NO PLAGIARISM DUE MONDAY NOVEMBER 11, 2019. ATTACHED ARE RESOURCES TO ASSIST WITH ASSIGNMENT.

The Attenuating Effect of Depression Symptoms on Negative-Affect Expression: Individual and Group Effects in Group Psychotherapy for

Personality Disorders

Daniel W. Cox and David Kealy University of British Columbia

Jeffrey H. Kahn Illinois State University

Katharine D. Wojcik University of British Columbia

Anthony S. Joyce University of Alberta

John S. Ogrodniczuk University of British Columbia

Across a breadth of psychotherapeutic approaches, feeling affect intensely and then talking about those feelings is a common means for increasing insight and other desired outcomes. While several naturalistic and laboratory studies have found that depression symptoms attenuate (i.e., weaken) the association between negative-affect intensity and negative-affect expression, depression’s attenuating effect has not been examined in a psychotherapeutic context. The first aim of the present study was to examine if depression symptoms’ attenuating effect on the association between negative-affect intensity and negative-affect expression extended into group psychotherapy. Our second aim was to examine group effects on patients’ negative-affect expression. Participants (N � 239) were patients consecutively admitted into a psychodynamic group-psychotherapy day treatment program for people with personality disorders. Patients indicated their negative-affect intensity and negative-affect expression each week that they were in treatment. Depression symptoms were assessed at baseline. Results indicated that depression symptoms attenuated (i.e., moderated) the association between negative-affect intensity and negative- affect expression. Further, while the association between patient intensity and expression increased over the course of treatment, the moderating effect of depression on this association did not vary over treatment. Regarding group effects, group negative-affect intensity was associated with higher levels of patient negative-affect expression. Inversely, group affect expression was associated with lower levels of patient affect expression. Patient depression symptoms did not moderate the association between group negative-affect intensity and patient negative-affect expression. Our findings indicate that while group affect intensity and affect expression impacts patients’ expression, depression’s attenuating effect on negative-affect expression extends to patient effects but not group effects.

Public Significance Statement Our findings indicate that in group therapy, people with higher levels of depression symptoms are less likely to express their distress to others. However, over the course of therapy, their willingness to express distress increases.

Keywords: affect expression, negative affect, depression, personality disorder, psychotherapy

Expressing affect is foundational to counseling and is valued across psychotherapeutic orientations (Whelton, 2004). Further, affect expression has been linked with a breadth of desired out-

comes including psychological health, physical health, and life functioning (Frattaroli, 2006). General consensus regarding the value of affect expression has resulted in researchers investigating

This article was published Online First January 31, 2019. Daniel W. Cox, Counselling Psychology Program, University of British

Columbia; David Kealy, Psychotherapy Program, Department of Psychiatry, University of British Columbia; Jeffrey H. Kahn, Psychology Department, Illinois State University; Katharine D. Wojcik, Counselling Psychology Pro- gram, University of British Columbia; Anthony S. Joyce, Department of

Psychiatry, University of Alberta; John S. Ogrodniczuk, Psychotherapy Pro- gram, Department of Psychiatry, University of British Columbia.

Correspondence concerning this article should be addressed to Daniel W. Cox, Counselling Psychology Program, University of British Colum- bia, 2125 Main Mall, Vancouver, BC V6T1Z4, Canada. E-mail: dan.cox@ubc.ca

T hi

s do

cu m

en t

is co

py ri

gh te

d by

th e

A m

er ic

an P

sy ch

ol og

ic al

A ss

oc ia

ti on

or on

e of

it s

al li

ed pu

bl is

he rs

. T

hi s

ar ti

cl e

is in

te nd

ed so

le ly

fo r

th e

pe rs

on al

us e

of th

e in

di vi

du al

us er

an d

is no

t to

be di

ss em

in at

ed br

oa dl

y.

Journal of Counseling Psychology © 2019 American Psychological Association 2019, Vol. 66, No. 3, 351–361 0022-0167/19/$12.00 http://dx.doi.org/10.1037/cou0000335

351

factors that inhibit expression to facilitate clinical intervention and improve patient outcomes.

Affect expression is an adaptive way to regulate negative affect (Kennedy-Moore & Watson, 2001). Research has revealed a ro- bust association between the intensity of affective experience and affective expression; however, depression symptoms have been found to attenuate (i.e., weaken) the association between negative- affect intensity and negative-affect expression in naturalistic (e.g., Garrison, Kahn, Sauer, & Florczak, 2012) and laboratory (e.g., Campbell-Sills, Barlow, Brown, & Hofmann, 2006) studies. The present study builds on this work by examining the attenuating effect of depression symptoms within a group psychotherapeutic context. We examined this association in a psychodynamic day treatment program for people with personality disorders—a treat- ment in which affect expression was emphasized (Piper, Rosie, Joyce, & Azim, 1996). Because of the group-oriented context of the day treatment program, we evaluated both patient and group effects on patients’ negative-affect expression.

Negative-Affect Expression

When people are upset, conveying their unpleasant experiences to others often reduces their subjective distress (i.e., negative affect; see Kennedy-Moore & Watson, 2001 for a review). This process has been observed when negative affect was expressed verbally and nonverbally. Meta-analytic evidence has indicated that the types of distress reduced by negative-affect expression are broad; examples include feelings of anger, sadness, and fear (Frat- taroli, 2006). The positive association between negative-affect intensity and negative-affect expression has been consistently found in a number of contexts and in a number of populations (e.g., Campbell-Sills et al., 2006; Garrison et al., 2012; Kahn & Garri- son, 2009). These studies found that the more affectively intense people’s experiences were, the more likely they were to convey their experiences to others. The apt analogy of the fever has been used to explain the link between intensity and expression (Stiles, 1995). Fevers covary with and fight infections, just as negative- affect expression covaries with and reduces negative affect. Due to expression’s link with reduced negative affect, affect expression has been conceptualized as an adaptive affect-regulating response to distress. Therefore, influences on the linear association between negative-affect intensity and negative-affect expression can disrupt people’s ability to regulate their affect. Because of these links between negative-affect expression and desired outcomes, re- searchers have attempted to understand what predicts affect ex- pression—and more specifically—why some people are less likely to express negative affect.

Expression in Psychotherapy

Affect expression has been associated with psychotherapeutic process and outcome across a breadth of therapeutic approaches (Orlinsky, Ronnestad, & Willutski, 2004). In a meta-analysis of psychodynamic psychotherapy, affect intensity and expression were found to independently predict patient improvement and differentiate those who succeeded compared with those who did not (Diener, Hilsenroth, & Weinberger, 2007). Another study found that patients’ tendency to discuss their emotions—which were measured at intake—predicted reductions in their perceived

stress and symptoms over the course of therapy (Kahn, Achter, & Shambaugh, 2001). The association between expression and in- sight has been supported in laboratory studies that have found that affect expression improved participants’ understanding of what and why they were feeling (e.g., Pennebaker & Seagal, 1999).

In psychotherapeutic models for patients with personality dis- orders, interpersonal processes generally—and affect expression specifically— have been noted as important due to the social context in which personality disorders are embedded (Paris, 2004; Piper et al., 1996). Affect expression has been considered a central task of psychodynamic psychotherapy, which has a long tradition of being employed in the treatment of patients with personality disorders. Within psychodynamic treatments, a primary goal is for patients to experience previously avoided affective experiences and adaptively express them within an interpersonal context (Svartberg, Stiles, & Seltzer, 2004). Through expression, patients gain personal insight and change (Piper et al., 1996). In a meta- analysis of psychodynamic psychotherapy, affect experience/ex- pression accounted for a 30% improvement in patients’ success rates (Diener et al., 2007).

To further facilitate the interpersonal processes that enable in- sight and change, psychodynamic psychotherapy for patients with personality disorders has been extended into group contexts (Piper et al., 1996). Several studies have demonstrated that group psy- chodynamic treatments for personality disorders have been linked with a number of desired outcomes including reduced anxiety, depression, somatic concerns, and interpersonal problems (see Abbass, Town, & Driessen, 2012 for a review). Within psychody- namic group theory, a major pathway to change is the evocation and expression of powerful feelings in the presence of others followed by cognitive integration that leads to increased awareness and clarity (Rutan, Stone, & Shay, 2014). Extending these mech- anisms of change, other group theorists and researchers have argued that intimate behaviors—affect expression, disclosure of affect and personal information, as well as authentic responses—are a mecha- nism of change that cut across group theories and approaches (Kelly & Barsade, 2001; Shadish, 1984). Through group member engage- ment in intimate behaviors, patients gain awareness and insight about themselves and about their relationships with others that facilitates therapeutic change. Consistently, intimate behaviors in group psychotherapy have been linked with self-discovery, insight, and other therapeutic changes (e.g., Leichtentritt & Shechtman, 2010; Shadish, 1984).

Within group psychotherapy there are several therapeutic fac- tors that facilitate patients’ engagement in affect expression and other intimate behaviors. For example, patients’ observing and then mimicking fellow group members can facilitate increased use of adaptive interpersonal behaviors (Lese & MacNair-Semands, 2000). Further, empathy from other group members has been linked with patients’ increased engagement in group processes (Johnson, Burlingame, Olsen, Davies, & Gleave, 2005). Collec- tively, these group therapeutic factors can enhance patients’ use of negative-affect expression by facilitating patients’ movement to- ward the norm of the group (Yalom & Leszcz, 2005).

Depression and Affect Expression

Due to the value of affect expression in psychotherapy, it is important to understand what inhibits patients’ expression so that

T hi

s do

cu m

en t

is co

py ri

gh te

d by

th e

A m

er ic

an P

sy ch

ol og

ic al

A ss

oc ia

ti on

or on

e of

it s

al li

ed pu

bl is

he rs

. T

hi s

ar ti

cl e

is in

te nd

ed so

le ly

fo r

th e

pe rs

on al

us e

of th

e in

di vi

du al

us er

an d

is no

t to

be di

ss em

in at

ed br

oa dl

y.

352 COX ET AL.

clinicians can assess for and attend to these inhibiting factors to facilitate successful treatment. Several naturalistic (e.g., Garrison et al., 2012) and laboratory (e.g., Campbell-Sills et al., 2006) studies have found that mood disorder symptoms—and depression symptoms specifically—were associated with reduced negative- affect expression. More specifically, symptoms attenuated the af- fect intensity-expression link. In other words, depression symp- toms have been associated with a reduced likelihood of using affect expression to regulate negative affect.

Emotion dysregulation theory has been used to explain the association between depression symptoms and reduced negative- affect expression. Within emotion dysregulation theory, people with higher levels of depression symptoms have a reduced will- ingness to experience psychological distress (Campbell-Sills et al., 2006). While the baseline psychological distress among those with elevated depression symptoms is greater than most, the perceived acute distress that accompanies negative-affect expression pro- duces further anxiety (e.g., Kahn & Garrison, 2009). Thus, those with elevated depression symptoms are more likely to use mal- adaptive emotion regulation strategies such as suppression and avoidance (see Aldao, Nolen-Hoeksema, & Schweizer, 2010 for a review).

Current Study

While there is substantial evidence that depression symptoms attenuate the association between negative-affect intensity and negative-affect expression (e.g., Campbell-Sills et al., 2006; Gar- rison et al., 2012; Kahn & Garrison, 2009), this association has not been examined within psychotherapy. This is a critical omission in the literature, given the centrality of affect expression to insight- oriented therapy and the vast number of psychotherapy patients who present with symptoms of depression. Therefore, the purpose of the present study was to examine if depression symptoms’ attenuating effect on the association between negative-affect in- tensity and negative-affect expression extends into psychotherapy. We examined this association within a psychodynamic day treat- ment (i.e., group psychotherapy) for people with personality dis- orders—a treatment in which affect expression was emphasized (Piper et al., 1996).

To examine whether the attenuating effect of depression symp- toms on the link between negative-affect intensity and negative- affect expression extends into group psychotherapy, we hypothe- sized that patients’ negative-affect intensity would be positively associated with negative-affect expression (Hypothesis 1), depres- sion symptoms would be inversely associated with negative-affect expression (Hypothesis 2), and that the association between negative-affect intensity and negative-affect expression would be attenuated (i.e., weaker) for patients with higher levels of depres- sion symptoms (Hypothesis 3).

We also examined group effects on patient affect expression. Based on group theory indicating that patients are influenced by and move toward the norms of the group, it seems likely that intimate behaviors such as group affect intensity and group affect expression would be associated with patients’ affect expression. Therefore, we hypothesized that group negative-affect intensity would be positively associated with patient negative-affect expres- sion (Hypothesis 4), group negative-affect expression would be positively associated with patient negative-affect expression (Hy-

pothesis 5), and patient depression symptoms would moderate the effect of group negative-affect intensity on patient negative-affect expression (Hypothesis 6).

By considering group and patient effects simultaneously, we were able to examine patient effects while controlling for the group and examine group effects while controlling for the patient. This approach, which is based in actor-partner interdependence models that have been used to study dyads (Kenny, Mannetti, Pierro, Livi, & Kashy, 2002), has become increasingly popular in group psychotherapy research (e.g., Kivlighan & Paquin, 2014; Miles, Paquin, & Kivlighan, 2011). By understanding the unique effects of the patient and the group, theoretical and clinical impli- cations are able to be more precise.

Method

Participants

Participants (N � 239) were patients consecutively admitted into an intensive day treatment program for people with person- ality disorders. Participants’ ages ranged from 18 to 68 (M � 37.15, SD � 11.03). The sample was 70.7% female; 89.1% Cau- casian, followed by 2.1% Aboriginal and 1.7% Asian; 19.7% had less than a high school education, 40.6% completed high school or had some college, and 39.8% had a college degree or more; 42.3% were partnered, 29.3% were single, 26.4% were divorced or sep- arated, and 2.1% were widowed. Regarding personality disorders, 13.8% had a Cluster A disorder, 39.3% had a Cluster B disorder, and 50.2% had a Cluster C disorder. The mean number of person- ality disorders was 1.32 (SD � 1.26) with 36.3% having been diagnosed with two or more personality disorders.

Data for the present study were collected as part of continued naturalistic research following controlled studies of the intensive day treatment program (e.g., Piper et al., 1996). The affiliated research ethics boards approved this study and all participants provided informed consent.

Treatment

Participants were enrolled in an 18-week intensive day treat- ment program designed to improve the mental health and func- tioning of people with personality disorders (Piper et al., 1996). The program was developed to provide a broad-based, multidi- mensional treatment for people with personality disorders rather than targeting a specific personality disorder. Personality disorder researchers have argued that nondisorder-specific personality dis- order treatments are preferred because there is overlap between disorders, many patients are diagnosed with more than one per- sonality disorder, and nonspecific treatments are more generaliz- able (e.g., Crits-Cristoff & Barber, 2004; Livesley, 2012). To be enrolled, participants had to meet criteria for a Diagnostic and Statistical Manual of Mental Disorders (4th ed.; DSM–IV; Amer- ican Psychiatric Association, 1994) personality disorder and be at least 18 years old. People were not considered appropriate for treatment if they were full-time employees or students, actively psychotic or suicidal, had acute substance abuse, or were involved in any other mental health treatment.

Treatment occurred solely in group contexts, including several small and large groups for approximately eight hours each day

T hi

s do

cu m

en t

is co

py ri

gh te

d by

th e

A m

er ic

an P

sy ch

ol og

ic al

A ss

oc ia

ti on

or on

e of

it s

al li

ed pu

bl is

he rs

. T

hi s

ar ti

cl e

is in

te nd

ed so

le ly

fo r

th e

pe rs

on al

us e

of th

e in

di vi

du al

us er

an d

is no

t to

be di

ss em

in at

ed br

oa dl

y.

353DEPRESSION AND NEGATIVE-AFFECT EXPRESSION

(Piper et al., 1996). Group composition varied so that each patient was in at least one group per day with every other patient. There were 30 to 35 patients in the program at all times, and patients were discharged on Fridays and admitted on Mondays. Treatment was primarily psychodynamically oriented; however, it did have some cognitive-behavioral components (e.g., relaxation training). The treatment utilized a therapeutic milieu with the goal of patients identifying, understanding, and modifying their maladaptive be- havior patterns.

Treatment was divided into three 6-week phases (Piper et al., 1996). Phase 1 emphasized preparation to enable patients to utilize treatment. Some groups during Phase 1 included a communication skills group and a self-awareness insight-oriented group. Phase 2 emphasized patient insight and change. Some groups during Phase 2 included a personal relations group and a media-review group in which therapists and patients would watch video of a previous psychodynamic group to facilitate insight into individual patient and group interactions. Phase 3 emphasized successful termina- tion. Some groups during Phase 3 included a reentry group and a vocational group.

Measures

Depression symptoms. To measure depression symptoms, we used the Depression Symptom subscale of the Brief Symptom Inventory (BSI; Derogatis, 2000). At intake, respondents indicated how much they were distressed by each item (e.g., “Feeling no interest in things”) from 0 (not at all) to 4 (extremely). We computed the mean of participants’ responses with higher scores indicating more severe depression symptoms. The measure dem- onstrated strong internal consistency (� � .90). Validity of the BSI Depression Symptom subscale has been supported via correlations with structured clinical interviews of depression as well as other self-report measures of depression symptoms (Bromberg, Beasley, D’Angelo, Landzberg, & DeMaso, 2003; Derogatis, 2000).

Negative affect variables. We used two single-item self-report measures to assess negative-affect intensity and negative-affect expression (McCallum, Piper, & Morin, 1993). On the Friday of each week that patients were in treatment, they indicated their negative-affect intensity (“I felt negative feelings this week [e.g., anxiety, sadness, anger, pessimism]”) and negative-affect expres- sion (“I expressed my negative feelings this week”) from 1 (very little) to 6 (very much) during treatment that week. The items were developed for repeated assessment of negative-affect intensity and negative-affect expression in group psychotherapy; thus, they are brief and easy to understand. Consistent with common definitions of negative affect, the negative-affect intensity item was written to broadly assess subjective distress. This is particularly relevant in the present sample due to personality disorders having been linked with difficulty separating emotional versus physical experiences (e.g., Joyce, Fujiwara, Cristall, Ruddy, & Ogrodniczuk, 2013). Further, the negative-affect expression item was written to be broadly inclusive of types of expression. This is consistent with research indicating that similar emotion-regulating processes occur within verbal and nonverbal affect expression (Kennedy-Moore & Watson, 2001).

While these negative affect items have been used in previous studies of group psychotherapy (e.g., McCallum et al., 1993; Piper, Ogrodniczuk, McCallum, Joyce, & Rosie, 2003) and were written

consistent with the present conceptualizations of negative-affect intensity and negative-affect expression, we conducted several analyses to examine the validity of these items in the present sample. When conducting these analyses, we used methods that took into account the nested nature of the data (e.g., multiple observations for each patient).

Regarding negative-affect intensity, we would expect that pa- tients with higher levels of pretreatment distress would experience greater negative affect during treatment. Further, we would expect that higher levels of pretreatment distress would have a stronger association with negative-affect intensity than negative-affect ex- pression. To test these proposed associations, we examined the relations between the affect intensity and affect expression items with the general severity index of the BSI (Derogatis, 2000) and the Dysfunctional Behaviors subscale of the Objectives Behavior Index (OBI; Marziali, Munroe-Blum, & McCleary, 1999). The BSI is a self-report measure of psychiatric symptom severity and the OBI is a structured clinical interview used to assess the frequency and severity of dysfunctional behaviors (e.g., suicide attempts, substance abuse) among patients with personality disor- ders. Consistent with our expectations, the correlations with negative-affect intensity and negative-affect expression, respec- tively, with the BSI were r � .194 (p � .0026) and r � .093 (p � .1518) and with the OBI were r � .215 (p � .0008) and r � .084 (p � .1956). The BSI and OBI being more strongly associated with negative-affect intensity than negative-affect expression provides support for the negative-affect intensity item measuring what it was intended to measure.

Regarding responses to the negative-affect expression item, we examined associations with two theoretically related variables. First, we would expect that patients with higher levels of pretreat- ment problems related to interpersonal coldness would express less affect during treatment. People who are interpersonally cold act in ways that indicate that they are interpersonally disconnected or detached, and are often perceived as uncaring. Therefore, we would expect cold interpersonal problem scores to be negatively correlated with negative-affect expression. Further, we would ex- pect that higher levels of pretreatment cold interpersonal problems would have a stronger association with negative-affect expression than negative-affect intensity. Presently, we used the Cold subscale of the 64-item Inventory of Interpersonal Problems, a psychometri- cally sound measure of interpersonal problems (Horowitz, Alden, Wiggins, & Pincus, 2000). Consistent with our expectations, the Cold subscale correlated with negative-affect expression, r � �.217, p � .0007, but not with negative-affect intensity, r � �.017, p � .7937. The second way that we evaluated the validity of the negative-affect expression item was to examine the association between patient-reported negative-affect expression and therapist-reported negative-affect expression. We would ex- pect that there would be an association between patient and ther- apist reports of affect expression. Therapists rated patients’ negative-affect expression (i.e., “She/he expressed his/her negative feelings this week [e.g., anxiety, sadness, anger, pessimism]”) from 1 (very little) to 6 (very much) during patients’ third, sixth, ninth, 12th, 15th, and 18th weeks in treatment. Correlations be- tween patients’ and therapists’ ratings of negative-affect expres- sion ranged from .89 (Week 15) to .63 (Week 6), indicating strong to good patient-therapist interrater agreement (Fleiss, Levin, & Paik, 2003).

T hi

s do

cu m

en t

is co

py ri

gh te

d by

th e

A m

er ic

an P

sy ch

ol og

ic al

A ss

oc ia

ti on

or on

e of

it s

al li

ed pu

bl is

he rs

. T

hi s

ar ti

cl e

is in

te nd

ed so

le ly

fo r

th e

pe rs

on al

us e

of th

e in

di vi

du al

us er

an d

is no

t to

be di

ss em

in at

ed br

oa dl

y.

354 COX ET AL.

Personality disorders. To assess personality disorders, we used the Structured Clinical Interview for DSM–IV Axis II per- sonality disorders (SCID-II; First, Gibbon, Williams, & Spitzer, 1998). The SCID-II has strong psychometric properties and is considered a gold standard semistructured interview for personal- ity disorders (Lobbestael, Leurgans, & Arntz, 2011). At intake, SCIDs were administered by five assessors who each had three or more years of experience using the SCID. Independent raters coded recorded SCID interviews, and the mean kappa (i.e., inter- rater reliability) was .68.

Results

In our study, there were several variables (e.g., negative-affect intensity, negative-affect expression) for which we had multiple observations per patient (i.e., observations nested within patients). These variables that had within-patient variability were Level 1 variables. We also had several variables for which we only had a single observation per patient (e.g., depression symptoms). These variables that only had between-patient variability were Level 2 variables. Because we had multilevel data, we used multilevel modeling (MLM). MLM is a form of regression that takes into account the nested nature (i.e., nonindependence) of data (Rauden- bush & Bryk, 2002). In the present study, we specified all of our Level 1 coefficients as random, which allowed the intercepts and slopes to vary across patients. Further, we estimated the Level 1 associations as latent variables to mitigate biased estimates that are native to nonlatent approaches (Preacher, Zyphur, & Zhang, 2010). All analyses were conducted in Mplus 7.11 (Muthén & Muthén, 2012).

Because patients in our study were enrolled and discharged each week, to evaluate the impact of the group (i.e., other patients) on each patient’s negative-affect expression, we drew from methods developed for studying rolling therapy groups (Tasca et al., 2010). To compute group scores, we created aggregate scores of all of the other group members (i.e., group score � total group score � individual score). Thus, we computed different group scores for each patient. Because the group (i.e., the patients concurrently enrolled) varied from week to week as patients were discharged and admitted, we computed different group scores for each patient each week. Group scores all had multiple observations per patient

over time because group scores were recalculated each week as different patients (i.e., other group members) were discharged and enrolled in the program. Therefore, all of the group variables (e.g., group negative-affect intensity, group negative-affect expression) were Level 1 variables because they all had within-patient vari- ability.

Multilevel models of rolling groups substantially differ from multilevel models of closed groups. In rolling groups, group de- pendency cannot be modeled at a higher level than the patient level because it does not exist in the same way as in closed groups (Tasca et al., 2010). In closed groups, where group admission stops once the group begins and patients do not cross over groups, common approaches (e.g., Kivlighan & Paquin, 2014; Miles et al., 2011) model within-patient effects at Level 1, between-patient effects at Level 2, and between-groups effects at Level 3. This three-level approach accounts for patients being nested within groups and allows for partitioning variance due to group effects. In rolling groups, composition is in flux; thus, patients are not nested within groups and between-groups variance cannot be partitioned. In the present study, group composition varied every week of the study since patients were discharged and enrolled every week. Further demonstrating the logic of not modeling group at Level 3, our time-varying (i.e., Level 1) variables were assessed once per week; thus, group membership varied at each observation of our time-varying variables. Therefore, consistent with recommenda- tions for studying rolling groups (i.e., Tasca et al., 2010), group effects were included as Level 1 effects and there were no Level 3 group effects.

Preliminary Analyses

Data were screened for skewness and kurtosis via their z distri- butions and inspection of histograms (Tabachnick & Fidell, 2013). Data appeared normally distributed. The 239 patients reported their negative-affect intensity and negative-affect expression 3,263 times (M � 13.65 reports per participant). Across patients’ weekly reports, their mean negative-affect intensity and negative-affect expression were 3.62 (SD � .80) and 4.34 (SD � .62), respec- tively. Correlations and descriptive statistics for the primary study variables are presented in Table 1.

Table 1 Bivariate Correlations and Descriptive Statistics for Primary Variables

Variable 1 2 3 4 5 6

Level 1 1. Patient negative-affect expression 2. Patient negative-affect intensity .408���

3. Group negative-affect expression �.022 .006 4. Group negative-affect intensity .011 .004 .583���

Level 2 5. Patient depression symptoms �.239��� �.037 .001 .011 6. Patient-mean negative-affect intensity .622��� .489��� .051 .052 �.007

M 3.630 4.369 3.534 4.202 2.117 4.341 SD 1.456 1.464 .298 .298 1.036 .780

Note. Correlations between Level 1 variables account for nesting. Correlations between Level 1 and Level 2 variables are based on aggregate Level 1 reports. ��� p � .001.

T hi

s do

cu m

en t

is co

py ri

gh te

d by

th e

A m

er ic

an P

sy ch

ol og

ic al

A ss

oc ia

ti on

or on

e of

it s

al li

ed pu

bl is

he rs

. T

hi s

ar ti

cl e

is in

te nd

ed so

le ly

fo r

th e

pe rs

on al

us e

of th

e in

di vi

du al

us er

an d

is no

t to

be di

ss em

in at

ed br

oa dl

y.

355DEPRESSION AND NEGATIVE-AFFECT EXPRESSION

Multilevel Models to Partition the Variance

We ran two 2-level unconditional models to partition the vari- ance in patients’ negative-affect expression and patients’ negative- affect intensity. For patients’ negative-affect expression, 22.1% of the variance (intraclass correlation coefficient [ICC] � .221) was due to differences between patients and 77.9% of the variance was due to differences between weeks in treatment (i.e., within pa- tients). For patients’ negative-affect intensity, 18.1% of the vari- ance (ICC � .181) was due to differences between patients and 77.9% of the variance was due to differences between weeks in treatment (i.e., within patients).

Growth Curve Analyses

To examine if any of our time-varying variables systematically increased or decreased over the course of treatment, we ran a series of growth curve analyses. For these analyses, week in treatment (i.e., time) was the independent variable.

When examining patients’ negative-affect expression, the linear slope term of patients’ week in treatment was not significant (� � 0.010, t � 1.733, p � .083). However, the linear slope term of patients’ week in treatment was significant for their negative-affect intensity, such that each additional week in treatment was associ- ated with a 0.015 decrease in negative-affect intensity (6-point scale). Week in treatment explained an additional 3.4% of the within-patient variance in negative-affect intensity compared with the unconditional model reported above. When examining group effects, the linear slope term of patients’ week in treatment pre- dicting the groups’ negative-affect expression was not significant (� � �0.014, t � �0.151, p � .880). However, the linear slope term of patients’ week in treatment was significantly associated with the groups’ negative-affect intensity (� � �0.009, t � �6.822, p � .0001) such that each additional week in treatment was associated with a 0.009 decrease in group negative-affect intensity (6-point scale). Week in treatment explained an additional 11.9% of the within-patient variance in group negative-affect intensity compared with the unconditional model reported above. The re- sults of these growth curve analyses indicate that as patients

progressed through treatment, both patient and group negative affect became less intense, but patient and group expression of negative affect did not change.

Because patients’ week in treatment (i.e., time) was associated with patients’ and groups’ negative-affect intensity, we included patient week in treatment as a Level 1 variable in all models. By controlling for (i.e., detrending) time, we removed the influence that time may have had on the association between our indepen- dent and dependent variables (Wang & Maxwell, 2015).

Primary Analyses

Patient negative-affect intensity. In Model 1, we investi- gated the association between negative-affect intensity and negative-affect expression (Table 2). We examined patients’ weekly negative-affect intensity and their mean negative-affect intensity. Consistent with recommendations (Hoffman & Stawski, 2009) and similar to previous studies of within-person affect expression (e.g., O’Loughlin, Cox, Kahn, & Wu, 2018), we com- puted the mean score of each patient’s weekly negative-affect intensity. Patient-mean negative-affect intensity can be conceptu- alized as patients’ trait negative-affect intensity and this Level 2 variable was used to control for between-patient differences on their negative-affect intensity. Patient negative-affect intensity was person-mean centered, and patient-mean negative-affect intensity was grand-mean centered. We specified random effects for the intercept and the two Level 1 slopes—patient weekly negative- affect intensity and time (Model 1 in Table 3).

Supporting H1, patients’ weekly negative-affect intensity was positively related to their weekly negative-affect expression, while controlling for patient-mean negative-affect intensity, such that a 1-point increase in negative-affect intensity was associated with a .45 increase in negative-affect expression— both were 6-point scales (see Table 2, Model 1). Patient-mean negative-affect inten- sity was also positively related to between-patient negative-affect expression such that a 1-point increase in person-mean intensity was associated with a 0.64 increase in affect expression. Model 1 explained an additional 21.8% of the within-patient variance and 27.2% of the between-patient variance in negative-affect expres-

Table 2 Fixed Effects for Multilevel Models Predicting Patients’ Negative-Affect Expression

Model 1 Model 2 Model 3 Model 4

Predictor � SE � SE � SE � SE

Intercept, �00 .675 � .283 .731� .285 .692� .289 .731 .292

Level 1 Patient week in treatment, �10 .018

��� .005 .018��� .005 .021��� .005 .021��� .005 Patient negative-affect intensity, �20 .454

��� .028 .462��� .028 .462��� .028 .364��� .040 Group negative-affect intensity, �30 .296

� .123 .024 .208 Group negative-affect expression, �40 �.298

�� .111 �.316�� .115 Patient Week in Treatment � Patient Negative-Affect Intensity, �50 .011

�� .004 Patient Week in Treatment � Group Negative-Affect Intensity, �60 .025 .019

Level 2 Patient-mean negative-affect intensity, �01 .643

��� .064 .633��� .065 .635��� .066 .628��� .066 Patient depression symptoms, �02 �.174

��� .049 �.017��� .050 �.170�� .051 Cross-level interaction

Patient Depression Symptoms � Patient Negative-Affect Intensity, �22 �.069 �� .024 �.066�� .024 �.066�� .024

Patient Depression Symptoms � Group Negative-Affect Intensity, �32 .029 .104 .029 .105

� p � .05. �� p � .01. ��� p � .001.

T hi

s do

cu m

en t

is co

py ri

gh te

d by

th e

A m

er ic

an P

sy ch

ol og

ic al

A ss

oc ia

ti on

or on

e of

it s

al li

ed pu

bl is

he rs

. T

hi s

ar ti

cl e

is in

te nd

ed so

le ly

fo r

th e

pe rs

on al

us e

of th

e in

di vi

du al

us er

an d

is no

t to

be di

ss em

in at

ed br

oa dl

y.

356 COX ET AL.

sion above and beyond the previously reported two-level growth- curve model with week in treatment as the sole predictor.

Patient depression. In Model 2, we examined patient depres- sion effects without including group effects (see Table 2). Specif- ically, we tested our hypotheses that higher levels of patient depression symptoms would be associated with lower levels of patient negative-affect expression (H2) and that the association between negative-affect intensity and negative-affect expression would be weaker (i.e., moderated) for patients with greater depres- sion symptoms (H3). Depression symptoms were included as a Level 2 (i.e., person-level) variable and were grand-mean centered. As with Model 1, random effects for the intercept and the Level 1 slopes for time and patient negative-affect intensity were modeled (see Model 2 in Table 3). Supporting H2, when negative-affect intensity was zero (i.e., the mean), depression symptoms were significantly negatively associated with negative-affect expression, such that a 1-point increase in depression symptoms (5-point scale) was associated with a 0.17-point decrease in negative-affect ex- pression (6-point scale). This finding indicates that at patients’ average levels of negative-affect intensity, patients’ depression symptoms were negatively related to their negative-affect expres- sion over the course of treatment.

To examine if depression moderated the association between affect intensity and affect expression, we included a cross-level interaction in Model 2. From this interaction we could determine if patient depression symptoms (Level 2) significantly moderated the association between negative-affect intensity (Level 1) and negative-affect expression (Level 1). Supporting H3, depression symptoms moderated the association between negative-affect in- tensity and negative-affect expression. Above and beyond Model 1, Model 2 accounted for an additional 9.1% of the between- patient variance in negative-affect expression and for 7% of the within-patient variance in the association between affect intensity and affect expression.

Follow-up tests of depression’s moderating effect indicated that the association between patient negative-affect intensity and ex- pression became weaker from low depression symptoms (� � 0.53, t � 8.10, p � .0001) to mean depression symptoms (� � 0.46, t � 7.00, p � .0001) to high depression symptoms (� � 0.38, t � 5.87, p � .0001; Figure 1). This finding is consistent with previous research in nonpsychotherapeutic contexts indicating that

depression symptoms disrupt the adaptive affect regulation strat- egy of expressing affect when distressed.

Group negative-affect intensity and negative-affect expression. In Model 3, we added the group effects of negative- affect intensity and negative-affect expression (see Table 2). As noted above, because the group composition varied from week to week, we computed different group scores for each patient for each week. Group scores all had multiple observations per patient because group scores were recalculated each week as different patients were discharged and enrolled in the program. Therefore, the two group variables (i.e., group negative-affect intensity, group negative-affect expression) were Level 1 variables because they had within-patient variability. Group variables were centered around the group mean for each patient. Because the group (i.e., other patients) was different for each patient, this allowed us to examine group effects within the context of the other patients who were in treatment at the same time as each patient. Random effects were specified for the intercept and the slopes for patient week in treatment, patient negative-affect intensity, group negative-affect intensity, and group negative-affect expression (see Model 3 in Table 3).

Supporting the patient effects reported above, when the group variables were entered into Model 3, all of the patient effects noted in Model 1 and Model 2 remained significant. This indicates that

Table 3 Random Effects for Multilevel Models Predicting Patients’ Negative-Affect Expression

Model 1 Model 2 Model 3 Model 4

Random effects Variance SE Variance SE Variance SE Variance SE

Residual variance in within-patient negative-affect expression, eij 1.588 ��� .033 1.589��� .032 1.580��� .033 1.574��� .033

Residual variance in between-patient negative-affect expression, r0i .353 ��� .053 .321��� .047 .321��� .047 .321��� .048

Random slopes Patient week in treatment, e10 .001

�� .000 .001�� .000 .001�� .000 .001�� .000 Patient negative-affect intensity, e20 .056

��� .012 .052��� .012 .054��� .012 .034� .016 Group negative-affect intensity, e30 .021 .248 .016 .338 Group negative-affect expression, e40 .200 .193 .014 .196 Patient Week in Treatment � Patient Negative-Affect Intensity, e50 .000 .000 Patient Week in Treatment � Group Negative-Affect Intensity, e60 .000 .002

� p � .05. �� p � .01. ��� p � .001.

0

0.5

1

1.5

Low Negative-Affect Intensiy High Negative-Affect Intensity

N e g a ti

v e -A

ff e c t

E x p re

ss io

n

Low Depression

Symptoms

Mean Depression

Symptoms

High Depression

Symptoms

Figure 1. Simple slopes of the association between patients’ negative- affect intensity and negative-affect expression for those with low (�1 SD), mean, and high ( 1 SD) depression symptoms.

T hi

s do

cu m

en t

is co

py ri

gh te

d by

th e

A m

er ic

an P

sy ch

ol og

ic al

A ss

oc ia

ti on

or on

e of

it s

al li

ed pu

bl is

he rs

. T

hi s

ar ti

cl e

is in

te nd

ed so

le ly

fo r

th e

pe rs

on al

us e

of th

e in

di vi

du al

us er

an d

is no

t to

be di

ss em

in at

ed br

oa dl

y.

357DEPRESSION AND NEGATIVE-AFFECT EXPRESSION

the previously presented patient effects persisted while controlling for group effects.

Supporting H4, group negative-affect intensity had a significant positive association with patient negative-affect expression, such that a 1-point increase in group negative-affect intensity was associated with a .30-point increase in patient negative-affect expression. However, inconsistent with H5, group negative-affect expression had a significant negative association with patient negative-affect expression, such that a 1-point increase in group negative-affect expression was associated with a .30-point de- crease in patient negative-affect expression. These findings indi- cate that patients expressed more when others in the group expe- rienced more intense negative affect; however, patients expressed less when others in the group expressed more. Above and beyond Model 2, Model 3 accounted for an additional 0.6% of the within- patient variance in patients’ negative-affect expression.

In the same model, we also examined if patients’ depression symptoms moderated the association between group negative- affect intensity and patient negative-affect expression. Inconsistent with H6, patients’ depression did not moderate the association between group intensity and patient expression (Table 2, Model 3). Further, this interaction did not explain any additional variance in between-patient affect expression.

Time in treatment. In Model 4, we further explored the role of time (see Table 2). As noted above, patient week in treatment (i.e., time) was significantly correlated with patient negative-affect intensity and group negative-affect intensity. Therefore, in Model 4 we examined if patient week in treatment interacted with either patient negative affect or group negative affect to predict patients’ expression. Random effects for these two Level 1 interactions were specified (see Model 4 in Table 3). Group negative-affect intensity did not interact with time to predict patients’ expression; however, patient negative-affect intensity did interact with patient week in treatment to predict patient negative-affect expression. Above and beyond Model 3, Model 4 explained an additional 0.4% of the variance in within-patient negative-affect expression.

Follow-up tests of this moderating effect indicated that the association between patient affect intensity and expression became stronger from early treatment (� � 0.38, t � 5.72, p � .0001) to midtreatment (� � 0.46, t � 3.10, p � .0022), to late treatment (� � 0.56, t � 2.49, p � .0135; Figure 2). These findings indicate that as patients progressed through treatment, the congruence between their experience of negative affect and their expression of negative affect increased.

To examine if there was a three-way interaction between patient depression symptoms, patient negative-affect intensity, and patient week in treatment, we ran an additional model (not displayed in Table 2). This three-way interaction did not predict patient negative-affect expression (� � 0.004, t � 0.913, p � .361), nor did it explain any additional variance.

Discussion

Depression attenuating the association between negative-affect intensity and negative-affect expression has been observed in cross-sectional (Kahn & Garrison, 2009), naturalistic longitudinal (Garrison et al., 2012), laboratory (Campbell-Sills et al., 2006), and now psychotherapeutic contexts. Our findings regarding pa- tient effects were consistent with nonpsychotherapy studies that

have examined depression’s attenuating effect on affect expres- sion. However, when we examined group effects, depression’s attenuating effects were not observed. Further, consistent with our hypothesis, group negative-affect intensity was positively associ- ated with patient expression. However, inconsisitent with our hypothesis, group negative-affect expression was negatively asso- ciated with patient expression.

Patient and Group Effects

While our finding that depression symptoms attenuated the association between intensity and expression is consistent with research in nonpsychotherapeutic settings, it is important to con- sider why depression symptoms’ attenuating effect on the associ- ation between affect intensity and affect expression persisted within a therapeutic group context. One possible explanation is that in some groups, a few group members may have been the focus of the group and expressed substantially more than other members. This norm of unequal expression may have facilitated the tendency of those with higher levels of depression symptoms to not express their feelings.

The effect of groups with an unequal focus between group members negatively impacting patients’ engagement in therapeutic process behaviors has been found in a study of interpersonal growth and trauma recovery groups (Miles et al., 2011). Consistent with our finding that group expression was negatively associated with patient expression, the authors found that group intimate behaviors (e.g., emotional disclosure, authentic responding) had a negative association with patient intimate behaviors. However, they found that this negative association was moderated by the consistency in which the other group members engaged in intimate behaviors. When group intimate behaviors were expressed at high levels but unevenly between group members (i.e., a few group members engaged in more intimate behaviors than the rest of the group), group intimate behaviors were associated with reduced patient intimate behaviors (Miles et al., 2011). However, when group intimate behaviors were at high levels of consistency (e.g., little difference in group members’ intimate behaviors), group intimate behaviors had little to no association with patients’ inti- mate behaviors.

0

0.5

1

1.5

Low Negative-Affect Intensiy High Negative-Affect Intensity

N e g a ti

v e -A

ff e c t

E x p re

ss io

n

Late Treatment

Mid Treatment

Early Treatment

Figure 2. Simple slopes of the association between patients’ negative- affect intensity and negative-affect expression when patients were early in treatment (i.e., Week 1 of 18), in the middle of treatment (i.e., Week 9 of 18), and late in treatment (i.e., Week 18 of 18).

T hi

s do

cu m

en t

is co

py ri

gh te

d by

th e

A m

er ic

an P

sy ch

ol og

ic al

A ss

oc ia

ti on

or on

e of

it s

al li

ed pu

bl is

he rs

. T

hi s

ar ti

cl e

is in

te nd

ed so

le ly

fo r

th e

pe rs

on al

us e

of th

e in

di vi

du al

us er

an d

is no

t to

be di

ss em

in at

ed br

oa dl

y.

358 COX ET AL.

In our context— groups for patients with heterogeneous person- ality disorders— unequal negative-affect expression between group members may have been particularly salient. In these types of groups, the personality characteristics of the group members are likely at extreme ends of the spectrum. For example, patients with avoidant personality disorder tend to be interpersonally passive while patients with narcissistic personality disorder tend to be interpersonally dominant (Paris, 2004). In this context, if the group norm is that the amount of expression between group members is unbalanced, the tendency of those with higher levels of depression symptoms to not express their feelings may be facilitated.

Another explanation for why we found that group negative- affect expression was inversely associated with patient negative- affect expression concerns our conceptualization and operational- ization of group expression. Presently, we operationalized group expression by aggregating group members’ ratings of their own behaviors. Our methods were similar to the previously discussed study of group intimate behaviors (Miles et al., 2011) in which group intimate behaviors were operationalized by aggregating group members’ ratings of their own intimate behaviors. Also similar to our study, group intimate behaviors were inversely related to patient intimate behaviors. However, in another study predicting patient intimate behaviors, patients’ perceptions of group engagement were positively associated with patients’ inti- mate behaviors (Kivlighan & Paquin, 2014). Applied to our con- text, aggregating other patients’ perceptions of their own behaviors may not adequately represent how the patient perceives and is impacted by the group norm. However, by assessing each patient’s perception of the group, we may better assess the group norm that is perceived by, and is influencing, each patient.

Time in Treatment

While not part of our initial research questions, we found that the association between negative-affect intensity and negative- affect expression increased over the course of treatment. This association is consistent with previous research (e.g., Miles et al., 2011) and likely indicates what we would expect from beneficial group treatments—patients early in treatment are less likely to express their feelings; however, as their comfort and willingness increases within a therapeutic environment, their affect expression increases. Our study extends this work on therapeutic engagement increasing over time in treatment by finding that the association between affect intensity and affect expression increased over time. When we examined the effect of time on affect intensity and expression individually, negative-affect intensity actually de- creased over time and negative-affect expression was not associ- ated with time. Building from the emotion dysregulation theory of affect expression (Campbell-Sills et al., 2006), dysregulation oc- curs when the linear association between affect expression and affect intensity is disrupted. Therefore, our finding that this link between intensity and expression strengthens over time demon- strates an increase in adaptive expression (i.e., expression when experiencing affect intensely) as patients progressed through treat- ment.

While the intensity-expression association became stronger over the course of treatment, depression’s attenuating effect on this association did not vary over time. This finding indicates that while patients’ adaptive use of affect expression in response to

affect intensity increased as they progressed through treatment, the impact of baseline depression symptoms on disrupting this adap- tive association persisted throughout treatment. While our findings indicate that depression symptoms’ attenuating effects persisted throughout treatment, this does not mean that therapy did not facilitate their use of adaptive expression. Rather, our findings indicate that over the course of treatment, patients increased their use of adaptive affect expression. And while patients with higher levels of baseline depression symptoms used less adaptive affect expression, their increase in adaptive expression over the course of treatment did not deviate from other patients. In summary, while group psychotherapy may not overcome the attenuating effects of depression symptoms on the association between negative-affect intensity and negative-affect expression, patients with high levels of depression symptoms—relative to low levels—seem to achieve therapeutic benefits at similar rates.

Practical Implications

The primary clinical implication of our findings is that clinicians should be aware of and consider whether depression symptoms are impacting their patients’ affect expression. It is important to note that several studies have indicated that depression does not impede the clinical value of affect expression. In fact, there is evidence that those who have greater depression symptoms ben- efit more from affect expression than those with fewer depression symptoms (e.g., Milbury et al., 2017; Schneider et al., 2010). This may be because those who rarely express affect benefit more from interventions that facilitate expression (Frattaroli, 2006). This ex- planation builds on the axiom that patients benefit most from interventions in which they deviate from their typical and prob- lematic interpersonal styles (Kiesler, 1983). The therapeutic value of this interpersonal deviation has been supported in treatment studies of depression (e.g., Constantino et al., 2012). Therefore, while depression does not inhibit the therapeutic value of affect expression, it does reduce its likelihood. Clinically then, it is therapeutically valuable to work with patients to overcome this inhibition of affect expression both within treatment and in other domains in their lives.

A second clinical implication is that it may be important for group therapists to attend to the relative time that each patient has to talk. While we did not examine this directly, our findings, in concert with previous group research (Miles et al., 2011), indicate that the group engaging in higher levels of affect expression— particularly when there is an imbalance and a few members are the focus of the group—may inhibit other patients’ expression. One way by which group therapists can attend to the balance of pa- tients’ expression is to facilitate patient expression directly via therapist intervention. For example, in the presence of a group member who monopolizes the group’s time, group leaders can help other members explore why they are allowing their opportunity for expression to be diminished by another member (Yalom & Leszcz, 2005). Therapists can also attend to patients’ expression indirectly by fostering an atmosphere in which group cohesion is oriented around group members’ mutual facilitation of one another’s expres- sion of affect, and through intervening when individual-, subgroup-, or group-level dynamics inhibit affect expression.

T hi

s do

cu m

en t

is co

py ri

gh te

d by

th e

A m

er ic

an P

sy ch

ol og

ic al

A ss

oc ia

ti on

or on

e of

it s

al li

ed pu

bl is

he rs

. T

hi s

ar ti

cl e

is in

te nd

ed so

le ly

fo r

th e

pe rs

on al

us e

of th

e in

di vi

du al

us er

an d

is no

t to

be di

ss em

in at

ed br

oa dl

y.

359DEPRESSION AND NEGATIVE-AFFECT EXPRESSION

Limitations and Future Directions

Our study had several limitations that have implications for future research. First, the study was done in a specific context (i.e., psychodynamic intensive-day treatment) for a specific population (i.e., patients with personality disorders). The attenuating effect of depression on the association between affect intensity and affect expression should be examined in other group and individual psychotherapy contexts with other populations to examine its generalizability. It is also important that future work examine mediators of the association between depression symptoms and expression in psychotherapy. While nonpsychotherapeutic research has indicated suppression as a particularly important mediator (e.g., Kahn & Garrison, 2009), continued examination of these explanatory effects is important to facilitate improved clinical recommendations. Similarly, there may be specific therapist or group effects (e.g., cohesion) that buffer the impact of depression on affect expression. Presently, since all of the patients were exposed to all of the therapists, we could not evaluate therapist effects. Identifying these effects is important for facilitating effective interventions and training group therapists. Another limitation is that our affect intensity and expres- sion measures each consisted of a single item. Measures with more items and even subscales would facilitate a more nuanced understand- ing of affect intensity and expression in psychotherapy. It is also important to consider the limitation of examining affect intensity and affect expression within the same week of treatment. While our methods were consistent with previous studies that examined depres- sion’s attenuating effect on the association between affect expression and affect intensity (e.g., Garrison et al., 2012), group effects may take longer to impact patients’ expression. Future research investigating group effects using time-lag models would facilitate a more nuanced understanding of group effects on patient expression.

References

Abbass, A., Town, J., & Driessen, E. (2012). Intensive short-term dynamic psychotherapy: A systematic review and meta-analysis of outcome re- search. Harvard Review of Psychiatry, 20, 97–108. http://dx.doi.org/10 .3109/10673229.2012.677347

Aldao, A., Nolen-Hoeksema, S., & Schweizer, S. (2010). Emotion- regulation strategies across psychopathology: A meta-analytic review. Clinical Psychology Review, 30, 217–237. http://dx.doi.org/10.1016/j .cpr.2009.11.004

American Psychiatric Association. (1994). Diagnostic and statistical man- ual of mental disorders (4th ed.). Washington, DC: Author.

Bromberg, J. I., Beasley, P. J., D’Angelo, E. J., Landzberg, M., & DeMaso, D. R. (2003). Depression and anxiety in adults with congenital heart disease: A pilot study. Heart & Lung: The Journal of Acute and Critical Care, 32, 105–110. http://dx.doi.org/10.1067/mhl.2003.26

Campbell-Sills, L., Barlow, D. H., Brown, T. A., & Hofmann, S. G. (2006). Acceptability and suppression of negative emotion in anxiety and mood disorders. Emotion, 6, 587–595. http://dx.doi.org/10.1037/1528-3542.6 .4.587

Constantino, M. J., Laws, H. B., Arnow, B. A., Klein, D. N., Rothbaum, B. O., & Manber, R. (2012). The relation between changes in patients’ interpersonal impact messages and outcome in treatment for chronic depression. Journal of Consulting and Clinical Psychology, 80, 354 – 364. http://dx.doi.org/10.1037/a0028351

Crits-Cristoff, P., & Barber, J. P. (2004). Empirical research on the treat- ment of personality disorders. In J. J. Magnavita (Ed.), Handbook of personality disorders: Theory and practice (pp. 513–527). Hoboken, NJ: Wiley.

Derogatis, L. R. (2000). BSI-18: Brief Symptom Inventory 18 — Administration, scoring, and procedures manual. Minneapolis, MN: Pearson.

Diener, M. J., Hilsenroth, M. J., & Weinberger, J. (2007). Therapist affect focus and patient outcomes in psychodynamic psychotherapy: A meta- analysis. The American Journal of Psychiatry, 164, 936 –941. http://dx .doi.org/10.1176/ajp.2007.164.6.936

First, M. B., Gibbon, M., Williams, J. B. W., & Spitzer, R. (1998). Computer-assisted Structured Clinical Interview for DSM–IV Axis II. Toronto, Ontario, Canada: Multi-Health System.

Fleiss, J. L., Levin, B., & Paik, M. C. (2003). The measurement of interrater agreement. In W. A. Shewart & S. Samuel (Eds.), WilksStatis- tical methods for rates and proportions (3rd ed., pp. 598 – 626). Hobo- ken, NJ: John Wiley & Sons, Inc. http://dx.doi.org/10.1002/0471445428 .ch18

Frattaroli, J. (2006). Experimental disclosure and its moderators: A meta- analysis. Psychological Bulletin, 132, 823– 865. http://dx.doi.org/10 .1037/0033-2909.132.6.823

Garrison, A. M., Kahn, J. H., Sauer, E. M., & Florczak, M. A. (2012). Disentangling the effects of depression symptoms and adult attachment on emotional disclosure. Journal of Counseling Psychology, 59, 230 – 239. http://dx.doi.org/10.1037/a0026132

Hoffman, L., & Stawski, R. S. (2009). Persons as contexts: Evaluating between-person and within-person effects in longitudinal analysis. Re- search in Human Development, 6, 97–120. http://dx.doi.org/10.1080/ 15427600902911189

Horowitz, L. M., Alden, L. E., Wiggins, J. S., & Pincus, A. L. (2000). Inventory of Interpersonal Problems: Manual. San Antonio, TX: The Psychological Corporation.

Johnson, J., Burlingame, G. M., Olsen, J. A., Davies, D. R., & Gleave, R. L. (2005). Group climate, cohesion, alliance, and empathy in group psychotherapy: Multilevel structural equation models. Journal of Coun- seling Psychology, 52, 310 –321. http://dx.doi.org/10.1037/0022-0167 .52.3.310

Joyce, A. S., Fujiwara, E., Cristall, M., Ruddy, C., & Ogrodniczuk, J. S. (2013). Clinical correlates of alexithymia among patients with person- ality disorder. Psychotherapy Research, 23, 690 –704. http://dx.doi.org/ 10.1080/10503307.2013.803628

Kahn, J. H., Achter, J. A., & Shambaugh, E. J. (2001). Client distress disclosure, characteristics at intake, and outcome in brief counseling. Journal of Counseling Psychology, 48, 203–211. http://dx.doi.org/10 .1037/0022-0167.48.2.203

Kahn, J. H., & Garrison, A. M. (2009). Emotional self-disclosure and emotional avoidance: Relations with symptoms of depression and anx- iety. Journal of Counseling Psychology, 56, 573–584. http://dx.doi.org/ 10.1037/a0016574

Kelly, J. R., & Barsade, S. G. (2001). Mood and emotions in small groups and work teams. Organizational Behavior and Human Decision Pro- cesses, 86, 99 –130. http://dx.doi.org/10.1006/obhd.2001.2974

Kennedy-Moore, E., & Watson, J. C. (2001). How and when does emo- tional expression help? Review of General Psychology, 5, 187–212. http://dx.doi.org/10.1037/1089-2680.5.3.187

Kenny, D. A., Mannetti, L., Pierro, A., Livi, S., & Kashy, D. A. (2002). The statistical analysis of data from small groups. Journal of Personality and Social Psychology, 83, 126 –137. http://dx.doi.org/10.1037/0022- 3514.83.1.126

Kiesler, D. J. (1983). The 1982 interpersonal circle: A taxonomy for complementarity in human transactions. Psychological Review, 90, 185– 214. http://dx.doi.org/10.1037/0033-295X.90.3.185

Kivlighan, D. M., & Paquin, J. D. (2014). Whose perceptions matter more: Mine or my group’s? An actor-partner interdependence analysis disag- gregating between-person and between-session effects of group climate perceptions on intimate behaviors. Journal of Counseling Psychology, 61, 333–339. http://dx.doi.org/10.1037/cou0000025

T hi

s do

cu m

en t

is co

py ri

gh te

d by

th e

A m

er ic

an P

sy ch

ol og

ic al

A ss

oc ia

ti on

or on

e of

it s

al li

ed pu

bl is

he rs

. T

hi s

ar ti

cl e

is in

te nd

ed so

le ly

fo r

th e

pe rs

on al

us e

of th

e in

di vi

du al

us er

an d

is no

t to

be di

ss em

in at

ed br

oa dl

y.

360 COX ET AL.

Leichtentritt, J., & Shechtman, Z. (2010). Children with and without learning disabilities: A comparison of processes and outcomes following group counseling. Journal of Learning Disabilities, 43, 169 –179. http:// dx.doi.org/10.1177/0022219409345008

Lese, K. P., & MacNair-Semands, R. R. (2000). The Therapeutic Factors Inventory: Development of a scale. Group, 24, 303–317. http://dx.doi .org/10.1023/A:1026616626780

Livesley, W. J. (2012). Integrated treatment: A conceptual framework for an evidence-based approach to the treatment of personality disorder. Journal of Personality Disorders, 26, 17– 42. http://dx.doi.org/10.1521/ pedi.2012.26.1.17

Lobbestael, J., Leurgans, M., & Arntz, A. (2011). Inter-rater reliability of the Structured Clinical Interview for DSM–IV Axis I disorders (SCID I) and Axis II disorders (SCID II). Clinical Psychology & Psychotherapy, 18, 75–79. http://dx.doi.org/10.1002/cpp.693

Marziali, E., Munroe-Blum, H., & McCleary, L. (1999). The Objective Be- havioral Index: A measure for assessing treatment response of patients with severe personality disorders. Journal of Nervous and Mental Disease, 187, 290 –295. http://dx.doi.org/10.1097/00005053-199905000-00004

McCallum, M., Piper, W. E., & Morin, H. (1993). Affect and outcome in short-term group therapy for loss. International Journal of Group Psy- chotherapy, 43, 303–319. http://dx.doi.org/10.1080/00207284.1993.117 32595

Milbury, K., Lopez, G., Spelman, A., Wood, C., Matin, S. F., Tannir, N. M., . . . Cohen, L. (2017). Examination of moderators of expressive writing in patients with renal cell carcinoma: The role of depression and social support. Psycho-Oncology, 26, 1361–1368. http://dx.doi.org/10 .1002/pon.4148

Miles, J. R., Paquin, J. D., & Kivlighan, D. M. (2011). Amount and consistency, two components of group norms: An actor partner interde- pendence analysis of intimate behaviors in groups. Group Dynamics: Theory, Research, and Practice, 15, 326 –342. http://dx.doi.org/10.1037/ a0024676

Muthén, L. K., & Muthén, B. O. (2012). Mplus user’s guide (7th ed.). Los Angeles, CA: Author.

O’Loughlin, J. I., Cox, D. W., Kahn, J. H., & Wu, A. D. (2018). Attach- ment avoidance, alexithymia, and gender: Examining their associations with distress disclosure tendencies and event-specific disclosure. Jour- nal of Counseling Psychology, 65, 65–73. http://dx.doi.org/10.1037/ cou0000245

Orlinsky, D. E., Ronnestad, M. H., & Willutski, U. (2004). Fifty years of psychotherapy process-outcome research: Continuity and change. In M. J. Lambert (Ed.), Bergin and Garfield’s handbook of psychotherapy and behaviour change (5th ed., pp. 307–389). New York, NY: Wiley.

Paris, J. (2004). Sociocultural factors in the treatment of personality dis- orders. In J. J. Magnavita (Ed.), Handbook of personality disorders: Theory and practice (pp. 135–147). Hoboken, NJ: Wiley.

Pennebaker, J. W., & Seagal, J. D. (1999). Forming a story: The health benefits of narrative. Journal of Clinical Psychology, 55, 1243–1254. http://dx.doi.org/10.1002/(SICI)1097-4679(199910)55:10�1243::AID- JCLP6 3.0.CO;2-N

Piper, W. E., Ogrodniczuk, J. S., McCallum, M., Joyce, A. S., & Rosie, J. S. (2003). Expression of affect as a mediator of the relationship between quality of object relations and group therapy outcome for patients with complicated grief. Journal of Consulting and Clinical Psychology, 71, 664 – 671. http://dx.doi.org/10.1037/0022-006X.71.4.664

Piper, W. E., Rosie, J. S., Joyce, A. S., & Azim, H. F. A. (1996). Time-limited day treatment for personality disorders: Integration of research and practice in a group program. Washington, DC: American Psychological Association. http://dx.doi.org/10.1037/10208-000

Preacher, K. J., Zyphur, M. J., & Zhang, Z. (2010). A general multilevel SEM framework for assessing multilevel mediation. Psychological Methods, 15, 209 –233. http://dx.doi.org/10.1037/a0020141

Raudenbush, S. W., & Bryk, A. S. (2002). Hierarchical linear models: Applications and data analysis methods (2nd ed.). Thousand Oaks, CA: Sage.

Rutan, J. S., Stone, W., & Shay, J. (2014). Psychodynamic group psycho- therapy (5th ed.). New York, NY: Guilford Press.

Schneider, S., Moyer, A., Knapp-Oliver, S., Sohl, S., Cannella, D., & Tar- ghetta, V. (2010). Pre-intervention distress moderates the efficacy of psy- chosocial treatment for cancer patients: A meta-analysis. Journal of Behav- ioral Medicine, 33, 1–14. http://dx.doi.org/10.1007/s10865-009-9227-2

Shadish, W. R., Jr. (1984). Intimate behavior and the assessment of benefits in clinical groups. Small Group Behavior, 15, 204 –221. http:// dx.doi.org/10.1177/104649648401500204

Stiles, W. B. (1995). Disclosure as a speech act: Is it psychotherapeutic to disclose? In J. W. Pennebaker (Ed.), Emotion, disclosure, and health (pp. 71–91). Washington, DC: American Psychological Association. http://dx.doi.org/10.1037/10182-004

Svartberg, M., Stiles, T. C., & Seltzer, M. H. (2004). Randomized, con- trolled trial of the effectiveness of short-term dynamic psychotherapy and cognitive therapy for Cluster C personality disorders. The American Journal of Psychiatry, 161, 810 – 817. http://dx.doi.org/10.1176/appi.ajp .161.5.810

Tabachnick, B. G., & Fidell, L. S. (2013). Using multivariate statistics (6th ed.). Boston, MA: Pearson.

Tasca, G. A., Ramsay, T., Corace, K., Illing, V., Bone, M., Bissada, H., & Balfour, L. (2010). Modeling longitudinal data from a rolling therapy group program with membership turnover: Does group culture affect individual alliance? Group Dynamics: Theory, Research, and Practice, 14, 151–162. http://dx.doi.org/10.1037/a0018778

Wang, L. P., & Maxwell, S. E. (2015). On disaggregating between-person and within-person effects with longitudinal data using multilevel models. Psy- chological Methods, 20, 63– 83. http://dx.doi.org/10.1037/met0000030

Whelton, W. J. (2004). Emotional processes in psychotherapy: Evidence across therapeutic modalities. Clinical Psychology & Psychotherapy, 11, 58 –71. http://dx.doi.org/10.1002/cpp.392

Yalom, I. D., & Leszcz, M. (2005). The theory and practice of group psychotherapy (5th ed.). New York, NY: Basic Books.

Received June 8, 2018 Revision received November 21, 2018

Accepted November 29, 2018 �

T hi

s do

cu m

en t

is co

py ri

gh te

d by

th e

A m

er ic

an P

sy ch

ol og

ic al

A ss

oc ia

ti on

or on

e of

it s

al li

ed pu

bl is

he rs

. T

hi s

ar ti

cl e

is in

te nd

ed so

le ly

fo r

th e

pe rs

on al

us e

of th

e in

di vi

du al

us er

an d

is no

t to

be di

ss em

in at

ed br

oa dl

y.

361DEPRESSION AND NEGATIVE-AFFECT EXPRESSION