Table of Contents
Results and Discussion
Preliminary and Descriptive Data Analysis
Factor Structure and Internal Consistency
Annex: Spanish Version of the Prasad-Baron Questionnaire
Background. Gender ideology plays an important role in human behaviour, and therefore its assessment, not only in particular areasbut considering a wide variety of situationsconcerning both public and private life, is necessary. Methods. This study presents the Spanish version of the questionnaire developed in 1996 by Prasad and Baron, a tool for measuring gender-role cognitions, and analyses its psychometric properties in a sample of 233 Spanish undergraduate students (75.1% females, 42.9% self-claimed feminists). Results. The questionnaire showed a 3-dimension structure -Equality, Belief and Principles- which coincided partially with the theoreticaloriginal one. Excellent reliability (Cronbach) results, of over .90, were found. Concurrent validity assessed with a measure of conservatism (E: r = -.253, p < .001; B: r = - .421, p < .001; P: r = -.123, p = .061) and with self-defined identification with feminism (E: F (2, 232) = 4.04, p =.019); B: F (2, 232) = 12.71, p < .001; P: F (2, 232) = 13.05, p < .001) was satisfactory. Scores in all dimensions differed by gender. Conclusions. This study reveals satisfactory psychometric properties of the Spanish version of the Prasad-Baron questionnaire, making it an appropriate tool to obtain a complete assessment of gender ideology.
Gender attitudes; Sexrole attitudes ;Psychological assessment ;Psychometrics
Many tools exist to assess gender attitudes and related factors, partly because of the context- specificity of gender-role ideology (McHugh & Frieze, 1997). However, we were interested in tools that are able to measure components of the gender ideology (such as attitudes and beliefs) as separate factors, from a general, cross-sectional perspective -that is, considering a wide variety of situations of daily life where sexism can arise-, and aimed to both men and women for the assessment of attitudes toward both men and women. In other words, we focused our bibliographical search on tools that allow to assess the attitudes toward differential gender roles of an individual of any gender, and also other components of their gender ideology, such as their believes about gender-based or sex-based differential characteristics that make differential gender roles appropriate. Taking these considerations into account, we assessed the most widely used questionnaires and scales.
Some of the available tools aim to identify attitudes and beliefs in particular areas, such as romantic beliefs related to benevolent sexism in couple relationships (Sprecher&Metts, 1989; Viki, Abrams & Hutchison, 2003); or focus on differentiating between forms of hostile and benevolent sexism, such as the Ambivalent Sexism Inventory (Glick & Fiske, 1996). Others are designed for specific gender groups only, such as the Gender-Equitable Men Scale (Pulerwitz& Barker, 2008), and/or others assess attitudes toward one gender group, such as the Ambivalence Toward Men Inventory (Glick & Fiske, 1999) or the Attitudes toward Women Scale or AWS (Spence &Helmreich, 1972; Spence, Helmreich&Stapp, 1973). Some authors have proposed to assess gender ideology as an identity instead of with attitudes scales: this is the case of a vignette-style tool (Kroska, 2000) designed to be used in another particular area (measuring variation in the meaning of a marital identity). Moreover, none of these tools allow the assessment of attitudes, beliefs and/or principles as separate components of gender ideology and the way in which they interact. Also, as far as we know, most of these questionnaires and scales are not available in the Spanish language.
Therefore, we think that there is a need for tools that allow experts to obtain a more cross- sectional and complete assessment of the gender ideology (and its components) of an individual, regardless of their gender. We carried out a study of psychometric tools that met these criteria, and found out that, in our opinion, the most complete tool currently existing was the one developed by Prasad and Baron (Prasad & Baron, 1996), which was never published, even though it was validated with a sample of 198 US students with apparently satisfactory results. As far as we know, the questionnaire had not been translated to any other language other than the original (English), and thus, after contacting with the original authors, we translated it into Spanish and adapted it by administrating it to college students of our area.
233 undergraduate students of the degrees in Psychology and Medicine at the UniversitatAutònoma de Barcelona. The students were contacted using the online social networking service Facebook, which has been reported in the literature to be the most popular Social Network Site (SNS) among Spanish college students (Arteaga-Sánchez, Cortijo&Javed, 2014).A virtual flyer with information about the study was published on every Facebook group involving undergraduate students from the target groups that could be identified online. This included groups, whether public or private, whose headline indicated clearly that they were meant for Medicine or Psychology undergraduate students of any year in the UniversitatAutònoma de Barcelona. An exhaustive search was performed on Facebook in order to identify all such groups.
The original Prasad-Baron Questionnaire. It consists of 106 items that the original authors sorted theoretically into 5 different scales - Belief, Moral, Equality, Reversal and Separation. Belief scale (22 items) is supposed to assess assumptions not based on rational knowledge but often on prejudices and cultural stereotypes on the characteristics of women and men; Moral scale (9 items) was conceived to measure moral principles on how lives and relationships between men and women should be; Equality scale (40 items) is thought to assess attitudes towards gender equality as opposed to the current situation of male privilege; Reversal scale (13 items) is supposed to measure attitudes towards an eventual situation of female privilege, and could be considered to assess attitudes favourable to affirmative action (previously called positive discrimination) even though not all items are clearly interpretable in this way; and Separation scale (6 items) is thought to assess attitudes towards the convenience or not of gender segregation in different social situations. The 5 scales are theoretically conceived to measure different components of the gender ideology of an individual, even though the authors report that they are correlated with each other, since psychological factors assessed in the different scales relate logically with each other (i.e., beliefs about what women and men are determine attitudes about whether gender equality is possible and/or desirable, moral principles on how women and men should live and relate to each other, and ideas on the desirableness of gender segregation or the need to establish affirmative action measures. The authors did not provide information on mean and standard deviation scores in each scale, although they reported the existence of gender and cultural differences.
The response to each item is coded by a 5-option Likert scale, where 1 is described as “Totally agree”, 2 as “Moderately agree”, 3 as “Neutral”, 4 as “Moderately disagree” and 5 as “Totally disagree”. The questionnaire includes inverse items in each scale, except for Belief, where all 22 items are expressed in positive terms (i.e. Men are better suited for higher education than women), and Reversal, where all 6 items express the need for measures that benefit female over male (i.e. Families should spend more money on the education of daughters than on the education of sons).
In the original article, 16 out of the 106 total items were not sorted into any of the 5 scales. However, in our study we aimed to test the factorial structure of the questionnaire, since the original authors did not provide any factorial data; therefore, we did not initially assume the theoretical structure proposed by the original authors, and we included all the items in the analysis.
Measurement of conservatism. Along with the questionnaire, and in order to assess concurrent validity, participants were asked to complete the Spanish version of a well-known 50-item conservatism questionnaire (OrtetiFabregat, 1991; Wilson & Patterson, 1968), in which the scores vary from 0 to 100, with 100 being the most conservative ideology. Conservatism has previously been linked to sexist attitudes (Christopher & Mull, 2006).
Additional data. Subjects were also asked to answer some basic sociodemographic questions (gender, age, degree, maximum completed education level, and political ideology) and some general questions in order to assess their conception of feminism and gender issues.
Review Procedure. The original questionnaire, provided with the original article, was carefully examined for mistakes and inconsistencies. We found that, in our opinion, one item was missing in the Equality scale in order to maintain the logical structure that is often found in this scale, where groups of 3 items are presented together, with an item assessing whether the subject agrees with the current situation of male supremacy, an item measuring whether they agree with the egalitarian situation, and finally there is the corresponding item assessing if they agree with an eventual situation of female benefit over male (belonging to the Reversal scale). For this reason, we added this missing item to the Spanish version (item 62), resulting in a 107-item
questionnaire and a 41-item Equality scale. We detected a few items which might not have the same meaning in our culture than in the American one (i.e. Women's sports at colleges should receive less funding than men's sports); however, given the fact that this was the first time when the factorial structure of the Prasad-Baron questionnaire would be assessed, we decided to keep in order to assess their behaviour. We did make minor changes for cultural reasons, such as slightly changing the formulation of the item Medicine is a more appropriate career for a woman than engineering, in which we replaced medicine for nursing.
Translation Procedure. The original questionnaire was translated into Spanish by a bilingual expert in psychology. This preliminary Spanish version was re-translated into English by an expert translator from the University Translation Service who did not know the original questionnaire. After that, the original version, the preliminary Spanish version, and the re- translated English version were reviewed independently by 4 bilingual experts in psychology and psychometrics not linked in any ways with this project. The reviewers were instructed to compare the 3 versions, and assess whether each item in the Spanish version could be considered equivalent to the original item, using the re-translation into English as an aid in order to identify eventual correspondence mistakes. They were provided with, and told to proceed according to, a recent review published in literature on good practices concerning the translation of psychometric tests (Muñiz, Elosua& Hambleton, 2013). Each reviewer compared each item in the 3 versions and assessed the correspondence between the original and the preliminary Spanish item, both in linguistic and psychological terms. Results were sent by each reviewer to the authors as comments written in a comparative table created for this purpose. All comments made by all reviewers were accurately considered by the 3 authors. The comments that were made by at least 3 out of the 4 reviewers were immediately taken into account and incorporated into the Spanish version of the questionnaire. The comments not meeting these criteria were examined carefully and taken into account if at least 2 out of the 3 authors considered them appropriate. After that, the complete resulting Spanish version was reviewed and answered by
the 3 authors, in order to identify linguistic mistakes (i.e. spelling, concordance or item numeration mistakes). All mistakes detected were corrected. The Spanish version incorporating all these modifications was considered the final Spanish version of the questionnaire, and was administrated to the subjects in order to assess its psychometric properties.
Administration Procedure. The online questionnaire administrated to the participants incorporated filters to prevent missing and inconsistent answers. Response to all questions was compulsory. The questionnaire was anonymous and participants could only be tracked by their e-mail address. All this information was provided before the beginning of the questionnaire, along with the informed consent, which the participants had to accept in order to access the rest of the questions.
Once all data were collected, it was transferred to an appropriate file for its statistical treatment and analysed using SPSS 19 software. We assessed descriptive data, factor structure, and internal consistency using factorial analysis, concurrent validity with the factors conservatism and self-reported identification with feminist ideology, and reliability (Cronbach). When analysing concurrent validity, we used correlations with conservatism and self-claimed feminism, instead of prediction of assumed gender or cultural differences as the original authors did, because in a patriarchal culture females also show negative attitudes towards female gender, especially those linked with benevolent sexism (Fields, Swan &Kloos, 2009; Glick & Fiske, 2001), and also because, unlike the original sample, ours did not include a representative sub-sample of subjects belonging to cultural minority groups. We also performed a specific data analysis aiming to detect eventual gender differences in the different factors.
Results and Discussion
Preliminary and Descriptive Data Analysis
The mean age (standard deviation) of the sample was 22.02 years (2.81). The gender distribution showed that 75.1% of the participants were female, 24.5% were male, and 0.4% did not identify with wither female or male gender (other gender). Just over half (50.6%) were Psychology students and 49.4% were Medical students. Politically, most of the students (56.2%) defined themselves using the expression left, followed by those who situated themselves in center-left (23.6%). As expected, since they were undergraduate students, the highest education level completed by the majority (75.1%) of the sample was precollege education (High School Diploma).
Despite the fact that 98.7% of the sample thought that they were not sexist, only 42.9% defined themselves as feminists, with 6% claiming that they were not sure on whether they were feminists or not, and 51.1% stating that they definitely did not identify themselves with feminism. When asked about the definition of feminism, 65.7% claimed that it is a movement that fights for gender equality, 54.5% considered that it is a movement that fights for equality between men and women, and only 5.2% thought that feminism is a movement composed only by women; however, 27.5% believed that feminism it is a movement that fights for the supremacy of women, and only 36.9% perceived feminism as being necessary. When asked about the meaning of the terms sex and gender, 67.4% answered correctly that they are not synonyms and claimed to know the difference between them, with 15% being aware that they have different meanings but did not know the difference, and 17.6% thought that they have equivalent meanings. Mean (standard deviation) score for conservatism was 21.19 (8.22).
Before performing factorial analysis, we calculated the scores in the 5 theoretical scales conceived by the original authors, which were as follows: M = 4.22, SD = 0.7 for Belief; M = 2.19, SD = 0.64 for Moral; M = 1.42, SD = 0.55 for Equality, M = 4.13, SD = 0.53 for Reversal, and M = 4.45, SD = 0.58 for Separation.
Factor Structure and Internal Consistency
In order to assess whether the real structure of the questionnaire coincided with that theoretically conceived by the authors, we performed factorial analysis. This was possible because a KMO (Kaiser-Meyer-Olkin) value of .925 was obtained, along with significant (χ² (2, N = 107) = 25315.94, p <.001) results in the Bartlett test (although the distribution of each item did not adjust to normality). We tested several Principal Component Analysis models, including 3-factor, 4-factor, 5-factor, and 6-factor structures, all of them with and without the 16 items that the original authors did not include in any scale. We evaluated all the models, taking into account the following psychometric criteria: extraction criteria (factors with values >1, sedimentation graphics, explained variance >45%, parsimony and parallel analysis) and interpretability criteria (percent of values >.3 loaded in the expected factor, in an unexpected factor, and cross-loadings), adjustment (study of the distribution of communalities), and residuals (<25% of residuals greater than .05). The solution that met all these criteria in the most satisfactory way was the 3-factor one including all the items. However, on the basis of the results (Table 1), it was decided to eliminate those items from the questionnaire that: (a) Did not show any loading higher than .3; (b) Did not behave as expected by loading clearly higher in one dimension where they did not theoretically fit, or cross-loaded in several dimensions, when this could not be theoretically explained; (c) Groups of items asking about the same issue that showed unexplainable contradictory loadings in different subscales, since we reasonably thought that in these cases all the items should belong to the same subscale; (d) Items that were considered to be inappropriate in our time and/or geographical context. Using these criteria, we eliminated 17 items.
Table 1. Factorial structure
Abbildung in dieser Leseprobe nicht enthalten
- Quote paper
- Elisabet Tasa-Vinyals (Author)Marisol Mora-Giral (Author)Rosa Maria Raich-Escursell (Author), 2015, Spanish Adaptation of the Prasad-Baron Questionnaire, Munich, GRIN Verlag, https://www.grin.com/document/309370