Pijc/~olog!ci;lR~~pcii.t.s. 2008,103.921-930. C I's!chological Reports 2008

DISCRIhlINATORY ATTITUDES:

A SCALE DEVELOPAIENT I N TURKISH '

S2irizmaty.-This study presented the de\-elopmmt of a reliable and \ d i d scale, s for use in research related to discrimiriatioii in a the Discriminatory A t t i t ~ ~ d eScale. Turkish context. First, related literature \\as searched and a pool o i 69 items \\-as ~ e n e r a t e dto cover the basic forms of discrimin,~rion.Then. data were collected from 8-10 college students and sublected to esplorarory fdcror analjsis aitl? varimau rotation. Iea\i~ig21 irenis. Y e s t . a confirmator! factor analysis \\as perforn~ed.which confirmed a ?-factor structure. These facrors v e r e named Discrimination Against H o m o sexuals. Discrimination Against Women. Discrimindtion Based on Race or Ethnicit! and Rslipion, and Discrimin~rionAgainst Foreigners. Correlational anal!ses \\-it11 ratings from the Left-hght Orientation Scale. Cronbach coefticient alpha, and cyperr opinion p r o ~ i d e de\iiience on 1-alidity and inrern;ll consistency and reliability of the Discriminatory ;Ittitudes Scale.

Discrimination on the basis of racial or ethnic group. sex, sexual orien, sociolotation, and religion h a ~ eall been studied in p ~ y c h o l o g ~education, gy, anthropology, and political science. Empirical studies of variables related to different kinds of discrimination, holvever, require having valid and reliable measures (Utsey & Ponterotto, 1996; Glick & Fiske, 1996: Rose11 & Hartman, 2001; Bodkin-Andreu.~,Craven, & hlarsh, 2001). An examination of the literature in Turkey indicates that a $el{. measures of specific kinds of discrimination have been developed or adapted for the Turkish context. Gumus and Go~llleksiz(19991 developed a scale assessing discriminatory attitudes towards Armenians. Jews, and foreigners in Turkey. Kuzgun and Sevim (2004) developed a survey of attitudes torards a o m en who had jobs. Finally, Sakalll-Ugurlu (2002) adapted Glick and Fiske's Xmbi\lalent Sexisnl Scale (1996) into Turkish. These separate scales cover several common attitudes but exclude attitudes towards different racial and group, women in general, homosexual individuals, ethnic groups as a and individuals from different religious backgrounds. LIoreover, in the adaptation of a questionnaire. a point that needs to be --

'Address correspondence t o Kukeri , i k u I'ural. Ph.D.. rissistant Professor of Lciuc,>tion. Aldnan LIenderes Uni\-ers~r>-. Egirinl F k . . E g i t ~ mHilimleri Bolumu, Aydin-Turk?! or e-mail ~rakarvural@ adu.edu.tri.

considered is that many discriminatory attitudes are culturally determined. \YThen a questionnaire, test, or scale is adapted into a different language, it is esse~ltialthat the culture in \\,hich it \\.as originally developed and the culture for which it is being adapted are similar (Greenfield. 1997: AIerenda. 20061. Considering that most di\rersity-related measures are developed in a western context and that the Turkish culture is nonwestem, development of a neur measure \vhich is rnore culturally applicable in a Turkish context is more appropriate. In light of the above-stated cultural differences and the lack of measures in Turkey. there is a need to develop a cult~irallyfocused and more comprehensive survey related to discrirnination. The main purpose of this study mas to de1,elop a single reliable and valid measure of a \vide range of discriininatory attitudes in a Turkish contest, includi~lgcotumon attitudes, such as those to~vardswomen, gays and lesbians, foreigners, racial and ethnic groups, and individuals from different religious backgrounds. Having a single general measure al1ou.s broader and refined assessment o i the entire range of discrimination. The specific items also allow cross-cultural comparison of specific aspects of' discrimination.

Defiizitl'on of Discri~?zllrliztlol? The term iiiscri~ni~~i~tiou is defined in the Osfoi~~i Atlcailccd Lenrize~s' Dktioizaq, (Cron-ther. 1995) as "iayainst/in fa~.orof somrbody) treating a person or group differently than others" ip. 3311. According to the Atiteric'ziz Heritilgt. DICtlO~zi~i:j. (199-11, discrimination means "to make a clear distinction: differentiate: to nlake distinctions on the basis of preference or prejudice" ( p . 4011. Sandhu and Asp! (1997)definecl discri~ninationas "unequal and unfavorable treatment or actions that when directed toward a target group or its members, limit the economic and political opportunities of that group" ( p . 3581. All these definitions share tbe follovinp: being prejudiced, being against somebody, granting pri\-ilepes, or introducing penalties based on differences. \Wrhile major strides hare been made to prevent discrimination. e\.ery society in the urorld still witnesses discriminatory attitudes \vithin its population. Turke!- is a particularl! interesting case gi\,en the variety of ethnic groups present and its historical position as a cultural crossroads betxveen \YTestern European and Middle Eastern cultures, both Christian and Islamic iPoyrazli. 2003: Kabasakal & Bodur. 20081. Turkey is of further interest as a secularized counts!-. although the majority of the population is yet hIuslim. Turke) having been the seat of the Isla~niccaliphate until the fall of the Ottoman Empire iL\Ianz, 20031. For these reasons, religious, ethnic, and sex discriinination all have been present and are currently a source of struggle within the society (see. e.g.. Gumus & C;omleksiz, 1999: Turkish Social Sciences Association, 2005). Recent del-elopments in terms of the possibility of

TL'IIKISII DI3CKII\llNA?'ORI' ATTITUDES SCALE

923

inclusion of Turkey in the European Union ca!1 for an examination of its culture that includes attitudes of discrimination. \Yhile Turkey is among the countries \vhich have made substantial improvements in the areas of human rights and democracy, research conducted within the past decade calls for continued intervention to impro\.e relationships anlong groups of people from different backgrounds (istanbul hIiilkiyeliler Yakfi Sosyal A r a ~ t i r n ~ a l a r hlerkezi. 1998; Cumus & Gomleksiz. 1999; Aksu. 2005; Turkish Social Sciences Association, 2005). The current stud!; hopefull! n-as undertaken to develop a tool for researchers to continue to stud! discriminator)- attitudes and to work towards finding methods to improve relationships among different groups.

X~ETHOD

Sample The items \\,ere tested on 810 college students majoring in elementary education at 10 public institutions in Turkey. Thirty-seven percent of the pasticipants were men and 60'4 \\-omen: 3% did not identify sex. Fift't!i-sis percent of the participants \yere freshmen and 14'?/0 were seniors. nrhose ages ranged from 16 to 29 years 112.1=20.1;S D = 1.9). Disc~z'mirzatot~j~ Attitudes Scill~.-The items for the scale \\-ere generated through a literature revie\\- conducted by a group of researchers. The search \vas based on terms like discrimination. prejudice, religion and political orientation, attitudes. and gender and the found resources \vere carefull! screened to generate items (c.p., Xllport, 1951; Kagitcibasi. 1973; Xlten~e),er,1988: Snizek & Neil, 1992; Xltemeyer & Hunsbesger. 1992: Llaluso, 1995: Tibon & Blumberg, 1999: Schiller, 2004). Three additional items \Yere adopted from previous studies [Kagitcibasi, 1973: Christie, i.t a!.. 1963, cited by Gurnus 6r Gornleksiz, 19993. This process led to selection of 69 items. Lrft-Right Poiitic~zlO~iew~il~ioi~ Scale.-Students' left-right political orientation nras measured through a single item adopted from Guldu and Donmez 12002). Students .\r.ere asked to indicate their political orientation based on an 11-point scale \\.ith 1.erba1 anchors of 1: Kadical left and 11: Radical right. Students \\-ere also given the option to check "this scale is not applicable to me" if the! thought this \\-as the case for them; 116 students marked this option. The number \\.ho rated their political orientation \vas 394. Stu'z't~izt PCI~(OI?U/ Profilt' FoI.wI.-T~I~squestionnaire \\.as prepared b!, the researchers to collect information on studenrs' ape. sex, the name of the uni\rersity the)- n.ere attending. anci their class standing. Pvucedtirc Ten different unilrersities representing different regions i11 Turkey were randomly selected. Both city colleges 15) and rural-area colleges ( 5i \\.ere in-

921

hl GOSILEKSIZ.

ET d l

cluded to represent more diverse groups. All schools were public institutions. Following governinent mandates, all students in elementary education majors in Turkey study the same curriculun~.with the exception of a few elective courses. Questionnaires \\-ere sent by mail along with instructions to the deans of the College of Education for each campus, after they had been informed about the study by telephone. Questionnaires were administered to the students by their instructors during class.

A~za/j'scs T o assess construct validity of the Discriminatory Attitudes Scale, data were analyzed by factor analysis to assess the scale's dimensionality; Pearson product-moment correlations further assessed the scale's construct validity; and Cronbach alpha was calculated to estimate the scale's internal reliability. A negative relation betn~eendiscriminatory attitudes and the left-right political orientation mas expected because individuals who are more oriented to\yards the right have been shoxvn to have more discriminatory attitudes, while those who are more oriented towards the left are more liberal and supportive of social change (Kohn. 1974; Guldu & Donmez. 2002: Jost, Glaser, Kruglanski, & Sulloway, 2003: Sibleq-, Robertson. & LY7ilson,2006). In addition, empirical research indicates that liberals are more likely to support egalitarian practices such as equal rights for ga!-s or women than d o conservatives (Jest, Nosek, & Gosling, 2008). RESULTS Several stages were followed in testing psychometric properties of the Discriminatory Attitudes Scale. First, the means, standard deviations, and item-total correlation coefficients of the 69 items xvere investigated. Then, the data were subjected to Principal Component Analysis xvith varimax rotation. Items with a mean score below 2 or above 4,n,ith a standard deviation below 1.00, and an item-total correlation coefficient value below .50 were excluded. Analysis yielded four different factors with Eigenvalues of 7.1. 1.9, 1.5, and 1.3. The percentages of variance explained were 33.896, 8.9%, i.-I%, and 6.09/b. This four-factor solution or structure accounted for 36% of the total variance. All comnlunalities were greater than .50, After the factor analysis, 21 items remained (Table 1). Next, these 21 items v e r e subjected to a confirmatory factor analysis. Use of both factor analysis methods on the same sample is an approved method of scale de\.eloprnent ( Joreskog & Sorbom, 1993, p. 22-27; Thompson, 2005, p. 110-114). The sample was split randomly into tu.0 halves and subjected to a Principal Component Analysis. For the first half of the sample, the LISREL results supported the +factor model: degrees of freedom (183)/Chi square 1570.65 ) = 0.32, root mean square error of approximation

TLKKISH DISCRIhIINATORY ATTITUDES SC:\LE

925

(R,1/ISEA=0.0711, G F I (0.891, AGFI (0.861, CFI (0.90).and NFI 10.851 were acceptable (Bentler, 1990). For the second half of the sample, the LISREL results also supported the 4-factor model: degrees of freedom 1183~/Chi square (409.24)=0.45, root mean square error of approximation I R ~ ~ T S E X = 0.054), G F I (0.911, AGFI (0.891, CFI (0.94). and NFI 10.891 were acceptable (Bentler, 1990). Con~parisonof factor loadings showed that the factor loadings for each half matched the pattern for the full data set. The same items loaded under the same factors in both halves and the full sample. This provides evidence that the findings are generalizable and valid. All of these results indicate that the items constructed to measure "discriminatory attitudes" belong together. Based on the items grouped together, the four factors were labeled Discrimination Against Homosexuals, Discrimination Against KTomen, Discrimination Based on Race or Ethnicity and Religion, and Discrimination Against Foreigners. Unpaired t tests on upper and lower quartile groups for each factor indicated that items in the subscales distinguished students based on the strength of their positive (or negatire) attitudes, providing additional evidence for the ralidity of the scalr. An expert in the subject matter, the first author, evaluated the items and attested to the face validit! and completeness of the scale. The final scale has 21 items urhich are rated on a ?-point Likert-type scale using anchors of' 1: Strongl!, disagree and 5: Strongly agree. LY7hen calculating the score, negative responses are reversed. Higher scores on the scale indicate Inore discriminatory attitudes. The Pearson product-n~omentcorrelations indicated that Discriminator!; Attitudes correlated highly \\.ith Left-IGght Political Orientation. Students who were more oriented towards the right displayed higher overall Discriminatory Attitudes (1. = 3 9 , p