10.1192/bjp.156.6.809Access the most recent version at DOI: 1990, 156:809-818.BJP 

A S Bellack, R L Morrison, J T Wixted and K T MueserAn analysis of social competence in schizophrenia.

Referenceshttp://bjp.rcpsych.org/content/156/6/809#BIBLThis article cites 0 articles, 0 of which you can access for free at:

permissionsReprints/

permissions@rcpsych.ac.ukto To obtain reprints or permission to reproduce material from this paper, please write

to this article atYou can respond http://bjp.rcpsych.org/cgi/eletter-submit/156/6/809

from Downloaded

The Royal College of PsychiatristsPublished by on August 22, 2012http://bjp.rcpsych.org/

http://bjp.rcpsych.org/site/subscriptions/ go to: The British Journal of PsychiatryTo subscribe to

British Journal of Psychiatry, 1990, 156, 809—818

An Analysis of Social Competence in Schizophrenia

ALAN S. BELLACK,RANDALLL. MORRISON, JOHN T. WIXTED and KIM T. MUESER

Twenty-one schizophrenics with prominent negative symptoms were compared with 37schizophrenics without them, 33 patients with major affective disorder and 20 non-patientcontrols on a battery of measures including a role-play test of social skills, the Social AdjustmentScale, and the Quality of Life Scale. The negative schizophrenics were most impaired on everysubscale of each measure, followed in order by the non-negative schizophrenics, affectivedisorder patients, and non-patient controls. The social skill measures were not correlated withpositive symptom levels, but were highly correlated with measures of community functioning.The results are consistent with the hypothesis that socialdysfunction results from focal deficitsin social skills.

Severe impairment of social functioning is one of thehallmarks of schizophrenia. Deterioration of socialrelations is one of the defining diagnostic criteriaspecified in DSM—III—R(American PsychiatricAssociation, 1987), and social isolation or withdrawal and marked impairment in major life rolefunctioning are listed as prominent prodromal andresidual symptoms. While there has been widespreadagreement about the existence of severe socialdisability for some time, there is surprisingly littledata on the precise nature or the basis of thedysfunction (Wallace, 1984).

The most prominent hypothesis is that socialimpairments are epiphenomena which result fromother more basic symptoms of the disorder, such asnegative or deficit symptoms. There is considerableevidence to suggest that such symptoms as avolition,anhedonia and alogia demarcate a distinct subtypeof schizophrenia (Andreasen, 1985; Carpenteret al, 1985; Crow, 1985). Patients whose illness ischaracterised by prominent negative symptoms havean increased frequency of structural brain impairment and decreased intellectual functioning, as wellas a poorer prognosis (Johnstone et al, 1978). Theyalso tend to have poor pre-morbid social competenceand impaired social relationships (Liddle, 1987;Johnstone et al, 1979).

These findings suggest that negative symptomsplay a central role in social disability. This possibilityhas high face validity in regard to such symptomsas blunted affect, avolition, anhedoma, and anergia.However, other negative symptoms, such as asocialityand alogia, seem to be as much reflections of socialimpairment as causes of it. Indeed, Carpenter et al(1988) recently distinguished between primary andsecondary negative symptoms, the former representingbasic and enduring aspects of the illness while thelatter are consequences of other factors. Thus,secondary negative symptoms could result from

social disability as well as vice versa. Thisdichotomy has important implications for understanding the nature of the disorder and itstreatment, but it has not yet been systematicallyexamined.

A second hypothesis about the basis of socialimpairment in schizophrenia is provided by thebehavioural model of social skills (Bellack &Morrison, 1982; Liberman, 1982; Trower et a!,1978). According to this model, effective socialperformance requires the skill to perform theappropriate response, as well as the cognitive/perceptual ability to determine what response isappropriate and the motivation to perform this(McFall, 1982). Thus, social skills are seen asnecessary (although not sufficient) to produceeffective social performance. The second hypothesissuggests therefore that schizophrenics have focaldeficits in such social skills, and previously it hasbeen presumed that they either fail to learnthese initially or lose them after long periods ofhospitalisation or social isolation. Unfortunately, theexisting literature provides no sound validation forthe skills model or for determining the precise roleof social skills deficits in lowered social competence.The literature on pre-morbid social competence andchildhood adjustment documents that many schizophrenics have long-standing impairments in rolefunctioning. However, these studies are primarilyretrospective and cannot elucidate the reasons forthe faulty performance (Zigler & Levine, 1981). Ithas frequently been demonstrated that social skillstraining, which is based on the skills model, increasesskill level and improves overall functioning (Morrison& Wixted,1989).Thisprovidesconvergentvalidationalsupport for the skillsmodel, but does not demonstratethat pre-treatment deficits were responsible forimpaired functioning or that improved functioningresulted from increased skill.

809

SchizophrenicsThoseNonnegativenon

negativewithaffectivedisorderpatient

controls(n=21)(n=37)(n=33)(n=20)Age1

(mean):years30.332.834.832.9Sex:male

1222187female9151513Education

(mean):numberofyears

11.812.013.112.8SES24.74.74.13.0Race:

black101368white11242712Age

of onsetofdisorder(mean):years

21.321.824.3—Durationofdisorder(mean):years

9.110.210.7—Currenthospital

isation(mean):days

29.431.928.6—

810 BELLACK ET AL

Although the most frequently cited source ofsupport for the skills model is behavioural observations of patients engaging in social interactions,few such studies have reported separate dataon a carefully diagnosed cohort of schizophrenics.Longabaugh et a! (1966) found notable deficits ina sample of schizophrenic in-patients. However, theirsubjects had all been hospitalised for at leastsix years, and could have been suffering frominstitutionalisation syndrome rather than skill deficiencies. Argyle (1981) reported that schizophrenicshave a variety of specific skills deficits, includinginappropriate facial expression, gestures and posture,low rewardingness, and poor synchronising. Lindsay(1982) also found that schizophrenics have specificperformance deficits, although his sample consistedof only three patients. However, Rutter (1977a,b;1978), in a study of paralinguistic behaviour, speechpatterning, and visual interaction patterns, found fewdifferences between schizophrenics and matchedgroups of non-patients and patients with otherdiagnoses.

Despite its widespread acceptance, the evidence insupport of the skills hypothesis is quite tenuous.Firstly, no studies have examined patients diagnosedaccording to current (DSM-III (American PsychiatricAssociation, 1980)or DSM-III-R) criteria. Secondly,it is not clear that schizophrenics have specific skilldeficits, as opposed to a general impairment in socialfunctioning which results from other aspects of thedisorder, such as negative symptoms. Thirdly, it isalso uncertain if the social impairments characteristicof schizophrenia are specific to the disorder, or areassociated with other chronic disorders as well. Thepurpose of the current investigation was threefold:(1) to determine if schizophrenics are more sociallyimpaired than other chronic patients or non-patients;(2) to compare the social competence of patients withand without negative syndrome; and (3) to examinethe relationship of social skills to positive andnegative symptoms. A secondary goal of the studywas to examine the validity of role-play tests of socialskills.

Patients and controls

withno obviousevidenceof organicbrainsyndrome,mentalretardation, or alcohol or drug abuse. Affective disorderpatientswererequiredto have at leasta six-monthhistoryof illness,so as to be comparableto the schizophrenicsonchromcity. Patients were interviewed within several weeksof theirindexhospitalisationwiththe Schedulefor AffectiveDisorders and Schizophrenia (SADS) (Spitzer & Endicott,1978)or the Structured Clinical Interview for DSM—II1(SCID)(Spitzer&Williams,1985)bya trainedpsychologist.psychiatrist or psychiatric nurse. Based on these interviewsand hospital records, a consensusdiagnosiswas reachedby the interviewerand one of the first two authors (ABorRM) in accordance with DSM-ffl-R criteria. Approximately20¾of the structured interviews were reviewed by athird clinicianin order to provide an independent checkon the reliability of diagnosis. There was a 100 percent agreementon primary DSM-III—Rdiagnosisfor thissample.

Non-patient subjects were recruited from the nonprofessional staff and surrounding community of thehospital. Non-patient volunteers were excluded fromparticipation if they had a history of treatment for apsychiatric disorder. Groups were matched as carefully aspossibleon age, socioeconomicstatus (SES)(Hollingshead& Redlich, 1958), race, and sex. ANOVAs (on age,education) and x2(on sex, race) were non-significant (NS),indicating that the groups were carefully matched. However,as most non-patient controls were employed, there was a

TABLE I

Demographic variables and information about course ofdisorder for patients and non-patients

Method

The subjects, who included 58 schizophrenic patients(21 with negative syndrome and 37 with non-negativesyndrome), 33 patients with major affective disorder (29bipolar, 4 unipolar), and 20 non-patient controls, wereselected from in-patient admissions to the psychiatry serviceof the Medical College of Pennsylvania (MCP) at theEastern Pennsylvania Psychiatric Institute (EPPI). Patientswere considered for the study if they were 18-50 years-of-age,

1. Patients included between 18 and 50 years-of-age.2. Hollingshead & Redlich, 1958.

811SOCIAL COMPETENCE IN SCHIZOPHRENIA

videotaped). Given these limitations, behaviour on thesetasks must be interpretedcautiously. However, they havebeen shown to be quite useful in evaluating the effects ofsocial skills training and in discriminating high- and lowskill groups from diverse diagnostic populations (Bellack,1979; 1983).

Allsubjectsparticipatedin a role-playtestof socialskillswhich was videotaped and retrospectively rated on a numberof specific behavioural components. Subjects were seatedin a videotape studio with a research assistant who portrayedan interpersonal partner. They enacted 12social encountersin which the subject was required to either initiate aconversation, resist unfair treatment, or express appreciationfor something the confederate did or said. Each interactioncontinued through three verbal interchanges (i.e. the subjectwasrequiredto makeat leastthreeresponsesto confederateprompts).

Role-play enactments were subsequently rated onappropriateness of: gaze, speech duration, meshing (smoothness of turn taking and conversational pauses), affect, and

two verbal content measures (request/compliance andpraise/appreciation). Independent ratings were made ofoverall social skill. All ratings were made by researchassistants who were blind to subjects' group status. Raters

weretrained on a library of videotapesfrom our previousresearch. They first learned to match criterion ratings, andthen practised rating independently until each behaviouralcategory was rated with a reliability of at least r = 0.80.Initial data tapes from this protocol were rated independentlyby theprimaryraters,and thenreviewedjointlyby the raters and one of the authors to further ensure theaccuracy and consistency of the ratings. One third of thevideotapes (drawn proportionately from each subject groupand different time points) were scored by an independentrater to providea reliabilitycheck. Reliabilitywashighforall categories (Pearson correlation coefficients range,0.74—0.89).

Procedure

Clinical records of all new admissions to the MCP/EPPIin-patient units were reviewed by a research staff member,and appropriate patients were then screened using a SADSor the SCIDto determinethe diagnosis.Patients acceptedinto the study participated in the interviewmeasuresandrole-play test on two separate days within one week of thediagnostic interview. Testing was initiated after the researchteam and the attending physician concurred that acutesymptoms had remitted sufficiently for the patient to leavethe ward and complete the research tasks. Non-patientsubjects were similarly scheduled to complete the interviewand role-play test measures within a one-week period.

Data analysis plan

The data analysis consisted of a series of steps designedto first categorise and describe our sample, and then testthe followinghypotheses:(1) negativeschizophrenicswillbe moresociallyimpairedthan non-negativepatients, whowill not differ from bipolars, but all three groups will beimpaired in relation to non-patient controls; (2) social

significant difference between them and the patient groupson SES. All subjects were paid for participation in the study.

Table I provides a summary of age, educational level,SES, race, and sex, as well as age at onset of disorder (i.e.first in-patient hospitalisation), duration of disorder, andduration of present hospitalisation for the groups. Allpatients were on medication at the time of testing. Overall,the patient samples are young, acute groups, with fewhospitalisations before the index assessment.

Structured Interviews

The diagnosticinterviewercompletedthe BriefPsychiatricRating Scale (BPRS) (Overall & Gorham, 1962). All subjectsthen participated in a secondstructured interviewwith aninterviewerwho was blind to diagnosis.That interviewerrated the subject on the Scale for the Assessment ofNegativeSymptoms(SANS)(Andreasen,1982;Andreasen& Olsen, 1982), and on two measures of social rolefunctioning in the community: the Social Adjustment Scale

II (SAS) (Weissman & Bothwell, 1976;Weissman & Paykel,1974), and the Quality of Life Scale (QLS) (Heinrichs etal, 1984).The SASis a self-reportand interviewerratingscale which yields scores on five dimensions: ‘¿�workadjustment', which includes adjustment as a student orhomemaker; ‘¿�householdadjustment', which deals withrelationships with household members; ‘¿�externalfamilyadjustment', or relationships with relatives outside of thehousehold; ‘¿�socialand leisure adjustment'; and ‘¿�generaladjustment', which is an overall measure. The QLS is a21-iteminterviewerrating scalewhichprovidesratings onfour dimensions: ‘¿�interpersonalrelations', which judgescapacity to form relationships as well as the extent of socialinteractions;‘¿�instrumentalrole functioning',whichfocusesprimarily on occupational, student, or homemaker roles;‘¿�intrapsychicfoundations', which relates to cognitive,affective, and motivational functioning; and ‘¿�commonobjects and activities', which assessesparticipation with theobjects and activities of every day life (e.g. readingnewspapers, shopping, possession of a wallet, keys,watch, etc.).

Interviewers were trained and supervised by the first twoauthors. Approximately 25°loof all interviews were ratedby a second, independent rater. Reliability was satisfactoryfor each subscaleof each measure(reliabilitycoefficientsrange r=0.77-0.9l) with the exception of ‘¿�householdadjustment' on the SAS (r=0.64).

Behavioural observations

Social skill is a construct which has defied easy measurement. Characteristically, it is defined operationally byperformance on an analogue test of social interaction (e.g.a role play or structured conversation). Adequate performance on the test implies that the individual has therequisite skills in his/her repertoire, although other factorsmay interfere with performance in real life interactions. Anindividual who does not perform adequately may notpossess the requisite skills,or may fail to manifest them

due to confounding factors (e.g. anxiety about being

SchizophrenicsThose withaffectiveNon-patientnegative(NS)non-negative (5)disorder (A)controls(C)(n=21)(n=37)(n=33)(n=20)SubscaleMean

RangeMean RangeMean RangeMean Range F'TukeyofBPRS(s.d.)(s.d.)(s.d.)(s.d.)

812 BELLACKET AL

competencemeasureswillhaveonlya moderatecorrelationwith symptom measures (suggesting that social dysfunctionis not solely a by-product of either positive or negativesymptoms); (3) role-play measures of social skills will berelated to role functioning in the community.

Results

Subclasslflcatlon of schlzophrenlcs

Schizophrenic subjects were retrospectively categonsed asnegative or non-negative according to the criteria ofAndreasen (1982). Negative subjects were characterised byscores of four (marked) or five (severe) on two or morecategories on the SANS. As all subjects were receivingneuroleptics and acute positive symptoms were at leastpartially remitted, we did not separately classify subjectsas having positive schizophrenia. Hence, the non-negativegroup includes patients who might fit Andreasen's positiveor mixed categories. Twenty-one patients met the criteriafor negative schizophrenia and 37 were classified as nonnegative. The negative group had a mean score of 3.27(range= 9-24) on the five SANS scales, compared to a meanof 1.67 (range = 1—15)for non-negatives. Analyses ofvariance indicated that the negative group exhibitedsignificantly more symptoms on each of the five SANScategories: ‘¿�affectiveflattening', F(l,51 d.f.)= 26.50,P<0.OOl;‘¿�alogia',F(l,51 d.f.)= 13.94,P<0.OOl;‘¿�avolitionapathy', F(1,5l d.f.)= 60.59, P<zO.OOl; ‘¿�anhedoniaasociality', F(l,5l d.f.)= 34.27, P<0.OOl; and ‘¿�inattention',F(l,5l d.f.)=8.l, P<0.005.

Medication and symptoms

All schizophrenics were receiving neuroleptics at the timethey were assessed. Some of the common side effects of

neuroleptics, including akinesia and pseudoparkinsonism,can result in impairments which are similar to negativesymptoms. As a partial check on the extent to which thenegative symptoms of our subjects may have been druginduced, we compared the dosage levelsof patients in the twoschizophrenicgroups.Dosageswerefirstconvertedto chlorpromazine equivalents (Davis & Glen, 1984). Negativesymptom patients were receiving a mean of 891.67 mg perday compared with 883.80mg per day for non-negativeones.This difference was not significant (t (49 d.f.)= 0.05).Furthermore, correlations between SANS scores and dosagelevel were uniformly low and non-significant (range—¿�0.10—0.03).

Descriptivestatisticsand results of analyses of variance forBPRSsummaryscoresarepresentedinTableII. Asexpected,the negativeschizophrenicgroup scoredsignificantlyhigherthan the other three groups on the ‘¿�anergia'factor, whichsubstantially reflects negative symptoms. The two schizophrenic groups did not differ on any other scale, indicatingthat the negative group did not exhibit more symptomsuniformly. The non-negative and negative schizophrenicshad mean scores of 12.41 and 14.14 respectively on theBPRS‘¿�thoughtdisorder'scale,indicatingthat mostof thesesubjects were still actively psychotic at the time they wereassessed.The affectivedisorderpatients werelessthoughtdisorderedthan either of the schizophrenicgroups at thetime of assessment (x=9.15, P<0.05), but 88°loof themwere psychotic at the time of admission, and 67°locontinuedto exhibitpsychoticsymptomsat the timeof testing(basedon BPRS scores of four or more on any of the individual‘¿�thoughtdisorder' items).

Group differences In social competence

The basic strategy for comparing social functioning of thefour groups was to conduct separate two-way, repeated

TABLE IIDescriptive statistics, F ratios, and Tukey comparisons for the Brief Psychiatric Rating Scale (BPRS)

Anxiety!depression 8.47 (4.11) 4—1810.08 (3.68) 4—17 8.47 (3.48) 4—20 7.40 (3.14) 4—14 2.69 S>C

Anergia 10.34 (4.45) 4—19 7.95 (3.60) 4—18 5.18 (2.04) 4—12 4.43 (0.84) 4—7l8.02** NS S>A,CNS>S

Thoughtdisorder 14.14 (5.62) 4—2512.41 (4.36) 4—23 9.15 (4.29) 4—19 4.60 (1.19) 4—821.56** NS S,A>C

NS S>AActivation 6.79 (3.08) 3—13 6.30 (2.26) 3—12 6.73 (3.00) 3—13 3.60 (0.88) 3—6 7.85 NS S,A>CHostility 7.86 (4.09) 3—197.68 (3.63) 3—146.21 (2.86) 3—124.05 (1.47) 3—76.77** NS S>C

Total 44.5 (11.1) 23—64 42.1 (9.6) 24—57 34.5 (8.0) 17—5123.0 (4.9) 17—3627.05@ NS S,A>CNS S>A

eP<0.05, e@P<0@®@1. 3, 107d.f.

Schizopnegative

Mean' (s.d.)hrenicsnon-negativeMean' (s.d.)Those

withaffective disorder

Mean' (s.d.)Non-patientcontrols

Mean'(s.d.)Role-play

test of socialskillsNon-verbalcomponent0.64 (1.00)0.10 (0.60)—0.17 (0.40)—0.56(0.31)Verbal

component0.46 (0.75)0.23 (0.86)—0.07 (0.69)—0.69(0.41)Overallratings0.76 (0.86)0.29 (0.80)—0.17 (0.71)— 1.14(0.67)Quality

of LjfeScaleInterpersonalrelations— 1.00 (0.47)—0.23 (0.70)—0.06 (0.76)1.53(0.60)Role

functioning—0.71 (0.39)—0.22 (0.74)—0.06 (1.00)1.54(0.37)Intrapsychicfoundations— 1.26 (0.53)— 1.19 (0.63)0.15 (0.75)1.45(0.40)Common

objects & activities—0.61 (0.95)—0.07 (0.75)—0.04 (0.97)1.23(0.75)SocialAdjustmentScaleWorkadjustment0.98 (0.21)0.29 (0.55)0.10 (0.87)— 1.39(0.57)Household

adjustment0.85 (0.57)0.27 (0.57)0.14 (0.92)— 1.35(0.83)External

Family adjustment0.83 (0.60)0.28 (0.88)0.23 (0.74)— 1.37(0.52)Social/Leisure

adjustment0.90 (0.32)0.29 (0.56)0.20 (0.77)— 1.65(0.63)General

adjustment0.92 (0.42)0.24 (0.70)0.11(0.69)— 1.64 (0.57)

813SOCIAL COMPETENCE IN SCHIZOPHRENIA

measures analyses on the role-play data, the SAS, and theQLS, in which group served as a between-subjects variable,and the subscalesof the threemeasuresof socialfunctioningservedas within-subjectsvariables.The statisticused wasa MANOVA approach to repeated measures analysis ofvariance (Bray & Maxwell, 1985), which was calculated bySPSS Advanced version V2.0 (Norusis, 1988). Theseanalyses allowed us to simultaneously examine both overallgroup differences within each domain and possibleinteraction effects, while controlling the number ofindependent tests conducted. As the subscaleson the variousinstruments varied somewhat in scoring range and format,raw scores were normalised and converted to z scores forentry into the MANOVA. Multivariate effects on eachanalysis were tested with Pillai's trace statistic. Becauseof the z transformation, each social competence subscalehad the same mean and standard deviation when collapsedacross subject groups. Therefore, the resultant Fratios weremeaningless and are not discussed below.

Role-play test data

Previous research on social skill suggeststhat the components of role-play behaviour are substantially related tooverall social competence, but that they do not individuallyaccount for a significant proportion of variance (Conger& Farrell,1981;Romano & Bellack,1980).Consequently,the componentscoresweresummedfor eachsubjectacrossnon-verbal(gaze, length, meshing,and affect) and verbal(praise/appreciation and request/compliance) behaviour

categories.This meant that the within-subjectsvariableson the role-play MANOVA consisted of overall social skillas well as non-verbal and verbal skill. The group effect washighly significant(F[3, 102d.f.] =23.83, PcZO.001), andthere was a significant group by subscale interaction(F [6,204 d.f.] = 2.85, P<0.05). As indicated in Table III,

the order of means was the same for each of the variables:negative schizophrenics exhibited the least skill, followedby non-negative schizophrenics, affective disorder patients,and non-patient controls respectively. Tukey post hoc tests(across the three skill variables) indicated that the patientgroups each exhibited significantly less skill than the nonpatient controls (P<O.05), and the negative schizophrenicsperformed significantly worse than both the non-negativesand the affective disorder patients (P<0.05). The lattertwo groups were not significantly different.

The significant interaction was examined by conductingone-way, repeated measures ANOVAs on each group. Asthese analyses were conducted on z scores generatedseparately for each skill category, they reflect comparativestandingof thegroupson eachmeasureratherthan absolutecomparisons between skill levels in each domain. The onlyone of the four analysesto reach significancewasthat fornon-patient controls (F [2, 17 d.f.J = 6.07, P<0.00l).They were rated as proportionately more skilful than thepatient groups on the overall measure than on either of thetwo component categories.

SAS and QLS data

Results for the QLS and SAS (Table III) were similar tothe results for the role-play test. On each subscale in eachof these instruments, negative schizophrenics were the mostimpaired, followed in order by non-negative schizophrenics,affective disorder patients and non-patient controls. TheMANOVA on the QLS yielded a highly significant groupeffect (F [3, 102d.f.] = 68.72, P<0.OOl) and a significantgroup by subscale interaction (F [9, 306 d.f.)= 2.99,P<0.05). Tukey post hoc tests on group means indicatedthat all three patient groups scored significantly worse thanthe non-patient controls. The negative schizophrenics wererated significantly worse than the other two patient groups,

TABLE IIIDescriptive statisticsfor socialfunctioning measures(role-play, SAS and QLS)

1. The meanswerecalculatedusingz scoretransformations.

Anxiety!AnergiaBriefPsychiatric

ThoughtRatingScale

ActivationHostilityTotaldepressiondisorderRole-play

test of socialskillsVerbalcomponents—0.17—0.020.00—0.020.07—0.04Non-verbalcomponents—0.030.240.080.16—0.030.11Overallratings—0.160.180.170.170.070.16Quality

of LifeScaleInterpersonalrelations—0.03—0.05— 0.44**— 0.40— 0.38—0.49**Role

functioning0.10—0.19—0.17—0.30—0.22—0.29Intrapsychicfoundations0.09—0.27— 0.43**—0.30— 0.35*—0.48@Common

objects &activities—0.07—0.08—0.12—0.05—0.20—0.20SocialAdjustmentScaleWork

adjustment—0.41'0.37'0.090.280.070.11Householdadjustment—0.260.170.320.37'0.130.27External

familyadjustment—0.070.200.290.220.420.40'Social/leisureadjustment—0.180.170.260.39'0.35'0.35'General

adjustment—0.250.170.230.300.34*0.28

814 BELLACKET AL

which did not differ from one another. As with therole-play data, the significant interaction was examinedwith a series of one-way, repeated measures ANOVAs. Theresults indicated that the interaction was accounted forby the low scores for negative schizophrenics on theintrapsychic foundations subscale (F [3, 17d.f.l = 5.94,P<0.00l). This scale, which consists of items such asmotivation, curiosity, and anhedonia, is more a measureof negative symptoms per se than of ‘¿�qualityof life'.Consequently, it is not surprising that negative schizophrenicswould be rated differentiallylower on this scalethan on the other subscales.

Each of the scales on the SAS can be coded ‘¿�notapplicable' if it is not relevant to the subject's currentlife situation. A majority of subjects in each of the threepatient groups received such scores for either the ‘¿�workadjustment'or ‘¿�externalfamilyadjustment'scales(e.g.theywere not working, did not have any contact with relatives,etc.). Consequently, the MANOVA was conducted on theremaining three scales. There was a highly significant groupeffect (F [3,89 d.f.] = 62.27, P<0.001), but the interactionwas not significant. As with the role play and QLS,Tukey tests indicated that the three patient groups werefunctioningsignificantlymore poorlythan the non-patientcontrols, and the negative schizophrenics were significantlymore impaired than either of the other two patient groups.UnivariateANOVAs were conducted on thosepatientswho

did receive scores on the ‘¿�work'and ‘¿�externalfamily' scales,and the three patient groups wererated significantlyloweron both: external family, F(3, 91 d.f.)= 30.23, P<0.001;work F(3, 75 d.f.)=45.37, P<0.001. The negativeschizophrenics were significantly worse than the other twopatient groups on ‘¿�workadjustment', but did not differon ‘¿�externalfamily'.

Social functioning and symptoms

Hypothesis two proposed that social dysfunction was notsolely a by-product of either positive or negative symptoms.In order to examine this, Pearson correlation coefficientswere calculated between the social functioning measures(role play, SAS, and QLS) and the symptom measures(BPRS and SANS) for all schizophrenics combined. Theresults for positive symptoms, as reflected by the BPRS,appear in Table IV. The correlations between BPRS factorscores and role-play measures were uniformly low and nonsignificant (range —¿�0.18—0.25), indicating that there waslittle relationship between positive symptoms and socialskills. The correlations with the SAS and QLS were morevariable. While most of the correlations were nonsignificant, there did appear to be a consistent relationshipbetween several BPRS subscalesand adequacy of interpersonal relationships, as reflected on both the SAS andQLS.

A similarpattern was reflectedon the SANS,althoughthe correlationswereconsistentlyhigher(seeTableV).ThecorrelationsbetweenSANSand role-playmeasuresrangedfrom0.08to 0.55.Not surprisingly,affectiveflatteninghada notable impact on non-verbal skills as did anergia in theBPRS. For the most part, SANS ratings accounted for lessthan 10—15%of the variance in role-play behaviour whichis consistent with the BPRS anergia ratings. In contrast,SANS scales were highly correlated with role functioning(i.e. SAS and QLS), in some cases accounting for more than50% of the variance.

Validity of role-play measures

In order to examine hypothesis three (the validity of therole-play test), a series of Pearson correlations was

TABLE IV

between social functioning measures (role-play, SAS and QLS) and a symptommeasures (BPRS)

Pearson correlation coefficients

‘¿�P<O.Ol,“¿�P<O.OOl.

AffectiveflatteningScale

for the AssAlogiaessmen:

of NegaAvoiltion

apathytive

SymptomsAnhedonia

asocialityInattentionRole-play

test of socialskillsVerbalcomponents0.190.080.180.090.12Non-verbalcomponents0.55―0.38'0.290.36'0.29Overallratings0.36'0.34'0.46―0.32'0.32'Quality

ofLifeScaleInterpersonalrelations—0.54―—0.22—0.53―—0.83―—0.36'Rolefunctioning—0.28—0.24—0.53―—0.43―—0.33'Intrapsychicfoundations—0.58―—0.48―—0.75―—0.74―—0.52―Common

objects &activities—0.08—0.28—0.40―—0.16—0.18SocialAdjustmentScaleWork

adjustment0.330.36'0.72―0.58―0.44'Householdadjustment0.37'0.190.59―0.64―0.16External

family adjustment0.39'0.020.210.48―0.51―Social/leisureadjustment0.52―0.36'0.58―0.79―0.46―General

adjustment0.47―0.34'0.54―0.66―0.50―

Role play test of social skillsVerbal Non-verbal Overall

components componentsratingsQuality

of LifeScaleInterpersonalrelations

—¿�0.42—0.47—0.58Rolefunctioning—¿�0.42—0.38—0.53Intrapsychicfoundations

—¿�0.50—0.57—0.70Commonobjects&activities

—¿�0.36—0.35—0.40Social

AdjustmentScaleWorkadjustment0.430.540.62Householdadjustment

0.360.470.60Externalfamilyadjustment

0.370.460.43Social/leisureadjustment

0.410.460.59Generaladjustment 0.430.470.59

815SOCIAL COMPETENCE IN SCHIZOPHRENIA

TABLE VPearson correlation coefficients between social functioning measures (role-play, QLS AND SAS) and a symptom

measure (SANS)

‘¿�P<O.Ol,“¿�P<O.OOl.

TABLE VI

Pearson correlations between role-play test and QLS andSAS for all subjects'

Discussion

The results of this investigation provide strongsupport for the contention that schizophrenics havepronounceddeficitsinsocialcompetence.They weresubstantially impaired on verbal, non-verbal, andoverall ratings of social skill, as well as on interviewratings of role functioning. They scored significantlyworse than non-patient controls on each measure.They also were consistently rated as more impairedthan the affective disorder patients, although thedifferences were not significant for the non-negativesubsample. This is the first study in which a carefullydiagnosed cohort of schizophrenics was clearly shownto have marked social disability in comparison withboth non-patients and a matched group of patientswith another chronic disorder.

The results also indicate that the impairments insocial functioning are not simply secondary consequences of negative or positive symptoms. Thecorrelations between the SANS and role functioningmeasures (SAS and QLS) were consistently high andsignificant, which is consistent with the hypothesisthat negative symptoms can result in social impairment. However, the analyses of variance distinguishedbetween patients with and without high levels of negative symptoms. Patients with prominent negativesymptoms consistently showed the greatest impairments, but non-negative patients also exhibitedsubstantial deficits on each measure. These dataindicated that poor social functioning can occur inthe absence of negative symptoms.

‘¿�Allcorrelations significant at P<0.001.

calculated between role-play measures and scores onthe SAS and QLS for all subjects combined. Theresulting correlation matrix is presented in Table VI.The correlations were uniformly high and significant (range0.35—0.69;the negative sign reflects the inverse scalingof the QLS).

816 BELLACKET AL

The SANS, SAS, and QLS were all rated by thesame interviewer, in part to ensure that the SANSrater had adequate information to make accuratejudgements. This raises the possibility that thecorrelations between the three may be somewhatinflated by halo effects. However, two factors vitiatethis concern. The anergia subscale of the BPRS washighly correlated with the SANS total score (r=0.46,P<0.OOl). As the BPRS was rated by a differentinterviewer, this correlation provides an independentcheck on the validity of the SANS ratings. Similarly,independent ratings of the SANS, SAS, and QLSwere conducted to verify inter-rater reliability. Themagnitude of these reliability coefficients (ranger= 0.77—0.91)also suggests that ratings of the twodomains were distinct and valid.

The role-play test provides data on social competence from a different domain: behaviouralobservation. Both the ANOVAs and the correlationssuggest that negative symptoms could have adeleterious effect on social skill. However, once again,non-negative patientsalso exhibitmarked skilldeficits

and the correlations between the SANS and the roleplay measures were modest in comparison to thosebetween the former and the SAS and QLS. Moreover,the primary effect of negative symptoms on social skillcomponents seemed to be on non-verbal behaviour,possibly reflecting the concordance between such behaviour and ratings of anergia. These data suggestthat negative symptoms have a more pronouncedeffect on performance of social roles than on thesocial skills required for effective functioning. Ofcourse, these correlational data do not specifycausality or the direction of effects. Negativesymptoms could be a manifestation of socialimpairment, as well as vice versa. Moreover, bothphenomena could be mediated by some thirdvariable. However, the data do demonstrate that socialskill is not simply a result of negative symptoms.

Pearson's correlations indicate that positive symptoms do not have a uniform effect on socialbehaviour. Some symptoms (e.g. hostility, suspiciousness and excitement) apparently have a highlydeleterious effect on interpersonal relationships butcore psychotic symptoms (e.g. hallucinations, delusions) do not have a marked effect. Similar fmdingshave been reported by Johnstone et al (1979) andLiddle (1987). Previous research suggests that nonnegative schizophrenia is often associated with goodsocial adjustment between episodes (Andreasen,1985). The validity of that conclusion apparentlydepends on which symptoms are examined. Schizophrenics may sometimes be able to control orcompensate for delusions and hallucinations duringsocial interactions, but cognitive and affectivesymptoms have a more deleterious effect.

While non-negative patients were more sociallycompetent than negative patients, they were defmitelynot functioning adequately. They were significantlyimpaired in comparison to non-patient controls onalmost every measure. It is possible that theirfunctioning would improve as symptoms continuedto remit. However, it seems unlikely that socialdysfunction of the magnitude found would totallyresolve. Moreover, the affective patients were alsosignificantly impaired, despite having a different setof symptoms. These data further support thehypothesis that social dysfunction is not alwayssecondary to a specific set of schizophrenic symptoms.It should also be pointed out that many of oursubjectsweresufferingfrom some degreeofpositiveand negative symptoms. This combination may bemore pernicious than either alone, producing highlevels of distress and robbing patients of the capacityto effectively cope with their discomfort.

Both positive and negative symptoms were highlycorrelated with role functioning, but had littlerelationship with social skill. Conversely, social skillwas highly correlated with role functioning. Together,these findings provide considerable support for thesocial skills hypothesis. Social skill, as reflected byrole-play behaviour, appears to have a significanteffect on social functioning in the community,independent of other symptoms of the disorder. Thisconclusion is consistent with the hypothesis proposedby Strauss et a! (1974) that social competencerepresents an independent component of schizophrenia, orthogonal to both positive and negativesymptoms. By ‘¿�independent'we do not imply thatthat social functioning is unaffected by positive andnegative symptoms: our data strongly argue to thecontrary. However, social skill deficits can also resultin social dysfunctions, including phenomena such asalogia and asociality which present as negativesymptoms. Carpenter et a! (1988) refer to such effectsas ‘¿�secondary'negative symptoms. Similarly, in somecases ‘¿�thoughtdisorder' might better be conceptualisedas a deficit in social cognition or a communicationproblem (Thomas et a!, 1987; Cutting & Murphy,1988). Thus, the poorer role functioning of some ofthe negative schizophrenics in this study might haveresulted from more extensive skill deficits rather thanfrom negative symptoms per se (i.e. ‘¿�primary'negative symptoms).

The data provide strong support for the validityof the role-play test. Role play performance was, toa considerable extent, independent of core psychoticsymptoms and was highly related to independentmeasures of social functioning and interpersonalrelationshipsinthecommunity. Previousresearchhas suggested that role-play behaviour is not a direct

817SOCIAL COMPETENCE IN SCHIZOPHRENIA

parallel of behaviour in the natural environment(Bellack, 1979, 1983). However, the current studydocuments that role-play behaviour provides anexcellent reflection of overall social functioning inthe community. The ratings of ‘¿�overallsocial skill'in particular provide a cost-efficient way of assessinga critical aspect of functioning which is not welltapped by more traditional assessment procedures.

Current data are not adequate to determine theaetiological factor underlying schizophrenics' socialimpairment. Problems of social learning, as proposedby the social skills model, is but one possibility:schizophrenia characteristically strikes during youngadulthood, preventing the patient from masteringcritical social/developmental tasks associated withheterosexual and job skills, separation from family,etc. Conversely, there is evidence that some interpersonal deficits may result from a structural orfunctional brain impairment. For example, schizophrenics have been reported to have marked deficitsin facial affect recognition, which might result froma right-hemispheric lesion or imbalance in hemisphericdominance (Morrison et a!, 1988). Data on earlychildhood disturbances and poor pre-morbid socialfunctioning suggests the possibility of an insidiouspathological process, rather than a problem in learning(Lewine et a!, 1978; Strauss eta!, 1974). In keepingwith recent thinking on the heterogeneity of the disorder, it may well be that there are multiple possiblepathways to social dysfunction. While data from thecurrent investigation cannot resolve this issue, bydocumenting the relative independence of social skillsdeficits they provide substantial justification forfurther research on both the nature and aetiology ofsuch deficits.

Acknowledgement

This research was supported by a grant from the National Instituteof MentalHealth (MH 38636)to the seniorauthor.

References

Ar@eaic@PsYcHIATRIcA@ocIAnoN(1980)DiagnosticandStatisticalManual of Mental Disorders (3rd edn)(DSM-III). Washington,DC: APA.

—¿�(1987)Diagnosticand StatisticalManualof MentalDisorders(3rd edn, revised) (DSM-III-R). Washington, DC: APA.

ANDREA5EN,N. C. (1982) Negative symptoms in schizophrenia:definitionandreliability.ArchivesofGeneralPsychiatry,39,784—788.

—¿� (1985) Positive and negative schizophrenia: a criticalevaluation. Schizophrenia Bulletin, 11, 380—389.

—¿� & OLsE?J,S. (1982) Negative vs. positive schizophrenia:definition and validation. Archives of General Psychiatry, 39,789—794.

Aaoyi@,M. (1981)Thecontributionofsocialinteractionresearchto socialskifistraining.In SocialCompetence,(edsJ. D. Wine& M. D. Smye), pp. 261—286.New York: Guilford.

BELLACK,A. S. (1979) A critical appraisal of strategies for assessingsocial skill. Behavioral Assessment, 1, 157—176.

—¿� (1983) Recurrent problems in the behavioral assessment of

social skill. Behaviour Research and Therapy, 21, 29-42.—¿� & M0REIs0N, R. L. (1982) Interpersonal dysfunction. In

International Handbook of Behavior Mod4fication and Therapy(eds A. S. Bellack, M. Hersen & A. E. Kazdin), pp. 717-748.New York: Plenum.

BEAY,J. H. & MAXWELL,S. E. (1985) Multivariate Analysis ofVariance (series no. 07—054).Sage university paper series onquantitative applications in the social sciences. BeverlyHills: SagePublications.

CARPENTER, W. T., HEINRICHS, D. W. & ALPHS, L. D. (1985)

Treatment of negative symptoms. Schizophrenia Bulletin, 11,440—452.

—¿�, HEn@aIcHs, D. W. & WAGMAN, A. M. I. (1988) Deficit and

nondeficit forms of schizophrenia. American Journal ofPsychiatry, 145, 578—583.

CoN0RR,J. C. & FARRELL,A. D. (1981) Behavioral componentsof heterosocial skills. Behavior Therapy, 12, 41-55.

CRow, T. J. (1985) The two-syndrome concept: origins and currentstatus. Psychological Bulletin, 11, 471—486.

Cu-I-rING,J. & MURPHY,D. (1988) Schizophrenic thought disorder:a psychological and organic interpretation. British Journal ofPsychiatry, 152, 310—319.

DAVIS, J. M. & Guni, B. (1984) Pharmacological treatment in the

care of schizophrenic patients. In Treatment and Care forSchizophrenia (ed. A. S. Bellack). New York: Grune &Stratton.

HEINRICHS, D. W., HANWN, T. E. & CARPENTER, W. T., Jr (1984)

The Quality of Life Scale: an instrument for rating theschizophrenic deficit syndrome. Schizophrenia Bulletin, 12,388—398.43, 330—340.

HOLLINGSHEAD, A. B. & REDLICH, F. C. (1958) Social Class

and Mental Illness: A Community Study. New York: JohnWiley.

J0HNsTONE, E. C., CROW, T. J., FRIm, C. D., et al (1978) Thedementia of dementia praecox. Acta Psychiatrica Scandinavica,57, 305—324.

—¿�, Fium, C. D., Gou, A., et al (1979) The outcome of severeacute schizophrenic illnesses after one year. British Journal ofPsychiatry, 134, 28-33.

LEWINE, R. R. J., WATT, N. F. & FRYER, J. H. (1978) A study of

childhood social competence, adult premorbid competence, andpsychiatric outcome in three schizophrenic subtypes. Journal ofAbnormal Psychology, 87, 294—302.

LIBERMAN,R. P. (1982) Assessment of social skills. SchizophreniaBulletin, 8, 62—83.

LIDDLE, P. F. (1987) The symptoms of chronic schizophrenia: a

re-examination of the positive-negative dichotomy. BritishJournal of Psychiatry, 151, 145—151.

LINDSAY,W. R. (1982) The effects of labelling: blind and nonblindratings of social skills in schizophrenic and nonschizophreniccontrol subjects. American Journal of Psychiatry, 139,216-219.

LONGABAUGH,R., ELDRED,S. H., BELL, N. W., et al (1966) Theinteractional world of the chronic schizbphrenic patient.Psychiatry, 29, 78-99.

MCFALL, R. M. (1982)A reviewand reformulationoftheconceptof socialskills.BehavioralAssessment,4, 1—33.

MORRISON,R. L., BELLACK,A. S. & MUESER,K. T. (1988) Facialaffect recognition deficits and schizophrenia. SchizophreniaBulletin, 14, 67—83.

—¿� & WIXTED, J. T. (1989) Social skills training. In A ClinicalGuide for the Treatment of Schizophrenia (ed. A. S. Bellack).New York: Plenum.

OVERALL,J. E. & 0o1uw4, D. R. (1962) The Brief PsychiatricRating Scale. Psychological Reports, 10, 799-812.

NoRusu, M. J. (1988) SPSS/PC+ Advanced Statistics V2.O.Chicago: SPSS Inc.

818 BELLACKET AL

ROMANO,J. M. & BELLACK,A. S. (1980) Social validation of acomponentmodelof assertivebehavior.Journal of Consultingand Clinical Psychology, 4*, 478—490.

Rw-rER, D. R. (1977a) Speech patterning in recently admitted andchronic long-stay schizophrenic patients.British Journal of Socialand ClinicalPsychology,16, 45—55.

—¿� (1977b) Visual interaction and speech patterning in admittedand acuteschizophrenicpatients.BritishJournal of SocialandClinical Psychology, 16, 357—362.

—¿�(1978) Visual interaction in schizophrenicpatients: thetimingoflooks.BritichJoumalofSocialandClinicalPrjvhology,17,281-282.

SPITZER,R. L. & ENDICOTT,J. (1978) Schedule for AffectiveDisorders and Schizophrenia (SADS). New York: New YorkState Psychiatric Institute.

—¿� & WILLIAMS, J. B. W. (1985) Structured Clinical Interview

for DSM-III-Psychotic Disorders Version(7/1/85). New York:NewYork State PsychiatricInstitute.

STRAUss, J. S., CARPENTER,W. T., Jr & B*xrico, J. J. (1974) Thediagnosis and understanding of schizophrenia. Part III.

Speculations on the processes that underlie schizophrenicsymptomsand signs.SchizophreniaBulletin,11, 61—69.

THoMAs,P., KiNG,K. & FRASER,W. I. (1987) Positive and negativesymptomsof schizophreniaand linguisticperformance.ActaPsychiatrica Scandinavica,76, 144-151.

TROWER, P., BRYANT, B. & ARGYLE, M. (1978)SocialSkillsand Mental Health. Pittsburgh: University of PittsburghPress.

WALLACE, C. J. (1984) Community and interpersonal functioning

in the course of schizophrenic disorders. Schizophrenia Bulletin,10, 233—257.

WEISSMAN,M. M. & BOTHWELL,S. (1976) Assessment of socialadjustment by patient self-report. Archives of GeneralPsychiatry, 33, 1111—1115.

—¿� & PAYKEL, E. S. (1974) The Depressed Woman: A Study ofSocialRelations.Chicago:Universityof ChicagoPress.

ZIGLER, E. & LEvINE, J. (1981) Premorbid competence in schizophrenia. What is being measured? Journal of Consulting andClinical Psychology, 49, 96—105.

A1an S. Bellack, PhD, Professor of Psychiatry, The Medical College of Pennsylvania, 32(X)Henry Avenue,Philadelphia, PA 19129, USA; Randall L. Morrison, PhD, The Medical College of Pennsylvania; JohnT. Wixted, PhD, The Medical College of Pennsylvania; Kim T. Mueser, PhD, The Medical College ofPennsylvania.

5Correspondence