Journal of Social Research & Policy, Vol. 7, Issue 1, July 2016

Validity of the construct of Right-Wing Authoritarianism and its measure in post-socialistic region: A case of the Czech Republic JOHANA CHYLÍKOVA1 The Institute of Sociology of the Czech Academy of Sciences

MARTIN BUCHTÍK The Institute of Sociology of the Czech Academy of Sciences

Abstract This article presents an analysis of the construct validity of the Right-Wing Authoritarianism (RWA) scale measured on a nationally representative sample of respondents in the Czech Republic. The authors – adhering to a paradigm of internationally comparative research – test whether the RWA scale, which was originally developed in the North American cultural milieu, can serve as a valid measurement instrument in a significantly different cultural environment. The test consists of an examination of the scale’s factorial structure, which has been set in the theory as unidimensional. To examine the possibility that the RWA measure was biased by a systematic method bias – the acquiescent response style (ARS) – several models were specified in confirmatory factor analysis. The analysis showed that the RWA scale in the Czech Republic is not unidimensional and that Czech respondents have a tendency to consciously agree both with the pro-authoritarian and anti-authoritarian items of the RWA scale and that this tendency is not caused by the ARS effect. The authors conclude that the construct of right-wing authoritarianism operationalized in the RWA scale cannot be validly measured in the Czech Republic and suggest a possible explanation of the different factorial structure of Czech RWA data. Ke ywords: Authoritarianism; RWA Scale; Construct Validity; Cultural Differences; Response Style.

Introduction “Authoritarianism is something authoritarian followers and authoritarian leaders cook up between themselves. It happens when the followers submit too much to the leaders, trust them too much, and give them too much leeway to do whatever they want.” (Altemeyer, 2006, p. 2). With the deteriorating state of political culture in the Czech Republic, in late 2012 we began to ponder what potential exists in the Czech population for the rise of various authoritarian political groups. We sought an internationally comparable instrument that could assess how large a part of the Czech electorate is predisposed to vote for subjects that profile themselves as 1 Postal Address: T he Institute of Sociology of the Czech Academy of Sciences, Jilská 1, 110 00, Prague, Czech Republic. E-mail Address: [email protected]

2 | JSRP

Johana Chylíkova, Martin Buchtík

authoritarian. In addition, in late 2012 and early 2013, the Czech Republic held its first ever direct presidential election and we were very interested in learning which candidates were likely to collect votes from people who are inclined towards an authoritarian way of thinking. We started to look for a widely used measure of authoritarianism and we found it in t he field of political psychology. The right-wing authoritarianism scale (hereinafter RWA scale) has been theoretically described as an instrument that identifies respondents who are predisposed to think in a rather authoritarian manner. Moreover, this scale appears to be the most reliable research tool since it is regarded as a state-of-the-art instrument (Feldman, 2001; Perry, Sibley & Duckitt, 2013) and is by far the most common measure for studying authoritarianism in psychological research. The RWA scale seemed to be the best choice for studying authoritarianism, especially when we took into consideration that it is fairly sociological in character, even though it is widely used in the field of political psychology. We, as sociologists, liked that the RWA scale’s items are expressions of attitudes and relate to social norms , and we wanted to make use of this feature in sociological research. Moreover, the RWA scale has not been previously tested in the Czech Republic (Altemeyer, 1981; 1996; 2006) and we liked the idea of being, in a way, pioneers in the field of researching authoritarianism in our country. From the beginning, our intention was to test whether the RWA scale, originally developed in the North American cultural milieu, is a measurement instrument suitable for use in the Czech Republic. Since we respect and appreciate the idea of comparative social research, we prefer to use internationally comparable analytical tools and focus on their comparability across different countries and cultures. We think that it is better to use current measurement instruments developed in other countries and examine their potential in a new cultural environment than to develop locally applicable instruments that do not allow for intercultural comparison. To explore whether the RWA scale is a valid instrument with which to measure authoritarianism among the population of the Czech Republic, we opted for an analysis of construct validity; i.e. factorial structure analysis. When testing a measurement instrument in a n ew environment, it is essential to find out whether the data correspond to the factorial structure set out in the theory. If data collected in the new environment do not produce the desired factorial structure, the concept under investigation is not empirically present and its measurement is invalid. We are aware that knowledge about the dimensionality of the data is not the only possible way to test validity; a scale’s validity can be investigated using other methods, such as criterion validity or convergent and discriminant validity. However, in this study we adhere to construct validity because it is the crucial prerequisite of valid measurement of a construct whose factorial structure is already established in theory. In this study we test whether Czech RWA data has the theoretically given factorial structure. According to the author of the RWA scale, the scale measures one construct that should manifest itself in a confirmatory factor analysis (CFA) as one factor composed of all items of the scale, or two highly correlated factors (Altemeyer, 1996). In our analysis we focus on confirming this structure and we control for possible method bias that might have occurred in the data and affect the results. This kind of bias is produced as a result of the acqu iescent response style (ARS) and is well known to systematically affect measures derived from Likert -type response scales, i.e. the response scale the RWA scale employs. By means of this additional analysis we sought to make sure that the ARS bias did not mask the real factorial structure in our data. Theoretical background The Origin and Development of the Concept of Authoritarianism The concept of authoritarianism that was introduced in the theory of the authoritarian personality marked a milestone in the development of social and political psychology. It was introduced into expert discourse in 1950 by the authors of The Authoritarian Personality (Adorno et al., 1950), who drew on the work of such scholars of the Frankfurt School as Max Horkheimer, Erich Fromm and Herbert Marcuse (e.g. Horkheimer, Fromm & Marcuse, 1936), all of whom

Validity of the construct of Right-Wing Authoritarianism …

3 | JSRP

considered describing the authoritarian personality as crucial for understanding the emergence of totalitarian power, something with which they had intimate, and oftentimes personally tragic, experience. The concept was further elaborated in the field of psychology in the 1980s by Altemeyer, who had become one of its prominent proponents (Altemeyer, 1981; 1996; 2006). The enduring importance of this theme is apparent from the fact that it was the subject of a thematic issue of the journal Political Psychology (Marcus, 2005). Authoritarianism is usually defined in concrete terms through the concepts of the authoritarian personality or authoritarian followers. The authoritarian personality is "a specific social character, which "is programmed to obey societal authorities, accept their norms and values, and releases its hatred towards members of weaker societal groups. It has been assumed that this type of character is especially susceptible to fascist movements and propaganda."(Stellmacher & Petzel, 2005, p. 246). “Authoritarian followers” are people who support the steps of various authorities, accept their opinions, and aggressively assert them over the opinions of others (Altemeyer, 2006, pp. 9-10). These authorities may be supportive of established ideas, which is what we deal with in this article, or, conversely, supportive of change.2 These individuals are characterized by a submission to authority and by the belief that strong authority offers a lifeline in a world that is dangerous and hard to predict (ibid.). In the discourse of political and social psychology, there is a general consensus that there are ways to reliably measure the leanings towards authoritarianism or an authoritarian personality trait; various batteries of self-reporting items have already been used in quantitative psychological research for this purpose (Funke, 2005, p. 196). The first instrument for measuring authoritarianism was the F-scale, developed in the late 1940s by a team at UC Berkeley (Adorno et al., 1950). According to its creators, the F-scale reflects nine traits of authoritarianism: conventionalism, authoritarian submission, authoritarian aggression, anti-intraception, superstition and stereotypes, power and “toughness”, destructiveness and cynicism, projectivity, and sex (Adorno et al. 1950, p. 228). Nevertheless, the scale’s authors admitted that these traits represent rather the most characteristic features of the authoritarian personality rather than an exhaustive list: “This does not imply that all the features of this personality pattern are touched upon in the scale, but only that the scale embraces a fair sample of the ways in which this pattern characteristically expresses itself.” (ibid.) The F-scale remained in use for quite a long time, and following criticism of the method, summed up, for instance, by Gul & Ray (1989), several new versions of the scale were developed. In the 1970s Altemeyer (1981) created the Right-Wing Authoritarianism (RWA) scale, which replaced the F-scale as the most used measure of authoritarianism. Apart from the RWA scale there exist two other measures of authoritarianism, but they are used only marginally. Oesterreich’s scale (2005) is fairly psychological in nature – it focuses on psychological personality traits and it intentionally avoids any attitudinal questions. This scale holds promise for psychological research, but it is not suited for standard sociological research. The other scale is the Group Authoritarianism Scale (GA scale) (Stellmacher & Petzel, 2005), which stems from Duckitt´s concept of authoritarianism (1989). Duckitt explains authoritarianism as a consequence of strong identification with a social group, which manifests itself as a stringent demand for social cohesion. Authoritarianism is then "an individual belief about the appropriate relationship that should exist between groups and their individual members" (Stellmacher & Petzel, 2005, p. 248). In Duckitt’s conception, the authoritarian personality has the following attributes: unconditional obedience to in-group leaders, intolerance of persons who do not conform to norms and rules, and emphasis on behavioral and attitudinal conformity (ibid). The GA scale has never

2 A predisposition towards submission to new and revolutionary authorities is a rather marginal phenomenon in society. An analogical tool that measures such a predisposition, the LWA scale, was successfully tested on a sample of political extremists, but the authors of that tool themselves acknowledge that it is of limited use in the general population (Van Hiel, Duriez & Kossowska, 2006).

4 | JSRP

Johana Chylíkova, Martin Buchtík

been widely used and the author of the concept himself even prefers working with the RWA scale (Duckitt, 2006; Duckitt & Sibley, 2007; Perry, Sibley & Duckitt, 2013). The RWA construct and the RWA scale The RWA construct was derived from the F-scale by reducing the nine facets of the F-scale to three; the construct is a “covariation of three attitudinal clusters in a person” (Altemeyer, 1996). The three attitudinal clusters are: “authoritarian submission (uncritical subjection to authority), authoritarian aggression (feeling of aggression towards norm violators) and conventionalism (strict adherence to conventional norms and values)” (Altemeyer, 1996, p. 7). The RWA scale is a direct operationalization of the RWA construct, i.e. the individual items of the RWA scale reflects the three clusters. At this point it is important to stress that although the items measure the three clusters, the construct and the scale itself are unidimensional (See Diagram 1). The unidimensional RWA construct is also bipolar, which means that it has “positive” (proauthoritarian) and “negative” (anti-authoritarian) tails. The bipolarity of the construct is assured by the design of the RWA scale, which is semantically balanced; this means that half of its items are formulated in support of the measured construct and half of the items are formulated against it. If you look at how items on the RWA in Table 4 are formulated, you can see that half of the items are statements that express authoritarian (or conservative) attitudes; these items are formulated in support of the measured construct. The other items express an anti-authoritarian (or liberal) outlook; these items are formulated against the measured construct.

Diagram 1: The Right Wing Authoritarianism construct The original RWA scale is a 30-item battery where each item uses a 9-point Likert type response scale (Altemeyer, 1996), half of the items express anti-authoritarian attitudes, and the other half express pro-authoritarian attitudes. Altemeyer subsequently modified the RWA scale and in its latest version there are 20 items (Altemeyer, 2006) with 10 pro-authoritarian items and 10 antiauthoritarian items. Although Altemeyer emphatically warns against any further reduction of the scale (“Hell, will be the final destination of any researcher who decides to use only part of the RWA scale…” (Altemeyer, 2006, p. 36)), there are a number of shorter versions that are of a more practical length for standard interviewing (Gray & Durrheim, 2006; Rattazzi, Bobbio & Canova, 2007; Zakrisson, 2005), which were formulated on the basis of a conceptual rationale, psychometric investigation or by random sampling (Zakrisson, 2005). From the perspective of survey methodology the RWA scale suffers several methodological flaws or at least deficiencies. These problems relate to the length and complexity of the items, but also to the ambiguity of the formulations. The RWA scale’s items are relatively long and complex sentences, and they usually have very complicated wording. For this reason the scale puts an enormous burden on respondents. Moreover, almost every item on the scale measures multiple traits at once, i.e. it usually requests an opinion on more than one thing (for example “Everyone should have their own lifestyle, religious beliefs, and sexual preferences, even if it makes them different from everyone else”). The items also usually capture two or three RWA attitudinal clusters

Validity of the construct of Right-Wing Authoritarianism …

5 | JSRP

at once (authoritarian aggression, authoritarian submission, conventionalism), so it is impossible to identify items that measure only one attitudinal cluster. In social survey methodology there is a consensus that questionnaire items that measure multiple traits at once, i.e. they are double- or multi-barreled, are methodologically defective (Babbie, 2007, p. 247). When preparing a questionnaire, sociologists usually make an effort to ensure that their items are unambiguous. For the psychologist Altemeyer, unambiguity is not a necessity; he believes that psychologists benefit from information based on a particular feeling that is evoked by an item (Altemeyer, 1996, p. 13). The RWA scale is widely used and review articles reveal it was used in 77 surveys between 1987 and 2010 (Altemeyer, 1996; Perry, Sibley & Duckitt, 2013). However, it is important to note that none of these studies were conducted on a representative sample of the national adult population.3 In most cases the respondents were students (of psychology) or other very specific groups. Even in other publications that we drew on in the preparation of this study , we were unable to find any data that came from a representative survey (Akrami & Ekehammar, 2006; Cohrs et al., 2005; Duckitt, 1993, 2001, 2006; Duckitt & Sibley, 2007; Gray & Durrheim, 2006; Kemmelmeier et al., 2003; Lambert, Burroughs & Nguyen, 1999; Mavor, Louis & Laythe, 2011; Rattazzi, Bobbio & Canova, 2007; Zakrisson, 2005). The RWA has been criticized by various researchers since as far back as the 1980s (Ray, 1985). Their criticisms have centered on the following arguments: (1) the scale primarily measures conservative ideology, not the personality traits of the individual (Ray, 1985; Funke, 2005; Stellmacher & Petzel, 2005); (2) the scale is problematic from a methodological perspective; the items measure multiple traits at once – they are “multi-barreled” (Funke, 2005); and (3) the measured concept is focused solely on the individual level and it ignores the social context (Duckitt, 1989); “authoritarianism and the relationship between authoritarian attitudes and authoritarian behavior is much more flexible and influenced by the social context than was originally proposed” (Stellmacher & Petzel, 2005, pp. 246). To counter these criticisms, researchers who use the RWA claim that the RWA scale is based on well-thought-out theoretical foundations, and they cite a number of surveys that have been performed in which its reliability has been repeatedly confirmed and its external validity proven using related instruments, i.e. showing the RWA scale’s correlation with other scales (e.g. the Big Five) (Akrami & Ekehammar, 2006; Altemeyer, 2006). The Dimensionality of the RWA scale Since the 1970s Altemeyer has conducted hundreds of RWA studies surveying students at the University in Manitoba, Canada. He has provided evidence of the validity of the scale using extensive correlational analyses with other scales (e.g. Attitudes towards Homosexuals Scale, The Posse against Radicals Scale, The Christians Orthodoxy Scale, The Big Five Scale) (Altemeyer, 1996), and he has also demonstrated the psychometric properties of the RWA scale based on the results of confirmatory factor analysis (CFA) and the Cronbach’s alpha consistency of the scale. As stated in a previous subsection, the RWA scale captures a unidimensional and bipolar construct; i.e. the scale measures one latent construct, and half of its items are formulated in favor of the measured construct and the other half against it. The unidimensionality of the scale has been proven in confirmatory factor analyses; in North American data , Altemeyer found a one-factor solution or two-factor solution with highly correlated factors (r=0.4 – r=0.7) (Altemeyer, 1996, p. 53). The two-factor solution, i.e. the division of a one-dimensional construct into two factors, is quite common in psychometrics when the dimensionality of a bipolar construct is being investigated. A scale measuring one construct with ite ms formulated in support 3 T he only mention of a representative sample of the population that we found in a general literature search was in a study by Mick McWilliams & Jeremy Keil, who in 2005 conducted research on a “reasonably representative sample” (Altemeyer, 2006), although it is not entirely clear what is meant by that phrase. Altemeyer indicates that: n=1.000, Cronbach’s alpha=0.9 and the average value of the converted 9-point scale is 90. Unfortunately, despite our extensive efforts, we were unable to find the primary source.

6 | JSRP

Johana Chylíkova, Martin Buchtík

of the measured construct and items formulated against it is often found to be composed of two highly correlated factors (Marsh, 1996; Cambré, Welkenhuysen -Gybels & Billiet, 2002). This outcome is commonly the result of an acquiescent response style (ARS) (Billiet & McClendon, 2000), a systematic method effect that will be described later in this text. From the scholarly literature we have analyzed it seems that the factorial structure of the RWA scale has not been significantly addressed by researchers so far. We searched studies in which the RWA scale had been used and looked for analyses of factorial structure. In Central and Eastern Europe the RWA scale has been used on several occasions to study authoritarian attitudes. In one instance, it was applied in the form of a 30-item scale (Kemmelmeier et al., 2003), but more frequently a shorter version was used, either one derived from the original RWA scale (Duriez, Van Hiel & Kossowska, 2005; Van Hiel & Kossowska, 2007), or a comp osite version combing the RWA scale and the F-scale (Todosijevic, 2008). There is also a CEE study of Altemeyer’s 30-item version of the RWA, but its original composition was modified: some items were removed, and some items from other scales were added (Krauss, 2002). Unfortunately none of these studies tested the construct validity. There is no scholarly literature based on research in other parts of the world either that has demonstrated the dimensionality of RWA data using confirmatory factor analysis (CFA) (Akrami & Ekehammar, 2006; Altemeyer, 1996; Duckitt, 1993, 2001, 2006; Duckitt & Sibley, 2007; Duriez & Van Hiel, 2002; Gray & Durrheim, 2006; Mavor, Louis & Laythe, 2011; Lambert, Burroughs & Nguyen, 1999; Zakrisson, 2005). We found only one study th at tested the dimensionality of data using an unmodified version of Altemeyer’s RWA scale; Mavor, Louis & Sibley (2010) used a non-representative sample of students and showed using confirmatory factor analysis that their data did not conform to a one-factor solution. Acquiescent Response Style The response scale of the RWA items is a Likert-type scale and it is well known that this type of response scale is susceptible to the response style bias (McClendon, 1991; Tourangeau, Rips & Rasinski, 2000). Response style refers to the tendency of a respondent to respond in some systematic way, regardless of the content of the given statement (Billiet & McClendon, 2000). A response style can manifest itself in various ways, for example as a tendency (a) to agree, i.e. acquiescence, (b) to disagree, i.e. nay-saying, or (c) to find middle ground on the scale, i.e. middle-saying (Tourangeau, Rips & Rasinski, 2000). In this study we focus on the acquiescent response style (ARS), which occurs when some respondents agree both with items that support the measured construct, and with items that are opposed to it (Groves, 2004). The acquiescent response style can derive from various sources: (a) One is so-called satisficing, a phenomenon that occurs when a respondent is unwilling to make the effort to answer questions and does not go through all the phases in the cognitive process of formulating a response when answering a question (Krosnick, 1991; Tourangeau, Rips & Rasinski, 2000). The respondent’s response is intended simply to “satisfy” the researcher’s need for an answer to a question. Satisficing tends to be more common when the way in which the questionnaire items are formulated is overly complicated and their meaning is unclear, or when respondents are being asked something they know nothing about (Tourangeau, Rips & Rasinski, 2000); (b) Tourangeau et al. used examples from research studies to demonstrate that some respondents tend, in the presence of an interviewer, to agree with the statements that are read to them because they look at the interviewer as an authority they would prefer not to oppose; in this case the effect of a complicated or unclear question is further heightened. Scholars have shown that the ARS correlates positively with age and negatively with education (ibid., Billiet & McClendon, 2000); in other words, people who have a tendency to agree with contradictory statements tend to be older and have a lower level of education. The acquiescent response style creates a systematic measurement error – ARS bias – which pushes values of the variables in one direction and distorts the results of analyses. The variance caused by this kind of measurement error may be one reason why it is not possible to confirm a

Validity of the construct of Right-Wing Authoritarianism …

7 | JSRP

one-factorial solution in CFA for unidimensional constructs (Billiet & McClendon, 2000). If analysts want to be sure about the rejection of a one-factorial structure in supposedly unidimensional data of a bipolar scale, they may introduce an ARS factor into the CFA model and observe whether the ARS bias was or was not present. Response style analysis is especially recommended in cross -cultural survey research (Billiet, Cambré & Welkenhuysen-Gybels, 2002), because respondents from different countries may employ a different response style or may be susceptible to a response style to different extents. Modeling the ARS enables researchers to uncover statistical relationships between variables without these method effects and thus to better recognize the empirical reality. There is no question that modeling the ARS with Czech RWA data is useful: in our study we test the RWA scale in an entirely different cultural background from that it has previously been tested in and we can assume that the response style of Czechs may differ from that of respondents in the North America. Moreover, we know that many RWA items are formulated as complicated sentences, and their meaning may not always have been clear to respondents. Finally, Altemeyer (1996) himself considered the possibility of the ARS bias in RWA data, claiming that his measurements exhibited a weak measurement bias that distorted responses towards more positive answers. However, he did not consider this effect to be significant enough to render measurement of the RWA construct invalid. The Present Study Data Collection and Methodolog y In our study we use the latest version of the RWA scale (Altemeyer, 2006), which contains 22 items (for the original wording of the statements, see the first column in Table 1) on a 9-point 4 Likert- type response scale. The first two items are not part of the scale per se, but are included in the questionnaire to allow respondents to practice working with the 9-point response scale. The RWA scale itself is made up of just 20 items, 10 of which are formulated as positively indicative of an authoritarian attitude (pro-authoritarian items) and 10 are formulated as positively indicative of a liberal attitude (anti-authoritarian items). The statements were translated from English into Czech on the basis of two separate translations that were subsequently revised by a third translator. In the final translation we primarily aimed to capture the intended meaning of the statements, so the translations were not always literal. We retained the original meaning of the statements, but given the different cultural environments in the Czech Republic and North America we had to make some mino r amendments (a similar approach has been employed by some other authors (Zakrisson, 2005; Krauss, 2002)) mainly to items F and M: there is no strong tradition of religious authorities in the Czech Republic and the practice of school prayer does not exist. We had to simplify other statements to ensure they were intelligible to all respondents. 5 In each statement we independently identified keywords indicative of the three RWA clusters: conventionalism, authoritarian submission, and authoritarian aggression. We retained these keywords and phrases, and we always tried to retain the same subject and predicate used in the statement (respecting the grammatical structures of the two languages). These revisions can be described as the essential minimum required in order for the statements to be used in a survey in the Czech Republic. The final wording of the statements is presented in the second column of Table 1 (back-translated into English).

4 Agreement with the statement: -4 = strongly disagree with the statement; -3 = strongly disagree; -2 = moderately disagree; -1 = slightly disagree; +1 = slightly agree; +2 = moderately agree; +3 = strongly agree; +4 = very strongly agree (see also Altemeyer, 2006, p. 11). 5 T his included minor revisions to statements C, F, J, K, O, R, S, V, and a more pronounced simplification of items E and P.

8 | JSRP

Johana Chylíkova, Martin Buchtík

Table 1: RWA scale: original and optimalised formulation Formulation in original RWA scale

Formulation used in Our Society survey

A

The established authorities generally turn out to be right about things, while the radicals and protestors are usually just “loud mouths” showing off their ignorance.

When there is a problem to be dealt with, the established authorities tend to be right, not the noisy opponents, who in reality know nothing about the problem.

B

Women should have to promise to obey their husbands when they get married.

When women get married they should have to promise they will obey their husbands.

C

Our country desperately needs a mighty leader who will do what has to be done to destroy the radical new ways and sinfulness that are ruining us.

Our country desperately needs a strong leader who will fight against the radicals and the immoral views that threaten our society.

D

Gays and lesbians are jus t as healthy and moral as anybody else.

Gays and lesbians are just as healthy and moral as anyone else.

E

F

G

H I

J

K

L

M

It is always better to trust the judgment of the proper authorities in government and religion than to listen to the noisy rabblerousers in our society who are trying to create doubt in people’s minds Atheists and others who have rebelled against the established religions are no doubt every bit as good and virtuous as those who attend church regularly. The only way our country can get through the crisis ahead is to get back to our traditional values, put some tough leaders in power, and silence the troublemakers spreading bad ideas.

It is always better to trust the judgment of the established authorities than the rabblerousers who try to create doubt in people's minds. Atheists and opponents of the church are equally as good and honest as people who regularly attend church. If our country is to cope with the crisis, it is necessary to return to traditional values, elect strong leaders, and silence the troublemakers spreading dubious ideas.

There is absolutely nothing wrong with nudist camps.

There is nothing at all wrong with nudist camps.

Our country needs free thinkers who have the courage to defy traditional ways, even if this upsets many people. Our country will be destroyed someday if we do not smash the perversions eating away at our moral fiber and traditional beliefs. Everyone should have their own lifestyle, religious beliefs, and sexual preferences, even if it makes them different from everyone else. The “old-fashioned ways” and the “oldfashioned values” still show the best way to live.

Our country needs free thinkers who have the courage to defy traditional values and customs, even if this upsets many people.

You have to admire those who challenged the law and the majority’s view by

Our society is going to fall apart unless we stop the attacks on our traditional values and morality. Everyone should have the opportunity to choose their own lifestyle and religion and select a partner according to their own sexual preference. "Good old values" and "established traditions" are still the best guidelines for how to live a good life. We should admire those people who challenge the views of the majority, defend a woman's right to abortion and animal

Validity of the construct of Right-Wing Authoritarianism …

N

O

P

Q

R

S

T U

V

9 | JSRP

protesting for women’s abortion rights, for animal rights, or to abolish school prayer.

rights, and are for limiting the authority of the church.

What our country really needs is a strong, determined leader who will crush evil, and take us back to our true path. Some of the best people in our country are those who are challenging our government, criticizing religion, and ignoring the “normal way things are supposed to be done.” God’s laws about abortion, pornography and marriage must be strictly followed before it is too late, and those who break them must be strongly punished. There are many radical, immoral people in our country today, who are trying to ruin it for their own godless purposes, whom the authorities should put out of action. A “woman’s place” should be wherever she wants to be. The days when women are submissive to their husbands and social conventions belong strictly in the past.

Our country needs a strong leader who can overcome dangerous opinions and put us back on the right path.

There is no “ONE right way” to live life; everybody has to create their own way.

There is no ''one right way'' to live one's life. Everyone has to find their own way.

Homosexuals and feminists should be praised for being brave enough to defy “traditional family values. This country would work a lot better if certain groups of troublemakers would just shut up and accept their group’s traditional place in society.

We should respect homosexuals and feminists for their courage in challenging traditional family values. Life would be much better in our society if certain groups of people stopped badgering us with their views and simply accepted their traditional role in society.

Some of the best people in the country are those who openly criticize government or the church and pay no mind to what ''is normal'' and what ''everyone else does''.

Abortion, pornography, and unmarried cohabitation should be strongly condemned because they are violations of God's laws. The authorities need to put those radicals and immoral people who are trying to destroy our country for their own purposes back in their place. Women don't belong ''in the kitchen'' but wherever they want to be. The days when they had to obey their husbands and adhere to social conventions belong in the past. It would be better for our society to return Our country will be great if we honor the to the values of our forebears, listen to the ways of our forefathers, do what the established authorities, and quash the authorities tell us to do, and get rid of the opinions of the people who are “rotten apples” who are ruining everything. undermining our society.

Our data source is the Czech Society survey of January 2013. The Czech Society survey is a long-term research project that continuously charts Czech public opinion on a broad range of issues. Data collection was conducted by the Public Opinion Research Centre at the Institute of Sociology of the Czech Academy of Sciences. The survey was con ducted in January 2013 (fro m the 14th to the 21st of the month) on a representative quota sample of 1,032 respondents aged 15 and over, where the quotas were sex, age, education, size of place of residence, and NUTS -3 regions based on data from the 2011 Population and Housing Census of the Czech Republic. The RWA battery was included in a standardized omnibus -type questionnaire. For descriptive statistics see Appendix 1. A unique feature of our study is that we tested the RWA scale on a representative sample of the national population. All the studies employing one of the original versions of the scale that we had available to us were conducted on samples of populations of university students.

10 | JSRP

Johana Chylíkova, Martin Buchtík

Analysis: Introduction The aim of our analysis was to find out whether it is possible to confirm the one-dimensional structure of the RWA scale using Czech data, i.e. we were looking for one substantial factor consisting of both pro-authoritarian and anti-authoritarian items or two highly correlated substantial factors, the first consisting of pro-authoritarian items, the second of anti-authoritarian items. As an auxiliary tool to evaluate our data we analyzed the consistency of the scale based on Cronbach’s alpha, which Altemeyer (1996) employs as the main instrument for evaluating the quality of the RWA scale. As we have suggested earlier, we also took into consideration the possibility that the Czech RWA data were systematically affected by an acquiescent response style (ARS) bias. We tested for the presence of this method effect using confirmatory factor analysis. Modeling the Acquiescent Response Style Using CFA Before we present the results of the analysis, it is necessary to describe how a potential ARS bias can be modeled. Billiet & McClendon (2000) propose us ing CFA to model the acquiescent response style as a common factor, thus taking it as a latent variable that influences all the items that measure the substantial construct. Such an analysis requires that the data are from a balanced scale, i.e. a battery of items that are semantically balanced. A semantically balanced scale contains roughly the same number of items formulated in support of the measured construct and roughly the same number of items formulated against the construct. Why is this prerequisite important? An item formulated in support of the construct is obviously in logical contradiction with a contra construct item; when a respondent then agrees with both pro - and contra-construct items, this is a clear indicator that s/he is using an acquiescent response style. Modeling ARS using CFA makes use of such instances and can thus estimate the size of the ARS bias in the data. There exist several rules for modeling ARS as a latent variable. In general, to model the acquiescent response style as a factor it must be assumed that each substantial construct measured with items using the same response scale has one latent variable representing the ARS (Billiet & McClendon, 2000). When data reflect one (balanced) construct, there is one ARS latent variable, and when data reflect two (balanced) constructs, there are two ARS latent variables, and so on. Usually the factor that reflects the measured construct is called a “content” fa ctor, and the factor reflecting the response style bias is called a “style” factor. According to Billiet & McClendon (ibid), introducing the style factor into a model with just (a) content factor(s) must necessarily improve the fit of the model. There also exist guidelines for identifying the ARS bias as an ARS latent variable in the model. The ARS latent variable must have certain characteristics for us to be able to identify the modeled factor as a genuine ARS factor (Billiet & McClendon, 2000): 1. The ARS factor has a non-zero variance smaller than the variance of the content factor (ibid, p. 611), 2. The ARS factor correlates positively with age and correlates negatively with education (ibid, p. 621), 3. The ARS factor has a strong positive correlation with the sum-agree variable, which is the sum of cases in which the respondent responded in agreement with the items in the battery (ibid, p. 621). 4. Cambré, Welkenhuysen-Gybels & Billiet (2002, p. 4) specify an additional characteristic of the ARS factor: its factor loadings should be smaller than those of the content factor. This characteristic stems from the assumption that the content factor affects the scale items significantly more strongly than the ARS factor. Since the supposedly unidimensional RW A scale is balanced it is possible to use the analysis proposed by Billiet and McClendon. We modeled two factors, the first being the RWA latent variable (factor RWA), corresponding to the measured construct, and the second being the alleged acquiescent response style (factor ARS). Our aim in the analysis was to determine whether the second latent variable in the model is genuinely the acquiescent bias. If we can’t identify the alleged acquiescence factor in the CFA model as a genuine acquiescence factor, we

Validity of the construct of Right-Wing Authoritarianism …

11 | JSRP

may conclude that the Czech RWA data were not distorted by a response style bias and that we can trust the outcome of models with only content factors. Results In the following text we present the results of the scale consistency analysis with the Cro nbach´s alpha and the results of the CFA analyses in which we tested various models. Scale Consistency: Cronbach’s α We used a standard procedure for analyzing the consistency of the scale: Cronbach’s α, an indicator calculated from inter-item correlations.6 The Cronbach’s α we obtained for the 20-it em RWA scale was 0.78. Unfortunately we could not compare our results with those of Altemeyer, because the author of the RWA scale has only published such results for the 30-item scale. The values of the α coefficients based on samples of university student populations in the United States and Canada since the 1970s ranged from 0.85 to 0.9 and in isolated cases reached 0.95 (Altemeyer, 1996). Coefficients published by other researchers using a 30-item scale also ranged from 0.8 to 0.95 (Altemeyer, 1996). However, scholars claim the α coefficient for scale consistency is sensitive to the number of items in the scale, and Cortina (1993) states that if there are more than twenty items on a scale even a seemingly h igh coefficient of 0.7 may indicate that the consistency is not very strong. Given this information, it can justifiably be assumed that the 30-item scale generates more favorable α coefficients because it has more items. A more accurate comparison of the scale consistency in our data and the consistency reported by Altemeyer not distorted by the effect of the scale length can be made on the basis of the values of the mean inter-item correlation in the scale. In Altemeyer’s first studies, where he obtained α coefficients between 0.85 and 0.88, the mean inter-item correlation he gave for the thirty-item scale was 0.18 (Altemeyer, 1996). In our data, the mean inter-item correlation is 0.15. A difference in correlations of 0.03 is not too large, so it is our belief that our measurements were not fundamentally less consistent than those of Altemeyer. Analysis of the factorial structure To model the RWA data in a CFA we used an R statistical program with a complementary Lavaan package. We used the Satorra–Bentler Maximum Likelihood Method (MLM) to estimate parameters in all models, as it is better suited for non-normally distributed data7 than Maximum Likelihood (Kline, 2005). At first we employed the model with only one common factor corresponding to the RWA construct (Model 1), then we introduced the second factor representing the acquiescent response style (ARS) (Model 2). Having obtained the results of Model 2, we proceeded to specify Model 3. This model contains two content factors, the first composed of the pro-authoritarian items, the second with the anti-authoritarian items of the scale. Model 3 turned out to be the most appropriate and offered evidence about the factorial structure of Czech RWA data. 1. Model 1 6 Cronbach’s alpha is used whenever it is assumed that the items on the scale measure one construct (Cortina, 1993). George & Mallery (2003) claim that an α coefficient greater than 0.8 indicates strong scale consistency and a value greater than 0.7 can be regarded as acceptable. Because one-half of the items of the RWA scale (anti-authoritarian items) are formulated with the opposite meaning of the measured construct than the other half (pro-authoritarian items), we necessarily had to recode the response scale of the anti-authoritarian items D, F, H, I, K, M, O, R, T and U so that all the items would relate to the measured construct (authoritarianism) in the same direction. 7 T he data obtained from the RWA scale on a representative population of the Czech Republic were not normally distributed and this was confirmed by the results of a Kolmogorov-Smirnov test and a Shapiro-Wilk test. This abnormality was confirmed by histograms and by a normalized and detrended Q-Q plot.

12 | JSRP

Johana Chylíkova, Martin Buchtík

Model 1a contains one latent variable (ξ1 , called RWA) that influences all twenty variables measured by the RWA scale. Aside from the standard practice of assigning a factor loading of 1 to the first item, there are no other restrictions applied to the factor loadings. The residuals of the measure d variables are not allowed to correlate. The results obtained from this analysis show that this model does not fit the data (Model 1a: RMSEA = 0.115; p value RMSEA = 0.000; SRMR = 0.131; CFI = 0.459). We inspected the modification indices, which suggested considering a number of correlations between items. We added correlations between items C and N, D and H, D and K, D and U, K and T and R and T. The fit of the modified model (Model 1b) proved unsatisfactory (Model 1b: RMSEA = 0.093, p value of RMSEA = 0.000; SRMR = 0.117; CFI = 0.659). We concluded that the model with one common RWA factor does not correspond to our data even after modifications. 2. Model 2 In Model 2 we tested whether the existence of one factor reflecting the RWA latent variable could have been masked by the ARS bias. We introduced a second factor into Model 1 and considered it to represent the alleged ARS. We took all 20 items and put them into the CFA. Model 2a has the following parameters: 1. In the model, there are two latent variables ξ, one of which is assumed to represent the RWA construct (ξ1; this factor is called RWA), and the other represents the acquiescent response style (ξ2; this factor is called ARS, i.e. the acquiescent response style). 2. All the items in the model load to the RWA latent variable; besides the standard practice of assigning the first item a factor loading of 1 (item C in this specific case), the factor loadings are left free. The correlations between the residuals of the measured items are set at 0. 3. All the items in the model load to the ARS latent variable; a restriction of 1 was imposed on the factor loadings of all the items loading to this factor. In this respect, we follow the guidelines given by Billiet & McClendon (2000), who claim that all the items in a battery are equally susceptible to ARS bias. 4. The correlation between the two latent variables is set at 0 as there is no theoretical assumption that the RWA factor should correlate with the ARS factor. Model 2a did not converge, and the reason for this was presence of item D, which was discovered to be problematic even in our preliminary exploratory factor analyses. We excluded item D fro m the analysis and got a converged solution with an unsatisfactory model fit (Model 2b: RMSEA = 0,073; p value of RMSEA = 0.000; SRMR = 0.078; CFI = 0.776). We employed modificatio n indices, which strongly suggested excluding item P from the analysis and introducing a correlation between items C and N8 . These subsequent adjustments led to a more satisfying model fit (Model 2c: RMSEA = 0.056; p value of RMSEA = 0.015; SRMR = 0,053; CFI = 0,875): Nevertheless, we must point out that this fit is not excellent.

8 T he justification for this subsequent respecification is that the correlation between the residuals of items C and N indicates that these items share a certain portion of their variance.

Validity of the construct of Right-Wing Authoritarianism …

13 | JSRP

Picture 1: Model 2c Table 2: Confirmatory factor analysis (MLM Estimator) – models overview p value Satorra Bentler ML N. observations CFI RMSEA SRMR RMSEA Model 1a 792 0.459 0.115 0.000 0.131 Model 1b

792

0.659

0.093

0.000

0.117

Model 2a

NC

NC

NC

NC

NC

Model 2b

813

0.776

0.073

0.000

0.078

Model 2c

813

0.892

0.055

0.057

0.05

Model 3a

792

0.760

0.077

0.000

0.08

Model 3b

805

0.904

0.053

0.178

0.053

The parameter estimates from Model 2c help to determine whether the data are affected by the acquiescent response style. We compared the attributes of the ARS latent variable with the preconditions set by Billiet & McClendon (2000). 1. 2.

The variance of the ARS factor is lower in value than the variance of the RWA factor (var (RWA) = 0.675; var (ARS) = 0.488). The ARS factor shows a very strong correlation (r = 0.9; p= 0.000) with the sumagree variable9 - see Picture 2. This indicates that the factor is made up of agreement responses to contradictory statements.

9 T he sumagree variable was introduced as a latent variable in Model 2c – see Picture 2. The sumagree latent variable had one indicator - the sumagree variable that comprised the sum of instances in which the respondent indicated points 6, 7, 8, or 9 as his or her answer on the response scale. T he range of the value of the var iable was between 0 and 18, as there were 18 items in the model. The reported correlation 0,9 is the correlation between the ARS latent variable and the sumagree latent variable.

14 | JSRP

3.

4.

Johana Chylíkova, Martin Buchtík

The relationship to the other observed variables, age and education, is not confirmed . Our data show that the ARS factor does not correlate with age (r= -0.044; p = 0.262), nor is there any correlation between the ARS factor and education (r = 0.014; p = 0.720). 10 When evaluating the ARS factor it is beneficial to check the factor loadings from the RWA and ARS latent variables, as suggested by Cambré, Welkenhuysen -Gybels & Billiet (2002). Standardized loadings from the ARS factor are all positive, while the RWA factor is loaded by items positively or negatively, depending on whether the item is pro-construct or anti-construct (See Table 3). The exclusively positive direction of the factor loadings from the ARS factor indicates that it influences all the items in the same direction, regardless of whether they are formulated as pro - or anti-authoritarian. This characteristic of the factor loadings of the ARS factor shows that the variance of this latent variable is made up of agreement responses to contradictory items on the scale. An important finding is the size of the loadings fro m the ARS factor - they are very high and half of them are greater than the factor loadings from the RWA latent variable. This means that these items are influenced by the ARS factor more than by the RWA factor and that the alleged ARS factor is very strong, stronger than the RWA factor.

Picture 2: Model 2c with the sumagree latent variable Note: The structure of Model 2c with education as the latent variable and Model 2c with age as the latent variable is the same, the difference is that the sumagree latent variable and its indicator is replaced with education, i.e. age, as the latent variab les, and its indicator. We can see that our ARS factor conforms to the first and the second precondition, i.e. it has a non-zero variance, lower than the variance of the content factor, and it strongly correlates with the sumagree variable. But the precondition about the correlation with age and education is not 10 T he age and education variables were introduced as latent variables in Model 2c. Each of the latent variables had only one indicator; the education latent variable was composed of an ordinal variable, “education”, which had four categories; the age latent variable was composed of an interval variable, “age”. T he reported correlations -0.044 and 0.014 are the correlations between the ARS latent variable and the age and education latent variables.

Validity of the construct of Right-Wing Authoritarianism …

15 | JSRP

confirmed, and also the size of the ARS factor loadings is not in accordance with assumptions. From these results it follows that the alleged ARS factor apparently does not represent a genuine acquiescent response style. Given the fact that the alleged ARS factor is so strong , we interpret its variance as a conscious agreement with both pro-authoritarian and anti-authoritarian statements. If respondents agreed consciously, it is justified to think of Czech RWA data as a two-dimensional construct, where the first dimension comprises pro -authoritarian items and the second anti-authoritarian items. We test the existence of the two dimensions in our data in a subsequent analysis. Table 3: Confirmatory factor analysis; factor loading for Model 2c Model 2c λ Standardized item of ξ

ξ1 (RWA)

ξ2 (ARS)

C

λ 11 (0.369)

λ 12 (0.312)

E

λ 21 (0.262)

λ 22 (0.382)

F

λ 31 (-0.356)

λ 32 (0.369)

G

λ 41 (0.581)

λ 42 (0.320)

H

λ 51 (-0.394)

λ 52 (0.326)

I

λ 61 (-0.297)

λ 62 (0.364)

J

λ 71 (0.473)

λ 72 (0.321)

K

λ 81 (-0.442)

λ 82 (0.432)

L

λ 91 (0.499)

λ 92 (0.351)

M

λ 101 (-0.356)

λ 102 (0.391)

N

λ 111 (0.475)

λ 112 (0.330)

Q

λ 131 (0.185)

λ 132 (0.321)

R

λ 141 (-0.327)

λ 142 (0.373)

S

λ 151 (0.637)

λ 152 (0.339)

T

λ 161 (-0.339)

λ 162 (0.477)

U

λ 171 (-0.309)

λ 172 (0.337)

V

λ 181 (0.590)

λ 182 (0.326)

3. Model 3 We entered all twenty items into a model with two factors, ξ1 and ξ2 , the first comprising proauthoritarian items and the second anti-authoritarian items. Other than the standard fixation of the first item of the factors to 1, and allowing the correlation between the two factors, Model 3a contains no other specifications. The model’s fit is not ideal (Model 3a: RMSEA 0.077; p value of RMSEA= 0.000; SRMR = 0.080; CFI= 0.760); therefore, restrictions from the modification indices were applied and two items – P and O – were excluded from the analysis. According to the modification indices we set Model 3b with correlations between C and N, between J and L, between D and U, and between

16 | JSRP

Johana Chylíkova, Martin Buchtík

C and G,11 and the fit proved to be much better (Model 3b : RMSEA = 0.053, p value of RMSEA = 0.178; SRMR = 0.053; CFI = 0.904). Such a fit is very good so we may consider Model 3b to reflect the data well. If we look at the estimated parameters, we can see that the factor loadings of most of the items have a reasonable size (see Table 4) and that the Pearson correlation between substantial factors is -0.128; p = 0.010. The fit indices of Model 3b are the best of all the models in our analyses, thus we conclude that the Czech RWA data are best represented with a model with two correlated substantial factors, the first pro-authoritarian, and the second anti-authoritarian (liberal).

Picture 3: Model 3b Table 4: Confirmatory factor analysis; factor loadings for Model 3b Model 3b λ Standardized Item of ξ

ξ1 Authoritarian

C

0.558

E

0.428

F G

ξ2 Liberal

0.507 0.732

H

0.524

I

0.403

J

0.508

K L M

0.663 0.481 0.551

11 T here are practical justifications for the restrictions applied from the modification indices: the correlating items have very similar content (see the table with statements).

Validity of the construct of Right-Wing Authoritarianism …

N

0.629

Q

0.335

R S

0.511 0.714

T

0.594

U

0.398

V

17 | JSRP

0.639

Discussion If we look at the content of the individual pro- and anti-authoritarian statements in the RWA scale, we can clearly see a difference between them. Pro-authoritarian statements are generally expressive of support for a strong leader, a return to traditional values and morality, and the need for authority.12 The anti-authoritarian items, on the other hand, express tolerance of homosexuals, atheists, sexual liberalism, and rebellion against traditional customs. 13 The crucial finding in the final model (Model 3b) is that two factors, pro-authoritarian and anti-authoritarian (liberal), correlate very subtly (r=-0.128). Such a result contradicts what Altemeyer sets as a factorial property of the RWA scale. Altemeyer reports correlation between two content factors in his data between 0.4 – 0.7, but the correlation in our data is much lower, i.e. in the Czech data, the proauthoritarian factor has little relation to the liberal factor. We can illustrate the difference between our and Altemeyer’s findings with an example. Let’s take the liberal item D, “ Gays and lesbians are just as healthy and moral as anyone else”, and the pro-authoritarian item J, “Our society is going to fall apart unless we stop the attacks on our traditional values and morality ”. In Altemeyer’s data item D is placed on the liberal tail of the RWA dimension (i.e. in the liberal factor, which is highly correlated with the authoritarian factor), item J on the authoritarian tail (i.e. in the authoritarian factor, which is highly correlated with the liberal factor). A North American authoritarian respondent usually agrees with J, but disagrees with D. However; in the Czech cultural background, there is no single dimension with liberal and authoritarian tails, but two almost independent dimensions, and this is why one Czech respondent often agrees with both items D and J (see Appendix 1 for the frequencies of answers to the individual items of the scale). In general, from our results it follows that Czechs often demand traditional values, morality, good old values, and established authorities, but apparently for them this is not in contradiction with tolerance of homosexuality and nudity, religious freedom, and abortion rights. We can conclude that in the Czech cultural background the RWA construct comprises two different dimensions that are only subtly related. From this it follows that the RWA construct, conceptualized and operationalized in the RWA scale, cannot be validly measured in the Czech Republic. Arriving at such a conclusion, we must ask why there is no single dimension in the Czech RWA data, while it reportedly exists in North American data. There are two answers to this question, both relate to culturally conditional social norms, and the first is linked to the second. The first answer relates directly to social norms and refers to a cultural difference between the Czech Republic and North America; the second takes into account the design of the RWA scale itself. 1.

Although the RWA concept is, in theory, composed of three attitudinal clusters – authoritarian aggression, authoritarian submission and conventionalism - it is a

12 T he exception among the pro-authoritarian items is item P, which, however, did not correlate with any of the other pro-authoritarian items and was ultimately not included in the final model 13 It is important to note that the anti-authoritarian items do not reflect issues of immigration and racism. It is reasonable to suppose that inclusion of such items might change the overall results.

18 | JSRP

Johana Chylíkova, Martin Buchtík

bipolar and unidimensional construct, i.e. a single dimension with a liberal and an authoritarian tail. The RWA scale embraces the whole continuum, which is defined on the basis of social norms and attitudes – individual items of the scale are expressions of attitudes to social norms, order in society, government, and socially desirable or defective phenomena. Thus, in Altemeyer’s conceptualization rightwing authoritarianism is above all a matter of conventions, social norms, and attitudes, and this conceptualization firmly ties the concept to the social and cultural backgrounds of a given country. Originally the North American RWA scale managed to include those conventions and attitudes that split North Americans between rather authoritarian and rather non-authoritarian. The essential problem of the RWA scale in the Czech Republic is that it does not include those norms and attitudes that differentiate between Czech authoritarians and non-authoritarians. If we wanted to follow up on Altemeyer’s work in the Czech Republic, we would have to identify social norms and attitudes that have the potential to uncover authoritarianism. Unfortunately, at the present time we don’t have enough information to do so. The difference in conventions and social norms can perhaps , partly, be explained by the low level of traditional religiosity among the Czech population. Czech society is one of the most secular national populations in the world – only 16% of Czechs believe in God (Eurobarometer, 2010, p. 204), and generally only 20% of the population in the Czech Republic has trust in the church (Kunštát et al., 2014). The traditional religious dimension of authoritarianism consequently plays only a negligible role in Czech society, and this may have had a big influence on how respondents interpreted many of the items on the RWA scale. The fact that Christian morality is not widespread in society is probably a factor in the high tolerance shown towards sexual minorities, sexual liberalism, and atheism, which was manifested in support for pro-liberal items in the RWA scale. There is also evidence showing that in the European post-communist region different aspects (i.e. distributive justice and the communist economy) of authoritarianism are accentuated (Krauss, 2002) and traditional conservatism is s till considered rather non-conformist. 2.

The second possible answer to the question why there is no single dimension in the Czech RWA data also relates to social norms, but it interacts with the very design of the RWA scale. The pro-authoritarian and anti-authoritarian items of the scale unevenly reflect the three attitudinal clusters behind the RWA concept. This imbalance may have a crucial impact on the dimensionality of the data. Antiauthoritarian items largely reflect only conventionalism, and capture t he other two clusters – aggression and submission – only marginally. Because of the emphasis on conventionalism, the anti-authoritarian items are tied to social norms in a given society much more than pro-authoritarian items, which predominantly capture aggression and submission. Pro-authoritarian items, in particular, are indicative of a demand for a strong leader who will set the country in order and get rid of troublemakers – an expression of both aggression and submission. This imbalance in the scale results in a situation where on the one side of a bipolar RWA dimension there are predominantly social norms , and on the other side there are expressions of a demand for a strong leader and the protection of social norms. In our opinion such an imbalance is a methodological flaw that affects results when the RWA scale is used in a country where there does not exist a theoretically anticipated relation between the traditionally conservative type of social norms, and a demand for a strong leader and the protection of society. It seems that such a relation exists in North America, but apparently not in the Czech Republic. From the results we have obtained it seems that Czech Republic is a country where most of the public have liberal attitudes, but at the same time demand a strong leader and order in society.

Validity of the construct of Right-Wing Authoritarianism …

19 | JSRP

The Czechs’ support for pro-authoritarian statements that call for a strong leader and getting rid of troublemakers may be a reflection of the disappointment with the political culture in the country. Without going into detail here, we can say that Czech politics in 2013 was in the midst of a crisis, which is apparent from the fact that in January 2013 only 12% of the population expressed trust in the government, 11% trust in the lower chamber of Parliament, and 26% in the President (Kunštát et al., 2014). Because we have no longitudinal measure of the RWA scale available to us, it is impossible to determine whether the Czechs’ call for a strong hand and order in society was emphasized as a consequence of the current political crisis or whether it is a long-lasting attribute of Czech society. We conclude our contribution to the study of authoritarianism with the lesson that even a "stateof-the-art" measurement instrument must be tested for validity in a different cultural environment and that its quality must not be taken for granted. Regrettably, if we look at scholarly studies that have used the RWA scale in countries other than the USA and Canada, we must say that this does not usually occur. Acknowledgements In undertaking this research the authors gratefully acknowledge funding from two scientific projects: the first is the CSDA, The Czech Social Science Data Archive, sponsored by the Ministry of Education, Youth and Sports of the Czech Republic, reg.no.: LM2015060; the second is “The Understanding of Intergroup Differences in the Discursive Processes of Public Opinion”, sponsored by the Czech Science Foundation, reg.no.: 13-10320S. References 1.

Adorno, T. W., Frenkel-Brunswik, E., Levinson, D.J., & Sanford, R.N. (1950). The Authoritarian Personality. New York: Harper.

2.

Akrami, N., & Ekehammar, B. (2006). Right-Wing Authoritarianism and Social Dominance Orientation, Their Roots in Big-Five Personality Factors and Facets. Journal of Individual Differences. 27(3), pp. 35-62. https://doi.org/10.1027/16140001.27.3.117

3.

Altemeyer, B. (1981). Right-wing authoritarianism. Winnipeg: University of Manitoba Press.

4.

Altemeyer, B. (1996). The authoritarian specter. Cambridge: Cambridge University Press.

5.

Altemeyer, B. (2006). The Authoritarians. Winnipeg: University of Manitoba.

6.

Babbie, E. (2007). The Practice of Social Research. Belmont, CA:Thomson Wadsworth.

7.

Billiet, J. B., & McClendon, M.J. (2000). Modeling Acquiescence in Measurement Models for Two Balanced Sets of Items. Structural Equation Modeling, 7(4), pp. 608628. https://doi.org/10.1207/S15328007SEM 0704_5

8.

Billiet, J., Cambré, B., & Welkenhuysen-Gybels, J. (2002). Equivalence of measurement instruments for attitude variables in comparative surveys, taking method effects into account: the case of ethnocentrism. In A. Ferlingoj & A. Mrvar (Eds.) Developments in social science methodology (pp. 53-72), Ljubljana: FDV.

20 | JSRP

9.

Johana Chylíkova, Martin Buchtík

Cambré, B., Welkenhuysen-Gybels, J., & Billiet, J. (2002). Is it content or style? An evaluation of two competitive measurement models applied to a balanced set of ethnocentrism items. International Journal of Comparative Sociology, 43 (1), pp. 1-20. https://doi.org/10.1177/002071520204300101

10. Cohrs, C. J., Moschner, B., Maes, J., & Kielmann, S. (2005). The Motivational Bases of Right-Wing Authoritarianism and Social Dominance Orientation: Relations to Values and Attitudes in the Aftermath of September 11, 2001. Personality and Social Psychology Bulletin, 31(10), pp. 1425-1434. https://doi.org/10.1177/0146167205275614 11. Cortina, J. M. (1993). What is coefficient alpha? An examination of theory and applications. Journal of applied Psychology, 78(1), pp. 98-104. https://doi.org/10.1037/0021-9010.78.1.98 12. Duckitt, J. (1989). Authoritarianism and group identification: A new view of an old construct. Political Psychology, 10(1), pp. 63–84. https://doi.org/10.2307/3791588 13. Duckitt, J. (1993). Right-Wing Authoritarianism Among White South African Students: Its Measurement and Correlates. The Journal of Social Psychology, 133(4), pp. 553563. https://doi.org/10.1080/00224545.1993.9712181 14. Duckitt, J. (2001). A dual process cognitive-motivational theory of ideology and prejudice. Advances in Experimental Social Psychology, 33(1), pp. 44-113. https://doi.org/10.1016/s0065-2601(01)80004-6 15. Duckitt, J. (2006). Differential Effects of Right Wing Authoritarianism and Social Dominance Orientation on Outgroup Attitudes and Their Mediation by Threat Fro m and Competitiveness to Outgroups. Personality and Social Psychology Bulletin, 32(5), pp. 684-696. https://doi.org/10.1177/0146167205284282 16. Duckitt, J., & Sibley, C.G. (2007). Right Wing Authoritarianism, Social Dominance Orientation and the Dimensions of Generalized Prejudice. European Journal of Personality, 21(2), pp. 113-130. https://doi.org/10.1002/per.614 17. Duriez, B., & Van Hiel, A. (2002). The march of modern fascism. A comparison of social dominance orientation and authoritarianism. Personality and Individual Differences, 32(7), pp. 1199-1213. https://doi.org/10.1016/S0191-8869(01)00086-1 18. Duriez, B., Van Hiel, A., & Kossowska, M. (2005). Authoritarianism and Social Dominance in Western and Eastern Europe: The Importance of the Sociopolitical Context and of Political Interest and Involvement. Political Psychology, 26(2), pp. 299320. https://doi.org/10.1111/j.1467-9221.2005.00419.x 19. Eurobarometer (2010). Special Eurobarometer, biotechnology. Bruxelles: Eurostat. 20. Feldman, S. (2001). Authoritarianism. In N. J. Smelser & P. B. Baltes (Eds.). International Encyclopedia of the Social & Behavioral Sciences (pp. 971). Amsterdam: Elsevier. 21. Funke, F. (2005). The dimensionality of right-wing authoritarianism: Lessons from the dilemma between theory and measurement. Political Psychology, 26(2), pp. 195-218. https://doi.org/10.1111/j.1467-9221.2005.00415.x

Validity of the construct of Right-Wing Authoritarianism …

21 | JSRP

22. George, D., & Mallery, P. (2003). SPSS for Windows step by step: A simple guide and reference. 11.0 update (4th. ed.). Boston: Allyn & Bacon. 23. Gray, D., & Durrheim, K. (2006). The validity and reliability of measures of righ t-wing authoritarianism in South Africa. South African Journal of Psychology, 36(3), pp. 500520. https://doi.org/10.1177/008124630603600305 24. Groves, R. M. (2004). Survey Error and Survey Costs. New Jersey: Wiley. 25. Gul, F., & Ray, J. (1989). Pitfalls in Using the F Scale to Measure Authoritarianism in Accounting Research. Behavioral Research in Accounting, 1(1), pp. 182-192. 26. Horkheimer, M., Fromm, M., & Marcuse, H. (1936). Studien über Autorität und Familie [Studies on authority and family] . Paris: Felix Alcan. 27. Kemmelmeier, M., Burnstein, E., Krumov, K., Genkova, P., Kanagawa, C., Hirshberg, M.S., Erb, H., Wieczorkowska, G., & Noels, K. A. (2003). Individualism, collectivis m, and authoritarianism in seven societies. Journal of Cross-Cultural Psychology, 34(3), pp. 304-322. https://doi.org/10.1177/0022022103034003005 28. Kline, R. (2005). Principles and Practice of Structural Equation Modeling. New York: The Guilford Press. 29. Krauss, W. S. (2002). Romanian Authoritarianism 10 Years after Communis m. Personality and Social Psychology Bulletin, 28(9), pp. 1255-126 4. https://doi.org/10.1177/01461672022812010 30. Krosnick, J. A. (1991). Response strategies for copint with the cognitive demands of attitude measures in surveys. Applied Cognitive Psychology, 5(3), pp. 213–236. https://doi.org/10.1002/acp.2350050305 31. Kunštát, D., Červenka, J., Ďurďovič, M., Tabery, P., & Vinopal, J. (2014). 25 let české demokracie očima veřejnosti [25 years old Czech democracy in the public eye] . Prague: Academia. 32. Lambert, A., Burroughs, T., & Nguyen, T. (1999). Perceptions of Risk and the Buffering Hypothesis: The Role of Just World Beliefs and Right-Wing Authoritarianism. Personality and Social Psychology Bulletin, 25(6), pp. 643-656. https://doi.org/10.1177/0146167299025006001 33. Marcus, G. (Ed.) (2005). Political Psychology, https://doi.org/10.1111/j.1467-9221.2005.00414.x

26(2),

pp.

167-332.

34. Marsh, H.W. (1996). Positive and negative global self-esteem: A substantively meaningful distinction or artifactors?. Journal of Personality and Social Psychology, 70 (4), pp. 810-819. https://doi.org/10.1037/0022-3514.70.4.810 35. Mavor, K. I., Louis, W.R., & Sibley, C.G. (2010). A bias -corrected exploratory and confirmatory factor analysis of right-wing authoritarianism: Support for a three-factor structure. Personality and Individual Differences, 48(1), pp. 28–33. https://doi.org/10.1016/j.paid.2009.08.006 36. Mavor, K. I., Louis, W.R., & Laythe, B. (2011). Religion, prejudice, and authoritarianism: Is RWA a boon or bane to the psychology of religion?. Journal for

22 | JSRP

Johana Chylíkova, Martin Buchtík

the scientific study of religion, 50 (1), pp. 22-43. https://doi.org/10.1111/j.1468 5906.2010.01550.x 37. McClendon, M. J. (1991). Acquiescence and recency response-order effects in interview surveys. Sociological Methods and Research, 20 (1), pp. 60-103. https://doi.org/10.1177/0049124191020001003 38. Oesterreich, D. (2005). Flight into Security: A New Approach and Measure of the Authoritarian Personality. Political Psychology, 26 (2), pp. 275 - 297. https://doi.org/10.1111/j.1467-9221.2005.00418.x 39. Perry, R., Sibley, C.G., & Duckitt, J. (2013). Dangerous and competitive worldviews : A meta-analysis of their associations with Social Dominance Orientation and Right Wing Authoritarianism. Journal of Research in Personality, 47(1), pp. 116-127. https://doi.org/10.1016/j.jrp.2012.10.004 40. Rattazzi, A. M. M., Bobbio, A., & Canova, L. (2007). A short version of the Right Wing Authoritarianism (RWA) Scale. Personality and Individual Differences, 43(5), pp. 1223-1234. https://doi.org/10.1016/j.paid.2007.03.013 41. Ray, J. J. (1985). Defective validity in the Altemeyer authoritarianism scale. The Journal of Social Psychology, 125(2), pp. 271-272. https://doi.org/10.1080/00224545.1985.9922883 42. Stellmacher, J., & Petzel, T. (2005). Authoritarianism as a group phenomenon. Political Psychology, 26(2), pp. 245-274. https://doi.org/10.1111/j.1467-9221.2005.00417.x 43. Todosijevic, B. (2008). The structure of Political Attitudes in Hungary and Serbia. East European Politics & Societies, 22(4), pp. 879-900. https://doi.org/10.1177/0888325408319103 44. Tourangeau, R., Rips, L.J., & Rasinski, K. (2000). The Psychology of Survey Response. Cambridge: Cambridge University Press. 45. Van Hiel, A., & Kossowska, M. (2007). Contemporary attitudes and their ideological representation in Flanders (Belgium), Poland, and the Ukraine. International Journal of Psychology, 42(1), pp. 16-26. https://doi.org/10.1080/00207590500411443 46. Van Hiel, A., Duriez, B., & Kossowska, M. (2006). The Presence of Left -Wing Authoritarianism in Western Europe and Its Relationship with Conservative Ideology. Political Psychology, 27(5), pp. 769-793. https://doi.org/10.1111/j.1467-9221.2006.00532.x 47. Zakrisson, I. (2005). Construction of a short version of the Right-Wing Authoritarianism (RWA) scale. Personality and Individual Differences, 39(5), pp. 863872. https://doi.org/10.1016/j.paid.2005.02.026 48. ***The R Project for Statistical Computing, Retrieved from http://www.r-project.org/; 49. ***Latent Variable Analysis http://lavaan.ugent.be/ Appendix 1: Descriptive statistics item n valid

Mean

Variance

frequencies (valid answers; in %)

Validity of the construct of Right-Wing Authoritarianism …

Std. Deviation C

23 | JSRP

-4

-3

-2

-1

0

1

2

3

4

1003 998

1,01 -1,20

2,26 2,24

5,10 5,00

4,8 3,8

4,9 3,5

6,2 5,6

8,1 9,2

12,5 17,3 16,4 14,9 15,1 15,9 11,1 16,0 16,0 18,7

E

993

1,24

1,83

3,37

0,9

2,5

4,5

8,9

18,0 15,7 20,5 19,1 9,8

F*

1002

-2,31

1,86

3,48

1,0

1,2

2,6

4,4

9,1

G

966

0,77

2,21

4,88

4,9

4,5

9,4

8,1

15,2 14,2 19,6 13,9 10,4

H*

1004

-1,85

2,15

4,61

2,9

2,5

3,8

5,0

10,8 11,6 15,8 17,9 29,8

I*

996

-1,44

1,86

3,47

1,4

2,0

4,3

5,8

15,7 18,9 18,3 20,0 13,7

J

972

0,60

2,17

4,70

3,3

5,2

10,6 12,4 17,4 12,3 14,9 14,4 9,4

D*

K*

8,3

8,9

16,5 20,9 36,1

1018

-2,64

1,67

2,77

0,7

0,8

1,7

2,3

6,4

L

998

1,08

1,92

3,70

1,8

3,0

5,9

9,2

15,3 19,5 18,8 16,9 9,4

14,7 22,2 42,3

M*

996

-1,86

1,79

3,21

1,0

1,6

2,3

4,4

13,0 14,7 20,9 21,7 20,5

N

989

1,00

2,18

4,73

4,2

3,2

8,2

7,9

12,9 18,1 16,8 15,6 13,0

O*

3,1

11,7 11,2 20,0 16,9 14,9 11,3 6,7

978

-0,48

2,04

4,15

4,1

P

1003

-1,52

2,29

5,24

26,9 15,6 16,1 8,6

12,5 6,7

Q

974

1,25

2,15

4,61

3,9

3,7

4,5

15,8 14,3 19,6 15,9 16,3

R* S

1021 997

-2,38 0,35

1,88 2,09

3,55 4,36

1,4 4,3

1,7 5,8

2,7 2,4 7,4 9,2 16,7 20,2 38,3 10,4 12,1 19,7 16,1 14,7 10,1 6,6

T*

1027

-2,76

1,47

2,16

0,3

0,6

1,2

1,3

U*

989

-0,09

2,03

4,13

6,8

5,7

9,2

11,0 25,8 16,5 13,0 7,8

4,2

V

977

0,21

2,12

4,51

5,8

6,3

10,3 10,5 22,9 14,8 14,0 8,7

6,4

6,0

5,6

7,8

7,7

3,4

2,7

15,7 27,4 40,3

* anti-authoritarian items n total = 1032; n valid (listwise) = 792 Scale: -4 = strongly disagree with the statement; -3 = strongly disagree; -2 = moderately disagree; 0 = strictly neutral -1 = slightly disagree; +1 = slightly agree; +2 = moderately agree; +3 = strongly agree ; +4 = very strongly agree