18

Click here to load reader

Why Are We Disclosing in This Group

Embed Size (px)

Citation preview

Page 1: Why Are We Disclosing in This Group

Why Are We Disclosing in This Group? Using Response SurfaceAnalysis to Examine Reasons for Self-Disclosure Among Taiwanese

Counseling Group Members

Yu-Kuang Kevin HsuNational Hsinchu University of Education

Jill Denise PaquinChatham University

Dennis M. Kivlighan, Jr.University of Maryland, College Park

We examined the relationship between an engaged group climate experienced by groupmembers and their reasons for self-disclosing over time. 96 group participants in 14,10-session counseling groups completed the Group Climate Questionnaire and SelfDisclosure Questionnaire (SDQ) after Sessions 2 through 9. Polynomial hierarchicallinear modeling and response surface analysis were used to examine the fit between agroup member’s perceptions of engagement and the other members’ perceptions ofengagement. Perceptual fit was used to predict 4 motivating factors for group memberself-disclosure represented by the subscales of the SDQ: (a) Relational Motivation, (b)Expectation of Help, (c) Other-Initiated, and (d) Discomfort with Silence. When thegroup member and the other group members agreed that engagement was high themember was more likely to self-disclose for relationally motivated reasons or becauseshe or he expected to receive help. When the group member and the other groupmembers agreed that engagement was low, the member was more likely to self-disclosebecause they were uncomfortable with silence. Implications for theory, practice, andresearch are discussed.

Keywords: group climate, group counseling, group process, response surface analysis, self-disclosure

A group’s climate, in any domain, can bethought of as the members’ perceptions of thegroup across several interactional dimensions(MacKenzie, 1983). MacKenzie suggests thatgroup climate encompasses features of thegroup environment that encourage “compatibletypes of interpersonal events” (p. 159). One

type of compatible interpersonal event is agroup member’s willingness or unwillingness toself-disclose in the group. In therapy groups,willingness to self-disclose is particularly im-portant to examine because as Yalom andLescsz (2005) suggest, self-disclosure sets inmotion all of the therapeutic factors in groupcounseling. Research shows that greater self-disclosure is related to group cohesion(Tschuschke & Dies, 1997) and to increasedgroup empathy, group intimacy, and group be-havior (Shechtman & Rybko, 2004). Greaterself-disclosure during group counseling is alsorelated to decreases in interpersonal problems(Toren & Shechtman, 2010). In addition, lim-ited self-disclosure impedes group progress(Doxsee & Kivlighan, 1994). Only one pub-lished study found no relationship betweengroup member self-disclosure and treatmentoutcome (Crits-Christoph, Johnson, Gibbons, &Gallop, 2013). In the drug counseling groups

This article was published Online First November 4,2013.

Yu-Kuang Kevin Hsu, Department of Educational Psy-chology and Counseling, National Hsinchu University ofEducation, Taiwan; Jill Denise Paquin, Graduate Programsin Counseling Psychology, Chatham University; Dennis M.Kivlighan, Jr., Department of Counseling, Higher Educa-tion, and Special Education, University of Maryland, Col-lege Park.

Correspondence concerning this article should be ad-dressed to Yu-Kuang Kevin Hsu, Department of Educa-tional Psychology and Counseling, National Hsinchu Uni-versity of Education, Taiwan. E-mail: [email protected]

Thi

sdo

cum

ent

isco

pyri

ghte

dby

the

Am

eric

anPs

ycho

logi

cal

Ass

ocia

tion

oron

eof

itsal

lied

publ

ishe

rs.

Thi

sar

ticle

isin

tend

edso

lely

for

the

pers

onal

use

ofth

ein

divi

dual

user

and

isno

tto

bedi

ssem

inat

edbr

oadl

y.

Group Dynamics: Theory, Research, and Practice © 2013 American Psychological Association2014, Vol. 18, No. 1, 20–37 1089-2699/14/$12.00 DOI: 10.1037/a0034671

20

Page 2: Why Are We Disclosing in This Group

examined, group member self-disclosure aboutthe past was not related to reductions in cocaineuse. However, the partially mixed results forself-disclosure may be due to how self-disclosure was operationalized. Group membersmay disclose for either therapeutic or nonthera-peutic reasons. It is, therefore, important to un-derstand why group members choose to self-disclose in groups.

Recently, Woo, Hsu, and Hung (2005) devel-oped the Self Disclosure Questionnaire (SDQ)to examine reasons that group members give formaking or withholding disclosures in groups.The SDQ has four factors that are related to agroup member’s reasons for making self-disclosures in group therapy: Relational Moti-vation, Expectation of Help, Other-Initiated,and Discomfort with Silence (Hsu, Woo, &Hung, 2009). In examining the construct valid-ity of the SDQ, Hsu (2011) found that Rela-tional Motivation for self-disclosure was themost often cited reason in early group sessionsand also increased over time, whereas disclo-sures made due to experiencing Discomfortwith Silence decreased over time.

In the current study, we examine how anengaged group climate is related to group mem-bers’ willingness to disclose as defined by theSDQ. Only a few studies have examined therelationship between an individual group mem-bers’ perception of group climate and their per-ceptions of therapeutic factors (Johnson et al.,2006; Joyce, MacNair-Semands, Tasca, &Ogrodniczuk, 2011; Kivlighan, Multon, &Brossart, 1996). These studies show that indi-vidual perceptions of engagement are related tohigher levels of the therapeutic factors exam-ined.

Unfortunately, MacKenzie (1983) did not of-fer specific hypotheses linking aspects of groupclimate to group members’ reasons for self-disclosing. The closest empirical findings, how-ever, are the positive relationships between per-ceptions of an engaged group climate with theSecure Emotional Expression therapeutic factor(Joyce et al., 2011). Other research usingMacKenzie’s (1983) Group Climate Question-naire suggests that group climate can be char-acterized as encouraging movement either to-ward or away from others (for a review seeHurley, 1997). In a highly engaged group, themembers are more likely to self-disclose be-cause they are moving toward others, whereas

in a group that is not engaged, self-disclosure isless likely because group members are movingaway from others. As noted above, however,self-disclosure is not simply a “yes or no” prop-osition. We believe Hurley’s (1997) movingtoward or away from distinction can be used topredict the reasons that group members have forself-disclosing.

Similar to Hurley (1997), Elliot and Thrash(2002) write that “the distinction between ap-proach and avoidance motivation” is “fundamentaland integral to the study of affect, cognition, andbehavior” (p. 804). Reasons for self-disclosingcan be classified as either moving toward oraway from the other group members. RelationalMotivation, includes items such as self-disclosing because “I felt a sense of camarade-rie.” Expectation of Help, includes items suchas self-disclosing because “I think the memberswould give me support if I disclosed.” Both ofthese factors can be conceptualized as ap-proach-oriented motivations for self-disclosure.The factor Other-Initiated includes items suchas self-disclosing because “Everyone looked atme and I felt pressure to speak,” and the factorDiscomfort with Silence, includes items such asself-disclosing because “I thought someoneshould speak to break the group’s silence.” Bothof these factors, in contrast, can be conceptual-ized as more reactive, avoidance-oriented mo-tives for self-disclosing. Therefore, we hypoth-esized that a more engaged group climate wouldbe positively related to approach-oriented mo-tivations for self-disclosure (Relational Motiva-tion and Expectation of Help) and negativelyrelated to reactive, avoiding motives for self-disclosing (Other-Initiated and Discomfort withSilence).

Person-Group Fit in Perceptionsof Group Climate

Most group climate research examines anindividual group member’s perceptions of howengaged the group is to some type of groupprocess or outcome. However, these individualperceptions of how engaged the group is alwaysoccur in the larger context of how engaged theother group members see the group. Simultane-ously examining both the individual groupmember’s perception of engagement and theperceptions of engagement of the other groupmembers is an important theoretical and meth-

21GROUP CLIMATE AND SELF-DISCLOSURE

Thi

sdo

cum

ent

isco

pyri

ghte

dby

the

Am

eric

anPs

ycho

logi

cal

Ass

ocia

tion

oron

eof

itsal

lied

publ

ishe

rs.

Thi

sar

ticle

isin

tend

edso

lely

for

the

pers

onal

use

ofth

ein

divi

dual

user

and

isno

tto

bedi

ssem

inat

edbr

oadl

y.

Page 3: Why Are We Disclosing in This Group

odological advance over studies that examineonly individual perceptions of engagement. Forexample, a member of one group might viewthe group as highly engaged and the other groupmembers might also see the group as highlyengaged. Contrast this with another group inwhich a participant views the group as highlyengaged while the other members of her groupsee the group as disengaged. We argue that theperceptions of engagement by the other groupmembers during a particular session provide animportant context for understanding and inter-preting an individual’s perception of engage-ment. This relationship can be understood interms of a group member’s perceptual fit withthe group (for a more extensive review of per-ceptual fit see Paquin, Kivlighan, & Drogosz,2013).

Perceptual fit is one aspect of person-group(P-G) fit which has been examined extensivelyin the organizational psychology literature. Inthe organizational literature, P-G fit refers to thecongruence between a worker and her immedi-ate peer or work group. P-G fit is positivelyrelated to both group processes (i.e., comfort inand commitment to their group; Vogel & Feld-man, 2009) and outcomes (i.e., goal attainment;Kristof-Brown, Zimmerman, & Johnson, 2005).Only three studies have examined P-G fit incounseling or therapy groups (Flowers, 1987;Lo Coco, Gullo, & Kivlighan, 2012; Paquin etal., 2013). The Lo Coco, Gullo, and Kivlighan(2012) study is the most germane to the presentresearch and is described below.

Edwards (1993) and Shanock, Baran, Gentry,Pattison, and Heggestad (2010) argue that, con-ceptually, P-G fit has three distinct concepts:agreement (fit), disagreement (lack of fit), anddirection of disagreement. Agreement existswhen a group member’s perception of engage-ment matches the other group members’ percep-tions of engagement. When agreement exists,the group member and the other group memberssee engagement along a continuum of low tohigh engagement. Disagreement exists when agroup member’s perception of engagement andthe other group members’ perceptions of en-gagement are opposite. Disagreement may beeither nondirectional or directional. Nondirec-tional disagreement is analogous to the absolutedifference in group member and other groupmember engagement ratings. The more thegroup members’ engagement rating diverge (in

either direction, positive or negative) from theaverage engagement rating of the other groupmembers, the greater the nondirectional disagree-ment. Directional disagreement is analogous tothe numerical difference in group member andother group member engagement ratings. In di-rectional disagreement, the group member’s en-gagement rating can be either higher or lowerthan the engagement ratings of the other groupmembers.

The Lo Coco et al. (2012) study is the onlygroup counseling study to examine both agree-ment and disagreement in perceptual fit. LoCoco et al. (2012) examined the consequencesof agreement and disagreement on perceptionsof group alliance in groups for clients witheating disorders. A member’s symptom reduc-tion was greater when the group member andthe other group members agreed that their alli-ance to the group-as-a-whole was strong. Whenthe other group members’ perceptions of thealliance to the group-as-a-whole were strongerthan the individual group member’s own per-ceptions of the alliance to the group, there wasgreater symptom reduction. Similar to Lo Cocoet al. (2012) who found that person-group fit inalliance to the group-as-a-whole was related togroup member outcomes, we believe that the fitbetween a member and the other group mem-bers’ perceptions of engagement will have sim-ilar effects on reasons for self-disclosing.

Hypotheses

Lo Coco et al. (2012) provided one theoreti-cal explanation for why agreement would berelated to outcome. This explanation also canaccount for why member and other group mem-bers’ agreement in perceptions of engagementwould be positively related to approach-oriented reasons for self-disclosing (RelationalMotivation and Expectation of Help) and neg-atively related to reactive, avoiding motives forself-disclosing (e.g., Other-Initiated and Dis-comfort with Silence).

According to Byrne’s (1971) similarity-attraction model, perceptual similarity has a re-inforcing effect. Applied to engagement ratings,when the group member and the other groupmembers agree that engagement is high, thegroup member and the other group members seethe group climate as encouraging approach-oriented behaviors. If the group member and the

22 HSU, PAQUIN, AND KIVLIGHAN

Thi

sdo

cum

ent

isco

pyri

ghte

dby

the

Am

eric

anPs

ycho

logi

cal

Ass

ocia

tion

oron

eof

itsal

lied

publ

ishe

rs.

Thi

sar

ticle

isin

tend

edso

lely

for

the

pers

onal

use

ofth

ein

divi

dual

user

and

isno

tto

bedi

ssem

inat

edbr

oadl

y.

Page 4: Why Are We Disclosing in This Group

other group members agree that their groupclimate is encouraging approaching behaviors,there should be mutual reinforcement for self-disclosing based on reasons related to relationalmotivation or the expectation of receiving help.When agreement about engagement is high butthe absolute level of engagement is low, weexpect that when members disclose, they will bedoing so because others have prompted or ini-tiated these disclosures, or because they areuncomfortable with silence during a session.This is because the group members agree thatthe group climate is encouraging avoidance.

Disagreement about the group’s level of en-gagement signals ambiguity in the group, thuscreating uncertainty for the group member as tohow to behave. Given this ambiguity, we be-lieve that the group member cannot predict howthe other group members will react to his or herself-disclosures. In groups where nondirectionaldisagreement about engagement exists, the am-biguity should lead the group member and theother group members to be less likely to self-disclose for reasons related to Relational Moti-vation and Expectation of Help, and will bemore likely to self-disclose for Other-Initiatedand Discomfort with Silence reasons.

Lastly, theorists have not addressed issuesrelated to the directionality of fit in the groupcounseling literature. Therefore, it is difficult tomake an a priori hypothesis about this relation-ship. A more general research question aboutwhether there is a relationship between the di-rection of disagreement regarding perceptionsof group engagement over time and a groupmember’s reasons for self-disclosing is appro-priate.

Method

Groups

Fourteen counseling groups occurring on acollege campus in Taiwan were included in thecurrent study. Three of the groups served col-lege students seeking services through the col-lege’s counseling center, and 11 groups servedboth students and nonstudent community par-ticipants. Groups were conducted, with super-vision, as part of a practicum training course ingroup counseling. Groups ranged in size fromfive to nine members (M � 6.86, SD � 1.29).Groups met for 10, 90-min weekly sessions.

Ten of these groups were co-led by two train-ees, and four groups were led by a single leader.The purpose of the groups was to resolve theproblems such as life adjustment, career deci-sion, or relationship issues as described in therecruitment materials. Groups were generallyunstructured and based on the interpersonal pro-cess approach (Yalom & Lescsz, 2005) whichencouraged members to listen, support, solveproblems, practice new behaviors in and outsideof group, and exchange feedback with one an-other. Structured activities were only used in thefirst and second sessions of the groups’ forma-tion in order to facilitate member involvement,build norms, and foster group cohesion. AfterSession 2, the group format shifted to a lessstructured environment in which problem solv-ing was identified as the primary goal. After thefirst session, the attendance rate for the samplewas 77.57% across sessions.

Group Members

A total of 96 group members participated inthe current study including 47 college students(49%) and 49 individuals from the communityat large (51%). There were 71 women (74%)and 25 men (26%) ranging in age from 16- to52-years-old (M � 28.77, SD � 9.44). Highestlevel of education of the sample was as follows:One participant completed elementary school(1%), six completed middle school (6.3%), sixcompleted high school (6.3%), 47 undergradu-ate college students (49%), 33 completed col-lege or university (34.4%), one completed amaster’s degree (1%), and educational attain-ment for two members was unknown. Partici-pants were excluded from participating if theymet diagnostic criteria for a serious psychiatricillness or had significant maladaptive problems.Informed consent was provided to all partici-pants during the screening process, and all gaveconsent. Assignment to a particular group wasbased on participants’ scheduling availability.

Group Leaders

Groups were facilitated by 23 graduate stu-dents in a counseling program either enrolled ina group practicum course or on internship.Eleven of the groups were led by masters-levelgraduate students, and three groups were led bypsychology interns (masters-level clinicians).Nineteen of the leaders were women and four

23GROUP CLIMATE AND SELF-DISCLOSURE

Thi

sdo

cum

ent

isco

pyri

ghte

dby

the

Am

eric

anPs

ycho

logi

cal

Ass

ocia

tion

oron

eof

itsal

lied

publ

ishe

rs.

Thi

sar

ticle

isin

tend

edso

lely

for

the

pers

onal

use

ofth

ein

divi

dual

user

and

isno

tto

bedi

ssem

inat

edbr

oadl

y.

Page 5: Why Are We Disclosing in This Group

were men, ranging in age from 24- to 48-years-old. Group leaders participated in training for 6weeks prior to leading the groups: practicumstudents participated in a 3-hr course and internstook part in a 10-hr workshop. Additionally, allgroup leaders met weekly for group supervi-sion.

Measures

Group Climate Questionnaire-Short Form(GCQ-S). The GCQ-S (MacKenzie, 1983),translated into Mandarin, consists of 12, partic-ipant-rated items on a 7-point Likert scale, rang-ing from 1 (not at all) to 7 (extremely). Whereasthe GCQ is comprised of three scales: Engage-ment, Avoidance and Conflict we only used theEngagement scale in this study. Construct va-lidity of the engaged scale has been tested ex-tensively, with demonstrated links to processand outcome (for a review of these studies seeMcClendon & Burlingame, 2010). The Chineseversion, used in the present study, was trans-lated and used in a previous study (Wang & Lin,2000). In a recent study, Wang, Chen, Wang,and Lin (2012) used confirmatory factors anal-ysis to analyze the factorial structure of theChinese version of the GCQ-S using data from744 group members, in 82 groups in Taiwan.The researchers showed a satisfactory model offit that support MacKenzie’s theory of the three-factor model. Coefficient alpha for the Engage-ment scale in the present study was .69 and theaverage item to total scale correlation was .46.

Self-Disclosure Questionnaire (SDQ).The SDQ for group members was developed byWoo et al. (2005) in order to elucidate whygroup members choose to disclose during groupsessions. Factor analysis of the SDQ was con-ducted using data from 174 participants across27, 10-session groups. The final version of theSDQ found four factors related to willing todisclose: Relational Motivation (eight items, co-efficient alphas .75) was defined as disclosingfor reasons related to the relationships a groupmember has with other group members. Expec-tation of Help (six items, .76) was defined asdisclosing based on the expectation of receivinghelpful assistance or feedback from the group.Other-Initiated (five items, .58) was defined asdisclosing for reasons based on prompting fromothers. Lastly, Discomfort with Silence (twoitems, .46) was defined as disclosures made in

an effort to break a silence during group ses-sions (Hsu, 2011; Hsu et al., 2009). A score of1 is given if a reason is offered; 0 if no reasonis offered. Therefore, the average score for eachsubscale was between 0 and 1. Construct valid-ity for the SDQ was established by confirmingexpected relationships between the four SDQfactors and time in group.

Coefficient alphas of these four factors in thisstudy were Relational Motivation .72, Expecta-tion of Help .71, Other-Initiated .58, and Dis-comfort with Silence .42. Lower coefficient al-phas (.42–.46) for the subscale Discomfort withSilence could be attributed to its only havingtwo items. However, the two items “I thoughtsomeone should speak to break the group’s si-lence” and “No one spoke up when the leaderinvited us to share” share very similar content.The SDQ and its items can be found in Appen-dix A.

Procedure

Data for the current study were gathered overthe course of 3 years between 2008 and 2011.Groups met once a week, for 90 min eachsession, for 10 sessions. Participants were askedto complete the SDQ and GCQ-S after Sessions2 through 10. A total of 665 data points weregathered from 96 participants in 14 groups. Six-hundred and 46 (97.14% of the total sample)completed both the SDQ and the GCQ-S andwere included in the analysis.

Data Analysis

We assessed P-G fit using polynomial regres-sion and response surface analysis (Shanock etal., 2010). First, the researcher conducts a poly-nomial regression (in our case polynomial hier-archical linear modeling, HLM) to obtain theregression (gamma) coefficients, which are usedto plot the three-dimensional (3D) response sur-face. We ran an HLM analysis with five specificpredictors: (a) the group member’s engagementperception, (b) the other group members’ aggre-gated perceptions of engagement, (c) a qua-dratic term that is formed by squaring the groupmember’s perceptions of engagement, (d) aninteraction term that is formed by multiplyingthe group member’s perceptions of engagementby the other group members’ perceptions ofengagement, and (e) a second quadratic termthat is formed by squaring the other group mem-

24 HSU, PAQUIN, AND KIVLIGHAN

Thi

sdo

cum

ent

isco

pyri

ghte

dby

the

Am

eric

anPs

ycho

logi

cal

Ass

ocia

tion

oron

eof

itsal

lied

publ

ishe

rs.

Thi

sar

ticle

isin

tend

edso

lely

for

the

pers

onal

use

ofth

ein

divi

dual

user

and

isno

tto

bedi

ssem

inat

edbr

oadl

y.

Page 6: Why Are We Disclosing in This Group

bers’ perceptions of engagement. Before con-structing the quadratic and cross product termsthe group member’s group climate rating andthe other group members’ group climate ratingswere centered at the midpoints of their respec-tive scales.

In our analyses the other group member’smean climate score is calculated without thetarget individual’s climate score. Because En-gagement and reasons for self-disclosing wereobtained at each group session, we used thesession number to control for the passage oftime. The specific HLM model examined isdepicted in Appendix B.

The coefficients from the HLM model areused to calculate test values for four lines alongthe response surface: (a) the slope of the line ofagreement (a group member’s Engagement �the other group members’ Engagement); (b) thecurvature along the line of agreement; (c) theslope of the line of disagreement (a group mem-ber’s Engagement � � the other group mem-bers’ Engagement); and (d) the curvature alongthe line of disagreement. The equations outlinedin Edwards and Parry (1993) and Edwards(1993) were used to calculate the: (a) slopealong the line of agreement (a1 � �200 � �300);(b) curvature along the line of agreement (a2 ��400 � �500 � �600); (c) slope along the line ofdisagreement (a3 � �200 � �300); and (d) cur-vature along the line of disagreement (a4 ��400 � �500 � �600). The lines of agreement anddisagreement and an evaluation of the responsesurface are used to test the hypotheses and to

answer the research questions. The coefficientsfrom the HLM analyses are an intermediary stepand are used to calculate the lines describedabove and to plot the response surface.

Hypothesis 1 concerning agreement in per-ceptions of an engaged group climate will besupported if there are significant positive slopesalong the lines of agreement for Engagement forthe Relational Motivation and Expectation ofHelp reasons for self-disclosing and significantnegative slopes along the lines of agreement forEngagement for the Other-Initiated and Dis-comfort with Silence reasons for self-disclos-ing. For example, when Relational Motivationis the criterion variable, a significant positiveslope along the line of agreement for engagedperceptions would mean that when the groupmember and the other group members agree thatengagement is high the group member has highratings on the Relational motivation scale. Hy-pothesis 1 is also examined by viewing theresponse surface. Figure 1 can provide an ex-ample of how the response surfaces are inter-preted to examine Hypothesis 1. The line ofagreement extends from the nearest to the far-thest corners of the plane. As seen in Figure 1ratings of relational motivations to disclose in-crease as the member and other group membersagree that engagement increases. The figurealso shows that at the highest levels of engage-ment the response surface turns down along theline of agreement. This turning down is anindication of curvature along the line of agree-ment. As noted above, the a1 coefficient will

Figure 1. Response surface for engagement and relational motivation as a reason toself-disclose.

25GROUP CLIMATE AND SELF-DISCLOSURE

Thi

sdo

cum

ent

isco

pyri

ghte

dby

the

Am

eric

anPs

ycho

logi

cal

Ass

ocia

tion

oron

eof

itsal

lied

publ

ishe

rs.

Thi

sar

ticle

isin

tend

edso

lely

for

the

pers

onal

use

ofth

ein

divi

dual

user

and

isno

tto

bedi

ssem

inat

edbr

oadl

y.

Page 7: Why Are We Disclosing in This Group

indicate if the slope observed along the line ofagreement in Figure 1 is significant; whereas thea2 coefficient will indicate if the curvaturealong the line of agreement in Figure 1 is sig-nificant.

Hypothesis 2 concerning disagreement inperceptions of an engaged group climate will besupported if there are significant negative cur-vature along the lines of disagreement for En-gagement for the Relational Motivation and Ex-pectation of Help reasons for self-disclosing andsignificant positive curvature along the lines ofdisagreement for Engagement for the Other-Initiated and Discomfort with Silence reasonsfor self-disclosing. Hypothesis 2 is also exam-ined by viewing the response surface. Figure 1also provides an example of how the responsesurfaces are interpreted to examine Hypothesis2. The line of disagreement extends from theleft corner to the right corner of the plane. Forlevels of relational motivation to disclose on theleft side of the plane, other group member rat-ings of engagement exceed member ratings ofengagement, whereas on the right side of theplane, member ratings of engagement exceedother group members ratings of engagement. Asseen in Figure 1 ratings of relational motiva-tions to disclose decrease on either side of theline of disagreement. This turning down is anindication of curvature along the line of dis-agreement. As noted above, the a4 will confirmif this curvature along the line of disagreementin Figure 1 is significant.

The research questions about the direction ofdisagreement will be answered by examiningthe slopes along the lines of disagreement. Thisresearch question is explored by also examiningthe line of disagreement. This time, however,we are interested if there is a positive or nega-tive slope along the line of disagreement. Inexamining Figure 1 it appears that the slopealong the line of disagreement is relatively flat.As noted above, the a3 coefficient will confirmif the slope3 along the line of disagreement inFigure 1 is significant.

Results

A pattern-mixture model was used to see ifmissing data affected engagement and reasonsfor self-disclosure. The results of this analysissuggested that the data could be consideredmissing at random. Four completely uncondi-

tional, three-level hierarchical linear modelswere run to partition the variance in RelationalMotivation, Expectation of Help, Other-Initiated, and Discomfort with Silence as rea-sons for self-disclosing into between-sessions,between-groups-members and between-groupsvariance components.

For Relational Motivation, 51.34% of thevariance was between sessions, 43.42% of thevariance was between group members, and5.23% of the variance was between groups. The�2 statistics for the between-members variancecomponent (df � 82, n � 646, �2 � 407.51)was significant (p � .001) but the between-groups variance component (df � 13, n � 646,�2 � 23.27) was not significant (p � .05),meaning that there was significant variance be-tween group members in their Relational Moti-vation for self-disclosing but not betweengroups. For Expectation of Help, 63.44% of thevariance was between sessions, 35.93% of thevariance was between group members, and0.63% of the variance was between groups. The�2 statistics for the between-members variancecomponent (df � 82, n � 646, �2 � 430.58)was significant (p � .001) but the between-groups variance component (df � 13, n � 646,�2 � 16.03) was not significant (p � .10),meaning that there was significant between-group member variance in Expectation of Helpbut not between groups. For Discomfort withSilence, 76.98% of the variance was betweensessions, 20.98% of the variance was betweengroup members, and 2.04% of the variance wasbetween groups. The �2 statistics for the be-tween-member variance component (df � 82,n � 646, �2 � 242.87) was significant (p �.001) but the between-groups variance compo-nent (df � 13, n � 646, �2 � 19.42) was notsignificant (p � .10), meaning that there wassignificant between group member variance inDiscomfort with Silence, but not betweengroups. Group was fixed in for these three as-pects of self-disclosure in subsequent analyses.

For Other-Initiated self-disclosure, 59.62%of the variance was between sessions, 30.67%of the variance was between group members,and 9.71% of the variance was between groups.The �2 statistics for the between-members vari-ance component (df � 82, n � 646, �2 �247.37) was significant (p � .001) and the be-tween-groups variance component (df � 13,n � 646, �2 � 36.78) was also significant (p �

26 HSU, PAQUIN, AND KIVLIGHAN

Thi

sdo

cum

ent

isco

pyri

ghte

dby

the

Am

eric

anPs

ycho

logi

cal

Ass

ocia

tion

oron

eof

itsal

lied

publ

ishe

rs.

Thi

sar

ticle

isin

tend

edso

lely

for

the

pers

onal

use

ofth

ein

divi

dual

user

and

isno

tto

bedi

ssem

inat

edbr

oadl

y.

Page 8: Why Are We Disclosing in This Group

.01), meaning that there was significant vari-ance both between group member and between-groups variance in Other-Initiated self-disclo-sure.

HLM Results

Table 1 displays the results of the hierarchicallinear polynomial model for P-G fit for an en-gaged group climate and the four reasons forself-disclosing. As member-rated engagementincreased, the group member was more likely todisclose due to Relational Motivation (� �0.03, S.E � 0.01, t (df � 67) � 2.28, p � .03)and Expectations for Help, � � 0.04, S.E �0.02, t (df � 67) � 2.37, p � .02. As othermember ratings of engagement increased, the

member was less likely to disclose due to Dis-comfort with Silence, � � �0.06, S.E � 0.03,t (df � 67) � �2.16, p � .03.

For Other-Initiated Self-Disclosure both qua-dratic terms were significant, � � �0.02, S.E �0.01, t (df � 67) � �2.25, p � .04 and � �0.06, S.E � 0.02, t (df � 67) � 2.67, p � .02.Because these quadratic terms were used todefine the response surface and calculate thecurvatures along the lines of agreement anddisagreement they will be discussed below.

Tests of the Hypotheses

To examine the effects of P-G fit in ratings ofengagement we examined the quadratic and in-teraction coefficients and the derived response

Table 1Hierarchical Linear Polynomial Model for P-G Fit for an Engaged GroupClimate and Reasons for Self-Disclosing

Effect Y SE T p

Relational motivationIntercept 0.37 0.03 11.90 0.00Session slope �0.00 0.00 �0.100 0.32Member engaged 0.03 0.01 2.28 0.03Other group members engaged 0.01 0.02 0.64 0.52Member engaged2 �0.01 0.01 �1.05 0.30Other member engaged2 �0.02 0.02 �0.90 0.37Member by other members Interaction �0.03 0.02 �1.38 0.17

Expectation of helpIntercept 0.23 0.03 8.62 0.00Session slope �0.01 0.00 �1.26 0.21Member engaged 0.04 0.02 2.37 0.02Other group members engaged 0.02 0.03 0.77 0.45Member engaged2 �0.01 0.01 �0.66 0.51Other member engaged2 �0.02 0.02 �0.61 0.54Member by other members Interaction �0.03 0.03 �1.13 0.26

Other-initiated self-disclosureIntercept 0.23 0.03 8.24 0.00Session slope 0.00 0.01 0.28 0.78Member engaged 0.01 0.01 0.90 0.39Other group members engaged �0.02 0.02 �0.80 0.44Member engaged2 �0.02 0.01 �2.25 0.04Other member engaged2 0.06 0.02 2.67 0.02Member by other members Interaction �0.03 0.02 �1.47 0.17

Discomfort with silenceIntercept 0.15 0.02 6.90 0.00Session slope 0.00 0.01 0.66 0.51Member engaged �0.03 0.02 �1.81 0.07Other group members engaged �0.06 0.03 �2.16 0.03Member engaged2 0.00 0.01 0.19 0.85Other member engaged2 0.01 0.03 0.30 0.76Member by other members Interaction �0.01 0.029 �0.46 0.64

Note. For Relational Motivation, Expectations for Help, and Discomfort With Silence, df �95; for Other-Initiated self-disclosure, df � 13.

27GROUP CLIMATE AND SELF-DISCLOSURE

Thi

sdo

cum

ent

isco

pyri

ghte

dby

the

Am

eric

anPs

ycho

logi

cal

Ass

ocia

tion

oron

eof

itsal

lied

publ

ishe

rs.

Thi

sar

ticle

isin

tend

edso

lely

for

the

pers

onal

use

ofth

ein

divi

dual

user

and

isno

tto

bedi

ssem

inat

edbr

oadl

y.

Page 9: Why Are We Disclosing in This Group

surface. The slope and curvature of the responsesurface along the lines of agreement and dis-agreement for engagement and reasons for self-disclosing are displayed in Table 2. We alsoprovide a graphical representation of the re-sponse surface for the relationship between en-gagement and reasons for self-disclosing in Fig-ures 1 through 4.

The slope of the response surface along theline of fit for Relational Motivation for self-disclosure shows the effect of fit at high and lowlevels of group member’s and other group mem-bers’ engaged ratings. As hypothesized, therewas a significant positive slope along the line offit (a1 � 0.05, p � .042). This positive slopewhich can be seen in Figure 1. This figureshows that, when the member and group agreethat engagement is increasing, a group mem-ber’s self-disclosures based upon RelationalMotivation are also increasing. The value for a2,(i.e., the curvature along the line of agreement)shows that the line of agreement is significantlycurvilinear for Relational Motivation for self-disclosing (a2 � �0.06, p � .006). In examiningFigure 1, the scores along the line of agreement dropoff at the higher levels of engagement. Although

reasons related to Relational Motivation forself-disclosing are increasing as the group mem-ber’s and the other group members’ ratings ofan engaged group climate are increasing, therate of this increase is less at the higher levels ofan engaged group climate (see Figure 1).

As hypothesized, there was a significant pos-itive slope along the line of fit (a1 � 0.06, p �.017) for Expectations of Help. This positiveslope is seen in Figure 2. The figure shows that,when the member and other group membersagree that engagement is increasing, the groupmember’s disclosures for Expectation of Helpare also increasing. The value for a2, (i.e., thecurvature along the line of agreement) showsthat the line of agreement is significantly curvi-linear for Expectation of Help (a2 � �0.05, p �.049). Although reasons related to Expecta-tions of Help for self-disclosing are increas-ing as the group member’s and the othergroup members’ ratings of an engaged groupclimate are increasing, the rate of this in-crease is less at the higher levels of an en-gaged group climate (see Figure 2).

Contrary to our hypothesis, there was not asignificant negative slope along the line of fit(a1 � �0.01, p � .825) for Other-Initiatedself-disclosures. The value for a2, (i.e., the cur-vature along the line of agreement) shows thatthe line of agreement is significantly curvilinearfor Other-Initiated reasons for self-disclosing(a2 � �0.07, p � .002). As seen in Figure 3ratings of Other-Initiated reasons for self-disclosures are at their lowest levels when themembers agree that engagement is low or theyagree that engagement is high. When the mem-ber and other group members agree that engage-ment is moderate, Other-Initiated reasons forself-disclosures are at their highest.

As hypothesized, there was a significant neg-ative slope along this line of fit (a1 � �0.10,p � .004) for Discomfort with Silence. Thisnegative slope along the line of agreement canbe seen in Figure 4. The figure shows that, whena member and the other group members agreethat engagement is increasing, self-disclosuresmade due to Discomfort with Silence are de-creasing. The value for a2, (i.e., the curvaturealong the line of agreement) shows that the lineof agreement is not significantly curvilinear forDiscomfort with Silence as a reason for self-disclosing (a2 � 0.00, p � .935). As seen inFigure 4, as member and other group member

Table 2Slopes and Curvatures of the Response SurfaceAlong the Lines of Agreement and Disagreementfor an Engaged Group Climate and Reasons forSelf-Disclosing

Effect coefficient SE t(38) p

Relational motivationa1 Slope along x � y 0.05 0.02 2.36 0.042a2 Curvature along x � y �0.06 0.02–2.74 0.006a3 Slope along x � �y 0.02 0.03 0.63 0.530a4 Curvature along x � �y 0.00 0.03 0.03 0.975

Expectations for helpa1 Slope along x � y 0.06 0.02 2.38 0.017a2 Curvature along x � y �0.05 0.03–1.98 0.049a3 Slope along x � �y 0.02 0.03 0.63 0.528a4 Curvature along x � �y 0.01 0.03 0.16 0.87

Other-initiateda1 Slope along x � y �0.01 0.02–0.22 0.825a2 Curvature along x � y �0.07 0.02–3.11 0.002a3 Slope along x � �y 0.03 0.03 1.07 0.283a4 Slope along x � �y �0.01 0.03–0.226 0.821

Discomfort with silenceA1 Slope along x � y �0.10 0.03–2.93 0.004a2 Curvature along x � y 0.00 0.03–0.08 0.935a3 Slope along x � �y 0.03 0.04 0.86 0.390a4 Curvature along x � �y 0.02 0.04 0.66 0.508

28 HSU, PAQUIN, AND KIVLIGHAN

Thi

sdo

cum

ent

isco

pyri

ghte

dby

the

Am

eric

anPs

ycho

logi

cal

Ass

ocia

tion

oron

eof

itsal

lied

publ

ishe

rs.

Thi

sar

ticle

isin

tend

edso

lely

for

the

pers

onal

use

ofth

ein

divi

dual

user

and

isno

tto

bedi

ssem

inat

edbr

oadl

y.

Page 10: Why Are We Disclosing in This Group

agree that engagement is at higher levels there isa steeper slope (increasingly less likelihood ofdisclosing because of discomfort with silence)than when the member and other group mem-bers agree that engagement is moderate or low.

The curvature along the line of “nonfit” isused to see how a lack of agreement between agroup member’s and the other group members’perception of group climate is related to thegroup member’s reasons for self-disclosing.The line of nonfit extends from the left corner tothe right corner of the x-y plane. We predictedthat when disagreement exists between a groupmember and her or his group concerning thelevel of Engagement, group members are less

likely to self-disclose for reasons related to Re-lational Motivation and Expectation of Help,and will be more likely to cite Other-Initiatedand Discomfort with Silence as reasons for self-disclosure. None of the predicted relationshipsbetween a lack of fit on the dimension of GroupEngagement and reasons for self-disclosingwere significant. In addition, the direction of therelationship of nonfit was not significant for anyof the reasons for disclosure.

Discussion

As described above, group member self-disclosure is the life blood of interactional

Figure 2. Response surface for engagement and the expectation of help as a reason forself-disclosing.

Figure 3. Response surface for engagement and other initiated reasons for self-disclosing.

29GROUP CLIMATE AND SELF-DISCLOSURE

Thi

sdo

cum

ent

isco

pyri

ghte

dby

the

Am

eric

anPs

ycho

logi

cal

Ass

ocia

tion

oron

eof

itsal

lied

publ

ishe

rs.

Thi

sar

ticle

isin

tend

edso

lely

for

the

pers

onal

use

ofth

ein

divi

dual

user

and

isno

tto

bedi

ssem

inat

edbr

oadl

y.

Page 11: Why Are We Disclosing in This Group

group counseling, as self-disclosure sets in mo-tion many of the therapeutic factors describedby Yalom and Lescsz (2005). Therefore, under-standing the correlates of various reasons groupmembers choose to self-disclose is an importantinterim step to gaining a better understanding ofhow to facilitate meaningful disclosure amonggroup members. In the current study, we testedMacKenzie’s (1983) contention that it is agroup’s climate that encourages the expressionof interpersonal events, including group-members’ reasons for self-disclosing.

We used the approach and avoidance moti-vations described by Hurley (1997) and Elliotand Thrash (2002) and Byrne’s (1971) similar-ity-attraction model to formulate specific hy-potheses concerning the relationships betweenfit in engagement and reasons for self-disclos-ing. As hypothesized, reasons for self-disclos-ing due to Relational Motivation and an Expec-tation of Help can be considered to be internallyprompted, approach-oriented motives whichcan further be conceptualized as being related toa more engaged group climate. Other-Initiateddisclosures or disclosures made due to a Dis-comfort with Silence can be construed as exter-nally prompted, and may fall on the spectrum ofavoidance-oriented motives—which can furtherbe conceptualized as being related to less en-gaged group climates. Our results suggest thatthe internal versus external distinction may bean important one for future research into rea-sons for self-disclosing in groups.

A group member’s perception of engagementwithin her or his group was consistently relatedto the reasons that group members gave forself-disclosing in their groups. Specifically, amember’s perception of an engaged group cli-mate was positively related to self-disclosuresmade because the group member believed thedisclosure would strengthen relationshipsamong members (Relational Motivation) orbenefit the group member making the disclosurein some way (Expectation of Help). This sug-gests that a member’s perception of engagementwithin her or his group is related to the reasonscited for making self-disclosures.

Finding a relationship between a group mem-ber’s own perception of an engaged group cli-mate and self-disclosure is not surprising. Anumber of other studies have found similar re-lationships between a member’s perceptions ofan engaged group climate and other aspects ofgroup process (see McClendon & Burlingame,2010). What makes the current study’s findingsunique is that these group member effects werefound while simultaneously controlling for theeffects of the other group members’ perceptionsof the groups’ engaged climate. Thus, there isincreased validity and reliability of the observedrelationship between group engaged climateand reasons for self-disclosure. Writing aboutresearch with dyads, Cook and Kenny (2005)argue that actor effects can only be measuredaccurately when partner effects are controlled.An analogous situation arises in group counsel-

Figure 4. Response surface for engagement and discomfort with silence as reason forself-disclosing.

30 HSU, PAQUIN, AND KIVLIGHAN

Thi

sdo

cum

ent

isco

pyri

ghte

dby

the

Am

eric

anPs

ycho

logi

cal

Ass

ocia

tion

oron

eof

itsal

lied

publ

ishe

rs.

Thi

sar

ticle

isin

tend

edso

lely

for

the

pers

onal

use

ofth

ein

divi

dual

user

and

isno

tto

bedi

ssem

inat

edbr

oadl

y.

Page 12: Why Are We Disclosing in This Group

ing research. Group members (actors) are con-tinuously being influenced by the other mem-bers of their groups (partners). Therefore, theeffects of the other group members’ (partner)climate perceptions must be controlled in orderto accurately measure the effects of the groupmember’s (actor) own climate perceptions onher or his reasons for self-disclosing. Futureresearch on group climate perceptions shouldutilize this framework for understanding theproblem of data interdependence to control forpartner effects when examining actor effects.

Other group members’ perceptions of an en-gaged group climate were negatively related toa group member disclosing due to her or hisdiscomfort with silence; in other words, a groupmember was more likely to disclose during ses-sions because she or he was uncomfortable withsilence when the other members of her or hisgroup perceived a low level of group engage-ment. However, there was no relationship be-tween other members’ perceptions of thegroup’s climate and the other three factors weexamined regarding reasons for self-disclosing(prompting from others, the expectation of re-ceiving help, and relational motivation). Failureto detect these additional relationships betweenother members’ ratings of an engaged groupclimate and a member’s reasons for self-disclosing was surprising. There are now sev-eral studies that have used the actor partnerinterdependence model to examine the effectsof the other group member’s perceptions orbehaviors in counseling groups on individualgroup member’s perceptions or behaviors. Ineach of these studies there have been significanteffects for the other group members (Kivlighan,2011; Kivlighan, Kivlighan, & Cole, 2012;Miles, Paquin, & Kivlighan, 2011; Paquin,Miles, Kivlighan, 2011). These studies havefound other group members effects on bothbehaviors (attendance and in-session “intimate”behaviors) and perceptions (session evaluations).Notably, intimate behaviors and absences areobservable behaviors, whereas reasons for self-disclosing are unobservable, internal processes.It is possible that behaviors are more susceptibleto the influence of other group members thanare internal reactions. However, Kivlighan(2011) found that only other group members’ratings of therapeutic factors (not the member’sown rating of therapeutic factors) were relatedto the member’s perception of session smooth-

ness. Ratings of session smoothness, similar toreasons for self-disclosure, are internal reac-tions to the session. Therefore, it is not clearwhy there were not more partner effects forother group members’ engaged climate ratingsand reasons for self-disclosure. From this study,it appears that a group member’s own percep-tion of the group’s engaged climate may bemore important than the other group members’perception of the group’s engaged climate interms of reasons for self-disclosing. All of theother studies examining partner effects haveused western group members. Therefore, thedifference in the prominence of partner effectsmay be cultural.

Fit With the Group (Agreement About theGroup Climate)

The fit (agreement) between the group mem-ber and other group members were significantpredictors of a group member’s reasons for self-disclosing. As hypothesized, when both themember and the other group members saw thegroup as engaged, the group member was morelikely to self-disclose for both relationally mo-tivated reasons and the expectation of receivinghelp, and less likely to self-disclose becausethey felt uncomfortable with silence. Agree-ment between group members on the level ofengagement present in the group also can beconceptualized as a “strong situation” (Mischel,1976). In strong situations, all group membersperceive a similar engaged group climate,thereby inducing uniform expectations for be-havior, including individual reactions to inter-personal events (Mischel, 1976). From this per-spective, a group with a strong engaged climate(i.e., high agreement or P-G fit) should produceuniform reactions from the group members.

When there is consistent agreement amongmembers that the group’s climate is highly en-gaged, the group members cite more “ap-proach” reasons for their self-disclosures. Whenthere is consistent agreement among membersthat the group’s climate is low in engagement (a“negative” group climate) then group membersoffer “avoidant” explanations for their self-disclosures. Therefore, agreement on climateperceptions is important because it is an indica-tion of a “strong” group climate with consistentexpectations for group member behavior.

31GROUP CLIMATE AND SELF-DISCLOSURE

Thi

sdo

cum

ent

isco

pyri

ghte

dby

the

Am

eric

anPs

ycho

logi

cal

Ass

ocia

tion

oron

eof

itsal

lied

publ

ishe

rs.

Thi

sar

ticle

isin

tend

edso

lely

for

the

pers

onal

use

ofth

ein

divi

dual

user

and

isno

tto

bedi

ssem

inat

edbr

oadl

y.

Page 13: Why Are We Disclosing in This Group

In groups in which there was agreementamong members that the group’s climate wasengaged, there was significant curvature alongthe line of fit (agreement) for Relational Moti-vation, Expectation of Help, and Other-Initiatedreasons for self-disclosing. The form of thiscurvature reached its highest levels when groupmembers were in agreement about the level ofEngagement, and then leveled off (see Figure1). This curvature can be best explained interms of the “law of diminishing returns” (Mill,1848). The law of diminishing returns refers tothe tendency for increases in a unit or construct(i.e., the level of engagement in a group’s cli-mate) to eventually result in a slowed rate ofreturn on effectiveness (i.e., the ability to fostera particular reason for self-disclosure) after acertain level of the construct (i.e., engaged cli-mate) has been achieved. Furthermore, whenthere is agreement between the group memberand the other group members, the law of dimin-ishing returns implies that there is a “goodenough” level of engagement in a group’s cli-mate.

We hypothesized that disagreement (nonfit)between members as to the group’s engagedclimate would be related to reasons for self-disclosure; however, in the opposite direction;specifically, that given a more ambiguous groupclimate, members would disclose for reasonsrelated to being asked by others to disclose orfeeling uncomfortable with silence. However,this was clearly not the case. Neither the degreeof “nonfit” nor the direction of “nonfit” wasrelated to any of the reasons that the membersgave for self-disclosing. In retrospect, ourhypothesis was not sound. When a memberand the other group members disagree aboutthe climate there is a “weak situation” (Mis-chel, 1976). In a weak situation, group mem-bers do not perceive the climate the sameway, and there are inconsistent behavioralexpectations. Mischel argues that individualdifferences—not the group climate—will de-termine behavior in ambiguous or “weak”situations. With a weak group climate, groupmembers will expect that any motivation forself-disclosure (e.g., internally vs. externallyprompted, approach or avoiding) is appropri-ate. When there is disagreement about theclimate, the reasons that group members givefor self-disclosing should be idiosyncratic and

therefore unpredictable based on the environ-ment.

Strengths and Limitations

This was only the second study to use poly-nomial analysis and response surface method-ology to examine person-group fit in counselinggroups. The results showed that perceptual fitconcerning engagement was an important pre-dictor of why people make self-disclosures ingroups. The advantage of using response sur-face analysis in the current study is that it al-lowed us to resolve a problem present in othercounseling research concerning the lack of in-formation about agreement or disagreement.Future research also should consider analyzingperceptual fit when analyzing group climate da-ta. Another strength of the study is that theresults are based on a relatively large number ofparticipants and groups; therefore, power is notan issue when interpreting nonsignificant find-ings.

One of the major strengths of the currentstudy is its use of a nonwestern sample. Specif-ically, the current study’s focus on Taiwanesecollege students’ and community members’ ex-periences participating in counseling groupsadds to the group literature in important ways.The issue of whether or not a group memberself-discloses during sessions may be a cultur-ally embedded phenomenon (DeLucia-Waack& Donigian, 2004). Specifically, whether or notgroup members choose to self-disclose may bedependent on cultural norms surrounding boththe process of self-disclosing (e.g., a desire torefrain from interrupting or putting someone onthe spot; a desire to communicate support non-verbally, etc.) and the content of what is beingdisclosed (e.g., sharing personal informationabout self, about one’s family; verbalizing emo-tions, etc.; DeLucia-Waack & Donigian, 2004).Future empirical research should examine towhat extent group members’ reasons for self-disclosing are linked with cultural context andidentity, and include an examination of cultur-ally different groups, including Asian Americanand non-Asian samples.

Another strength of the current study is thatboth group climate and self-disclosure datawere collected across the life of the group—something identified by group researchers asbeing crucial to developing an understanding of

32 HSU, PAQUIN, AND KIVLIGHAN

Thi

sdo

cum

ent

isco

pyri

ghte

dby

the

Am

eric

anPs

ycho

logi

cal

Ass

ocia

tion

oron

eof

itsal

lied

publ

ishe

rs.

Thi

sar

ticle

isin

tend

edso

lely

for

the

pers

onal

use

ofth

ein

divi

dual

user

and

isno

tto

bedi

ssem

inat

edbr

oadl

y.

Page 14: Why Are We Disclosing in This Group

important group processes (Kivlighan, Cole-man, & Anderson, 2000; Zaccaro, Cracraft, &Marks, 2006).

There are also several limitations that must beconsidered in interpreting these results. First,both group climate and reasons for self-disclosure are participant ratings. Therefore,mono-method bias should be considered wheninterpreting our results. Second, the research iscorrelational so we cannot draw causal implica-tions about the relationships between group cli-mate perceptions and reasons for self-disclos-ing. Third, although the study provided a lookinto the reasons clients choose to make self-disclosures, we did not examine the type, quan-tity, or quality of disclosures made. Future re-search should explore these dimensions ofself-disclosure and whether they are related toimportant process and outcome variables. Fu-ture studies should also examine reasons whygroup members choose not to disclose ingroups. Lastly, unlike the Chinese version of theGCQ-S which is commonly used and its valid-ity is well-established, the SDQ has only beenused in three studies (Hsu, 2011; Hsu et al.,2009; Woo et al., 2005). Therefore, future re-search is needed in order to further demonstrateits’ reliability and validity.

Implications for Practice

The “strong situation. ” Contrary to ourhypothesis, when there was disagreementamong members concerning the level of groupengagement, there was no relationship betweenclimate and the reasons members gave for self-disclosing. This lack of a relationship may beattributable to the ambiguousness such a“weak” climate creates, and thus group mem-bers were reporting all sorts of reasons, includ-ing approach and avoiding-oriented reasons formaking self-disclosures. This finding lends sup-port to what may be an important advantage ofestablishing a “strong” versus “weak situation”with regard to group climate (Mischel, 1976).Specifically, one could surmise that a groupleader able to establish a “strong climate” (notnecessarily a “good” climate, but a “strong”one) is one in which all group members per-ceive the presence of an unambiguous groupnorm. The presence of such a clear, stronggroup norm increases the likelihood that mem-bers are engaging in behaviors such as self-

disclosing because of that norm, rather than dueto idiosyncratic reasons based on individual dif-ferences. If a weak situation is present, thegroup’s ability to influence (either positively ornegatively) individual client behavior is attenu-ated and therefore group therapy’s most power-ful mechanism of change—the influence of thegroup on its members—is rendered ineffectual.

Going further, Yalom and Lescsz (2005) con-tend that the group leader’s major job is to builda therapeutic group climate. If group leaderswant their group members to self-disclose forapproach-oriented motives rather than avoid-ance-oriented motives, then they would do wellto create a climate where all of the group mem-bers are in agreement that the group is highlyengaged. Importantly, the current results sug-gested that when the group climate was per-ceived as engaged and that there was agreementabout this by both the individual and the othermembers of her or his group, group membersdisclosed for more internally prompted reasonsas opposed to externally prompted reasons.

An engaged, “strong” climate. Our resultsfurther indicated that group facilitators may bewise to consider not only the disclosures beingmade during sessions, but the motivations be-hind these disclosures; specifically, whether agroup member is disclosing for externally ver-sus internally motivated reasons, and whetherthe disclosures are in service of moving away ormoving toward goals.

Group leaders might also consider whether abehavioral norm (and whether it is strong or weak) isbeing created and reinforced when, for example, agroup member is prompted (once or repeatedly)to disclose by the leader or other members.Although self-disclosure may indeed be the life-blood of group counseling, externally promptedgroup member self-disclosures may have differ-ent treatment implications for the individualmaking them, as well as for other individualmembers and the group as a whole. Consideringsuch factors can provide important data forleaders and members, and can be used tochange, create, or maintain group norms thatfoster more approach-oriented behaviors and“moving toward” goals.

A final implication to consider is the impor-tance of identifying group members whoseviews of the group climate are at variance withthose of the other group members. This type of“nonfit” is not productive for the member or for

33GROUP CLIMATE AND SELF-DISCLOSURE

Thi

sdo

cum

ent

isco

pyri

ghte

dby

the

Am

eric

anPs

ycho

logi

cal

Ass

ocia

tion

oron

eof

itsal

lied

publ

ishe

rs.

Thi

sar

ticle

isin

tend

edso

lely

for

the

pers

onal

use

ofth

ein

divi

dual

user

and

isno

tto

bedi

ssem

inat

edbr

oadl

y.

Page 15: Why Are We Disclosing in This Group

her or his fellow group members as it appears tobe linked with absence and early dropout(Paquin et al., 2013); therefore, group leaderscan identify these “outliers” early and intervenein ways that maximize group engagementamong all members.

References

Byrne, D. (1971). The attraction paradigm. NewYork, NY: Academic Press.

Cook, W. L., & Kenny, D. A. (2005). The Actor-Partner Interdependence Model: A model of bidi-rectional effects in developmental studies. Interna-tional Journal of Behavioral Development, 29,101–109. doi:10.1080/01650250444000405

Crits-Christoph, P., Johnson, J. E., Connolly Gib-bons, M., & Gallop, R. (2013). Process predictorsof the outcome of group drug counseling. Journalof Consulting and Clinical Psychology, 81, 23–34.doi:10.1037/a0030101

DeLucia-Waack, J. L., & Donigian, J. (2004). Psy-choeducational group vignette #2: Asian Americangroup—To disclose or not to disclose? In J. L.DeLucia-Waack & J. Donigian (Eds.), The prac-tice of multicultural group work: Visions and per-spectives from the field (pp. 165–189). Belmont,CA: Brooks/Cole-Thomson Learning.

Doxsee, D. J., & Kivlighan, D. M., Jr. (1994). Hin-dering events in interpersonal relations groups forcounselor trainees. Journal of Counseling and De-velopment, 72, 621–626. doi:10.1002/j.1556-6676.1994.tb01692.x

Edwards, J. R. (1993). Problems with the use ofprofile similarity indices in the study of congru-ence in organizational research. Personnel Psy-chology, 46, 641–665. doi:10.1111/j.1744-6570.1993.tb00889.x

Edwards, J. R., & Parry, M. E. (1993). On the use ofpolynomial regression equations as an alternativeto difference scores in organizational research.Academy of Management Journal, 36, 1577–1613.doi:10.2307/256822

Elliot, A. J., & Thrash, T. M. (2002). Approach-avoidance motivation in personality: Approachand avoidance temperaments and goals. Journal ofPersonality and Social Psychology, 82, 804–818.doi:10.1037/0022-3514.82.5.804

Flowers, J. V. (1987). Client outcome as a function ofagreement or disagreement with the modal groupperception of curative factors in short-term, struc-tured group psychotherapy. International Journalof Group Psychotherapy, 37, 113–118.

Hsu, K. (2011). Investigation for the factors changesof members’ self disclosure in the process of coun-seling groups. Journal of Educational Psychology,42, 655–676.

Hsu, K., & Woo, S., & Hung, Y. F. (2009, August).Developing an instrument for assessing members’participation: Self-disclosure questionnaire.Poster presented at the American PsychologicalAssociation national meeting, Toronto, Canada.

Hurley, J. R. (1997). Interpersonal theory and mea-sures of outcome and emotional climate in 111personal development groups. Group Dynamics:Theory, Research, and Practice, 1, 86–97. doi:10.1037/1089-2699.1.1.86

Johnson, J. E., Pulsipher, D., Ferrin, S. L., Burlin-game, G. M., Davies, D., & Gleave, R. (2006).Measuring group processes: A comparison of theGCQ and CCI. Group Dynamics: Theory, Re-search, and Practice, 10, 136–145. doi:10.1037/1089-2699.10.2.136

Joyce, A. S., MacNair-Semands, R., Tasca, G. A., &Ogrodniczuk, J. S. (2011). Factor structure andvalidity of the Therapeutic Factors Inventory–Short Form. Group Dynamics: Theory, Research,and Practice, 15, 201–219. doi:10.1037/a0024677

Kivlighan, D. M., Jr. (2011). Individual and groupperceptions of therapeutic factors and session eval-uation: An actor–partner interdependence analysis.Group Dynamics: Theory, Research, and Practice,15, 147–160. doi:10.1037/a0022397

Kivlighan, D. M., Jr., Coleman, M. N., & Anderson,D. C. (2000). Process, outcome and methodologyin group counseling research. In S. D. Brown &R. W. Lent (Eds.), Handbook of counseling psy-chology, (3rd ed., pp. 767–796). New York, NY:Wiley.

Kivlighan, D. M., Jr., Kivlighan, D. M., III, & Cole,O. D. (2012). The group’s absence norm and com-mitment to the group as predictors of group mem-ber absence in the next session: An actor–partneranalysis. Journal of Counseling Psychology, doi:10.1037/a0025506

Kivlighan, D. M., Jr., Multon, K. D., & Brossart,D. F. (1996). Helpful impacts in group counseling:Development of a multidimensional rating system.Journal of Counseling Psychology, 43, 347–355.doi:10.1037/0022-0167.43.3.347

Kristof-Brown, A. L., Zimmerman, R. D., & John-son, E. C. (2005). Consequences of individual’s fitat work: A meta-analysis of person-job, person-organization, person-group, and person-supervisorfit. Personnel Psychology, 58, 281–342. doi:10.1111/j.1744-6570.2005.00672.x

Lo Coco, G., Gullo, S., & Kivlighan, D. M., Jr.(2012). Examining patients’ and other group mem-bers’ agreement about their alliance to the group-as-a-whole and changes in patient symptoms usingresponse surface analysis. Journal of CounselingPsychology, 59, 197–207. doi:10.1037/a0027560

MacKenzie, K. R. (1983). The clinical application ofa group climate measure. In R. R. Dies & K. R.MacKenzie (Eds.), Advances in group psychother-

34 HSU, PAQUIN, AND KIVLIGHAN

Thi

sdo

cum

ent

isco

pyri

ghte

dby

the

Am

eric

anPs

ycho

logi

cal

Ass

ocia

tion

oron

eof

itsal

lied

publ

ishe

rs.

Thi

sar

ticle

isin

tend

edso

lely

for

the

pers

onal

use

ofth

ein

divi

dual

user

and

isno

tto

bedi

ssem

inat

edbr

oadl

y.

Page 16: Why Are We Disclosing in This Group

apy: Integrating research and practice (pp. 159–170). New York, NY: International UniversitiesPress.

McClendon, D. T., & Burlingame, G. M. (2010).Group climate: Construct in search of clarity. InR. K. Conyne (Ed.), The Oxford handbook ofgroup counseling (pp. 164–181). New York, NY:Oxford University press.

Miles, J. R., Paquin, J. D., & Kivlighan, D. M., Jr.(2011). Amount and consistency, two behaviors ingroups. Group Dynamics: Theory, Research, andPractice, 15, 326–342. doi:10.1037/a0024676

Mill, J. S. (1848). Principles of political economy.London, UK: John W. Parker.

Mischel, W. (1976). Towards a cognitive socialmodel learning reconceptualization of personality.In N. S. Endler & D. Magnusson (Eds.), Interac-tional psychology and personality (pp. 166–207).New York, NY: Wiley.

Paquin, J. D., Kivlighan, D. M., Jr., & Drogosz, L. M.(2013). Person–group fit, group climate, and out-comes in a sample of incarcerated women partic-ipating in trauma recovery groups. Group Dynam-ics: Theory, Research, and Practice, 17, 95–109.doi:10.1037/a0032702

Paquin, J. D., Miles, J. R., Kivlighan, D. M., Jr.(2011). Predicting group attendance using in-session behaviors. Small Group Research, 42,177–198. doi:10.1177/1046496410389493

Shanock, L., Baran, B. E., Gentry, W. A., Pattison,S., & Heggestad, E. D. (2010). Polynomial regres-sion with response surface analysis: A powerfulapproach for examining moderation and overcom-ing limitations of difference scores. Journal ofBusiness and Psychology, 25, 543–554. doi:10.1007/s10869-010-9183-4

Shechtman, Z., & Rybko, J. (2004). attachment styleand observed initial self-disclosure as explanatoryvariables of group functioning. Group Dynamics:Theory, Research, and Practice, 8, 207–220. doi:10.1037/1089-2699.8.3.207

Toren, Z., & Shechtman, Z. (2010). Association ofpersonal, process, and outcome variables in groupcounseling: Testing an exploratory model. GroupDynamics: Theory, Research, and Practice, 14,292–303. doi:10.1037/a0018872

Tschuschke, V., & Dies, R. R. (1997). The contribu-tion of feedback to outcome in long-term grouppsychotherapy. Group, 21, 3–15. doi:10.1007/BF02872750

Vogel, R. M., & Feldman, D. C. (2009). Integratingthe levels of person-environment fit: The roles ofvocational fit and group fit. Journal of VocationalBehavior, 75, 68–81. doi:10.1016/j.jvb.2009.03.007

Wang, L. F., & Lin, M. J. (2000). The developmentof the “Scale of Group Therapeutic Factors”. Chi-nese Journal of Guidance and Counseling, 9,1–24.

Wang, Y. L., Chen, M. J., Wang, L. F., & Lin, M. J.(2012). Application of multilevel latent variablemodels on perceived and contextual group climate.Chinese Journal of Guidance and Counseling, 32,33–55.

Woo, S., Hsu, Y. K., & Hung, Y. F. (2005, Septem-ber). Why disclose or not? Investigating the rea-sons for members’ self-disclosure in counselinggroups. In C. C. Chao (Chair), Research of variouspsychotherapeutic approaches. Symposium con-ducted at the 7th Pacific Rim Regional Congress ofGroup Psychotherapy & 4th Asia Pacific Confer-ence on Psychotherapy, Taipei, Taiwan.

Yalom, I. D., & Leszcsz, M. (2005). The theory andpractice of group psychotherapy (5th ed.). NewYork, NY: Basic Books.

Zaccaro, S. J., Cracraft, M., & Marks, M. (2006). Col-lecting data from groups. In F. T. L. Leong & J. T.Austin (Eds.), The psychology research handbook:A guide for graduate students and research assis-tants (pp. 227–238). Thousand Oaks, CA: Sage.

(Appendices follow)

35GROUP CLIMATE AND SELF-DISCLOSURE

Thi

sdo

cum

ent

isco

pyri

ghte

dby

the

Am

eric

anPs

ycho

logi

cal

Ass

ocia

tion

oron

eof

itsal

lied

publ

ishe

rs.

Thi

sar

ticle

isin

tend

edso

lely

for

the

pers

onal

use

ofth

ein

divi

dual

user

and

isno

tto

bedi

ssem

inat

edbr

oadl

y.

Page 17: Why Are We Disclosing in This Group

Appendix A

SDQ: Reasons for Self-Disclosure, Four Factors, and Associated Items

Factor I: Relational Motivation(Eight Items)

2. Group members showed their concernwhen someone else shared her/his difficulties orproblems.

11. Group members gave some responses tothe person who talked about her/himself, andseeing this helped me to share something per-sonal.

12. I believed the group members would keepmy personal issues private.

18. I felt a sense of camaraderie.21. The group climate was active and all of us

were discussing our personal issues.7. I shared after someone else did because I

could relate to her/his experience.16. I hope to be known more after my shar-

ing.13. I think I could share a little bit for under-

standing the responses of other members.

Factor II: Expectation for Help (Six Items)

6. I expected the members could give mesome feedback.

10. I thought the leader would assist me withclarifying my concerns/issues.

20. I expected that the leader could give mesome feedback.

14. I thought the group members would assistme with clarifying my concerns/issues.

1. I want to deal with my personal difficultiesand problems sooner rather than later.

17. I think the members would give me sup-port if I disclosed.

Factor III: Other-Initiated (Five Items)

9. The leader looked at me.4. The leader verbally invited me to share.15. The leader nonverbally invited me to

share.19. The group was taking turns sharing and it

was my turn.3. Everyone looked at me and I felt pressure

to speak.

Factor IV: Discomfort With Silence(Two Items)

8. I thought someone should speak to breakthe group’s silence.

5. No one spoke up when the leader invitedus to share.

(Appendices continue)

36 HSU, PAQUIN, AND KIVLIGHAN

Thi

sdo

cum

ent

isco

pyri

ghte

dby

the

Am

eric

anPs

ycho

logi

cal

Ass

ocia

tion

oron

eof

itsal

lied

publ

ishe

rs.

Thi

sar

ticle

isin

tend

edso

lely

for

the

pers

onal

use

ofth

ein

divi

dual

user

and

isno

tto

bedi

ssem

inat

edbr

oadl

y.

Page 18: Why Are We Disclosing in This Group

Appendix B

Multilevel Model

The Level 1 model was:Y (Current Session’s Reason for Self-

Disclosing) � 0i � 1i (Session Number) � 2i(Group Member’s Engagement rating) � 3i(Other Group Members’ Average EngagementRatings) � 4i (Group Member’s EngagementRating2) � 5i (Product of the Group Member’sEngagement Rating and Other Group Members’Engagement Ratings) � 6i (Other GroupMembers’ Average Engagement Ratings2) � ε

The Level 2 model was:0i � 00 � r01i � 10 � r12i � 20 � r23i � 30 � r34i � 40 � r45i � 50 � r56i � 60 � r6

The Level 2 model shows that there were nogroup member level predictors of the groupmember’s reason for self-disclosing.

The Level 3 model was:00 � �000 � U0010 � �100 � U0120 � �200 � U0230 � �300 � U0340 � �400 � U0450 � �500 � U0560 � �600 � U06The Level 3 model shows that there were no

group level predictors of the group member’sreason for self-disclosing.

Received October 22, 2012Revision received August 22, 2013

Accepted September 9, 2013 �

37GROUP CLIMATE AND SELF-DISCLOSURE

Thi

sdo

cum

ent

isco

pyri

ghte

dby

the

Am

eric

anPs

ycho

logi

cal

Ass

ocia

tion

oron

eof

itsal

lied

publ

ishe

rs.

Thi

sar

ticle

isin

tend

edso

lely

for

the

pers

onal

use

ofth

ein

divi

dual

user

and

isno

tto

bedi

ssem

inat

edbr

oadl

y.