COMMUNICATION ARTICLE 10.1177/0093650204263436 Jones • Emotional SupportRESEARCH and Emotional • Change June SUSANNE JONES
2004
Putting the Person Into Person-Centered and Immediate Emotional Support Emotional Change and Perceived Helper Competence as Outcomes of Comforting in Helping Situations The goal of this study was to assess the effects of comforting messages using both experienced emotional change by the help recipient and help recipients’ judgments of the helper’s competence. A hypothesized path model proposed relationships between two comforting message factors (verbal person centeredness and nonverbal immediacy) and the two outcomes (emotional change and evaluations of helper competence). Data were generated from an experiment in which 258 participants disclosed a mildly upsetting event to a confederate trained to display different levels of person centeredness and nonverbal immediacy. Participants subsequently completed a set of instruments tapping both their emotional state and the perceived competence of the helper on four dimensions (help motivation, supportiveness, conversation management, and expressiveness). A modified model with two added paths fit the data well and revealed not only that people felt significantly better but also that they viewed the helper as more supportive and caring after having received person-centered comforting messages. Nonverbal immediacy only influenced evaluations of perceived helper competence, such that immediate helpers were perceived as more competent than nonimmediate helpers. Keywords: affective improvement; competence; emotional support; nonverbal immediacy; person centeredness
Emotional support from friends and family is an important, if not the most important, resource people rely on in times of stress and emotional hurt (Burleson & MacGeorge, 2002; Cutrona & Russell, 1990). Caring and compasCOMMUNICATION RESEARCH, Vol. 31 No. 3, June 2004 338-360 DOI: 10.1177/0093650204263436 © 2004 Sage Publications
338
Jones • Emotional Support and Emotional Change sionate emotional support helps people cope with upsetting events and hurtful emotions (Stroebe & Stroebe, 1996), makes for more satisfying relationships (Barbee, Rowatt, & Cunningham, 1998; Cutrona & Russell, 1990), and has curative psychological and physiological benefits (Gump, Polk, Kamarck, & Shiffman, 2001). Insufficient emotional support is particularly predictive of relationship dissatisfaction (Wan, Jaccard, & Ramey, 1996) and may even be harmful to one’s health (Rook & Underwood, 2000; Strittmatter & Bengel, 1996). In the context of communication, Burleson (2003) defines emotional support as consisting of “specific lines of communicative behavior enacted by one party with the intent of helping another cope effectively with emotional distress” (p. 552). The positive effects of helpful support (and the negative effects of unhelpful support) have led communication researchers to identify those message properties that make for more or less effective emotional support. Most studies seeking to identify relevant comforting message properties have largely relied on two paradigms: the naturalistic paradigm and the message perception paradigm (Burleson & MacGeorge, 2002). Researchers using a naturalistic framework commonly develop descriptive typologies of support behaviors based on retrospective self-reports of people who are coping with some acute or chronic stress (e.g., Dakof & Taylor, 1992; Hays, Magee, & Chauncey, 1994). Research using this paradigm has generated a rich array of effective and ineffective emotional support behaviors people experience in everyday life. For instance, in their study of perceived social support among cancer patients, Dakof and Taylor (1992) classified behaviors, such as being physically present, showing empathy and concern, and expressing acceptance as helpful esteem or emotional support behaviors. Unhelpful behaviors included critical responses or minimizations. Although these typologies are highly ecologically valid, they are somewhat limited because people’s recollections of supportive behaviors are subject to change, imprecise, and colored by contextual interpretations (Goldsmith, 1992). Furthermore, detecting theoretically important conceptual properties of comforting messages is difficult because the support typologies are often too diverse and contextually specific (Burleson, 2003). Researchers using a message perception paradigm have relied on more deductive methodologies and have commonly presented participants with a set of emotional support messages that instantiate some feature of theoretical interest. These comforting messages are then embedded in hypothetical dialogues or scenarios, and participants are asked to evaluate the messages on a set of criteria, such as helpfulness, effectiveness, appropriateness, or sensitivity. For example, Burleson and Samter (1985, Study 2) asked participants to evaluate the helpfulness, sensitivity, appropriateness, and effective-
339
COMMUNICATION RESEARCH • June 2004 ness of comforting messages that varied in person centeredness, a construct that captures the degree to which a comforter verbally expresses empathy and validates the distressed person’s feelings (see also Goldsmith & MacGeorge, 2000; Jones & Burleson, 1997). However, these message evaluations lack ecological validity because the elicited responses reflect evaluations of comforting messages participants read with respect to hypothetical situations, and as Burleson and MacGeorge (2002) note, “There is obviously a difference between actually experiencing a message when upset and making judgments about messages directed at hypothetical others” (p. 391). In an attempt to overcome the limitations of research reflecting a naturalistic or a message perception perspective, Jones and Guerrero (2001) relied on experimental procedures and examined manipulated variations of comforting messages in actual emotional support situations. In their experiment, participants disclosed an emotionally upsetting event to a confederate who was trained to respond with comforting messages that displayed different levels of verbal person centeredness. In addition, confederates responded with cues that reflected various levels of nonverbal immediacy, a concept that reflects empathy, interpersonal warmth, and psychological closeness and that encompasses behaviors such as smiling, eye gaze, and direct body orientation. After the conversation, participants evaluated the helpfulness, sensitivity, appropriateness, and effectiveness of more or less person-centered and immediate comforting behaviors exhibited by the confederate. In line with related studies, participants evaluated messages high in verbal person centeredness and nonverbal immediacy as most appropriate, useful, helpful, and effective (for a review see Burleson & MacGeorge, 2002).
Goals of the Current Project Jones and Guerrero’s (2001) experiment targeted comforting message evaluations in face-to-face conversations and generated important information about whether messages that instantiated two theoretically important comforting-message features were perceived as qualitatively better. However, Burleson’s (2003) definition of emotional support points to the importance of emotional change: Emotional support messages primarily aim to help upset persons work through and cope with difficult emotions. Indeed, making people feel better is the primary intention of most help providers. Therefore, the current study takes an important step beyond the Jones and Guerrero experiment and examines whether more or less helpful verbal and nonverbal emotional support messages actually improve recipients’ emotional states and shape their perceptions of the helper’s competence. To that end, the current study focuses most squarely on the recipient’s perspective
340
Jones • Emotional Support and Emotional Change and is among the first studies to assess two important recipient outcomes of emotional support, namely actual emotional impact and perceptions of helper competence. Thus, the present study aims to establish what kinds of messages both help emotionally upset people feel better and also make for more or less satisfying social interactions. It seems reasonable to use recipients’ reported emotional change and perceived helper competence as criteria to assess the effects of comforting messages that vary in person centeredness and nonverbal immediacy. Burleson (1994, 2003) identified several message outcomes relevant for assessing emotional support, including both immediate instrumental effects (how well messages improve emotional states) and immediate relational effects (how well recipients judge the helper). Recipient reports of experienced emotional outcomes reflect immediate instrumental outcomes that tell us whether more or less effective comforting messages actually alleviate recipient’s sadness and upset. Judgments of message sources tap the competence impressions recipients have of the helper and reflect the immediate relational outcomes of a specific emotional support event. Competence judgments tell us about the impressions recipients have of the helper and provide information about whether the recipient perceived the helper as caring and supportive.
Properties of Comforting Messages Person centeredness. Comforting messages have frequently been evaluated for the extent to which they vary in levels of person centeredness. Person centeredness reflects an awareness of and adaptation to the subjective, affective, and relational aspects of the interactants and the communicative contexts (Burleson, 2003, p. 395). Applegate (1980) and Burleson (1982) developed a nine-level hierarchical coding scheme that empirically assesses personcentered (PC) qualities of comforting messages. Low PC messages implicitly or explicitly deny the feelings of the upset person, whereas moderately PC messages implicitly recognize distressed feelings of the person and express sympathy. Highly PC comforting messages explicitly acknowledge, contextualize, and elaborate the feelings of the distressed person. Numerous message perception studies indicate that people evaluate messages that display high levels of person centeredness as more appropriate, sensitive, effective, and helpful than messages that display lower levels of person centeredness (Burleson & Samter, 1985; Jones & Burleson, 1997; for a detailed summary see Burleson, 2003; Burleson & MacGeorge, 2002). Nonverbal immediacy. Impressive empirical evidence points to the important role of nonverbal immediacy (NI) in the comforting process (Dolin &
341
COMMUNICATION RESEARCH • June 2004 Booth-Butterfield, 1993; Lewis, Derlega, Shankar, Cochard, & Finkel, 1997; Winstead, Derlega, Lewis, Sanchez-Hucles, & Clarke, 1992). NI cues fulfill three functions in the emotional support process (P. A. Andersen, 1999; Patterson, 1995). First, NI cues often convey positive affect and liking (Wiener & Mehrabian, 1968). Nonimmediacy cues, in contrast, can convey negative affect and dislike because a person is exhibiting avoidance rather than approach. Second, NI cues stimulate physiological arousal (P. A. Andersen, Guerrero, Buller, & Jorgensen, 1998; Burgoon & Aho, 1982). Increased physiological arousal is important because the interpretation of arousal is connected to the experience of emotions, such as warmth, care, love, and liking. Finally, NI cues communicate interpersonal warmth and closeness by creating a sense of psychological connection and relational intimacy (P. A. Andersen, 1989; Mehrabian, 1971). NI behaviors should influence the emotional support process for at least two reasons. First, comforting behaviors consist of many emotion cues, such as smiling and crying, that are communicated primarily through nonverbal channels (P. A. Andersen & Guerrero, 1998; Noller, 1984). Second, because NI cues increase both physical and psychological closeness, these cues might help a distressed person feel connected to another person. People often report that the general presence and care of the comforter is an important aspect of emotional support (Dakof & Taylor, 1990).
Outcomes of Comforting Messages Immediate instrumental outcomes: Affective improvement. Surprisingly little empirical evidence documents the immediate emotional impact of comforting messages. Highly PC and NI behaviors should make people feel better for various reasons. First, these comforting behaviors have consistently been evaluated as more sensitive and helpful than comforting messages that deny or ignore the feelings of the distressed person (Burleson & Samter, 1985; Jones & Burleson, 1997; Samter, Burleson, & Murphy, 1987). Second, these support behaviors might facilitate the emotional change process because they explicitly encourage people to disclose their difficult feelings. Several lines of research (e.g., cognitive appraisal theory, inhibition-confrontation theory) suggest that the disclosure of emotion aids in the coping process and has psychosomatic benefits (Lazarus & Folkman, 1984; Niederhoffer & Pennebaker, 2002). For instance, in a series of studies, Pennebaker and colleagues have shown that unlike low disclosers (or inhibitors), high disclosers experience a significant drop in heart rate and skin conductance levels immediately after having disclosed a traumatic event (Pennebaker & Beall, 1986; Pennebaker, Hughes, & O’Heeron, 1987; Pennebaker & O’Heeron, 1984; for a
342
Jones • Emotional Support and Emotional Change review see Niederhoffer & Pennebaker, 2002). Drops in heart rate and skin conductance are associated with reduced stress and arousal levels and might also indicate improved affective states. Finally, PC and NI comforting messages create a supportive, discursive environment within which distressed people feel safe to talk about their feelings without negative repercussions (Burleson & Goldsmith, 1998). Immediate relational outcome: Perceived helper competence. Competence evaluations reflect impressions that recipients have of the helper after a specific comforting conversation and may have important relational consequences (Segrin, 1999; Spitzberg, 2003; Spitzberg & Cupach, 2002). Samter et al. (1987), for instance, found that PC comforters were better liked and were viewed as more attractive than non-PC comforters. Four competence dimensions might be particularly salient in the emotional support context: help motivation, supportiveness, expressiveness, and conversation management. A key component of emotional support is a helper’s initial motivation to help. Because emotional support can be potentially face threatening to the recipient (Goldsmith, 1992), the comforter’s nonverbal and verbal messages should reflect a willingness to help. At times, helpers might be unmotivated to help and might express this with dismissing or even denigrating remarks (Lehman & Hemphill, 1990). Supportiveness is a second important competence dimension in emotional support situations because this reflects comforters’ ability to express positive regard, empathy, understanding, and sensitivity— all crucial factors when providing help. Expressiveness is a third competence dimension and shows comforters’ ability to effectively use empathic facial animations and vocal affect while offering help. These behaviors are important because they indicate the degree to which comforters are perceived to be involved in the conversation. The ability to manage conversations is a final important competence component. Helpers might be viewed as competent conversation managers if they avoid interruptions and lengthy pauses, provide appropriate conversational acknowledgements (e.g., “Uh-huh”), and offer encouragements that allow recipients to elaborate on their story (Spitzberg & Cupach, 1984, 2002; Spitzberg & Hecht, 1984). Highly PC and Ni comforting messages are likely to be viewed as competent because they convey genuine and honest help intentions. When comforters use this kind of emotional support, they indicate both that they are paying attention and also that they are willing to offer emotional support. Finally, PC and NI comforting messages convey high levels of sensitivity to the needs and feelings of the upset person. PC and NI comforting messages show that the comforters are listening to the distressed person and are taking her or his concerns seriously. Non-PC and nonimmediate comforters, on the other hand,
343
COMMUNICATION RESEARCH • June 2004
Expressiveness
Supportiveness Nonverbal Immediacy
Helper Competence
Help Motivation
Person Centeredness
Figure 1:
Affective Improvement
Conversational Management
Conceptual Path Model for Perceived Helper Competence and Reported Affective Improvement as a Function of Helpers’ Emotional Support That Varies in Person-Centeredness and Nonverbal Immediacy.
express a lack of concern, and they might be perceived as disapproving and uncaring. In addition, recipients might feel belittled, irritated, or less capable of dealing with their difficult emotions.
Model Figure 1 presents the conceptual model tested in the current study. As this figure illustrates, affective improvement is an observed variable that was directly measured, whereas helper competence is an unobserved (latent) construct that was assessed with four indicators (help motivation, supportiveness, expressiveness, and conversation management). The relationship between the latent construct and its indicators is commonly referred to as the measurement model and is tested with factor analytic procedures to determine whether the latent construct is represented well by its indicators. The relationships between the exogenous variables and the dependent variables (i.e., helper competence and affective improvement) are referred to as the structural equation model that is tested with multiple regression procedures. The present structural equation model hypothesized that helpers who provide emotional support that is more PC and NI will be viewed as more competent than helpers who provide less PC and less NI support. Similarly, the model predicts that recipients who receive more PC and NI support will report feeling better than people who receive less PC and less NI support. The experimental design of the study ensured the independence of message person centeredness and NI. However, it was expected that recipients’ reported affective improvement scores would covary with perceived helper scores.
344
Jones • Emotional Support and Emotional Change
Method Participants A total of 264 students from upper-division communication classes at a large university in the southwestern United States participated in this study for extra credit or research experience points. Five students were eliminated from the study because they recognized the nature of the study, and one student withdrew from the study for religious reasons. The average age of the participants was 21.7 years (range = 18 to 49 years). The majority of the sample consisted of White or European Americans (n = 210), but it also included Mexican Americans, Latinos/Latinas, or Hispanics (n = 21); Asian or Asian Americans, (n = 12); and African Americans (n = 7). Eight participants belonged to other ethnic groups.
Research Design and Procedure The study was based on a 3 (PC: high, moderate, and low) × 3 (NI: high, moderate, and low) design. A brief summary of the experimental setup is provided below; the specifics can be obtained from Jones and Guerrero (2001). The experimental setup required participants to identify a recent emotionally upsetting event that they were comfortable talking about with a stranger in a 5-min conversation. After the conversation, participants completed a set of scales, which included items pertaining to their emotional improvement and their judgments of the helper’s competence. The experiment utilized two female and two male confederates. All confederates were trained to enact each of nine possible combinations of person centeredness and NI. Prior to each interaction, confederates drew a slip from an envelope to determine which condition they would enact. Confederates displayed their randomly assigned comforting condition as soon as participants arrived at the test site.
Manipulations of Person Centeredness and Nonverbal Immediacy With respect to the moderate conditions, past research indicates that most people are both moderately immediate in their everyday interactions with others (Burgoon, Stern, & Dillman, 1995) and use simple expressions of condolence (“Gee, I’m sorry.”) as the most frequent form of emotional support (Burleson, 2002). Therefore, confederates were told to act as they would in a
345
COMMUNICATION RESEARCH • June 2004 typical interaction with a stranger and to use moderately PC messages in the form of expressions of condolence (e.g., “I’m sorry to hear that.”) or questions to clarify the details of the distressing event (e.g., “How well did you know her before that happened?”). In the high conditions, confederates were told to increase their verbal and nonverbal comforting expressions dramatically from what they would do in a typical interaction. In the high PC conditions, confederates were trained to focus their verbal attention on the emotions expressed by the participants. For example, confederates were trained to use comforting statements that expressed empathy (e.g., “I understand. I feel so bad for you.”), encourage participants to talk about their feelings (e.g., “Man, how are you feeling right now?”), or express acceptance of the other’s feelings (e.g., “I don’t blame you for feeling that way.”). In the high NI conditions, confederates were instructed to lean forward or move closer to the participant (but to avoid touch), to orient their body positions completely toward the participants, to increase eye contact to approximately 80% to 90% of the time, to smile a lot when appropriate, and to put “lots of warmth” in their voices. To ensure that confederates’ high levels of NI would be interpreted as positive (i.e., expressing warmth, liking, care), confederates were trained particularly to display high levels of positive affect, facial animation, and vocal warmth. In the low conditions, confederates were instructed to decrease their verbal and nonverbal comforting behaviors markedly from what they would typically do in interactions with strangers. They were told to reduce eye contact to approximately 20% to 30% of the time, to avoid smiling, and to look around the room. Confederates were also trained to express either a lack of interest or boredom and to appear distracted or tired during the conversation. In the low PC conditions, confederates used statements that encouraged the emotionally distressed person to forget about her or his feelings (e.g., “I think you ought to get over it.”) or that minimized feelings (e.g., “Oh, come on, it’s not the end of the world. It can’t really be that bad. You’ll get over it.”). Confederates were also encouraged to switch the conversation to an unrelated topic or to begin talking about personal concerns (e.g., “Guess what happened to me?”).
Coding and Manipulation Checks for Person Centeredness and Nonverbal Immediacy Three coders assessed confederate PC and NI cues from 84% (n = 216) of the videotaped conversations. Two primary coders rated PC and NI levels for all 216 confederates, whereas the secondary coder rated PC cues for 118 confed-
346
Jones • Emotional Support and Emotional Change erates and NI cues of 98 confederates. The general findings of the manipulation check are reported below; detailed findings are reported by Jones and Guerrero (2001). Person centeredness. The PC scale consisted of five 7-point semantic differential scales identifying fundamental features of person centeredness (i.e., self-centered vs. other-centered, invalidates vs. validates, judges vs. empathizes, disregards vs. acknowledges, and unconcerned vs. concerned). Interitem reliability was α = .98. Interrater reliability (based on Ebel’s intraclass r) was .95. The PC manipulation was analyzed with a 3 (person centeredness: high, moderate, low) × 3 (NI: high, moderate, low) analysis of variance (ANOVA), with coder ratings of person centeredness as the dependent measure. The ANOVA detected a significant main effect for person centeredness, F(2, 207) = 696.04, p < .001, η2 = .87, and a follow-up contrast revealed a significant linear effect for person centeredness, F(1, 213) = 1,416.41, p < .001, η2 = .86. As expected, confederates in the high PC condition were rated as most PC (M = 6.36, SD = 1.11), followed by confederates in the moderate PC condition (M = 4.04, SD = .18). Confederates in the low PC condition were rated lowest (M = 1.28, SD = .83). Nonverbal immediacy. NI was measured with a modified version of J. F. Andersen, Andersen, and Jensen’s (1979) NI instrument. Interitem reliability was α = .91. Interrater reliability (based on Ebel’s intraclass r) was .98. The NI manipulation was analyzed with a 3 (NI: high, moderate, low) × 3 (person centeredness: high, moderate, low) ANOVA, with coder ratings of NI as the dependent measure. As expected, the ANOVA yielded a significant effect for NI, F(2, 207) = 1906.03, p < .001, η2 = .94, and a follow-up contrast revealed a significant linear effect for NI, F(1, 213) = 3,548.75, p < .001, η2 = .94. Confederates in the high NI condition were rated as the most immediate (M = 6.65, SD = .23), followed by those in the moderate NI condition (M = 3.98, SD = .80). Confederates in the low NI condition were rated as the least immediate (M = 1.50, SD = .32).
Dependent Measures: Affective Improvement and Helper Competence Indicators The five dependent measures (affective improvement, help motivation, supportiveness, expressiveness, and conversation management) were constructed with items from Clark et al.'s (1998) 13-item Comforting Responses scale and Cupach and Spitzberg’s (1981) 27-item Ratings of Alter Competence
347
COMMUNICATION RESEARCH • June 2004 Table 1 Principal Axis Analysis for Clark et al.’s (1998) Comforting Responses Scale Factor loadings Scale Items I feel more optimistic now that I have talked with my conversational partner. I understand the situation better now that I talked about it with my conversational partner. My conversational partner made me feel better about myself. I feel better after talking with my conversational partner. Talking with my conversational partner about the event helped me get my mind off it. I felt that my conversational partner was putting me down. a My conversational partner’s comments were appropriate. The way my conversational partner talked to me irritated me. My conversational partner doesn’t seem to think that I can handle my own problems. a My conversational partner seemed really concerned about me. Eigenvalues Variance accounted for
I
II
.80
–.18
.74 .70 .69
–.09 –.26 –.35
.58 –.05 –.32 –.32
–.13 .74 .71 .60
–.07 –.34
.57 .55
4.57 45.73%
1.57 15.76%
a. Reverse coded before factor analysis.
(RAC) scale. The Comforting Responses scale measures people’s responses to different comforting strategies, and it usually generates two factors that tap affective improvement and negative helper evaluations. The RAC scale measures a person’s impression of the conversational partner’s competence in a specific conversational episode, and it consists of evaluations of specific behaviors (e.g., monotonous voice), as well as more general behaviors (e.g., supportiveness). The scale has frequently generated two factors that tap other-orientation and expressiveness (Spitzberg, 1988). Items for both scales were measured with 7-point Likert-type scales that ranged from 1 (very strongly disagree) to 7 (very strongly agree). Principal axis factor analyses with varimax rotations were performed on the Comforting Responses and RAC scales. A factor was defined by eigenvalues over 1.0, and items were defined as belonging to a factor if primary loadings were above .5 with no secondary or tertiary loadings greater than .40. The Comforting Responses scale generated two factors that explained 61.5% of the variance, whereas the RAC scale revealed an underlying threefactor structure that explained 62.1% of the variance. Factor loadings for the Comforting Responses and RAC scales are shown in Tables 1 and 2, respectively.
348
Jones • Emotional Support and Emotional Change Table 2 Principal Axis Factor Analysis for Cupach and Spitzberg’s (1981) Ratings of Alter Competence (RAC) Scale Factor Loadings Scale Items She or he was sensitive to my needs and feelings in the conversation. She or he was supportive. She or he was sympathetic. a She or he ignored my feelings. She or he was a good listener. She or he gave positive feedback. She or he understood me. She or he was polite. She or he was cooperative. She or he could easily put herself or himself into another person’s shoes. She or he was respectful. a Her or his voice was boring and monotonous. a She or he lacked self-confidence. a She or he spoke too slowly. a She or he appeared tired and sleepy. Her or his facial expressions were abnormally a blank and restrained. a She or he spoke too rapidly. She or he had an accurate self-perception. She or he was assertive. She or he was versatile. She or he was trustworthy. She or he was confident. Eigenvalues Variance accounted for
I
II
III
.83 .78 .77 .75 .73 .72 .71 .64 .60
.20 .15 .19 .40 .33 .08 .27 .38 .40
.21 .24 .30 .13 .25 .31 .30 .16 .17
.60 .57 .14 .25 .19 .13
.06 .35 .74 .70 .67 .60
.36 .27 .38 .24 .26 .06
.17 .22 .25 .15 .24 .40 .30
.57 .53 .09 .30 .20 .06 .15
.35 –.08 .59 .56 .53 .50 .50
10.13 46.03%
2.10 9.47%
.45 6.63%
a. Reverse coded before factor analysis.
Affective Improvement. Affective improvement was the first factor extracted from the Comforting Responses scale, and it accounted for 45.73% of the variance in help recipients’ comforting responses. This factor featured five items that represent frequently experienced positive emotional outcomes (e.g., “I feel better after talking with my conversational partner.”). The scale was internally consistent (α = .85). Helper competence indicators. The first helper competence factor, help motivation, was drawn from the Comforting Responses scale, and it contained five recipient judgments of the helper’s willingness to offer support. With the
349
COMMUNICATION RESEARCH • June 2004 exception of two items, all items reflected unwillingness on the part of the helper (e.g., “I felt my conversational partner was putting me down.”). Reliabilities for this subscale were α = .80. The supportiveness factor was drawn from the RAC scale, and it contained 11 items. This factor accounted for most of the variance in helper competence (57.03%) and included evaluations of supportiveness (“He or she was supportive.”), empathy (“He or she could easily put himself or herself into another person’s shoes.”), compassion (“He or she was sympathetic.”), and validations (“He or she gave positive feedback.”). Reliabilities for the supportiveness subscale were excellent at α = .95. The second factor extracted from the RAC scale contained seven expressiveness items that consisted of the recipient’s evaluations of the helper’s involvement in the conversation (e.g., “He or she appeared tired and sleepy.” “He or she was awkward in the conversation.”). All items were reverse coded, and reliabilities for the expressiveness subscale were acceptable at α =.88. The third factor that was extracted from the RAC scale was conversation management, and it contained five items that reflected a more global assessment of the general dynamics of the conversation (e.g., “He or she was versatile.”). Reliabilities for this subscale were acceptable at α = .77.
Results Table 3 presents the zero-order correlations for all variables involved in the experiment. As expected, person centeredness and NI were uncorrelated, whereas the dependent measures were moderately to highly correlated with one another.1 The hypothesized path model (see Figure 1) was tested with analysis of moment structure (AMOS) 4.0 (Arbuckle, 1997) and a maximum-likelihood estimation procedure. Because there is currently little consensus concerning the best index of overall fit for path models, several commonly reported indexes are reported. The Comparative Fit Index (CFI) (Bentler, 1990) and the Tucker-Lewis Index (TLI) (Bentler & Bonett, 1980), also referred to as the Bentler-Bonnett Non-Normed Fit Index (NNFI), are two such indexes. Because they are normed, CFI and TLI indexes range from 0 to 1, and values close to 1 indicate good model fit. The χ2 test is a “badness of fit index” that should ideally generate a nonsignificant value for well-fitting models. The χ2 test tends to be sensitive to sample sizes. The larger the sample size (> 200), the more likely the rejection of the model, and thus, the more likely the occurrence of a Type II error. The Root Mean Square Error of Approximation (RMSEA) is less sensitive to sample size, because it has a known (noncentral) χ2 distribution; a RMSEA ≤ .05 suggests good model fit (Hoyle & Panter, 1995;
350
351
(.98) .02 .45* .31* –.45* .26* .25* (.91) .24* .35* –.21* .25* .10 (.95) .63* –.83* .69* .60* (.88) –.61* .59* .32*
(.80) –.59* –.52*
Nonverbal Help Immediacy (NI) Supportiveness Expressiveness Motivation
(.77) .54*
Conversation Management
Note. N = 258. Measures in parentheses are Cronbach’s alpha internal consistency coefficients. All correlation coefficients are two-tailed. *p < .01
Person centeredness NI Supportiveness Expressiveness Help motivation Conversation management Affective improvement
Person Centeredness
Table 3 Intercorrelations Among Variables in the Path Model
(.85)
Affective Improvement
COMMUNICATION RESEARCH • June 2004 .17*
.74 .48*
Expressiveness
.59* .29
.86* Supportiveness Nonverbal Immediacy
.27* Helper Competence
-.78*
.10
.40 Help Motivation
.52*
.58* .86*
Person Centeredness
.25*
Affective Improvement
Conversation Management
.36
1.26*
-.15*
Figure 2. Adjusted Path Model. Note. Path values are standardized regression weights for predicted relationships and covariances for the two added paths. Values associated with endogenous (helper competence) and indicator variables (supportiveness, expressiveness, conversational management, demeaning attitude, affect improvement) are error variances. χ2 (4, N = 258) = 4.29, p = .37, CFI = 1.00, TLI = .99, RMSEA = .01. *p < .001
Hu & Bentler, 1993, 1995). Specific parameter estimates, testing the predicted relationships among the variables, were provided in the form of standardized regression weights (β) and error variance estimates.
Model Modifications An initial evaluation of the hypothesized model indicated a lack of fit between the model and sample covariance matrices, χ2(7, N = 258) = 33.46, p < .001, CFI = .32, TLI = .90, RMSEA = .12. A univariate Lagrange Multiplier (LM) test suggested two significant model modifications by adding covariance paths between PC and perceived conversation management, and between NI and perceived expressiveness (covariance coefficients = –.15 and .17, p < .001, respectively). The adjusted model, presented in Figure 2, indicated excellent fit between the hypothesized path model and the sample data, χ2(4, N = 258) = 4.29, p = .37, CFI = 1.00, TLI = .99, RMSEA = .01. Specific findings with respect to the hypothesized relationships among the variables in the adjusted model are summarized below.
352
Jones • Emotional Support and Emotional Change Affective Improvement As predicted, recipients felt better when the helper provided more PC emotional support (β = .25, p < .001). Notably, emotional change was exclusively a function of having received highly PC emotional support. Recipients did not report feeling significantly better as a function of the confederate’s level of nonverbal immediacy (β = .10, ns). It is also noteworthy both that error variance estimates were highest for affective improvement and that the two message factors accounted for only 7% of variance in affect improvement.
Helper Competence Overall, person centeredness and NI explained 34% of the variance in helper competence. As predicted, helpers who provided more PC and immediate emotional support were viewed as more competent (β = .52 and .27, p < .001, respectively). Interestingly, PC comforting exerted a much stronger influence on helper competence judgments than did NI. Finally, helper competence was well represented by the four indicators. Squared multiple correlations (R2) provide information about the percentage of variance in the dependent measure that is accounted for by the factor (helper competence) and were as follows: supportiveness = .75; expressiveness = .30; conversation management = .56; help motivation = .61. In addition, all competence factors were significant indicators for helper competence. That is, managing conversations, being motivated to help, being expressive, and being supportive are all factors that shape a recipient’s competence judgments of the helper.
Discussion The current study set out to examine whether PC and NI comforters both make people feel better and are judged as more competent compared to comforters who use less PC and less NI comforting messages. In an experimental setup, participants were asked to talk about an emotionally upsetting event with a confederate who was trained to exhibit different levels of PC and NI. Participants subsequently completed scales assessing both their emotional changes and their perceptions of the helper’s competence in terms of the helper’s supportiveness, expressiveness, conversation management, and help motivation. Path analytic procedures were used to examine a model that predicted positive relationships between the two manipulated comforting message factors (PC and NI) and help recipients’ reported levels of emotional change, as well as their judgments of the helper’s competence.
353
COMMUNICATION RESEARCH • June 2004 Results of the path-analytic procedures indicate that highly PC comforters made participants feel better than less PC comforters. These results extend past research on PC comforting (see Jones & Guerrero, 2001). PC comforting messages are not only evaluated as most helpful but also have an immediate and highly desirable effect—these kinds of messages seem to alleviate the difficult feelings experienced by upset people. In addition, PC comforters were also viewed as more competent in terms of help motivation, supportiveness, expressiveness, and conversation management. Not only are PC helpers better liked and viewed as more attractive (Samter et al., 1987), but also results of the current study clearly demonstrate that recipients view a PC comforter as someone who is willing to help, who acknowledges difficult emotions, and who explicitly encourages recipients to talk about their emotions. Similar results hold for competence impressions recipients have of immediate comforters: Immediate comforters were viewed as more competent than nonimmediate comforters. It is noteworthy that person centeredness was the only message factor that influenced recipients’ affective improvement. In addition, NI predicted helper competence to a far lesser extent than did person centeredness. There may be several reasons for the weak results associated with NI. First, people in the current study were strangers to one another. It is likely that NI behaviors exert a stronger influence on immediate instrumental and relational outcomes in established interpersonal relationships because relational partners are able to evaluate nonverbal information more precisely and accurately. Second, verbal emotional support might foster more emotional change than NI because of its informational value. U.S. culture is commonly categorized as a low-context culture in which information is mainly carried out through verbal expressions rather than contextual or environmental cues that are mostly nonverbal in nature. PC messages carry explicit emotional information that might fit the U.S. cultural context. Third, and most important, Jones and Guerrero (2001) suggest that person centeredness and NI might serve different functions in the emotional support process. Although high levels of NI communicate primarily interpersonal warmth and availability, highly PC messages often carry explicit statements that encourage the disclosure of difficult emotions. Having the opportunity to disclose difficult feelings in a safe environment, such as the one provided by highly PC comforters, might have set in motion a reappraisal process that is necessary for emotional change, as has been suggested by Burleson and Goldsmith (1998). These assumptions have yet to be examined with respect to person centeredness and NI, but an examination of the affective-improvement measure used in the current study provides initial evidence that PC comforting might set in motion and facilitate a reappraisal
354
Jones • Emotional Support and Emotional Change process. Recipients reported not only that they felt better but also that they were able to “get their mind off ” the upsetting event. Recipients also reported that they felt more optimistic after having talked with the confederate. These statements imply that highly PC messages might help people feel and think better. These messages not only lead to emotional improvement but may also enable people to gain a different perspective on their emotional experiences and feel confident about their abilities to deal with the event. The estimated model deviated from the hypothesized model with respect to two added paths. First, the person centeredness of confederates’ emotional support was directly and negatively correlated with perceptions of conversation management (r = –.15, p < .001). That is, people who used highly PC messages were perceived as low in conversation management. An explanation for this negative relationship might lie in the items that represented conversation management (e.g., “She had an accurate self-perception.”; “She was assertive.”; “She was confident.”). Users of low PC messages might well be viewed as more assertive and confident than comforters who use highly PC messages because low PC messages often forthrightly express advice or critical opinions without much facework. The second added path suggested a significant fit between NI and perceived expressiveness (r = .17, p < .001). NI and perceived expressiveness might simply overlap in terms of content. The reverse-coded items that made up expressiveness (e.g., “She spoke too slowly.”; “She appeared tired and sleepy.”) seem to reflect nonimmediacy items that are important in the emotional support process.
Limitations and Future Research Directions The current study has several limitations, as does all research. First, assessments of emotional change and helper competence are but two important immediate outcomes in the emotional support process. Future research needs to assess additional immediate instrumental and relational outcomes, as well as long-term effects of more or less PC and immediate emotional support messages. Second, the current study targeted only four competence dimensions that characterize perceived helper qualities and that are relevant in the provision of emotional support. It is entirely possible that there are additional helper characteristics that influence supportive interactions. Third, the two empirically determined modifications that were added post hoc to the model should be viewed as tentative. Because of these post hoc modifications, cross-validation of the model is necessary. Indeed, replicating the model with another sample would provide the strongest evidence for the validity of the predicted relationships. Finally, assessing emotional change
355
COMMUNICATION RESEARCH • June 2004 reliably and validly is undoubtedly a challenge, and future research should employ multiple measures of emotional change. In addition to posttest affective improvement scales, researchers might use emotional pretest and posttest emotion adjective measures (see Watson & Tellegen, 2002) and pretest and posttest codings of facial expressions (e.g., Ekman & Friesen’s [1978] Facial Action Coding system).
Conclusion The model tested here provides initial evidence that communicative enactments of emotional support have a direct and immediate emotional impact on help recipients who experience emotional distress. Simply put, PC messages make people feel better. Reports of emotional change and competence perceptions are two criteria by which researchers can distinguish more or less comforting emotional support messages. As with all scientific research, and particularly with structural equation modeling, conditions for associations between variables are a function of strong theoretical arguments, manipulations of putative causes, and logic (Hoyle & Panter, 1995). What corroborates the results of the current study is not only the strong theoretical foundation but also the use of an experimental design with uncorrelated independent variables. All things being equal, changes in the dependent measures can therefore be attributed to certain levels of the independent measures.
Notes 1. Multivariate analysis of variance (MANOVA) was conducted to demonstrate that only the linear effects of each independent variable contributed to the dependent measures. As expected, significant main effects emerged only for person centeredness, Λ = 2 .73, F(10, 492) = 8.30, p < .001, partial η = .14, and for nonverbal immediacy (NI), Λ = 2 .84, F(10, 492) = 4.47, p < .001, partial η = .08. People in the high and moderate person centeredness (PC) conditions reported feeling significantly better than people in the low PC condition, whereas affective improvement was not significantly influenced by NI manipulations. In addition, recipients evaluated the highly PC and nonverbally immediate comforter as more supportive, expressive, conversationally competent, and help motivated than the low PC and nonimmediate comforter.
References Andersen, J. F., Andersen, P. A., & Jensen, A. D. (1979). The measurement of nonverbal immediacy. Journal of Applied Communication Research, 7, 153-180.
356
Jones • Emotional Support and Emotional Change Andersen, P. A. (1989, May). A cognitive valence theory of intimate communication. Paper presented at the meeting of the International Network of Personal Relationships, Iowa City, IA. Andersen, P. A. (1999). Nonverbal communication: Forms and functions. Mountain View, CA: Mayfield. Andersen, P. A., & Guerrero, L. K. (1998). The bright side of relational communication: Interpersonal warmth as a social emotion. In P. A. Andersen & L. K. Guerrero (Eds.), Communication and emotion (pp. 305-324). Orlando, FL: Academic Press. Andersen, P. A., Guerrero, L. K., Buller, D. B., & Jorgensen, P. F. (1998). An empirical comparison of three theories of nonverbal immediacy exchange. Human Communication Research, 24, 501-535. Applegate, J. L. (1980). Person-centered and position-centered teacher communication in a day care center. Studies in Symbolic Interactionism, 3, 59-96. Arbuckle, J. (1997). Amos 4.0. Chicago: SmallWaters Corporation. Barbee, A., Rowatt, T. L., & Cunningham, M. R. (1998). When a friend is in need: Feelings about seeking, giving, and receiving social support. In P. A. Andersen & L. K. Guerrero (Eds.), Communication and emotion (pp. 282298). Orlando, FL: Academic Press. Bentler, P. M. (1990). Comparative fit indexes in structural models. Psychological Bulletin, 107, 238-246. Bentler, P. M., & Bonett, D. G. (1980). Significance tests and goodness-of-fit in the analysis of covariance structures. Psychological Bulletin, 88, 588-606. Burgoon, J. K., & Aho, L. (1982). Three field experiments on the effects of violations of conversational distance. Communication Monographs, 49, 71-88. Burgoon, J. K., Stern, L. A., & Dillman, L. (1995). Interpersonal adaptation: Dyadic interaction patterns. Cambridge, UK: Cambridge University Press. 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. Sarason (Eds.), Communication of social support: Messages, interactions, relationships, and community (pp. 3-29). Thousand Oaks, CA: Sage. Burleson, B. R. (2003). Emotional support skills. In J. O. Greene & B. R. Burleson (Eds.), Handbook of communication and social interaction skills (pp. 551-594). Mahwah, NJ: Erlbaum. Burleson, B. R., & Goldsmith, D. J. (1998). How the comforting process works: Alleviating emotional distress through conversationally induced reappraisals. In P. A. Andersen & L. K. Guerrero (Eds.), Communication and emotion (pp. 246-275). Orlando, FL: Academic Press. Burleson, B. R., & MacGeorge, E. L. (2002). Supportive communication. In M. L. Knapp & J. A. Daly (Eds.), Handbook of interpersonal communication (3rd ed., pp.374-422). Thousand Oaks, CA: Sage. Burleson, B. R., & Samter, W. (1985). Consistencies in theoretical and naive evaluations of comforting messages. Communication Monographs, 52, 104-123.
357
COMMUNICATION RESEARCH • June 2004 Clark, R. A., Pierce, A. J., Finn, K., Hsu, K., Toosley, A., & Williams, L. (1998). The impact of alternative approaches to comforting, closeness of relationships, and gender on multiple measures of effectiveness. Communication Studies, 49, 224-239. Cupach, W. R., & Spitzberg, B. H. (1981). Relational competence: Measurement and validation. Paper presented at the Western Speech Communication Association Conference, San Jose, CA. Cutrona, C. E., & Russell, D. W. (1990). Types of social support and specific stress: Toward a theory of optimal matching. In B. R. Sarason, I. G. Sarason, & G. R. Pierce (Eds.), Social support: An interactional view (pp. 319-366). New York: John Wiley. Dakof, G. A., & Taylor, S. E. (1992). Victims’ perceptions of social support: What is helpful from whom? Journal of Personality and Social Psychology, 58, 80-89. Dolin, D., & Booth-Butterfield, M. (1993). Reach out and touch someone: Analysis of nonverbal comforting behaviors. Communication Quarterly, 41, 383-393. Ekman, P., & Friesen, W. V. (1978). Investigator’s guide to the Facial Action Coding system. Palo Alto, CA: Consulting Psychologists Press. Goldsmith, D. (1992). Managing conflicting goals in supportive interaction— An integrative theoretical framework. Communication Research, 19, 264286. Goldsmith, D. J., & MacGeorge, E. (2000). The impact of politeness and relationship on perceived quality of advice about a problem. Human Communication Research, 26, 234-263. Gump, B. B., Polk, D. E., Kamarck, T. W., & Shiffman, S. M. (2001). Partner interactions are associated with reduced blood pressure in the natural environment: Ambulatory monitoring evidence from a healthy, multiethnic adult sample. Psychosomatic Medicine, 63, 423-433. Hays, R. B, Magee, R. H, & Chauncey, S. (1994). Identifying helpful and unhelpful behaviours of loved ones: The PWA’s perspective. AIDS Care, 4, 379-392. Hoyle, R., & Panter, A. (1995). Writing about structural equation models. In R. Hoyle (Ed.), Structural equation modeling: Concepts, issues, and applications (pp. 158-176). Thousand Oaks, CA: Sage. Hu, L. T., & Bentler, P. M. (1993). Can fit indexes in covariance structure analysis be trusted? Psychological Bulletin, 112, 351-362. Hu, L. T., & Bentler, P. M. (1995). Evaluating model fit. In R. Hoyle (Ed.), Structural equation modeling: Concepts, issues, and applications (pp.7699). Thousand Oaks, CA: Sage. Jones, S. M., & Burleson, B. R. (1997). The impact of situational variables on helpers’ perceptions of comforting strategies. Communication Research, 24, 530-555. Jones, S. M., & Guerrero, L. A. (2001). The effects of nonverbal immediacy and verbal person centeredness in the emotional support process. Human Communication Research, 27, 567-596. Lazarus, R. S., & Folkman, S. (1984). Stress, appraisal, and coping. New York: Springer.
358
Jones • Emotional Support and Emotional Change Lehman, D. R., & Hemphill, K. J. (1990). Recipients’ perceptions of support attempts and attributions for support that fail. Journal of Social and Personal Relationship, 7, 563-574. Lewis, R. J., Derlega, V. J., Shankar, A., Cochard, E., & Finkel, L. (1997). Nonverbal correlates of confederates’ touch: Confounds in touch research. Journal of Social Behavior and Personality, 12, 821-830. Mehrabian, A. (1971). Silent messages. Belmont, CA: Wadsworth. Niederhoffer, K. G., & Pennebaker, J. W. (2002). Sharing one’s story: On the benefits of writing or talking about emotional experience. In C. R. Snyder & S. J. Lopez (Eds.), Handbook of positive psychology (pp. 573-583). Oxford, UK: Oxford University Press. Noller, P. (1984). Nonverbal communication and marital interaction. Oxford, UK: Pergamon. Patterson, M. L. (1995). A parallel model of nonverbal communication. Journal of Nonverbal Behavior, 19, 3-29. Pennebaker, J. W., & Beall, S. K. (1986). Confronting a traumatic event: Toward an understanding of inhibition and disease. Journal of Personality and Social Psychology, 95, 274-281. Pennebaker, J. W, Hughes, C. F., & O’Heeron, R. C. (1987). The psychophysiology of confession: Linking inhibitory and psychosomatic processes. Journal of Personality and Social Psychology, 52,781-793. Pennebaker, J. W, & O’Heeron, R. C. (1984). Confiding in others and illness rate among spouses of suicide and accidental-death victims. Journal of Abnormal Psychology, 93, 473-476. Rook, K. S., & Underwood, L. G. (2000). Social support measurement and interventions: Comments and future directions. In S. Cohen, L. G. Underwood, & B. H. Gottlieb (Eds.), Social support measurement and intervention (pp. 311-334). New York: Oxford University Press. Samter, W., Burleson, B. R., & Murphy, L. B. (1987). Comforting conversations: Effects of strategy type on evaluations of messages and message producers. Southern Speech Communication Journal, 52, 263-284. Segrin, C. (1999). Social skills, stressful live events, and the development of psychosocial problems. Journal of Social and Clinical Psychology, 18, 1434. Spitzberg, B. H. (1988). Communication Competence: Measures of perceived effectiveness. In C. Tardy (Ed.), A handbook for the study of human communication: Methods and instruments for observing, measuring, and assessing communication processes (pp. 67-106). Norwood, NJ: Ablex. Spitzberg, B. H. (2003). Methods of interpersonal skill assessment. In J. O. Greene & B. R. Burleson (Eds.), Handbook of communication and social interaction skills (pp. 93-134). Mahwah, NJ: Erlbaum. Spitzberg, B. H., & Cupach, W. R. (1984). Interpersonal communication competence. Beverly Hills, CA: Sage. Spitzberg, B. H., & Cupach, W. R. (2002). Interpersonal skills. In M. L. Knapp & J. A. Daly (Eds.), Handbook of interpersonal communication (3rd ed., pp. 564-611). Thousand Oaks, CA: Sage. Spitzberg, B. H., & Hecht, M. L. (1984). A component model of relational competence. Human Communication Research, 10, 575-599.
359
COMMUNICATION RESEARCH • June 2004 Strittmatter, R., & Bengel, J. (1996). Alltagswissen über gesundheitliche Protektivfaktoren: Eine qualitative Studie [Common knowledge on positive health factors: A qualitative study]. Psychotheraphie, Psychosomatik, Medizinische Psychology, 46, 68-75. Stroebe, W., & Stroebe, M. (1996). The social psychology of social support. In E. T. Higgins & A. W. Kruglanski (Eds.), Social psychology: Handbook of basic principles (pp. 597-621). New York: Guilford. Wan, C. K., Jaccard, J., & Ramey, S. L. (1996). The relationship between social support and life satisfaction as a function of family structure. Journal of Marriage and the Family, 58, 502-513. Watson, D., & Tellegen, A. (2002). Aggregation, acquiescence, and the assessment of trait affectivity. Journal of Research in Personality, 36, 589-597. Wiener, M., & Mehrabian, A. (1968). Language within language: Immediacy, a channel in verbal communication. New York: Appleton-Century-Crofts. Winstead, B. A., Derlega, V. J., Sanchez-Hucles, J., & Clarke, E. (1992). Friendship, social interaction, and coping with stress. Communication Research, 19, 193-211. Susanne Jones, Ph.D., Arizona State University, is an assistant professor in the Department of Communication Studies, University of Minnesota, 225 Ford Hall, 224 Church Street SE, Minneapolis, MN 55455; phone: (612) 6260592; fax: (612) 624-6544;
[email protected]. Data for this study were collected as a part of the author’s dissertation. I would like to thank the two blind reviewers for their insightful comments, Laura Guerrero who directed my dissertation, and the students who served as confederates and coders in the experiment.
360