how strongly does this person outwardly display emotions whether stimulus "appropriate" or not?). It is worth noting, however, that the communication accuracy paradigm implicitly invokes a gen- eral definition ofexpressivity. That is, researchers are interested in how accurately people express a variety of emotions in a va- riety ofemotion-provoking situations. Under the rubric "flat affect," low emotional expressiveness has long been regarded as a central feature of schizophrenia (Bleuler, 1911/1950). Simply defined, flat affect refers to the lack of outward emotional expression seen in some schizo- phrenic patients. Currently, flat affect is an important part of several diagnostic schemes (e.g., Abrams & Taylor, 1978), is one of the key negative symptoms (Andreasen, 1983), and is an im- portant prognostic indicator (Fenton & McGlashan, 1991 Knight, Roff, Barnett, & Moss, 1979). Flat affect is most often measured with time-consuming clinical ratings. These ratings are typically made at only one time period, such as during a structured interview, and therefore do not necessarily provide an accurate representation of a patient's dispositional expres- siveness. More recently, systematic investigations of emotion in schizophrenia have provided support for a generalized expres- sivity deficit, across both positive and negative emotions (e.g., Berenbaum & Oltmanns, 1992 Kring, Kerr, Smith, & Neale, 1993). Expressiveness has also been implicated in psychiatric disor- ders other than schizophrenia. For example, several of the per- sonality disorders contain diagnostic criteria pertaining to emo- tional expression (Diagnostic and Statistical Manual of Mental Disorders, 3rd ed., rev. [DSM-III-R] American Psychiatric As- sociation, 1987). Frequent, dramatic, yet rapidly shifting, and shallow expression of emotion is a hallmark of the histrionic personality disorder. Constricted expression of emotion is a cri- terion for both schizotypal and schizoid personality disorders. These diagnostic criteria refer more to general expressiveness than to particular aspects of expression. In the health psychology literature, several relationships be- tween expressiveness and specific diseases have been reported. For example, Hollaender and Florin (1983) found that children with bronchial asthma showed fewer and shorter facial expres- sions of anger, joy, and surprise in a stressful achievement situa- tion compared with children without the condition. Friedman, Hall, and Harris (1985) found that Type A men defined as low in expressiveness by their scores on the Affect Communication Test (ACT Friedman, Prince, Riggio, & DiMatteo, 1980) were more unhealthy and aggressive, whereas, highly expressive Type A's were comparatively healthy and popular. In a study of breast cancer patients, Watson, Pettingale, and Greer (1984) showed that patients, as compared with controls, tended to experience more anxiety and disturbance in a stressful situation but were more likely to control or inhibit their expressive reactions, espe- cially anger. Temoshok and her colleagues (e.g., Temoshok et al., 1985) found that reduced expressiveness in skin cancer patients was associated with poorer immunological functioning at the site of the tumor, greater tumor thickness, and quicker tumor growth. In sum, emotional expressiveness is currently enjoying a re- newed interest among researchers. Complicated theories re- garding the role ofexpressivity have promoted both general and specific aspects of the construct. Refining the measurement of the construct is an important next step toward theoretical and empirical advancement of our understanding of expressivity's role in nonverbal communication, psychopathology, and health. Measuring Emotional Expressiveness Currently, expressiveness is most often measured either by using judges' ratings of communication accuracy or with mea- sures that tend to be conceptually ambiguous, have little or no established reliability and validity, demand time-consuming clinical ratings, or focus only on specific features ofexpressive- ness. Although a standardized self-report measure ofthe general predisposition to express emotion offers obvious advantages, several currently available measures appear inadequate or in- complete. For example, the Courtald Emotional Control Scale (Watson & Greer, 1983) contains items pertaining only to the controlled expression of negative emotions. The Self-Monitor- ing Scale (Gangestad & Snyder, 1985) assesses the responsive- ness of expressive behaviors to social cues. Although these con-

936 A. KRING, D. SMITH, AND J. NEALE structs are interesting in their own right, they differ in impor- tant ways from the more global expressivity construct central to most theories. The relevance of these distinctive features of emotional expressiveness to the general construct will remain speculative until a global index is available. Then it will be pos- sible to ask how the control ofemotionfiguresinto dispositional expressiveness, whether people are equally expressive of both positive and negative emotions, and whether specific modes of expression improve the prediction of psychological and health outcomes beyond that possible from general expressiveness alone. Embarking on a test construction project is typically moti- vated by the need to fill an existing measurement void. Given its widespread interest, it is not surprising that other self-report measures of expressiveness have been developed. For example, the ACT (Friedman et al., 1980) was designed to identify ex- tremely expressive, charismatic, or dynamic individuals. Items on the ACT include "When I hear good dance music, I can hardly keep still" and "I am terrible at pantomime as in games like charades." Given its acronym and the fact that it was de- signed to measure charisma, it is not surprising that valida- tional strategies included an examination ofthe measure's rela- tionship to acting ability and theatrical experience. The ACT was not, however, validated with a study of spontaneous expres- sion. Although the content of the ACT may overlap with items more specifically related to expressiveness, it was developed more as a measure of extreme expressivity or charisma, and thus its applicability as a general index of dispositional expres- siveness may be limited. The Emotional Expressivity subscale of the Social Skills Inventory (Riggio, 1986) was developed as part of an inventory of social skills and as such focuses on skill- related aspects of expressivity. In this context, expressivity is defined as a learned skill that is inherent to and necessary for social interaction. Our conceptualization of expressivity posits that it is a stable, individual-difference variable rather than the combination of various skills useful in interpersonal situations. King and Emmons (1990) developed the Emotional Expres- sivity Questionnaire (EEQ) as an adjunct to a measure of am- bivalence over emotional strivings. Although few details regard- ing final item selection were given, 16 items remained on the final version, and factor analysis of the interitem correlations yielded three factors: Expression of Positive Emotion (7 items), Expression of Intimacy (5 items), and Expression of Negative Emotion (4 items). Although in this article we are building a case for a unidimensional, generalized measure of individual differences in emotional expressivity, even if the three specific factors of the EEQ are deemed of interest, care should be taken in use ofthe EEQ to measure these factors. Apart from reported sex differences, there is no currently available validity evidence for these factors. However, the total scale was significantly cor- related with peer ratings of expressiveness. The total score may well index generalized dispositions to express emotions, but this fact would substantially diminish the importance ofthe Positive and Negative Expressivity factors that we emphasize herein.1 Recognizing the need for a general index of expressivity, the purpose of the present study was to develop a self-report mea- sure of individual differences in the degree to which people out- wardly express their emotions. Unlike other measures that tap ability and motivational aspects of expressiveness, the current instrument is intended as a general measure of expressiveness, the nontest personality, health, and mood correlates of which are empirically determined rather than either built-in or pre- sumed. Although emotional expression is related to the experi- ence of emotion (see Adelmann & Zajonc, 1989, for a review), the present scale was designed to be primarily a measure of ex- pressivity that would not be redundant with existing measures of emotional experience. A variety of samples and validation strategies were used, including measures ofconceptually similar and dissimilar personality variables, measures of spontaneous expressiveness, and other ratings of expressivity. Construction of the EES Construction of the EES followed a deductive strategy in that the construct ofemotional expressiveness was defined and items were then generated to fit the definition (Burisch, 1984). Spe- cifically, emotional expressiveness was defined as the extent to which people outwardly display their emotions, regardless of emotional valence or channel of expression. As discussed ear- lier, ofprimary importance in this study was the development of a measure assessing generalized expressivity. In addition, item generation and selection processes were guided by the notion that "a single scale ought to measure a single construct" (Briggs & Cheek, 1986, p. 109). In other words, the goal of scale con- struction was to generate a set of items pertaining to expressive- ness rather than other aspects of emotion (e.g., experience) and to select the best set of items that were homogeneous with re- spect to the construct of expressiveness (Nunnally, 1978). Orig- inally, we generated 40 items that sampled this domain of emo- tional expressiveness. The response format for the EES is a 6- point Likert scale (1 = never true and 6 = always true), allowing ratings of the extent to which each item applies to each partici- pant. The EES was given to six samples of either college students or adult community residents. What follows is a description of each of these samples, along with the presentation of studies conducted to (a) establish the psychometric properties of the scale, (b) establish initial convergent and discriminant validi- ties, (c) examine nontest correlates ofthe EES by examining its relation to spontaneous emotional expressiveness, and (d) assess the relationship between self- and other ratings of expressive- ness. The sample comprising each of these studies is indicated as each study is described. Sample Characteristics Sample A The initial sample consisted of 237 female and 136 male un- dergraduates (M age of 18.39 years, SD = 2.25) enrolled in an introductory psychology course at the State University of New York at Stony Brook. These participants also completed several other questionnaires (described below) selected to provide ini- 1 None of these considerations about the EEQ bear on the appropri- ateness of the King and Emmons (1990) measure of conflict/ambiva- lence over emotional expressivity, which was, in fact, the centerpiece of their article.