Asian Journal of Psychiatry 3 (2010) 145–151

Exploratory and confirmatory factor validation of the Dysfunctional Attitude Scalefor Malays (DAS-Malay) in Malaysia

Firdaus Mukhtar a,*, Tian P.S. Oei b

a Department of Psychiatry, Faculty of Medicine, Health Sciences, University Putra Malaysia, 43400 Serdang, Selangor, Malaysiab School of Psychology, University of Queensland, 4120 Brisbane, Australia

A R T I C L E I N F O

Article history:

Received 26 February 2009

Received in revised form 2 June 2010

Accepted 18 July 2010

Keywords:

Factor analysis

Dysfunctional Attitude Scale

Malaysia

A B S T R A C T

The aim of this study was to investigate the factor structure of the Malay version of the Dysfunctional

Attitude Scale (DAS-Malay) in clinical and nonclinical populations. The DAS is a self-report inventory

derived from Beck’s cognitive theory of depression to measure beliefs constituting a predisposition to

depression. The 40-item DAS-Malay was completed by 315 university students, 495 members of the

general community, 167 medical patients, and 113 patients diagnosed with major depressive disorder.

Through principal axis factoring, with varimax rotation, two factors were extracted; performance

evaluation and self-control. Correlation with depression and other variables indicated that the DAS-

Malay held good concurrent validity. In addition, sensitivity and specificity of the total scores of the DAS

were evident in this study. The results showed that the DAS-Malay possesses satisfactory psychometric

properties suggesting that this instrument is appropriate for use as a cognitive measure in a Malay

cultural context although several issues require consideration.

� 2010 Elsevier B.V. All rights reserved.

Contents lists available at ScienceDirect

Asian Journal of Psychiatry

journal homepage: www.e lsev ier .com/ locate /a jp

1. Introduction

In Western populations, the Dysfunctional Attitude Scale (DAS)(Weissman and Beck, 1978) has been one of the leading cognitiveinstruments in clinical research for measuring levels of cognitivevulnerability to depression (Beck et al., 1991; Dozois et al., 2003;Free et al., 1991; Kwon and Oei, 1994; Oei et al., 2006, 1999; Oeiand Sullivan, 1999) for more than 20 years. The DAS is one of thecognitive assessments that has a direct link or direction tosymptoms of depression (Hill et al., 1989). For instance, when aperson encounters a negative life event (e.g. fail in exam) and his/her dysfunctional attitudes include beliefs such as ‘‘If I fail theexam, I am a total failure’’ and ‘‘I must past the exam or else life isworthless’’, the interaction between the negative life event anddysfunctional attitude will lead to negative thoughts about self,the world and the future which eventually result in themanifestation of depressive symptoms. Valid and reliable cogni-tive assessments are essential in order to study the relationshipsamong life events and their link between behaviour and emotion(Hollon and Kendall, 1980). Besides that, cognitive assessment isalso essential to evaluate treatment outcomes of a cognitivebehavioural approach (Beck et al., 1991).

* Corresponding author. Tel.: +60 3 8947 2543; fax: +60 3 8941 4629.

E-mail addresses: drfirdaus@medic.upm.edu.my, fir_cp@yahoo.com

(F. Mukhtar).

1876-2018/$ – see front matter � 2010 Elsevier B.V. All rights reserved.

doi:10.1016/j.ajp.2010.07.007

In the West, the psychometric properties of the DAS have beenwell-established through earlier studies. Internal consistencycoefficients for the DAS have yielded results in the range of .79–.93 for university populations (Dobson and Breiter, 1983; Filipet al., 2005; Weissman and Beck, 1978) and .85 for a random adultpopulation (Oliver and Baumgart, 1985). Specifically, studies thathave utilised these procedures have revealed correlation coeffi-cients of .73 and .97 (Nelson et al., 1992; Oliver and Baumgart,1985). Besides the evidence of reliability, the concurrent validity ofthe DAS has been supported in the form of positive relationshipswith other concurrently administered measures of depressivecognition (Dobson and Breiter, 1983; Dobson and Shaw, 1986;Oliver and Baumgart, 1985). For instance, a positive relationshipwas found between the DAS and the Automatic ThoughtsQuestionnaire (ATQ) (Chioqueta and Stiles, 2004). In addition,depressed individuals scored higher on the DAS relative tonondepressed individuals (Chioqueta and Stiles, 2004; Hautzingeret al., 1985; Hill et al., 1989; Jutta, 2004).

In terms of factor structure of the DAS, Power et al. (1994)developed a short form of the DAS which consisted of threesubscales with 24 items to measure three types of cognitivevulnerability; achievement, dependency, and self-control. Confir-matory Factor Analysis (CFA) confirmed that most of the itemsloaded onto the hypothesised factors. In a study with bothdepressed and nondepressed populations, Carro et al. (1998) alsofound three factors (success/acceptance, perfectionism, and auton-omy) among Spanish populations. However, Parker et al.’s (1984)

F. Mukhtar, T.P.S. Oei / Asian Journal of Psychiatry 3 (2010) 145–151146

study of a group of Australian general practice patients found fourfactors with 25 items, which they termed externalised self-esteem,analytic self-esteem, tentativeness, and need for approval.

Since its establishment, it has been investigated in a number ofdifferent cultures ranging from Chinese (Chen et al., 1998), Dutch(Filip et al., 2005), Norwegian (Chioqueta and Stiles, 2004),Swedish (Ohrt and Thorell, 1998), German (Hautzinger et al.,1985; Jutta, 2004), Spanish (Carro et al., 1998) and Turkish (Sahinand Sahin, 1992), thus exemplifying its worldwide popularity.

In the context of Malaysia, even though Cognitive BehaviourTherapy (CBT) has been practiced in clinical settings, mostclinicians only use measures that assess symptoms of depression(e.g. Beck Depression Inventory) predominantly and it seems thatother important variables such as cognition is overlooked. In ourlocal study, one other measure of cognition, such as ATQ-Malay,has been validated in Malaysia (Oei and Mukhtar, 2008) with goodfactor structure and psychometric properties, which give evidencethat the area of cognition can be explored among Malays inMalaysia.

To date, no study has been designed to assess the psychometricproperties of the Malay version of the DAS (DAS-Malay) inMalaysia. Consequently, valid and reliable cognitive measures areimportant for two reasons; first, to establish the adequacy of thisinstrument specially designed to assess depressive cognitions in aMalay cultural context and second, to further investigate thetheoretical and empirical validity of the cognitive behaviouralapproach in Malaysia.

Therefore, in light of the current psychometric status of the DAS,the objectives of the present study were to (a) examine the factorstructures of the DAS in the Malay population in Malaysia, and (b)provide evidence of the psychometric properties of this scale inMalays so that the DAS could be used with confidence in Malaysia.It is hypothesised that the analyses of the psychometric propertiesof the DAS will yield a good and reliable measure for the Malays,consistent with nonWestern findings.

2. Method

2.1. Participants

A total of 1090 participants were recruited for this study. Thesample consisted of 315 (28.9%) students, 495 (45.4%) members ofthe general community, 167 (15.3%) patients from a primary careunit, and 113 (10.4%) patients diagnosed with major depressivedisorders from a psychiatric clinic. Of the participants, 75.2% of theparticipants were women. The participants’ ages ranged from 18 to63 years, with a mean of 26 years. The educational backgrounds ofthe participants included high school certificate level (47.6%),diploma/certificate level (17.1%), and university degree (32.5%); 1%of participants had only completed primary school and 1.8% did notspecify their level of education. All participants answered thequestionnaires themselves with minimal guidance from researchassistants because the questionnaire was in their own language(i.e. Bahasa Melayu).

2.2. Measures

The DAS is a 40-item measure of the presence of dysfunctionalattitudes to life, which predispose an individual to depression(Clark et al., 1999). The items such as ‘‘I am nothing if a person Ilove does not love me; If a person asks for help, it is a sign ofweakness’’ with responses attached to a 7-point Likert scaleranging from ‘‘totally agree’’ to ‘‘totally disagree’’. Total scores areobtained by summing across the 40 items, yielding scores thatrange from a minimum of 40 to a maximum of 280. The scale hasacceptable internal consistency (Cronbach’s alpha = .86) and

concurrent validity and was able to discriminate betweendepressed and nondepressed samples (Hill et al., 1989). Higherscores indicate a greater propensity for belief in purportedlydepressotypic attitudes.

Beck Depression Inventory (BDI)-Malay (Oei and Mukhtar,2008) is a validated version of the original BDI (Beck et al., 1961)with 20 items that provide an indication of the level of depressedmood. Participants respond to questions in relation to howthey felt over the past week. Higher total scores indicate moresevere depressive symptoms. The BDI-Malay does not constitutea clinical diagnosis, but has been widely used as a tool in theassessment process, and for discerning changes in mood duringtreatment. This measure has been reported to possess acceptablepsychometric properties with internal consistency (Cronbach’salpha) ranging from .71 to .91 as well as acceptable validityindices (Oei and Mukhtar, 2008).

ATQ-Malay (Oei and Mukhtar, 2008) is a 17-item questionnairemeasuring the frequency of negative automatic thoughts. Respon-dents rate the frequency of the 17 negative thoughts on a 1–5 scale.Specifically, participants were asked how frequent negativeautomatic thoughts such as ‘‘I’m a loser’’ have occurred in thepast week. Higher scores indicate increased severity of negativethoughts. Internal consistency of this measure ranges from .83 to.93 (Oei and Mukhtar, 2008).

The Zung Self-Rating Depression Scale-Malay (Zung SDS-Malay) isa translated version of the original Zung SDS (Zung, 1965), whichwas designed for assessing depression in patients whose primarydiagnosis was of a depressive disorder. The 20 items address eachof the four most commonly found characteristics of depression: itspervasive effect, its physiological equivalents, other disturbances,and psychomotor effects, such as ‘‘I have trouble sleeping at night’’and ‘‘I have trouble with constipation’’. Ten items are wordedpositively and 10 items are worded negatively. Each item is scoredon a scale of 1–4 (‘‘a little of the time’’ to ‘‘most of the time’’) withreverse scaling for the negatively worded items. Range of totalscore is from 20 to 80, within which most people with depressionscore between 50 and 69, while a score of 70 and above indicatessevere depression (Zung, 1965).

The WHO Quality of Life-BREF (WHOQOL-BREF) version inBahasa Malaysia (WHOQOL-BREF Malay) (Hasanah et al., 2003),consisting of 26 items, has been validated in Malaysia, withindications of good discriminant validity, construct validity,internal consistency (.64–.80) and test-retest reliability (.49–.88). The scale is a valid and reliable assessment of quality of life,especially for those with illness. Four domains that can beextracted from WHOQOL-BREF are physical and psychological

health, social, and environment, which assesses general quality oflife.

The Beck Hopelessness Scale-Malay (BHS-Malay) is a translatedversion of the original BHS (Beck and Steer, 1988) with a 20-itemscale for measuring negative attitudes about the future. Thescale’s manual claims internal consistency ranging from .82to .93 and a test-retest reliability of .69 (Beck and Steer,1988); examples of questions asked are ‘‘My future seems darkto me’’ and ‘‘I can’t imagine what my life would be like in tenyears’’.

2.3. Procedure

Student sample. A total of 315 undergraduates from twouniversities (studying medicine, dentistry, allied health sciences,and human sciences) participated voluntarily in this study topartially satisfy a research requirement of their course. All datawere collected through group administrators. Each subject wasprovided with questionnaires as described above, with anexplanation and accompanying directions for their use. A number

F. Mukhtar, T.P.S. Oei / Asian Journal of Psychiatry 3 (2010) 145–151 147

of nonethnically Malay students participated but their data werenot included in the analysis to ensure that the conditions of thisstudy were met.

General community sample. 495 out of 1500 members of thegeneral public participated in this study by completing ques-tionnaires with signed informed consent and returning these inenvelopes supplied. The envelopes were randomly distributed inpublic places by research assistants.

Medical patients sample. The Malay medical patients wererecruited from primary care clinics; an obesity clinic; ear, nose andthroat clinics; and community care clinics. They completedquestionnaires distributed by research assistants and returnedthem in envelopes supplied. Participants were excluded if theywere current drug or alcohol abusers, had a history of organicallybased cognitive dysfunction, demonstrated reading difficulties,were not fluent in Bahasa Malaysia, or were not ethnic Malays.

Patients with major depressive disorders. Malay patients withdepression were invited via mail, phone, or through referral frompsychiatrists who had been informed of the study. A letter ofinvitation and information regarding the study was provided andthose participants who were willing to participate presented at thepsychiatric clinic for the intake procedure assessment. The secondauthor, who works as a clinical psychologist, further evaluated thediagnosis of major depressive disorder using a structured clinicalinterview from the Diagnosis and Statistical Manual of MentalDisorder-4th Edition (DSM-IV) to ascertain participants’ eligibility.Participants were included if they were diagnosed as sufferingfrom major depression or dysthymia as defined by the DSM-IV.Patients were excluded if their depression was secondary toanother major psychiatric disorder (e.g. schizophrenia), if theywere currently abusing drugs or alcohol, had a history of anorganically based cognitive dysfunction, demonstrated readingdifficulties, or were not fluent in Bahasa Malaysia.

Translating and back-translating procedure. Four bilingualpsychologists with a Master’s degree or higher translated theMalay version of all the instruments (except WHOQOL-BREF) usingback-translating procedures. A professional language interpreterwas recruited to proofread the translated questionnaires to ensuretheir overall suitability and to resolve word ambiguity issues aftertranslation. The back-translated versions were similar to theoriginal versions and to each other. Minor differences concerninguse of colloquial expressions in both languages were reconciled.

Ethical Approval. All participants gave their signed informedconsent before undertaking the assessment. Ethical approval wassought from the research ethics committees belonging to theMinistry of Health of Malaysia and all the hospitals and institutionsparticipating in this study.

2.4. Statistical analyses

The Statistical Program for the Social Sciences (SPSS version14.0) and AMOS version 6.0 were used to analyse data in this study.Cronbach’s alpha coefficients (a) were computed to evaluate thereliability of the questionnaire, and correlations were calculated toexamine the concurrent validity of the BDI, using the total sample.

The CFA model fit was evaluated using multiple fit indices assuggested by Kline (1998). The selected indices were the chi-square statistics (x2), the root mean square residuals (RMR;Bentler, 1990), the comparative fit index (CFI) (Bentler, 1990), thegoodness of fit index (GFI), and the root mean square error ofapproximation (RMSEA; Browne and Cudeck, 1993). A good modelfit is indicated by values of .90 or higher for the CFI and GFI. RMRvalues less than .05 reflect a close fit while values of .1 or lowerindicate a reasonable fit for the RMR (Bentler, 1990). For theRMSEA, values of .05 or lower indicate a close fit while values lessthan .08 indicate an acceptable fit (Browne and Cudeck, 1993).

3. Results

3.1. Exploratory Factor Analysis (EFA)

It was decided to divide the total sample (N = 1090) into twogroups by using the odd-even split method (Group A [n = 545];Group B [n = 545]). Group A was used for EFA to establish the factorstructure and Group B was used for CFA to confirm the DAS-Mfactor structures found in Group A. Following Tabachnick andFidell’s (2001) suggestion, a principal axis factoring with varimaxrotation was employed in an initial exploratory analysis. Uponexamination of the correlation matrices, a substantial number ofcorrelations lower than .33 were found suggesting favourability ofthe data set (Tabachnick and Fidell, 2001). Favourable values of theKaiser–Meyer–Olkin value (.89) and a significant value (p < .001)of Bartlett’s Test of Sphericity also suggested that the data weresuitable for factor analysis.

In addition, examination of the multiple correlation squared(R2) indicated that the multicollinearity and singularity were not athreat in this data set. Varimax rotation was utilised to maximisethe dispersion of the loadings within factors so that loading asmaller number of variables loaded highly on each factor, thusresulting in more interpretable factor clusters (Field, 2000).

Hence, principal axis factoring with varimax rotation wasconducted. A number of criteria were used to determine the mostappropriate number of factors to retain: (a) minimum eigen-valuesof 1, (b) minimum factor loadings of .40, (c) minimal factorialcomplexity (multiple loading), and (d) meaningful interpretationof factors. The results indicated that two factors rotated to a simplestructure using the Direct Oblimin method with Kaiser normal-isation.

Using the previous criteria, two factors were extracted. Itemdescriptions, factor loadings, communality estimates, and factorintercorrelations are presented in Table 1. These two factorsaccounted for 35.34% of variance in scores. Factor 1, which waslabelled performance evaluation, accounted for 26.52% of thevariance while Factor 2 accounted for 8.81% of the variance. Factor2 was subsequently labelled self-control.

3.2. Confirmatory Factor Analysis

Preliminary CFA. CFA was subsequently utilised to examine theconstruct validity of the two-correlated factor models extractedfrom EFA analysis. Initially, 20 predictors (i.e., items) were used totest the model (Table 2). However, most of the fit indices suggesteda moderate fit (x2 = 517.1, df = 169, p < .001; CFI = .89; GFI = .91;RMR = .16; RMSEA = .06). Given these results, it was deemed thatthe model required minor respecification. Inspection of R2

suggested that the number of specified latent factors was thepotential source of the problem. Specifically, a substantial numberof R2 values less than .5 were found which cast doubt on the extentto which these items contributed to the factor they were designedto measure (Kline, 1998). Strong intercorrelations among thelatent factors also suggested that the number of latent factors wasalso problematic. Further inspection of the standardised residualscovariance revealed several pairs of items with a large residual(>2.38). Deletion of item 11 resulted in an improved model fit.

Final analysis of CFA. In a subsequent analysis, based on theresults of modified indices, 19 items were included in a final CFA totest the fit of the model (Table 2). Using Maximum Likelihoodprocedures to estimate the model, most of the indices indicated agood fit (x2 = 441.1, df = 151, p < .001; CFI = .90; GFI = .90;RMR = .15; RMSEA = .05). Inspection of nonstandardised regres-sion weights indicated significant loadings for all items. Latentfactor intercorrelation of .44 implied an acceptable level ofdiscriminant validity.

Table 1Component matrix of Exploratory Factor Analysis and Cronbach’s Alpha of the DAS-Malay.

Item Title Factor 1 Factor 2 h2

10 If I fail at my work, then I am a failure as a person. (Jika saya gagal dalam pekerjaan, bermakna saya gagal sebagai

seorang manusia)

.73 .56

9 If I do not do as well as other people, it means I am an inferior human being. (Kalau saya tak buat sebagus orang lain,

bermakna saya manusia yang lemah)

.68 .47

8 If a person asks for help, it is a sign of weakness. (Jika seseorang itu meminta pertolongan, itu tanda orang lemah) .67 .50

16 I am nothing if a person I love does not love me. (Saya memang tak berguna jika orang yang saya sayang tidak

cintakan saya)

.61 .38

14 If I fail partly, it is as bad as being a complete failure. (Kalau saya gagal tarafnya adalah seteruk orang yang gagal) .61 .41

26 If I ask a question, it makes me look inferior. (Kalau saya tanya soalan, ia akan menunjukkan kelemahan diri saya) .57 .39

1 It is difficult to be happy unless one is good looking, intelligent, rich and creative. (Susah untuk rasa gembira

melainkan seseorang itu cantik, bijak, kaya dan kreatif)

.53 .28

19 My value as a person depends greatly on what others think of me. (Nilai saya sebagai seorang manusia bergantung

pada apa yang orang lain fikir mengenai saya)

.52 .28

5 Taking even a small risk is foolish because the loss is likely to be a disaster. (Mengambil risiko sekecil manapun

adalah bodoh kerana kerugiannya aan menjadi lebih parah)

.51 .28

7 I cannot be happy unless most people I know admire me. (Saya tak boleh gembira kecuali mereka yang saya kenal

mengagumi saya)

.48 .25

31 I cannot trust other people because they might be cruel to me. (Saya tak percaya pada sesiapa pun kerana mereka

mungkin akan berlaku kejam pada saya)

.47 .18

13 If someone disagrees with me, it probably indicates he does not like me. (Kalau seseorang tak setuju dengan saya,

kemungkinan itu tanda dia tak suka saya)

.42 .18

2 Happiness is more a matter of my attitude towards myself than the way other people feel about me.

(Kegembiraan lebih bergantung pada sikap diri saya berbanding apa yang orang lain rasa terhadap saya)

.71 .52

12 Making mistakes is fine because I can learn from them. (Membuat kesilapan tak mengapa kerana saya boleh

belajar daripadanya)

.68 .54

17 One can get pleasure from an activity regardless of the end result. (Seseorang akan mendapat keseronokan

daripada suatu aktivti walau apapun kesudahannya)

.61 .37

35 I do not need the approval of other people in order to be happy. (Saya tak perlukan kebenaran daripada orang

lain untuk rasa gembira)

.52 .29

24 My own opinions of myself are more important than other’s opinion of me. (Pendapat saya mengenai diri saya

lebih penting berbanding dengan pandangan orang lain terhadap diri saya)

.51 .28

37 I can be happy even if I miss out on many of the good things in life. (Saya masih boleh bergembira walaupun

terlepas banyak perkara baik dalam hidup ini)

.50 .26

30 It is possible for a person to be scolded and not get upset. (Boleh jadi seseorang itu apabila dimarahi dia tidak akan marah) .48 .25

Eigen-values 5.92 2.39

% All 19 items (35.34%) 26.52 8.82

A All 19 items (.86) .86 .79

Factor 1, Performance Evaluation; Factor 2, Self-control; % of variance, Percentage of variance; a, Cronbach’s alpha coefficient. All items in Factor 2 are reversed scored.

Table 2Results of the comparison of different factorial models for DAS-Malay.

Model No of items/factors x2 df x2/df ratio CFI GFI RMR RMSEA

Parker et al. (1984) 25 (4 factor) 1178.36 274 4.30 .66 .83 .22 .08

Power et al. (1994) 26 (3 factor) 1277.71 298 4.29 .64 .82 .21 .08

Cane et al. (1986) 25 (2 factor) 1267.67 274 4.63 .69 .82 .21 .08

Group A 20 (2 factor) 517.1 169 3.06 .89 .91 .16 .06

Group A 19 (2 factor) 441.1 151 2.92 .90 .92 .15 .05

Group B 19 (2 factor) 451.2 151 2.98 .90 .90 .16 .05

x2, chi-square statistic; df, degree of freedom; CFI, comparative fit index; GFI, goodness of fit index; RMR, root mean square residual; RMSEA, root mean square error of

approximation.

F. Mukhtar, T.P.S. Oei / Asian Journal of Psychiatry 3 (2010) 145–151148

Group B. The two-factor model was further evaluated using anindependent (i.e., Group B) validation sample (N = 545). Results ofthe CFA for Group B are also displayed in Table 2. Using MaximumLikelihood estimation procedures, good fit indices were obtained(x2 = 483.8, df = 118, p < .001; CFI = .92; GFI = .90; RMR = .04;RMSEA = .07), suggesting the stability of its factor structure. Thesefindings confirmed the CFA results from Group A and suggest thatthe factor structure for the DAS-Malay is valid and stable.

CFA testing for Parker et al.’s (1984) model. The first modelexamined was Parker et al.’s (1984) four-factor model. Results ofthe fit indices were obtained (x2 = 1178.36, df = 274, p < .001;CFI = .66; GFI = .83; RMR = .22; RMSEA = .08) and the modelprovided a poor fit to the data. The analyses showed that Parkeret al.’s (1984) four-orthogonal factors model with 25 items was nota good fit to the data set.

CFA testing for Power et al.’s (1994) model. The second modeltested was Power et al.’s (1994) three-orthogonal factors model.Results of the fit indices were obtained (x2 = 1277.71, df = 298,

p < .001; CFI = .64; GFI = .82; RMR = .21; RMSEA = .08) and themodel provided a poor fit to the data. The analyses showed thatPower et al.’s (1994) three-factor model with 26 items was not agood fit to the data.

CFA testing for Cane et al.’s (1986) model. The third model testedwas Cane et al.’s (1986) two-orthogonal factors model. Results ofthe fit indices were obtained (x2 = 1267.67, df = 274, p < .001;CFI = .69; GFI = .82; RMR = .21; RMSEA = .08) and indicated thatthis model also provided a poor fit to the data. The analyses showedthat Cane et al.’s (1986) two-factor model with 25 items was not agood fit to the data.

3.3. Reliability and validity of the DAS for the whole sample

(N = 1090)

3.3.1. Internal consistency

Using the whole sample, internal consistency and validityanalysis were conducted. The reliability of the two latent factors

Table 6Discriminant analyses of DAS-Malay subscale (performance evaluation).

Group Nonclinical Clinical Total

Nonclinical 959 (98.2%) 18 (1.8%) 977 (100%)

Clinical 64 (56.6%) 49 (43.4%) 113 (100%)

92.5% of original grouped cases correctly classified.

Table 3Intercorrelations for DAS-M total scores and subscale with BDI-Malay, ATQ-Malay, Zung SDS, BHS and WHOQOL-BREF.

Variables BDI-Malay ATQ-Malay Zung SDS BHS WHOQOL-BREF

DAS-Malay total .64** .53** .68** .51** �.68**

Performance evaluation .52 .49 .53 .33 �.53

Self-control .55 .36 .61 .56 �.62

** Correlation is significant at the .01 level (1-tailed).

Table 7Discriminant analyses of DAS-Malay subscale (self-control).

Group Nonclinical Clinical Total

Nonclinical 966 (98.9%) 11 (1.1%) 977 (100%)

Clinical 36 (31.9%) 77 (68.1%) 113 (100%)

95.7% of original grouped cases correctly classified.

F. Mukhtar, T.P.S. Oei / Asian Journal of Psychiatry 3 (2010) 145–151 149

was evaluated. Table 1 shows the internal consistency values of theDAS-Malay. Using Cronbach’s alpha to estimate the reliabilitycoefficient, a moderate to high alpha was obtained for the overallscale (.86) and the subscales (Factor 1 = .86; Factor 2 = .79).

3.3.2. Concurrent validity

Concurrent validity was evaluated using the Pearson correla-tion coefficient. Descriptions of the scales and their intercorrela-tions are shown in Table 3. Evidence of concurrent validityrevealed a significantly marked degree of correlation of the DAS-Malays total scores with the BDI-Malay (r = .64), the Zung SDS(.68), and the QOL (�.68), and moderate correlation with the ATQ-Malay (r = .53) and the BHS (r = .51). Furthermore, the results alsorevealed a significantly moderate degree of correlation betweenperformance evaluation/dependency and the BDI-Malay (r = .52),the Zung SDS (r = 53), the WHOQOL-BREF (�.53), and the ATQ-Malay (.49) but low correlation with the BHS (r = .33). The finalsubscale of the DAS-Malay, self-control, showed a significantlymarked degree of correlation with the Zung SDS (.61) and theWHOQOL-BREF (�.62), a moderate relationship with the BDI-Malay (.55) and the BHS (.56), and low correlation with the ATQ-Malay (.36). This illustrates that the DAS-Malay has moderateconcurrent validity.

3.3.3. Discriminant validity

A direct discriminant analysis using all 19 items of the DAS-Malay was performed (see Table 4). Wilks’s Lambdas, evaluatedusing the F-test approximation, were significant at .001 for allitems. The clinical group demonstrated significantly higher meanscores on the DAS-Malay than subjects in the three nonclinicalgroups (Table 5). Classification subanalysis used the linearcombination of all 19 items. In terms of sensitivity, the analysisindicated that the high total scores of the DAS-Malay detected that51.3% of the patients had depression. Meanwhile, the specificityanalysis found that .9% of the nonclinical subjects had high DAS-Malay scores. Additionally, 99.1% of all subjects in the nonclinicalgroup (N = 977) and 48.7% of clinical group had low DAS-Malayscores. The overall percentage of correctly classified cases was94.1%.

Table 4Mean and standard deviation of DAS-Malay total and subscales scores for clinical

and nonclinical group (students, general community and medical patients).

Group Total DAS-M

mean (SD)

Factor 1

mean (SD)

Factor 2

mean (SD)

1. Clinical group 95.5 (20.8) 56.15 (18.4) 39.35 (11.7)

2. Nonclinical group 57.3 36.4 20.9

p-value sig. at .001*** Factor 1, performance evaluation; Factor 2, self-control.

Table 5Discriminant analyses of DAS-Malay total score.

Group Nonclinical Clinical Total

Nonclinical 968 (99.1%) 9 (.9%) 977 (100%)

Clinical 55 (48.7%) 58 (51.3%) 113 (100%)

94.1% of original grouped cases correctly classified.

Table 6 reports the performance evaluation and dependencyfactor. The analysis indicated that this subscale was able to detect43.4% of the clinical subjects and 1.8% of nonclinical subjects whoreported high scores in factor one of the DAS-Malay. As expected,98.2% of the nonclinical group and 56.6% of the clinical groupreported low scores of this factor. The overall percentage ofcorrectly classified cases was 92.5%. Table 7, shows that 68.1% ofthe clinical subjects and 1.1% of the nonclinical subjects had highscores in the self-control subscale. The overall percentage ofcorrectly classified cases was 95.7%. All items in self-control havereversed scoring. This result suggests that the DAS-Malay items areable to discriminate between clinical and nonclinical subjects.

4. Discussion

The purpose of the present research was to assess the factorstructure, reliability, and validity of the DAS-Malay among Malaysin Malaysia. This paper showed that the EFA identified twosubscales, namely performance evaluation and self-control. Al-though this result was not clearly consistent with previous studies,some items loaded on similar factors. Specifically, all items (items10, 9, 8, 14, 1, and 13) loading in Factor 1 (performance evaluation)were consistent with Western study (Cane et al., 1986) and Easternstudy (Sahin and Sahin, 1992) that specifically measure how aperson gives a low remarks on their own performance based oncomparison to other people’s standard. Some items that related inperformance evaluation factor are ‘‘If I do not do as well as otherpeople, it means I am an inferior human being’’ and ‘‘If I fail at mywork, then I am a failure as a person’’. In this factor, a person whoscores more in these items is likely to suffer from depression.Meanwhile, items for Factor 2, which is self-control, wereconsistent with Power et al.’s (1994) study where it was calledachievement instead. Three items (item 17, 24, and 30) that relatedto this theme were ‘‘One can get pleasure from an activityregardless of the end result’’, ‘‘My own opinions of myself are moreimportant than other’s opinion of me’’, and ‘‘It is possible for aperson to be scolded and not get upset’’ where a lesser score in thisitems means that the more vulnerable that person’s feelings anddysfunctional attitude which could possibly lead to depression.

These two factors shared almost 35.34% of the variance with the19-item subscale. In the CFA, none of the proposed factorial models

F. Mukhtar, T.P.S. Oei / Asian Journal of Psychiatry 3 (2010) 145–151150

from previous studies indicated a good fit to the data set. It isinteresting to note that the final CFA results from Group A showedthat the 19-item two-factor data provided an acceptable fitcompared to other models tested. Cross-validation of Group B inthe CFA also showed that the model demonstrated a good fit to thedata.

The DAS-Malay proved to have sufficient internal consistencyand reliability, although the Cronbach’s alpha value is lower thanfound in previous studies (Chen et al., 1998; Chioqueta and Stiles,2004; Filip et al., 2005; Ohrt and Thorell, 1998; Oliver andBaumgart, 1985; Weissman and Beck, 1978). Two subscales of theDAS-Malay, performance evaluation and self-control, yieldedsatisfactory reliability coefficients. The results of the presentstudy further verified the concurrent validity of the DAS (Filip et al.,2005; Nelson et al., 1992; Ohrt and Thorell, 1998; Weissman andBeck, 1978). Chioqueta and Stiles (2004) and Chen et al. (1998) alsoreported a positive relationship between the DAS and the BDI inthe Norwegian and Chinese population. Our findings in the currentstudy also showed that the total DAS-Malay scores have goodsensitivity and specificity in discriminating clinical and nonclinicalpatients (Chioqueta and Stiles, 2004; Hautzinger et al., 1985; Jutta,2004;).

This is one of the first studies to validate the DAS-Malay for usewith large and different populations and to subject two subscalesto factor analysis. It presents clear evidence that the DAS-Malay issufficiently reliable and a valid measure of depression symptoms.The major strengths of the present study include the use of a largesample (N = 1090 for both Group A and B), the use of EFA and CFAmethodology and the direct application of a theoretically derivedmeasure to a clinical setting and a specific sample. Furthermore,cross-validation of different samples using CFA strengthens thefindings of this study. The limitation of this study was that theparticipants were mainly female and further investigation forgender differences is warranted.

In terms of application of cognitive behavioural approach inMalaysia, findings of this study have its uniqueness. This may be aninfluence from Malay culture where all of them are Muslim and inIslamic teaching states that followers should be responsible fortheir action as that will be asked for justification by Allah in thehereafter. Therefore, the new version of the DAS is more relevantand applicable to measure the dysfunctional assumptions that leadto depression with less repetition and redundancy of the items. Inaddition, assessment of dysfunctional assumptions of self, othersand the world could give more avenues for a person to explore andvalidate their perception of depression that is occupied by culturalbeliefs such as evil spirits or black magic. On top of this, this toolcould assist clinician and researcher to offer more appropriatetreatment and service to patients with mood disorders.

In conclusion, the findings show that the DAS-Malay ismarginally acceptable and has satisfactory psychometric proper-ties for both CFA and reliability results. Hence, further validation isworthy of attention in future research in order to measurecognitive vulnerabilities of depression among Malays in Malaysiain a more precise and culturally sensitive way.

Role of funding source

This study is part of doctorate project at the University ofQueensland. Funding for this study was provided by theUniversity; no other funding sources were involved.

Contributors

Second author designed the study and provides professionaladvice and guidance on statistical analysis. First author managedthe literature searches, data collection, prepare analyses and wrote

the first draft of manuscript. All authors contributed to and haveapproved the final manuscript.

Conflict of interest

There are no conflicts of interest in this study as far as theauthors are concerned.

Acknowledgements

Special thanks to all participants in this study and Dr Asad Khanfrom University of Queensland, Australia for supportive advice onstatistic.

References

Beck, A.T., Steer, R.A., 1988. Beck Hopelessness Scale. Psychological Corp, SanAntiano: TX.

Beck, A.T., Ward, C.H., Mendelson, M., Mock, J., Erbaugh, J., 1961. An inventory formeasuring depression. Arch. Gen. Psychiat. 4, 53–63.

Beck, A.T., Brown, G., Steer, R.A., Weissman, A.N., 1991. Factor analysis of theDysfunctional Attitude Scale in clinical population. J. Consult. Clin. Psychol.3, 478–483.

Bentler, P.M., 1990. Comparative fit indices in structural models. Psychol. Bull. 107,238–246.

Browne, M.W., Cudeck, R., 1993. Alternative ways of assessing model fit. Sociol.Methods Res. 21, 230–258.

Cane, D.B., Olinger, J., Gotlib, I.H., Kuiper, N.A., 1986. Factor structure of the Dysfunc-tional Attitude Scale in a student population. J. Clin. Psychol. 42, 307–309.

Carro, I.L., Bernal, I.L., Vea, H.B., 1998. Depression in Cuba: validation of BeckDepression Inventory and the Dysfunctional Attitudes Scale with Cuban popu-lation. Av. Psicol. Clin. Latinoam. 16, 111–120.

Chen, Y., Xu, J., Yan, S., Xian, Y., Li, Y., Chang, X., Llang, G.T., Ma, Z., 1998. A preliminarystudy of the Dysfunctional Attitude Scale. Chin. Ment. Health J. 12, 265–267.

Chioqueta, A.P., Stiles, T.C., 2004. Psychometric properties of the Norwegian Versionof the Dysfunctional Attitude Scale (Form A). Cogn. Behav. Ther. 33, 83–86.

Clark, D.A., Beck, A.T., Alford, B.A., 1999. Scientific Foundations of Cognitive Theoryand Therapy of Depression. John Wiley, New York.

Dobson, K.S., Breiter, H.J., 1983. Cognitive assessment of depression: reliability andvalidity of three measures. J. Abnorm. Psychol. 92, 107–109.

Dobson, K.S., Shaw, B.F., 1986. Cognitive assessment with major depressive dis-orders. Cogn. Ther. Res. 10, 13–24.

Dozois, D.J.A., Covin, R., Brinker, J.K., 2003. Normative data on cognitive measures ofdepression. J. Consult. Clin. Psychol. 71, 71–80.

Field, A., 2000. Discovering Statistics using SPSS for Windows. Thousand Oaks,London.

Filip, R., Dirk, H., Kris, V.B., Paul, E., 2005. The Dutch version of the DysfunctionalAttitude Scale-Form A. Gedragstherapie 38, 285–293.

Free, M.K., Oei, T.P.S., Sanders, M.R., 1991. Treatment outcome of a group cognitivetherapy program for depression. Int. J. Group Psychother. 41 (4), 533–547.

Hasanah, C.I., Naing, L., Rahman, A.R.A., 2003. World Health Organization quality oflife assessment: brief version in Bahasa Malaysia. Med. J. Malaysia 58, 79–88.

Hautzinger, M., Luka, U., Trautmann, R.D., 1985. Dysfunctional Attitude Scale: aGerman version of the Dysfunctional Attitude Scale. Diagnostica 31, 312–323.

Hill, C.V., Oei, T.P.S., Hill, M.A., 1989. An empirical investigation of the specificity andsensitivity of the Automatic Thoughts Questionnaire and Dysfunctional Atti-tudes Scale. J. Psychopathol. Behav. Assess. 11, 291–311.

Hollon, S.D., Kendall, P.C., 1980. Cognitive self-statements in depression: develop-ment of an automatic thoughts questionnaire. Cogn. Ther. Res. 4, 383–395.

Jutta, J., 2004. The factor structure of the Dysfunctional Attitude Scale (DAS) in anon-clinical sample. Diagnostica 50, 115–123.

Kline, R.B., 1998. Principles and Practice of Structural Equation Modeling. TheGuildford Press, New York.

Kwon, S.-M., Oei, T.P.S., 1994. Roles of two levels of cognition in the development,maintenance and treatment of depression. Clin. Psychol. Rev. 143, 331–358.

Nelson, L.D., Stern, S.L., Cicchetti, D.V., 1992. The Dysfunctional Attitude Scale: howwell can it measure depressive thinking? J. Psychopathol. Behav. Assess. 14,217–223.

Oei, T.P.S., Bullbeck, K., Campbell, J.M., 2006. Cognitive change process duringcognitive behaviour therapy for depression. J. Affect. Disord. 92, 231–241.

Oei, T.P.S., Llamas, M., Devilly, G.J., 1999. The efficacy and cognitive processes ofcognitive behaviour therapy of panic disorder with agoraphobia. Behav. Cogn.Psychother. 27, 63–88.

Oei, T.P.S., Mukhtar, F., 2008. Exploratory and confirmatory factor analysis andpsychometric properties of Automatic Thoughts Questionnaire-Malay. HongKong J. Psyc. 18, 3, 92–100.

Oei, T.P.S., Sullivan, L.M., 1999. Cognitive changes following recovery from depres-sion in a group cognitive behaviour therapy program. Aust. New Zeal. J.Psychiatr. 33, 407–415.

Ohrt, T., Thorell, L.-H., 1998. Dysfunctional Attitude Scale (DAS) Psychometrics andnorms of the Swedish Version. Scand. J. Psychol. 27, 105–113.

F. Mukhtar, T.P.S. Oei / Asian Journal of Psychiatry 3 (2010) 145–151 151

Oliver, J.M., Baumgart, E.P., 1985. The Dysfunctional Attitude Scale: psychometricproperties and relation to depression in an unselected adult population. Cogn.Ther. Res. 9, 161–167.

Parker, G., Bradshaw, G., Blignault, I., 1984. Dysfunctional attitudes: measurement,significant constructs and links with depression. Acta Psychiatr. Scand. 70,90–96.

Power, M.J., Katz, R., McGuffin, P., Duggan, C.F., Lam, D., Beck, A.T., 1994. TheDysfunctional Attitude Scale (DAS): a comparison of forms A and B andproposals for a new subscaled version. J. Res. Perspect. 28, 263–276.

Sahin, N.H., Sahin, N., 1992. How dysfunctional are the dysfunctional attitudes inanother culture? Br. J. Med. Psychol. 65, 17–26.

Tabachnick, B.G., Fidell, L.S., 2001. Using Multivariate Statistics. Allyn and Bacon,Boston.

Weissman, A.N., Beck, A.T., 1978. Development and validation of the DysfunctionalAttitude Scale: A preliminary investigation. In: Paper Presented at the Paperpresented at the Meeting of the Association for the Advancement of BehaviorTherapy, Chicago.

Zung, W.W.K., 1965. A self-rating depression scale. Arch. Gen. Psychiat. 12, 63–70.