Sex differences in the provision of skillful emotional support: the mediating role of...

This study focuses on associations between sex, self-efficacy in the domain of emotional support, and the skillfulness (person centeredness) of emotional support messages. In particular, it examines whether self-efficacy mediates the effect of sex on emotional support skill, and whether targets'

responsibility for their problems moderates this mediating effect. Participants (N = 715) produced messages in response to scenarios depicting distressed friends who were or were not responsible for their problems, and completed a measure of self-efficacy. Compared to men, women produced emotional support messages with a higher level of person centeredness and reported greater self-efficacy in the domain of providing emotional support. Regression analyses indicated that self-efficacy mediated approximately 30% of the sex-related variance in person centeredness. Target responsibility did not moderate this mediating effect.

**********

* Different kinds of evidence combine to indicate that women, on average, are more skillful providers of emotional support than are men (for reviews see Goldsmith & Dun, 1997; Kunkel & Burleson, 1998). Indirect evidence comes from studies indicating that support is sought from women more frequently than from men (e.g., Akiyama, Elliott, & Antonucci, 1996), that women are preferred as support providers (e.g., Kunkel & Burleson, 1999), and that women value the ability to provide comfort more than do men (e.g., Burleson, Kunkel, Samter, & Werking, 1996). Direct evidence comes from studies that have examined the quality of men's and women's supportive efforts. To date, four studies concerned with sex differences in the quality of emotional support messages have found that girls' and women's messages, on average, are more person-centered than those of boys and men (Borden, 1979; Burleson, 1982; Hale, Tighe, & Mongeau, 1997; Samter, this issue).

As a characteristic of emotional support messages, person centeredness is defined as "the extent to which the feelings and perspective of a distressed other are explicitly acknowledged, elaborated, and granted legitimacy" (Burleson, 1994, p. 12). Highly person-centered messages exhibit the greatest degree of acknowledgment, elaboration, and legitimation, whereas messages low in person centeredness condemn, challenge, or ignore a target's feelings and perspective (Applegate, 1980; Burleson, 1982). A considerable program of research has found that greater person centeredness is more effective, not just in theory, but in fact. Messages higher in person centeredness are evaluated as better at reducing emotional distress, and result in more positive evaluations for the support provider (e.g., Jones & Guerrero, in press; for reviews see Burleson, 1994; Burleson & Goldsmith, 1998).

Thus, studies indicating that women produce emotional support messages with a greater degree of person centeredness than do men uphold the claim that women are more skilled as providers of emotional support. This greater degree of skill has been broadly attributed to the differential socialization of the sexes (e.g., Kunkel & Burleson, 1998, 1999). However, little research has tested whether the consequences or correlates of this socialization explain sex differences in skill. The current study focuses on one potentially explanatory factor: self-efficacy.

SKILL, SELF-EFFICACY, AND SEX DIFFERENCES

The production of highly person-centered emotional support messages is a challenging communicative task. Highly person-centered messages reflect a more complex set of perceptions and intentions, pursue broader sets of interaction goals, and in an important sense, do more work than less person-centered messages (O'Keefe & Delia, 1982; Samter, Burleson, & Basden-Murphy, 1989). In order to produce these more sophisticated messages, support providers must more fully recognize and appreciate the feelings and perspectives of others, and utilize that information to produce messages that validate those feelings and perspectives (Burleson & Samter, 1985; see also Burleson & Caplan, 1998).

Considerable evidence indicates that, across life domains, challenging tasks are more easily accomplished when self-efficacy is high. Self-efficacy refers to one's belief in the personal capacity to accomplish a task (Schunk, 1995). Greater self-efficacy typically results in greater motivation to undertake a task, stronger and more persistent effort at the task, and higher levels of achievement (Gibbons & Weingart, 2001; Schunk, 1995). Thus, greater self-efficacy with respect to emotional support provision should promote greater skill by making support providers more likely to accept rather than resist the challenge of providing support, and encouraging them to exert the necessary cognitive and communicative effort.

Several kinds of evidence combine to suggest that women, on average, have greater self-efficacy in the domain of emotional support than do men. Part of this evidence comes from research indicating that women, across the life span, obtain greater experience at support provision than do men. For example, support recipients indicate that women provide more emotional support to them than do men (e.g., Akiyama et al., 1996), and women report being the provider of more support to distressed friends and spouses than men do (e.g., Antonucci & Akiyama, 1987; George, Carroll, Kersnick, & Calderon, 1998). Further, observational studies of children indicate that girls are more likely to undertake comforting tasks than are boys (Eisenberg, Murphy, & Shepard, 1997; Zahn-Waxler, Friedman, Cole, Mizuta, & Hiruma, 1996).

In complex forms of communication, as with most complex tasks, "practice makes perfect," or at least better (see the review by Greene, in press). The more opportunities one has to provide emotional support, the more familiar one becomes with the task, the more one is able to observe the (in)effectiveness of strategies, and the more feedback one can obtain from targets or observers. Therefore, in general, increased experience should lead to greater task familiarity and skill, which should in turn enhance the perception that one is capable of successfully performing the task. Thus, women's greater experience providing emotional support is one indication that they may have greater self-efficacy in this domain.

A second reason to believe that women may be more self-efficacious in the area of emotional support skill concerns sex-role stereotypes. Self-efficacy is known to be based, in part, on perceptions of one's personal qualities (Gibbons & Weingart, 2001; Schunk, 1995). Females are typically thought (by both males and females) to be more "expressive" or "feminine" than are males, possessing more of such characteristics as empathy, nurturance, and caring (e.g., Fabes & Martin, 1991; Spence & Buckner, 1995). In addition, highly person-centered emotional support messages are viewed by both sexes as more feminine than masculine in character (Kunkel & Burleson, 1999). Thus, compared to males, females may have greater self-efficacy at emotional support provision because they believe they have more of the necessary sex-typed qualities, or view the task as more suited to their sex-role.

Finally, one study to date provides direct empirical evidence of women's greater self-efficacy at emotional support tasks. In a single study directly examining women's self-efficacy, Clark (1993) found that women faced with a comforting task "anticipated greater effectiveness prior to constructing their messages, rated their performances higher following message production, and attributed greater ability to themselves than to men" (p. 565). In sum, the evidence that (a) women produce emotional support messages with greater person centeredness than do men, (b) women have greater self-efficacy in the domain of emotional support than do men, and (c) self-efficacy affects skill provided the basis for the following hypotheses:

H1: Women will produce emotional support messages with a greater degree of person centeredness than will men.

H2: Women will report greater self-efficacy at providing emotional support than will men.

H3: Self-efficacy will mediate the relationship between sex and the production of person-centered messages.

A MODERATING FACTOR: THE CHALLENGE OF TARGET RESPONSIBILITY FOR THE PROBLEM

As previously argued, self-efficacy is relevant to the skillful provision of emotional support because, in general, providing emotional support is a complex communicative task. However, an extension of this argument is that self-efficacy may have the strongest effect on skill when the task of providing emotional support is especially challenging. Although the successful performance of a moderately difficult task certainly requires motivation and effort, these factors should be even more critical when undertaking truly difficult tasks. Since motivation and effort are primary means through which self-efficacy impacts on skill (e.g., Schunk, 1995), it seems probable that the relationship between self-efficacy and skill will be stronger for tasks that are more challenging.

Many factors may influence the level of challenge posed by situations in which supportive communication is required. However, one relevant influence may be the support-seeker's level of responsibility (i.e., blame) for his or her problem. Higher levels of support-seeker responsibility are known to reduce helpers' and support providers' motivation to provide assistance (see review by Weiner, 1995). For example, recent research indicates that higher target responsibility for the problem is associated with lesser intention to provide emotional support (MacGeorge, 2001), and greater perceived appropriateness of less person-centered messages (Jones & Burleson, 1997). Aside from motivation, awareness of a support-seeker's responsibility may make the production of highly person-centered messages more cognitively effortful. Because any blaming, criticism, or derogation of the target will typically reduce the overall person centeredness of an emotional support message, producing a highly person-centered message requires that the support provider inhibit this type of content or determine how to convey it in a message that still validates the target's feelings and perspective.

Thus, the production of highly person-centered messages is likely to become more challenging for support providers as the support-seeker's responsibility for his or her problem increases. This, in turn, may allow a greater role for providers' self-efficacy in determining the skill with which emotional support is provided to high rather than low responsibility targets. On the basis of this reasoning, the following hypothesis was proposed:

H4: The association between self-efficacy and person centeredness will be greater when targets are high in responsibility for their problems than when they are low in responsibility.

Further, if the association between self-efficacy and skill is stronger under high responsibility conditions, this stronger association should allow self-efficacy to mediate more of the hypothesized effect of sex on skill. Accordingly, the following hypothesis was addressed:

H5: Self-efficacy will mediate more of the effect of sex on person centeredness when targets are high in responsibility for their problems than when they are low in responsibility.

METHOD

Procedure

Participants were students (293 males, 422 females, average age = 19.3 years) from communication classes at a large Midwestern university and two medium-sized Eastern universities. They participated on a voluntary basis at the beginning or end of a class period, or in sessions conducted outside of class. Each participant was given a packet with one of 12 hypothetical scenarios and a set of questionnaires. Each of the scenarios depicted a friend as distressed due to a personal problem and as responsible or not responsible for that problem. After reading a scenario, each participant responded to seven questionnaires, some of which were not pertinent to this study and will not be discussed further. The first of the three questionnaires used in this study asked participants to produce a message directed at the friend depicted in the scenario. The second asked participants to report their self-efficacy at the provision of support. The third contained a manipulation check for the responsibility manipulations. At the end of the questionnaire packet participants were also asked to indicate their age and sex.

Responsibility Manipulations

Target responsibility for the problem was manipulated through the set of hypothetical scenarios. These portrayed "close friends" as discussing upsetting problems in their lives with the participants, and as having high control over the origin of their problems (high responsibility) or low control over the origin of their problems (low responsibility). (1) In order to increase the generalizability of the findings, the scenarios also incorporated a situation-replication factor that could be treated as a random factor in the statistical analysis. Both high and low responsibility manipulations were presented within six situations that differed with regard to the specific problem being experienced by the friend. The six problems, chosen to be representative of college student experiences, were as follows: failing an audition, not receiving a raise at work, being rejected by a friend, having parents threaten the removal of a privilege, having damaged a borrowed car, and having a fraternity or sorority event that one planned go badly. An independent groups design was employed; thus, each participant read and responded to only one of the 12 scenarios. (2)

The success of the responsibility manipulations was assessed with three, five-point Likert-style items examining perceptions of the target's responsibility ("The problem being experienced by [friend's name] is his [her] fault," "[Friend's name] is responsible for causing this problem," and "[Friend's name] has only himself [herself] to blame for this problem".) The alpha reliability for the three items was .92; an index for perceived responsibility was created from the mean of the items. A 2 (high/low responsibility) X 6 (situation-replication) ANOVA was performed in which perceived responsibility was the dependent variable and the situation replication was treated as a random factor. The significant effect for responsibility [F(1, 5) = 336.62, p < .001, [[eta].sup.2] = .50] (3,4) indicated that friends depicted in high responsibility conditions were viewed as being more responsible for their problems than friends in low responsibility conditions (Ms = 3.91 and 2.11, respectively).

Measurement

Person-centered emotional support skill. Participants were instructed to "write what you would actually say" in response to the friends in the hypothetical scenarios. These responses were coded for person centeredness within the hierarchical system developed by Applegate (1980) and Burleson (1982). This system classifies emotional support messages at one of nine levels. Messages that deny the target's feelings by criticizing them, challenging their legitimacy, or telling the target how to feel or act are coded at levels 1, 2, and 3, respectively. Messages that implicitly recognize the target's feelings by attempting to distract the target, offer expressions of sympathy, or present non-feeling centered explanations of the situation are coded at levels 4, 5, and 6, respectively. Messages explicitly recognizing and legitimizing the target's feelings by helping the target to articulate them, elaborating reasons why the feelings might be felt, or assisting the target to see how the feelings fit in a broader context are coded at levels 7, 8, and 9, respectively. More detailed descriptions of this coding system are available in a variety of published sources (e.g., Applegate, 1980; Burleson, 1982).

Coding procedures were modeled on those detailed by Samter (1989). Each message was coded according to its level of person centeredness. Messages that appeared to represent more than one level were coded based on their dominant focus. In rare cases where the dominant focus was difficult to determine, messages were coded into the highest level represented in the message. After practice sessions employing convenience samples from approximately 5% of the data, coding reliability was assessed based on 125 messages selected randomly from this data set and a similar data set for which coding was conducted simultaneously. These 125 messages were coded independently by the first and third authors. The level of exact agreement was 70%, and the intraclass correlation between each rater's codes was .82. Coding disagreements in the reliability sample were resolved by the first author, and the remaining coding was split evenly between the first and third authors.

Self-efficacy. A six-item, Likert-style, self-report measure was developed to assess participants' self-efficacy at the provision of emotional support. Three items asked participants to rate their own efficacy: "I consider myself to be quite skilled at helping friends when they are upset about problems they are having," "I believe that I am quite good at providing support to friends who are distressed about negative events in their lives," "I think that I am quite effective at comforting friends when they are unhappy about bad things that have happened to them." Three additional items asked participants to indicate how they believed their friends would rate their efficacy: "My friends would say that I am a good source of comfort when they are unhappy about problems they are having," "My friends would say that I know how to support them when they are upset about negative events in their lives," "My friends would say that I know how to help them when bad things have happened to them."

A principle axis factor analysis with oblique rotation was conducted to assess the structure of the items for the scale. The factor analysis extracted one factor with an eigenvalue greater than one (4.20), accounting for 70% of the variance. All six items loaded on the single factor with loadings ranging from .70 to .85. A reliability analysis was then conducted to determine whether any of the items loading less strongly on the single factor should be dropped to improve reliability. The alpha reliability for the six items was .91, and would not have been improved by the removal of any items. Therefore, a self-efficacy scale was created from the mean of the six items.

RESULTS

Hypotheses 1 and 2 predicted that women would produce emotional support messages with a greater degree of person centeredness and that women would report greater self-efficacy at providing emotional support than men. Independent groups t-tests indicated that women's messages exhibited a greater degree of person centeredness (M = 4.31, sd = 1.50) than men's (M = 3.63, sd = 1.33), t(713) = - 6.23, p < .001, [[eta].sup.2] = .05, and that women reported a higher level of self-efficacy at providing emotional support (M = 4.30, sd = 0.56) than did men (M = 3.84, sd = 0.64), t (713) = - 10.35, p < .001, [[eta].sup.2] = .13. Thus, both Hypothesis 1 and Hypothesis 2 were supported.

Hypothesis 3 predicted that the effect of sex on person centeredness would be mediated by self-efficacy. To assess this hypothesis, it was necessary that three preliminary conditions be met: (1) sex had to be associated with person centeredness, (2) sex had to be associated with self-efficacy, and (3) self-efficacy had to be associated with person centeredness. The support obtained for Hypotheses 1 and 2 indicated that the first two conditions were met. A Pearson correlation was calculated to examine the association between self-efficacy and person centeredness; person centeredness was found to increase as self-efficacy increased, r = .13, p < .005. Therefore, the preliminary conditions for a mediational analysis were met.

However, Hypothesis 5 predicted that the mediating effect of self-efficacy would be moderated by the target's responsibility for the problem. Testing this moderating capacity would require separate mediational analyses for the high and low responsibility conditions. To determine whether preliminary conditions for these separate analyses were met, a z-test was conducted to determine whether the strength of association between self-efficacy and person centeredness differed between the two responsibility conditions (a test of Hypothesis 4). The z-test indicated that the correlation between self-efficacy and person centeredness under low responsibility conditions (r = .11) was not significantly different from the correlation under high responsibility conditions (r = .16), z (715) = - 0.66, p > .05. Thus, because Hypothesis 4 was not supported, there was also no support for Hypothesis 5 and, correspondingly, no need to conduct separate mediation analyses for high and low responsibility conditions.

Accordingly, Hypothesis 3 was tested with a set of two hierarchical regression analyses with person centeredness as the dependent variable (see Table 1). In the first equation, sex was entered at the initial step and self-efficacy was entered second, whereas the second equation reversed the order of entry for these two variables. Sex and self-efficacy collectively explained a small portion of the variance in person centeredness, [R.sup.2] = .054, F(2,712) = 20.23, p < .001. When entered in the regression equation first, sex explained approximately 5% of the variance ([R.sup.2] = .052); entering self-efficacy did not significantly increase the explained variance ([R.sup.2] = .054). However, when self-efficacy was entered first, it accounted for approximately 2% of the variance ([R.sup.2] = .016). Thus, approximately 30% of the sex-related variance in person centeredness was mediated by self-efficacy.

DISCUSSION

The central purpose of this study was to examine relationships between sex, self-efficacy in the domain of emotional support provision, and skill in this domain. In particular, the research was motivated by the interest in determining whether self-efficacy mediated any of the effect of sex on person centeredness. The current study also examined whether this mediating capacity of self-efficacy was moderated by the support-seeker's responsibility for the problem, a factor thought to affect the level of challenge inherent in providing emotional support.

The present research adds to the evidence that women, on average, produce somewhat more skillful emotional support messages than do men. Consistent with four other studies (Borden, 1979; Burleson, 1982; Hale, Tighe, & Mongeau, 1997; Samter, this issue), the messages of women in the current research exhibited a higher level of person centeredness than those of men. However, it is important to note that the sex difference was not large ([[eta].sup.2] = .05), suggesting that the skill difference between men and women should not be overemphasized. Certainly this size of difference provides no evidence to support claims that men and women differ radically in their typical means of providing emotional support (e.g., the "Different Cultures Perspective"; Wood, 1997).

The current study also found that women perceived greater self-efficacy in the domain of emotional support than did men. This finding is consistent with research indicating that women, on average, have more experience providing emotional support than do men (e.g., Akiyama et al., 1996), and with studies indicating that providing person-centered emotional support is viewed as a more feminine than masculine activity (e.g., Kunkel & Burleson, 1999). The finding also constitutes a conceptual replication of Clark (1993), who found that women perceived themselves as more effective at comforting others. One important direction for future research is exploring the origins of self-efficacy, determining its relationship to experience, sex-role stereotypes, and other factors. Since both Clark's study and the current research were conducted with college students, subsequent studies might examine self-efficacy in an older sample (where marriage and fatherhood may have strengthened some men's confidence in caring and nurturing others), or in a population with less education (where sex-role stereotypes may be a more powerful influence).

Consistent with findings in other skill domains, in the domain of emotional support self-efficacy was found to have a positive relationship with skill. However, the current study also found the association to be a weak one. This is attributable, at least in part, to the high average and limited range of self-efficacy scores. The average for men and women combined was 4.11 on a 5-point scale, and the standard deviation was 0.63, indicating that most of the study's participants reported a relatively high level of self-efficacy. The ability to make one's friends feel better is a highly valued communication skill (e.g., Burleson et al., 1996), so self-efficacy scores may have been inflated because it is socially desirable to report that one possesses this skill. In addition, the measurement of self-efficacy after participants had produced their messages may have led them to feel more efficacious than they normally would (especially since there was none of the negative feedback that could have occurred in an actual encounter). Future research should address these methodological limitations of the current study.

Although the sex difference in skill was relatively small, and the association between self-efficacy and skill was weak, the current study nonetheless supported the contention that some of the sex difference in emotional support skill is explained by sex differences in self-efficacy. One direction for future research is examining whether there are conditions under which men may experience self-efficacy more equal to that of women and thus exhibit less of a skill deficit relative to women. In an intriguing study, Hale et al. (1997) found that the level of person centeredness in men's and women's messages was much more similar when they were responding to targets experiencing "major" (serious, unusual) problems than when they were responding to targets experiencing "daily" (less serious, typical) problems. Faced with the task of responding to more significant situations, men produced messages that were nearly as person-centered as those of women. Hale et al. suggest that serious situations may permit men to feel that providing emotional support is "heroic" (see also Eagly & Crowley, 1986), motivating them to exert "manly" effort to reduce distress. An alternative but complimentary view is that increased seriousness of the situation reduces the perceived femininity of the task, which in turn enhances men's self-efficacy (and skill).

Finally, the present study found no evidence that the mediating role of self-efficacy was moderated by the target's responsibility for the problem, a factor conceptualized as affecting the challenge of providing highly person-centered emotional support. There are several potential explanations for the finding that responsibility had no moderating effect, each suggesting possible directions for further study. One explanation lies in the weak association between self-efficacy and skill. If the improvements in measurement techniques previously discussed were to detect a stronger association, moderating influences might also be more evident. However, an alternative explanation is that high target responsibility does not increase the challenge associated with providing emotional support, or increases it for some but not others. Considerable research indicates that individuals vary in their capacity to successfully process and respond to competing or contradictory social information (see Burleson & Caplan, 1998). This suggests that some support providers will be able to integrate the information that a target is a distressed and needy friend with the information that he or she is to blame for the problem. These individuals will probably also perceive the challenge of producing a highly person-centered message in these situations. Others, however, will tend to select one piece of information on which to focus. These individuals may perceive that high target responsibility absolves them of the need for much sensitivity to the target's feelings and perspectives, thus making it easier (or at least no more difficult) to produce a message (albeit a less person-centered one!). Future research should examine whether high target responsibility is typically viewed as a challenge, as well as investigating other factors (e.g., target's level of distress, support provider's familiarity with the problem) that may increase or reduce challenge and affect the relationship between self-efficacy and skill.

TABLE 1
Regression Analyses Assessing the Mediating Effect of Self-Efficacy
on the Association Between Sex and Person Centeredness

                    Sex Entered 1st          Sex Entered 2nd

                  [R.sup.2]               [R.sup.2]
Predictor          Change     F Change     Change     F Change

Sex                 .052      38.85 ***     .038      28.74 ***
Self-Efficacy       .002       1.57(ns)     .016      11.27 **

NOTE:N = 714. *** = p < .001, ** = p < .01.

ENDNOTES

(1.) To test hypotheses beyond the scope of the current study, the scenarios were designed to manipulate perceptions not only of responsibility, but also of stability (the consistency with which the support-seeker was responsible for problems in his or her life), and effort (the extent to which a support-seeker had attempted to help him or herself). However, preliminary analyses indicated that stability and effort had inconsequential effects on the variables examined in this research, so analyses were conducted collapsing across these factors.

(2.) The complete scenarios are available from the first author.

(3.) Because the situation-replication factor was treated as a random factor in this analysis, the estimate of variance due to random factors included not only the variance owing to individual responses (as in a standard F-test), but also variance deriving from the interaction of the research factors and the situation-replication factor. For a fuller discussion of the rationale for this form of the F-test, see Jackson (1992), and Jackson & Brashers (1994).

(4.) A partial [[eta].sup.2] indicates the proportion of the total variance explained by a factor after other factors and interactions has been partialled out (e.g., Cohen, 1988; Tabachnick & Fidell, 1996). Therefore, in a standard fixed-effect model, the partial [[eta].sup.2] is the ratio of the treatment sum of squares to the treatment sum of squares plus the error (within-groups) sum of squares. However, in a mixed model (fixed and random effects) such as that employed here, the interactions between fixed and random factors also constitute "error" against which the significance of any fixed effect must be tested (e.g., Jackson, 1992; Jackson & Brashers, 1994). Therefore, this partial [[eta].sup.2]s is the ratio of the treatment sum of squares to the sum of (1) the treatment sum of squares, (2) the error (within-groups) sum of squares, and (3) the "error" sums of squares resulting from the interactions between the fixed factors and the random (situation-replication) factor.

REFERENCES

Akiyama, H., Elliott, K., & Antonucci, T. C. (1996). Same-sex and cross-sex relationships. Journals of Gerontology Series B-Psychological Sciences and Social Sciences, 51B(6), P374-P382.

Antonucci, T. C., & Akiyama, H. (1987). An examination of sex differences in social support among older men and women. Sex Roles, 11/12, 737-749.

Applegate, J. L. (1980). Adaptive communication in educational contexts: A study of teachers' communicative strategies. Communication Education, 29, 158-170.

Borden, A. (1979). An investigation of the relationships among indices of social cognition, motivation, and communicative performance. Unpublished doctoral dissertation, University of Illinois, Champaign, IL.

Burleson, B. R. (1982). The development of comforting communication skills in childhood and adolescence. Child Development, 53, 1578-1588.

Burleson, B. R. (1994). Comforting messages: Significance, approaches, and effects. In B. R. Burleson, T. L. Albrecht, & I. G. Samson (Eds.), Communication of social support.. Messages, interactions, relationships, and community (pp. 3-28). Thousand Oaks, CA: Sage.

Burleson, B. R., & Caplan, S. E. (1998). Cognitive complexity. In J. C. McCroskey, J. A. Daly, M. M. Martin, & M. J. Beatty (Eds.), Communication and personality: Trait perspectives (pp. 233-286). Cresskill, NJ: Hampton Press.

Burleson, B. R., & Goldsmith, D. J. (1998). How the comforting process works: Alleviating emotional distress through conversationally induced reappraisals. In P. A. Anderson & L. K. Guerrero (Eds.), Handbook of communication and emotion: Research, theory, applications, and contexts (pp. 245-280). San Diego, CA: Academic Press.

Burleson, B. R., Kunkel, A. W., Samter, W., & Werking, K. J. (1996). Men's and women's evaluations of communication skills in personal relationships: When sex differences make a difference--and when they don't. Journal of Social and Personal Relationships, 13, 201-224.

Burleson, B. R., & Samter, W. (1985). Consistencies in theoretical and naive evaluations of comforting messages. Communication Monographs, 52, 103-123.

Clark, R. A. (1993). Men's and women's self-confidence in persuasive, comforting, and justificatory communicative tasks. Sex Roles, 28, 553-567.

Eagly, A. H., & Crowley, M. (1986). Gender and helping behavior: A meta-analytic review of the social psychological literature. Psychological Bulletin, 100, 283-308.

Eisenberg, N., Murphy, B. C., & Shepard, S. (1997). The development of empathic accuracy. In W. Ickes (Ed.), Empathic accuracy (pp. 73-116). New York: Guilford.

Fabes, R. A., & Martin, C. L. (1991). Gender and age stereotypes of emotionality. Personality and Social Psychology Bulletin, 17, 532-540.

George, D. M., Carroll, P., Kersnick, R., & Calderon, K. (1998). Gender-related patterns of helping among friends. Psychology of Women Quarterly, 22, 685-704.

Gibbons, D. E., & Weingart, L. R. (2001). Can I do it? Will I try? Personal efficacy, assigned goals, and performance norms as motivators of individual performance. Journal of Applied Social Psychology, 31, 624-648.

Goldsmith, D. J., & Dun, S. A. (1997). Sex differences and similarities in the communication of social support. Journal of Social and Personal Relationships, 14, 317-337.

Greene, J. O. (in press). Models of communication skill acquisition and development. In J. O. Greene & B. R. Burleson (Eds.), Handbook of communication and social interaction skills. Mahwah, NJ: Erlbaum.

Hale, J. L., Tighe, M. R., & Mongeau, P. A. (1997). Effects of event type and sex on comforting messages. Communication Research Reports, 14, 214-220.

Jackson, S. (1992). Message effects research. New York: Guilford.

Jackson, S., & Brashers, D. E. (1994). Random factors in ANOVA (Vol. 98). Thousand Oaks, CA: Sage.

Jones, S. M., & Burleson, B. R. (1997). The impact of situational variables on helpers' perceptions of comforting messages: An attributional analysis. Communication Research, 24, 530-555.

Jones, S. M., & Guerrero, L. A. (in press). The effects of nonverbal immediacy and verbal person centeredness in the emotional support process. Human Communication Research.

Kunkel, A. W., & Burleson, B. R. (1998). Social support and the emotional lives of men and women: An assessment of the different cultures perspective. In D. J. Canary & K. Dindia (Eds.), Sex differences and similarities in communication (pp. 101-125). Mahwah, NJ: Erlbaum.

Kunkel, A. W., & Burleson, B. R. (1999). Assessing explanations for sex differences in emotional support: A test of the different cultures and skill specialization accounts. Human Communication Research, 25, 307-340.

MacGeorge, E. L. (2001). Support providers' interaction goals: The influence of attributions and emotions. Communication Monographs, 68, 72-97.

O'Keefe, B.J., & Delia, J. G. (1982). Impression formation and message production. In M. Roloff & C. Berger (Eds.), Social cognition and communication (pp. 33-72). Beverly Hills, CA: Sage.

Samter, W. (1989). Communication skills predictive of interpersonal acceptance in a group living situation: A sociometric study. Unpublished doctoral dissertation, Purdue University, West Lafayette, IN.

Samter, W. (this issue). How gender and cognitive complexity influence the provision of emotional support: A study of indirect effects. Communication Reports.

Samter, W., Burleson, B. R., & Basden-Murphy, L. (1989). Behavioral complexity is in the eye of the beholder: Effects of cognitive complexity and message complexity on impressions of the source of comforting messages. Human Communication Research, 15, 612-629.

Schunk, D. H. (1995). Self-efficacy, motivation, and performance. Journal of Applied Sport Psychology, 7, 112-137.

Spence, J. T., & Buckner, C. (1995). Masculinity and femininity: Defining the undefinable. In P. J. Kalbfleisch & M. J. Cody (Eds.), Gender, power, and communication in human relationships (pp. 105-140). Hillsdale, NJ: Erlbaum.

Weiner, B. (1995). Judgments of responsibility: A foundation for a theory of social conduct. New York: Guilford Press.

Wood, J. T. (1997). Gendered lives: Communication, gender, and culture (2nd ed.). Belmont, CA: Wadsworth.

Zahn-Waxler, C., Friedman, R. J., Cole, P. M., Mizuta, I., & Hiruma, N. (1996). Japanese and United States preschool children's responses to conflict and distress. Child Development, 67, 2462-2477.

Erina L. MacGeorge (Ph.D., University of Illinois, 1999) is an Assistant Professor of Communication at George Washington University (2130 H. St., NW, Washington, DC, 20052; elm@gwu.edu). Ruth Anne Clark (Ph.D., University of Wisconsin, 1964) is a Professor of Speech Communication at the University of Illinois. Seth J. Gillihan (M.A., George Washington University, 2001) is a doctoral student in Psychology at the University of Pennsylvania. The authors gratefully acknowledge assistance in data collection from Wendy Samter and Charlie Pavitt, and in data collection and entry from student research assistants Sara Bakker, Jessica Bureau, Beth Domenick, Lindsey Ferris, Erin Hagerty, Jennifer Hopkins, Gretchen Lehman, Kim McGraw, Nicole Mike, Kate Stephens, and Cathy Southammakosoae. The authors also wish to thank the anonymous reviewers for very helpful criticism. An earlier version of this article was presented at the International Communication Association convention, May 2001, Washington, D.C. Address correspondence concerning this article to the first author.