Accuracy of National Stereotypes in Central Europe: Outgroups Are Not Better than Ingroup in Considering Personality Traits of Real People
Abstract
In a study on national stereotypes in central Europe—composed of Austria, the Czech Republic, Germany, Poland and Slovakia—2241 participants rated their autostereotype (a typical representative of their own country) and heterostereotypes (typical representatives of the other countries) by using National Character Survey (NCS). Existing data from 17377 participants including self‐reports or observer ratings on Revised NEO Personality Inventory and NCS were compared with the national autostereotypes and heterostereotypes. Although national autostereotypes converged with personality traits of real people in Poland and an adult subsample in the Czech Republic, national heterostereotypes did not correspond to personality traits of real people in any of the studied countries. National stereotypes were shared within as well as across countries. In heterostereotypes, raters from similar cultural backgrounds speaking similar languages agreed better as compared with raters from more distant cultures. Target country played a role in agreement of raters from different countries, showed in the highest convergence between autostereotypes and heterostereotypes of a typical German. Sharing of national stereotypes is influenced by political and economic significance of the target country. Although national autostereotypes clearly differentiated between typical representatives of central European countries, the comparison of personality profiles of their inhabitants showed remarkable resemblance. Copyright © 2013 European Association of Personality Psychology.
INTRODUCTION
Stereotypes can be defined as characteristics associated with a certain social category. Stereotypes help people organize and make sense of their social environment. However, they can be misleading if they do not correspond to reality. False stereotypes can contribute to ungrounded negative attitudes. Thus, it is important to study to what extent stereotypes give a real picture of characteristics distinctive of a given group. Stereotypical characteristics encompass typical physical features, behaviours or personality traits (Terracciano et al., 2005). Within personality psychology, stereotypes are conceived as personality traits typical of representatives of a certain social category (e.g. nation, gender and age).
In our research, we focus on stereotypes of typical country representatives in central Europe, composed of Austria, the Czech Republic, Germany, Poland and Slovakia. National stereotypes in all countries were measured from ingroup as well as outgroup perspectives. Ingroup ratings of personality traits typical of one's own country representatives are called national autostereotypes. Outgroup ratings of other countries representatives are called national heterostereotypes. The main aim of our study is to compare both national autostereotypes and heterostereotypes with personality traits of real people living in the given countries to estimate whether national stereotypes are accurate.
ACCURACY OF NATIONAL STEREOTYPES
The present study builds upon previous research examining the accuracy of national stereotypes (Allik, Mõttus, & Realo, 2010; Realo et al., 2009; Terracciano et al., 2005). The study by Terracciano et al. (2005) encompassing 49 cultures around the world found that national stereotypes do not reflect personality traits of real people. National stereotypes correlated with self‐reports only in two countries (Japan and Poland). To control for possible biases present in self‐reports, national stereotypes were additionally compared with another measure of real people, observer ratings. Nevertheless, observer ratings converged with national stereotypes solely in four countries (Australia, Lebanon, New Zealand and Poland). On the other hand, Realo et al. (2009) showed that national autostereotypes rated by respondents from the Baltic Sea region were moderately related to personality traits of real people. In all countries—Belarus, Estonia, Finland, Latvia, Lithuania and Poland—the relationship between national autostereotypes and personality traits of real people was positive; in half of the countries, statistically significant.
The main difference between the research projects by Terracciano et al. (2005) and Realo et al. (2009) lies in methods used for assessing national stereotypes and personality traits of real people. Whereas Terracciano et al. (2005) measured national stereotypes with a 30‐item National Character Survey (NCS) and personality traits of real people with a 240‐item Revised NEO Personality Inventory (NEO‐PI‐R; Costa & McCrae, 1992), Realo et al. (2009) and Allik et al. (2010) used the same method, the NCS, for both national stereotypes and real people ratings. Furthermore, in the Realo and colleagues' research (2009), the same participants completed both stereotype and real people ratings, which could have positively biased the found level of agreement.
Krueger and Wright (2006) suggested that the disagreement between national stereotypes and personality traits of real people in the study of Terracciano et al. (2005) may have resulted from ratings of national stereotypes only from an ingroup perspective. National stereotypes rated from an outgroup perspective can provide more accurate estimates of personality traits of people in the given country. Previous research indeed showed that the process of stereotyping is sensitive to self‐enhancement motives of individuals aiming to retain a positive image of one's ingroup and oneself (Fein & Spencer, 1997). In this way, other motivations can bias participants' judgments, especially judgement concerning their own country.
In order to test the prediction that stereotypes judged from the outside perspective are more accurate estimates of real people characteristic in comparison to autostereotypes, Realo et al. (2009) let participants from six countries neighbouring Russia rate a typical Russian. Although the participants from different countries had relatively similar views on a typical Russian, their ratings did not correspond with self‐reported personality traits of real Russians. Nevertheless, the evidence on the inaccuracy of national heterostereotypes concerned only one country with a dominant position in a specific geographic region. One of the goals of our study was to revisit the correspondence between ratings of national heterostereotypes and real people in different cultural settings.
CRITERIA FOR STEREOTYPE ACCURACY
A problematic point in research on stereotype accuracy represents the establishment of accuracy criteria (Operario & Fiske, 2001; Funder, 1999). In studies dealing with accuracy of national stereotypes, personality traits of real people living in the given country are used. The choice of the criterion generally involves five issues. First, it is necessary to decide what assessment of personality traits of real people should be used—self‐reports or observer ratings. For the sake of validity, both forms are often included (Terracciano et al., 2005). The second issue concerns within‐ or between‐participants designs of studies comparing stereotypical and real‐people ratings. The within‐participants design, when one group of respondents provides both types of ratings, can be distorted by the tendency to rate oneself more positively as compared with ratings of other people, the so‐called self‐enhancement bias (Kwan, John, Robins, & Kuang, 2008).
The third choice pertains to methods for ratings of stereotypical and real‐people characteristics as the agreement between them can be influenced by the correspondence of the scales. Although the NEO‐PI‐R, usually administered for ratings of real people, corresponds with NCS, it is composed of a different number of items with a different wording. Konstabel, Lönnqvist, and Walkowitz (2012) found moderate agreement between self‐reports provided on NCS and NEO‐PI‐R (ranging between .19 and .74, median = 0.55). If NCS and NEO‐PI‐R correlate only moderately when identical targets are rated, then an even lower level of agreement can be expected when using them for ratings of different targets (real people vs. stereotypes). In order to eliminate the difference between measuring instruments, the NCS can be used also for self‐reports of real people, meaning both stereotypical and real people profiles are inferred from the same method (Allik et al., 2010; Realo et al., 2009).
The sample composition represents a fourth decision to be made in studies on stereotype accuracy. Whereas ratings of national stereotypes can be well generalized across different groups of people, independent of their gender, age and place of residence (Hřebíčková & Kouřilová, 2012), ratings of real people differ with respect to participants' age (McCrae, Terracciano, & 78 members of the PPCP, 2005) and gender (Costa, Terraciano, & McCrae, 2001; Schmidt, Realo, Voracek, & Allik, 2008). Although McCrae, Terracciano, Realo, and Allik (2007) found no difference in self‐reports of participants from northern and southern Italy, Ashton (2007) presumes that a region of residence also plays a role in ratings of one's own personality as individuals compare themselves with people living close to them. An optimal choice of real people ratings seems to be the normative sample from the NEO‐PI‐R manual balanced with respect to respondents' age, gender, education and place of residence. Bearing in mind the conclusion of the previous paragraph, a methodologically optimal design would be to use both NEO‐PI‐R normative sample and NCS for self‐reports of real people.
The fifth issue involves the choice of a normative sample for transformation of raw scores to T‐scores once a profile‐correlation approach is applied. More acceptable is the use of international norms that are available for the ratings of national stereotypes along the scales of NCS (N = 3989 participants from 49 cultures; Terracciano et al., 2005) or NCS self‐reports (N = 1320 participants from six countries; Realo et al., 2009). However, the international norms for NEO‐PI‐R have not yet been published. For transformation of NEO‐PI‐R raw scores, the American norms are used (Terracciano et al., 2005), which can be considered ethnocentric. On the other hand, authors who directly compared NEO‐PI‐R data standardized along US and international norms reported that both standardizations led to very similar results (see footnote 4, p. 412, McCrae, Terracciano, & 79 members of the PPCP, 2005).
CONSISTENCY AND TEMPORAL STABILITY OF NATIONAL STEREOTYPES
The correspondence with reality is not the only aspect that can enrich the understanding of stereotypical beliefs. Even if traits of real people did not play role in formation of stereotypes, would it maintain that stereotypes are just ad hoc estimates or are they shared within or even across countries? In order to contribute to the answer, we wanted to examine the agreement of national stereotypes rated by different groups of participants (i) within one country and (ii) across different countries in the central European region.
In order to find out whether ratings of national stereotypes converge within particular countries, we tested their consistency and temporal stability. The consistency was operationalized as a similar perception of national stereotypes among participants that differ with respect to their age or place of residence. Previous research is composed of university students from seven regions in the Czech Republic showed a high level of agreement in autostereotype as well as heterostereotypes ratings among participants living in different regions, irrespective of their central or border position. On the other hand, the comparison of students and adults yielded a lower level of agreement (Hřebíčková & Kouřilová, 2012). With the reported results, the perception of national stereotypes seems to be influenced by participants' age rather than by their place of residence. However, Realo et al. (2009), who included students as well as working adults into their samples from Estonia and Latvia, found no significant difference between ratings of younger and older participants. The present study is composed of Czech participants of different age and, as such, can provide answers pertaining to characteristics underlying the consistency of national autostereotypes and heterostereotypes.
Another way how to estimate whether national stereotypes within a particular country change or persist is to compare their perceptions in different periods. Recent empirical findings suggest that national autostereotypes and heterostereotypes are rather stable over time. Respondents from different countries agreed in ratings of a typical American before and after the Iraq invasion (Terracciano & McCrae, 2007). Similarly, Realo et al. (2009) showed that Polish and Estonian autostereotypes did not dramatically change over 5 years. However, research comparing national and ethnic stereotypes over longer periods found changes in stereotype content (Ferjenčík, 2006; Madon et al., 2001). In order to contribute to the contentious debate on temporal stability of national stereotypes, we compared ratings of national autostereotypes in four out of five countries under study in an interval of 5 years. With the exception of Austria, all remaining countries from central Europe took part in an international project, Personality Profiles of Cultures, in which autostereotypes were collected in 2003 (Terracciano et al., 2005).
AGREEMENT BETWEEN NATIONAL STEREOTYPES
Having determined the agreement between ratings of national stereotypes within countries, the next goal of our research was to compare stereotypical beliefs across different countries in central Europe. We were interested in the correspondence between national heterostereotypes, that is, the outgroup perspectives concerning one particular country, as well as between national heterostereotypes and their corresponding autostereotypes. The level of agreement between ‘outgroupers’ can provide information about factors that determine sharing of stereotypes. Although recent studies have not brought evidence on the accuracy of national heterostereotypes, they found a good agreement among people from different countries in their views on characteristics of typical country representatives. A comparison of national heterostereotypes concerning a typical Russian rated by participants from six countries neighbouring Russia showed a high level of agreement (Realo et al., 2009). Also Boster and Maltseva (2006) found that Europeans from 15 locations across Europe shared similar views of each other's national character. Nations that were geographically close to each other were viewed similarly, especially if they were situated relatively far from the rater.
Previous studies comparing national heterostereotypes with corresponding autostereotypes revealed a similar pattern as the comparison of heterostereotypes. Ratings of a typical American from the ingroup perspective agreed with how a typical American was perceived by participants from 49 countries (Terracciano & McCrae, 2007). Also the view on a typical Russian converged among participants from countries neighbouring to Russia and Russians themselves (Realo et al., 2009). Because of the specificity of central Europe where two different language groups, Germanic and Slavic, divide the five countries into two distinct cultural entities, we could examine other factors besides geographical proximity that underlie the sharing of national stereotypes. As language is obviously a tool for stereotype transmission, we expect that participants from countries speaking similar languages will agree in their view on traits typical for representatives of central European countries better as compared with participants from more distant language groups.
As the last step in comparison of national stereotypes, we wanted to look into the differences in ratings of national autostereotypes. In order to examine whether potential differences in autostereotypes copy existing differences in real people, we first compared autostereotypical ratings from the five studied countries and then personality profiles of people living in the countries. As literature suggests that stereotypes are exaggerated beliefs (Operario & Fiske, 2001), we presume that profiles of personality traits of real people from different countries will resemble each other more than ratings of national autostereotypes.
CENTRAL EUROPEAN REGION
The central European region represents a fascinating place for a study of national stereotypes as it is composed of a number of countries with strong national identities neighbouring to each other in a relatively small and densely populated area. Between the countries examined in our research, many similarities and also differences exist that influence the mutual perception of their members. There is an obvious resemblance between Czechs, Poles and Slovaks in that they speak very similar languages from the Slavic branch of Indo‐European languages (Stone, 1990). Germans and Austrians speak the same language, belonging to Germanic branch of Indo‐European languages. Although linguistic similarities do not necessarily imply a cultural similarity (Bobrownicka, 1995), in case of the Czech Republic, Slovakia and Poland, similarities in languages are also accompanied by a common fate in both distant and more recent history. Their cultural heritage connects them to Western Europe (Huntington, 1996) despite the fact that after World War II they were all assigned to the sphere of influence of the former Soviet Union for more than 40 years. Austrians, Czechs and Slovaks also share a common past, all once belonging to the Austro‐Hungarian Empire that fell apart after the World War I. Furthermore, Czechs and Slovaks had lived in one state, the former Czechoslovakia, for most of the 20th century. A dominant role, not only in the central European region but also on a whole European scale, is reserved for the largest country being studied, Germany. Also Poland belongs among the most populous European nations, whereas the Czech Republic, Austria and Slovakia, with its populations of 10, 8, and 5 million inhabitants, respectively, epitomize rather small states. Germany is also a leading economical power in the region. Austria has, likewise, a long tradition of a free market; whereas, in the three post‐communist countries, an ongoing economic transformation is still taking place.
THE PRESENT STUDY
In the current study, participants from five central European countries rated personality characteristics typical of their country inhabitants (national autostereotype) as well as of inhabitants of the four other countries in the central European region (national heterostereotypes). The Czech subsample further is composed of working adults in order to compare the perception of national stereotypes between two different age groups. In our project, we considered all of the aforementioned issues pertaining to the choice of accuracy criteria. Unlike the study of Realo et al. (2009), where always the same respondents rated national stereotypes as well as real people traits, our sample also included different respondents rating stereotypes and real people (between‐participants design) to control for the self‐enhancement bias. Furthermore, we compared the ratings of national stereotypes not only with self‐reports but also with observer ratings of real people. For ratings of personality traits of real people, Czech, German and Polish normative samples from the NEO‐PI‐R manual were used (Hřebíčková, 2004; Ostendorf & Angleitner, 2004; Siuta, 2007). To our knowledge, no such NEO‐PI‐R normative sample exists in Slovakia or Austria. In the Czech subsample, in addition to the ratings of real people on the basis of the normative sample from the NEO‐PI‐R, a different group of participants provided self‐reports on the NCS.
In our research, we pursued several interconnected goals. We wanted to examine (i) the accuracy of national autostereotypes and heterostereotypes, comparing them with ratings of real people living in central Europe; (ii) consistency and temporal stability of national stereotypes within countries; and (iii) the agreement of national stereotypes across the five central European countries.
METHOD
Participants
Altogether, 2241 university students (75% women) from five central European countries (Austria, the Czech Republic, Germany, Poland and Slovakia) rated national autostereotypes and heterostereotypes. Students were recruited mostly from the universities in the border regions of the given countries and the Czech Republic. The Czech sample also is composed of 100 working adults.
Ratings of national stereotypes
The Austrian sample consisted of 396 students, age range 18–65 years (M = 25.02, SD = 7.20; 75% women). The Czech sample consisted of 726 students, age range 18–54 years (M = 23.16, SD = 4.98; 75% women) and 100 adults (M = 40.46, SD = 15.06; 57% women). The German sample consisted of 329 students, age range 18–63 years (M = 23.73, SD = 5.00; 70% women). The Polish sample consisted of 281 students, age range 17–53 years (M = 22.7, SD = 3.64; 86% women). The Slovak sample consisted of 509 university students, age range 16–66 years (M = 24.39, SD = 6.58; 76% women). All participants completed the NCS five times in order to provide their autostereotype and heterostereotypes concerning the four remaining countries under study.
Ratings of real people
For NEO‐PI‐R ratings of real people living in Austria, data were taken from a publication by McCrae (2002, pp. 120–125). Self‐reports were provided by 444 college‐age and adult people (66% women); further details on age of participants were not given. No observer ratings were available for the Austrian subsample.
For ratings of real people living in the Czech Republic, self‐reports and observer ratings on NEO‐PI‐R were taken from the Czech NEO‐PI‐R professional manual (Hřebíčková, 2004; p. 139 self‐report, p. 140 observer rating). Self‐reports were provided by 2288 people, age range 14–83 years (M = 26.04, SD = 12.21; 55% women). Observer ratings were provided by 812 people. Participants rated targets ranging from 15 to 81 years (M = 36.10, SD = 14.82; 59% women). Self‐reports on NCS were provided by 994 university students from all parts of the Czech Republic, age range 19–53 years (M = 23.95, SD = 5.63; 75% women).
For ratings of real people living in Germany, data were taken from the German NEO‐PI‐R professional manual (Ostendorf & Angleitner, 2004; p. 95 self‐report, p. 129 observer rating). Self‐reports were based on 11724 people, age range 16–91 years (M = 29.92, SD = 12.08; 64% women). Observer ratings were provided by 1547 people, age range 16–91 years (M = 33.70, SD = 12.81; 67% women).
For ratings of real people living in Poland, data were taken from the Polish NEO‐PI‐R professional manual (Siuta, 2007; p. 85). Self‐reports were provided by 324 people, age range 30–79 years (a mean age and standard deviation were not specified in the manual, 57% women). Norms for observer ratings were not included into the Polish NEO‐PI‐R manual.
For ratings of real people living in Slovakia, self‐reports were not available. Thus, we used observer ratings on NEO‐PI‐R provided by 238 Slovak university students, age range 18–23 years (M = 20.16, SD = 1.40; 50% women).11
We are grateful to Emília Ficková from the Institute of Experimental Psychology, Slovak Academy of Sciences, for providing the raw scores and SDs on 30 facets and five domains of NEO‐PI‐R observer ratings from Slovak participants.
Participants rated 240 targets (a mean age and standard deviation were not specified in the available materials; 50% women). A part of the Slovak data were already published in study by McCrae, Terracciano and 78 members of the PPCP (2005).
Procedure
The NCS (Terracciano et al., 2005) was used for the ratings of national autostereotypes and heterostereotypes. NCS consists of 30 bipolar items intended to parallel the facets of the NEO‐PI‐R (Costa & McCrae, 1992). For example, a facet of Neuroticism, N3: Depression, was assessed by asking how likely, on a 5‐point scale, a typical country representative is depressed, sad and pessimistic, versus content and optimistic. The NCS was already available in Czech, German, Polish and Slovak (Terracciano et al., 2005). The questionnaires were administered online. In each subsample, respondents first rated a typical member of their own nation and subsequently typical country representatives of the remaining four central European countries presented in a random order.
The NEO‐PI‐R (Costa & McCrae, 1992) is a 240‐item measure based on the five‐factor model. It contains 30 facet scales, six for each of the basic personality factors: Neuroticism, Extraversion, Openness to Experience, Agreeableness and Conscientiousness. Each of the 30 facets consists of eight items (see Figure 1 for the overview of NEO‐PI‐R facets). Responses are given on a 5‐point Likert‐type scale from strongly disagree to strongly agree. Form R for observer rating with items rephrased in the third person was used in the Czech, German and Slovak sample. Another group of Czech participants provided self‐reports on NCS (see section Ratings of real people).
Data analysis
In our study, we were interested in agreement across whole profiles of personality traits, not in agreement on particular traits. Any two profiles may be similar not only because their distinctive features are well matched but also because they both reflect an average profile. For example, when raters are inclined to endorse Neuroticism less then Extraversion, Openness, Agreeableness or Conscientiousness, independent of specific targets they rate, there are automatically intercorrelations among the rated profiles. These spurious correlations are caused by what Cronbach (1955) called ‘generalized other’. In order to eliminate this tendency, all NCS stereotype scores were converted into T‐scores (M = 50, SD = 10) using mean scores and SDs of national stereotype ratings from 3989 people in 49 different cultures (Terracciano et al., 2005). Standardized scores show how much a typical country representative was perceived to be above or below these international mean values. Self‐reports on NCS were standardized using international norms (Realo et al., 2009). Self‐reports and observer ratings on NEO‐PI‐R were standardized using American norms (Costa & McCrae, 2002).22
Two reviewers suggested to use international norms for standardization of self‐reports and observer ratings on NEO‐PI‐R. We have addressed Robert McCrae, a coauthor of the PPOC project, who confirmed that the international norms for NEO‐PI‐R self‐reports are not available. Separate international norms for observer ratings from PPOC project (McCrae, Terracciano and 79 members of the Personality Profiles of Cultures Project, 2005; N = 12.156) are available for men and women in two different age groups. An optimal way would be to standardize subsamples within their respective age and sex, and then to combine them. However, we could not standardize the raw scores this way because of the missing demographic data from real people.
Profile agreement was calculated as an intraclass correlation (ICC) across the 30 facets, using the double‐entry method (Griffin & Gonzales, 1995) with the p‐value on the basis of the non‐doubled n of 30. Double‐entry ICCs are similar to Pearson's correlations, but besides profiles' shapes, they are also sensitive to differences in profile elevation, and scatter and thus are more conservative than Pearson's r (McCrae, 2008). Recently, objections were raised against the use of ICCs in profile similarity research as they represent an omnibus index confounding elevation, scatter and shape (for more details, see Furr, 2010). The discussion about suitability of one or the other index for estimation of profile similarity has not reached unequivocal conclusion so far (McCrae et al., under review). In order to introduce the results of our study in a succinct manner given the large number of relationships that we tested, we decided to use ICCs only. Another reason for the choice of ICCs except for their conservativeness in estimation of profile similarity was that ICCs were also employed in prior studies on national stereotypes accuracy (e.g. Realo et al., 2009; Terracciano et al., 2005). This way, results from similar stream of research can be more easily compared.
Internal consistency
Analysis of the 2343 individual responses indicated Cronbach's α for the five scales of the NCS autostereotypes, heterostereotypes and self‐reports. The reliability coefficients are shown in Table 1. Reliability coefficients were acceptable for a six‐item scale with one exception, Neuroticism in the heterostereotype of a typical Pole.
| Internal consistency | Factorial agreement | ||||||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| N | E | O | A | C | Median | Explained variance % | N | E | O | A | C | ||
| Autostereotype (NCS) | 0.56 | 0.63 | 0.57 | 0.69 | 0.79 | 0.63 | 47.76 | 0.94 | 0.95 | 0.92 | 0.96 | 0.98 | |
| Au | 0.62 | 0.59 | 0.53 | 0.77 | 0.83 | 0.62 | 51.14 | 0.61 | 0.86 | 0.87 | 0.96 | 0.93 | |
| Cz | 0.56 | 0.71 | 0.61 | 0.70 | 0.71 | 0.70 | 48.25 | 0.88 | 0.95 | 0.92 | 0.97 | 0.98 | |
| Heterostereotypes (NCS) | Ge | 0.56 | 0.54 | 0.53 | 0.75 | 0.82 | 0.56 | 49.79 | 0.92 | 0.88 | 0.87 | 0.93 | 0.94 |
| Po | 0.47 | 0.70 | 0.56 | 0.78 | 0.72 | 0.70 | 49.83 | 0.91 | 0.94 | 0.94 | 0.97 | 0.98 | |
| Sk | 0.59 | 0.71 | 0.62 | 0.74 | 0.75 | 0.71 | 50.82 | 0.94 | 0.94 | 0.92 | 0.96 | 0.97 | |
| Median | 0.56 | 0.67 | 0.57 | 0.75 | 0.77 | ||||||||
| Self‐report (NCS) | 0.71 | 0.72 | 0.56 | 0.62 | 0.76 | 51.98 | 0.90 | 0.90 | 0.91 | 0.93 | 0.93 | ||
- Note: N, neuroticism; E, extraversion; O, openness to experience; A, agreeableness; C, conscientiousness; Au, Austria; Cz, Czech Republic; Ge, Germany; Po, Poland; Sk, Slovakia; NCS, National Character Survey.
Factor structure
A principal components analysis followed by a varimax normalized rotation was conducted on the 30 items in the entire sample of 2343 respondents for the (i) ingroup ratings, (ii) ratings of a typical Austrian, (iii) ratings of a typical Czech, (iv) ratings of a typical German, (v) ratings of a typical Pole, (vi) ratings of a typical Slovak and (vii) self‐reports. The percentage of explained variance in the five‐factor solution is given in Table 1. In order to examine how well the NCS factor solutions replicated the NCS structure found in previous research (Terracciano et al., 2005), an explanatory principal component analysis was followed by Procrustes targeted rotations and by computation of an index of factorial agreement across the cultural groups. The varimax normalized factor loadings were targeted towards the factor structure of the NCS ingroup ratings obtained from the international sample, composed of 49 different countries (Terracciano et al., 2005). After the target rotations were performed, the factorial agreement with the international data was estimated for all seven conditions. The mean coefficients of congruence between the five‐factor structures of the international data and our sample are shown on the right side of Table 1. All congruence coefficients for autostereotype and self‐report structures were above 0.90; somewhat lower coefficients of congruence were found for heterostereotypes. Among 20 congruence coefficients, six were below 0.90. But even these six lower coefficients of congruence do not suggest total randomness as the means of the distribution of factor congruencies on the basis of Procrustes rotations of randomly permuted data range from 0.32 to 0.34 for the five factors of personality (McCrae, Zonderman, Costa, Bond, & Paunonen, 1996).
Interjudge reliability
The reliability of single raters was calculated as ICC according to the following formula: [ICC(1,1) = (MSR − MSW)/[MSR + (k − 1) MSW]] and of the culture‐average ratings [ICC(1,k) = (MSR − MSW)/MSW], where k is the mean number of raters per culture, MSR is the mean square for facets and MSW is the mean square for residual sources of variance of an ANOVA model with cultures as the independent variable (Terracciano et al., 2005). Interjudge reliability between single raters showed a low agreement with regards to the 30 facets of national autostereotypes, with coefficients ranging from 0.01 (N6: Vulnerability) to 0.27 (C2: Order) with a median value of 0.09. The median is lower than the ones reported by Terracciano et al. (2005) and Realo et al. (2009) (.17 and 0.14, respectively). The lower values of interjudge reliability of single ratings of national autostereotypes determined in our study (median = 0.09) can indicate that the respective national autostereotypes in central European region resemble more than autostereotypes in Baltic Sea region or even around the globe. Another reason for the differences in interjudge reliability between the results of our and previous studies can be the higher number of raters per culture. As the average number of raters per culture in denominator increases, the ICC values decrease. Reliability coefficients for culture‐averaged autostereotypes facets ranged from 0.83 to 0.99, with a median value of 0.98; and reliability coefficients for domains from 0.95 to 0.99, with a median value of 0.98. The aggregation across a large sample of judges yields highly reliable values, indicating that the perception is shared by the group as a whole.
Interjudge reliability between single raters was even lower in the 30 facets of national heterostereotypes, with median values of 0.05 (Czech), 0.02 (German), 0.04 (Austrian), 0.03 (Slovak) and 0.01 (Polish). The reliability of the culture‐averaged ratings of the 30 facets of national heterostereotypes (given as median values) was 0.94 (Czech), 0.90 (German), 0.93 (Austrian), 0.90 (Slovak) and 0.77 (Polish), again showing a good agreement in aggregated samples.
RESULTS
Accuracy of national stereotypes
The main goal of our research was to compare national stereotypes rated from ingroup and outgroup perspectives with personality traits of real people living in the countries under study. To assess personality traits of real people, we employed self‐reports and observer ratings on NEO‐PI‐R and, in the Czech subsample, additionally, self‐reports on NCS. We computed two kinds of agreement: ICCs between national autostereotypes and ratings of real people represented by self‐reports and observer ratings (when available) established autostereotype accuracy. To determine the heterostereotype accuracy, we computed ICCs between ratings of real people and heterostereotypes from the respective countries. Results are given in Table 2. ICCs indicating the level of autostereotype accuracy are printed in bold. ICCs indicating the level of heterostereotype accuracy are printed in italics.
| Real people | Autostereotypes | Heterostereotypes according to | ||||||||
|---|---|---|---|---|---|---|---|---|---|---|
| NEO (R) | NEO (S) | NCS (S) | NCS | Austrians | Czechs | Germans | Poles | Slovaks | ||
| Real Austrians | NEO‐PI‐R (S) | −0.17 | ||||||||
| Austrian heterostereotype according to | Czechs | −0.32 | 0.50****
p < .01; |
0.48****
p < .01; |
0.92******
p < .001. |
0.91******
p < .001. |
||||
| Germans | −0.18 | −0.07 | 0.35**
p < .05; |
0.31 | ||||||
| Poles | −0.31 | 0.32 | 0.95******
p < .001. |
|||||||
| Slovaks | −0.19 | 0.19 | ||||||||
| Real Czechs | NEO‐PI‐R (S) | 0.81******
p < .001. |
0.04 | 0.31 | ||||||
| NEO‐PI‐R (R) | 0.14 | 0.28 | ||||||||
| NCS (S) | 0.06 | |||||||||
| Czech heterostereotype according to | Austrians | −0.01 | −0.02 | −0.01 | −0.20 | 0.86******
p < .001. |
0.53****
p < .01; |
0.08 | ||
| Germans | 0.11 | 0.09 | −0.04 | −0.29 | 0.63******
p < .001. |
0.17 | ||||
| Pole | −0.26 | −0.22 | −0.11 | −0.50****
p < .01; |
0.50****
p < .01; |
|||||
| Slovaks | −0.37**
p < .05; |
−0.12 | −0.41**
p < .05; |
−0.60******
p < .001. |
||||||
| Real Germans | NEO‐PI‐R (S) | 0.62******
p < .001. |
−0.32 | |||||||
| NEO‐PI‐R (R) | −0.41**
p < .05; |
|||||||||
| German heterostereotype according to | Austrians | −0.48****
p < .01; |
−0.32 | 0.87******
p < .001. |
0.84******
p < .001. |
0.82******
p < .001. |
0.82******
p < .001. |
|||
| Czechs | −0.43**
p < .05; |
−0.25 | 0.73******
p < .001. |
0.95******
p < .001. |
0.95******
p < .001. |
|||||
| Poles | −0.46****
p < .01; |
−0.33 | 0.76******
p < .001. |
0.95******
p < .001. |
||||||
| Slovaks | −0.41**
p < .05; |
−0.19 | 0.72******
p < .001. |
|||||||
| Real Poles | NEO‐PI‐R (S) | 0.61******
p < .001. |
||||||||
| Polish heterostereotype according to | Austrians | 0.23 | 0.12 | 0.68******
p < .001. |
0.90******
p < .001. |
0.51****
p < .01; |
||||
| Czechs | 0.25 | 0.07 | 0.68******
p < .001. |
0.53****
p < .01; |
||||||
| Germans | 0.21 | −0.02 | 0.70******
p < .001. |
|||||||
| Slovaks | −0.07 | −0.31 | ||||||||
| Real Slovaks | NEO‐PI‐R (R) | −0.08 | ||||||||
| Slovak heterostereotype according to | Austrians | 0.13 | 0.01 | 0.32 | 0.98******
p < .001. |
0.66******
p < .001. |
||||
| Czechs | 0.00 | 0.00 | 0.34 | 0.68******
p < .001. |
||||||
| Germans | 0.13 | 0.02 | 0.68******
p < .001. |
|||||||
| Poles | −0.10 | 0.02 | ||||||||
- Note: All stereotype ratings (autostereotypes and heterostereotypes) were standardized using international norms (Terracciano et al., 2005). NEO‐PI‐R self‐reports: Czechs from Hřebíčková, 2004, Table 16, p. 78, N = 2288; Poles from Siuta, Table 10, p. 58, N = 603; Austrians from McCrae, 2002, Appendix 1, s. 120, N = 444; Germans from Ostendorf, Angleitner, 2004, Table 19, p. 95, N = 11724. Self‐reports were standardized using American normative data (Costa & McCrae, 1992, Appendix B, Table B‐1, p. 75, N = 1000). NEO‐PI‐R observer ratings: Czechs from Hřebíčková, 2004, Table 16, p. 78, N = 812; Slovaks from Ficková, unpublished manuscript, N = 238; German from Ostendorf & Angleitner, 2004, Table 43, p. 129, N = 1547. Observer ratings were standardized using American normative data (Costa, McCrae, 1992, Table B‐2, p. 76, N = 300). NCS self‐reports in the Czech sample were standardized using international norms (Realo et al., 2009). NEO (R), NEO‐PI‐R Form R; NEO (S), NEO‐PI‐R Form S; NCS, National Character Survey; NCS (S) National Character Survey used for self‐report. The level of autostereotype accuracy is in bold; the level of heterostereotype accuracy is in italics.
- * p < .05;
- ** p < .01;
- *** p < .001.
In Poland, Slovakia and Austria, only one measure of real people's personality traits was available. Thus, from these countries, only one indicator of autostereotype accuracy could be counted. In Germany and the Czech Republic where both self‐reports and observer ratings on NEO‐PI‐R exist, we obtained two indicators of autostereotype accuracy. In the Czech subsample, another group of respondents provided self‐report on NCS leading to three indexes of real people. Because of similarities between stereotypical ratings provided by Czech university students and working adults (median = .77), we state data only from university students in Table 2 so that the statistics are comparable across student samples from different countries. Notable differences between Czech students and adults are commented on in the text.
Across all subsamples of our study, eight ICCs specifying autostereotype accuracy in five central European countries ranged from −.41 to .61 (median = −.01). We found a moderate agreement between the Polish autostereotype and NEO‐PI‐R self‐reports on the basis of the Polish normative sample (ICC = .61, p < .001). Figure 1 illustrates the convergence between national autostereotypes and self‐reports of real Poles. Polish respondents described themselves as well as a typical Pole as high in Neuroticism, low in Agreeableness and average in Extraversion, Openness to Experience, and Conscientiousness.
The autostereotype of a typical Czech rated by adults correlated .13 (n.s.) with self‐reports, .16 (n.s.) with observer ratings both provided on NEO‐PI‐R and .51 (p < .01) with self‐reports provided on NCS, with a median value of .16. The three ICCs in the Czech subsample of university students ranged from .06 to .31, with a median value of .28 (Table 2). In Slovakia, Austria and Germany, all ICCs were negative.
Concerning the heterostereotype accuracy, across all subsamples of our study, 32 ICCs ranged from −.48 to .25 (median = −.15), showing that national heterostereotypes do not reflect personality traits of real people in central Europe. Median values of ICCs indicating the heterostereotype accuracy were −.25 for the heterostereotype of a typical Austrian, −.07 for the heterostereotype of a typical Czech, −.37 for the heterostereotype of a typical German, .22 for the heterostereotype of a typical Pole and .06 for the heterostereotype of a typical Slovak.
Consistency and temporal stability of national stereotypes
To answer whether national stereotypes are perceived the same way by different participants within one country, we examined their consistency. In the Czech subsample, two different groups—students and adults—rated national autostereotypes and heterostereotypes. The comparison of younger and older adults revealed the following levels of agreement: ICC for autostereotype = .61; Austrian heterostereotype = .92; German heterostereotype = .98; Polish heterostereotype = .60; and Slovak heterostereotype = .87 (all ps < .001). The correlations indicated a high level of agreement between Czech university students and adults in their perception of the Germanic neighbours and Slovaks. Participants' education could be another factor influencing ratings of national stereotypes. To test its effect, we divided Czech adults into two subgroups according to the level of completed education (58% with basic and high school education and 42% with university degree). The agreement between these two groups in ratings of national autostereotypes was .86; in ratings of national heterostereotypes, it ranged from .95 (a typical German) to .67 (a typical Slovak) (median = .80), showing similarities between ratings of people with different education backgrounds.
Furthermore, to determine whether stereotypical beliefs endure in time, we tested a temporal stability of national stereotypes across different periods in four out of the five countries studied. The ICCs between the profiles of national autostereotypes on the basis of the data from 2003 and 2008 were .87 for Poland, .82 for the Czech Republic, .81 for Germany and .78 for Slovakia (all ps < .001). Our results point to the fact that the perception of national autostereotypes in these four central European countries is relatively stable and has not changed considerably during the last 5 years. Additionally, we tested the temporal stability of Slovak national heterostereotypes because Czech participants rated a typical Slovak also in 2003. ICC between Slovak heterostereotypes from 2003 and 2008 was .70, (p < .001), indicating an agreement in perception of a typical Slovak by Czech raters at different points of time.
Agreement between national stereotypes
In order to verify whether national stereotypes are shared across different countries, we compared national heterostereotypes, national autostereotypes with their corresponding heterostereotypes and, lastly, national autostereotypes rated by participants living in different countries.
Agreement between national heterostereotypes
The comparison between heterostereotypes of five central European countries resulted in 25, out of a total of 30, statistically significant profile correlations, providing evidence on a convergence of outgroup perceptions. The highest level of agreement was found in ratings of a typical German (median = .90; see the right side of Table 2). A high level of agreement was also found in ratings of Austrian heterostereotypes. Five out of six ICCs were statistically significant, ranging from .31 (German vs. Slovak raters) to .95 (Polish vs. Slovak raters) (median = .70). Participants from the three Slavic countries agreed more closely in ratings of a typical Austrian when compared with German participants who perceived Austrians in a different way.
Regarding Slavic heterostereotypes, the highest level of agreement was found in ratings of a typical Pole. All ICCs were statistically significant, ranging from .51 to .90 (median = .67). Austrians and Germans reached high level of agreement in their view on a typical Pole (ICC = .90), Slovak (ICC = .98) and Czech (ICC = .86). In heterostereotype ratings of a typical Slovak, four out of six ICCs reached statistical significance. Czech participants did not agree with German and Austrian ratings of a Slovak heterostereotype. In a similar manner, Slovaks differed from Austrians and Germans in ratings of the Czech heterostereotype. Regarding both Czech and Slovak heterostereotypes, we found no correspondence between Czechs or Slovaks and their Germanic neighbours. Slovaks rated Czechs differently from Germanic nations particularly on traits corresponding to dominance—high on E3: Assertiveness and E4: Activity, and low on A4: Compliance and A5: Modesty.
Agreement between national autostereotypes and heterostereotypes
Ingroup and outgroup perception of national stereotypes converged only in case of a typical German (median = .75). A moderate agreement was found between Austrian autostereotype and heterostereotype rated by Czech university students. All other ICCs were negative or did not reach statistical significance. Results of the agreement between national autostereotypes and heterostereotypes in the five countries are shown in the middle of Table 2.33
In the subsample of Czech, working adults ICC ranged from −.43 till −.01(median = −.19).
Agreement between national autostereotypes
Ten ICCs on the basis of autostereotype ratings from the five central European countries belonging to two different language groups ranged from −.03 to .62 (median = .13). A significant agreement was found between Czechs and Poles (ICC = .62, p < .001), Austrians and Germans (ICC = .55, p < .01), Czechs and Austrians (ICC = .50, p < .01) and Austrians and Poles (ICC = .45, p < .05). Interestingly, Slovak autostereotype did not correlate with any of the autostereotypes from the four central European countries (median = .04). We found no convergence of German autostereotype with either Czech (ICC = .07, p = n.s.) or Polish autostereotypes (ICC = .19, p = n.s.). The country with the highest agreement with autostereotypes of the remaining countries was Austria (median = .48), followed by Poland (median = .32), the Czech Republic (median = .29), Germany (median = .13) and, lastly, Slovakia (median = .04). Figure 2(a and b) shows profiles of autostereotypes separately for (a) Slavic and (b) Germanic countries.
In order to estimate whether differences in autostereotypical ratings are based on real differences, we compared them with profiles of personality traits of real people living in the given countries. In line with our hypothesis, the ratings of real people were considerably more similar than ratings of national autostereotypes. The comparison of self‐reports and observer ratings in the five countries yielded eight ICCs, ranging from .33 to .90 (median = .68). We found statistically significant correlations between real Slovaks and both Czechs (ICC = .76, p < .001) and Germans (ICC = .60, p < .001). Furthermore, ratings of real Germans also resembled real Austrians (ICC = .90, p < .001), as well as real Czechs and Poles (both ICC = .71, p < .001).
DISCUSSION
In our research, we pursued several interconnected goals pertaining to national stereotypes in central Europe. The first goal of our research was to compare national autostereotypes and heterostereotypes with personality traits of real people living in the countries, in which we conducted our study. Second, we examined consistency of stereotypical ratings across different groups of participants, and temporal stability of national stereotypes within particular countries. Finally, we tested the agreement between national stereotypes across the five countries to find a possible mechanism of stereotype formation and sharing.
Accuracy of national stereotypes
In order to determine the accuracy of national stereotypes, we compared profiles of stereotypical characteristics rated from ingroup and outgroup perspectives with profiles of personality traits of real people. Regarding national autostereotypes, we found a significant agreement with personality traits of real people living in the central European region only in Poland and the Czech subsample of working adults. The agreement between ratings of Polish autostereotype and real Poles had been previously described in two other independent studies (Terracciano et al., 2005; Realo et al., 2009). Polish autostereotype copies the ratings of real Poles that are high on Neuroticism, low on Agreeableness and average on the remaining three dimensions of the five‐factor model. Polish participants ascribed to their typical country representative a high level of Neuroticism as compared with 48 countries around the globe (McCrae et al., 2007). There is a clear norm of spontaneity in Poland when it comes to expressing emotions. Wierzbicka (1994) suggested that in Polish culture, emotional spontaneity is valued higher than a desire to make someone else feel good, thus encouraging an overt display of both good and bad emotions. However, it indeed seems that bad feelings are expressed more readily. In a study by Dolinski (1996), Polish participants defined their mood as worse than usual in contrast to Americans who felt better than usual. Disagreement, disapproval or irritation could be expressed quite openly, which might shock foreigners from more ‘agreeable’ cultures. As Ronowicz (1995, p. 80) put it: ‘in Poland, an argument is not only considered a good way of exchanging ideas, but also an enjoyable form of conversation’.
In the Czech subsample, we found a moderate agreement between Czech autostereotype rated by working adults and personality traits of real people, both rated on NCS. On the other hand, the comparison of real people and Czech national autostereotype rated by university students yielded no convergence, although the same method was again used for both ratings. Adult participants, as compared with the student subsample, can probably take a better account of real people's traits when judging stereotypical characteristics of a typical Czech. The same pattern emerged in Estonia where profiles of real people converged with national autostereotype rated by adults but not by university students. However, in Latvia, national autostereotype did not agree with ratings of real people in either adult or student subsamples (Realo et al., 2009). Other countries so far included into studies on the accuracy of national autostereotypes could not provide an answer about the role of age differences in the ability to consider characteristics of real people because the samples are always composed of university students only.
As already emphasized in the Introduction, not only the age of respondents but also the method used for ratings of real people can influence the magnitude of agreement between stereotype and real people profiles. In the Baltic Sea project, all ratings were made using the same instrument—the NCS. National autostereotypes were moderately related to personality traits of real people, suggesting a possible effect of the method (Realo et al., 2009). Also our data showed possible problems with method equivalency. Whereas NEO‐PI‐R self‐reports and other reports correlate as high as .81 and .62 in the Czech and German subsamples, the correlations between the NCS self‐reports and NEO‐PI‐R self‐reports and observer ratings in the Czech subsample were almost near zero (ICC = .04 and .14, respectively). This shows that the NEO‐PI‐R and NCS (when used for self‐reports) do not exactly measure the same thing (Konstabel et al., 2012). On the other hand, different methods used for ratings of a typical Pole and real people living in Poland consistently lead to a significant agreement between stereotype and real people traits (Realo et al., 2009; Terracciano et al., 2005).
In the Czech subsample of our research, we used both NCS and NEO‐PI‐R for ratings of real people in order to determine the effect of method on the level of agreement. Personality traits of real people rated using NCS correlated with the Czech autostereotype rated using NCS by working adults but not by university students, indicating the effect of age. Yet autostereotypical ratings by adults did not agree with ratings of real people by using the NEO‐PI‐R, indicating the effect of method. It seems that the same method used for ratings of real people and national stereotypes can contribute to the level of agreement between them, qualified by respondents' age. However, in our results, the effect of age and method combine, and thus, it is difficult to disentangle the two factors that influence the agreement between ratings of national stereotypes and real people. Broadly speaking, caution is needed when generalizing the findings on the accuracy of national autostereotypes measured within one age group of respondents and one method for ratings of stereotypical and real people's traits.
The comparison of national autostereotypes with characteristics of real people suggests that the agreement between self‐reported personality traits and perception of a typical ingroup member is a variable in itself. Its magnitude can vary across nations from negative, close to zero to highly positive as previously concluded by Realo et al. (2009). This way, Polish respondents seem to be more capable of considering the characteristics of real people as a high level of agreement between Polish autostereotypes and self‐reports was found in our study and in two other studies (Realo et al., 2009; Terracciano et al., 2005). The relationship between national autostereotypes and characteristics of real people in the Czech Republic was positive but non‐significant, and in Germany even negative, as found in Terracciano's et al. (2005) study as well as our study. Although autostereotypes agreed with real people ratings at least in Poland and the Czech Republic, we found no significant convergence between real people and national heterostereotypes. In the two Germanic countries, heterostereotypes correlated with real people personality traits negatively. Similar results about moderate correspondence of real people ratings with national autostereotypes, and low or no correspondence with heterostereotypes were reported by Realo et al. (2009). Although members of the six countries in the Baltic Sea region shared their views on a typical Russian, none of the heterostereotypes agreed with personality traits of real Russians. One of the most interesting findings of our study is that the outgroup perspective does not provide a more accurate, unbiased view on characteristics of people living in given countries as compared with the ingroup perspective.
Consistency and temporal stability of national stereotypes
Other goals of our study involved the consistency, temporal stability and agreement of national stereotypes. We tested whether stereotypes are just ad hoc estimates or whether they are shared across different groups of participants in one country or even across countries. We found a high level of agreement between national stereotypes rated by Czech university students and adults. Our results support previous findings on a minor role of age differences in stereotype ratings from Estonia and Latvia (Realo et al., 2009). However, the level of agreement between Czech younger and older adults partially depended on the target of their ratings. The two age groups agreed more on heterostereotypes concerning the four neighbouring countries as compared with the Czech autostereotype. Czech university students perceived a typical representative of their country in a less socially desirable way than adults. We also examined a possible effect of completed education on the ratings of national stereotypes. Czech adults with different education backgrounds saw autostereotypes and heterostereotypes in a similar way. In heterostereotypes, the level of agreement between students and adults was also influenced by the target country. The highest level of agreement was found in their ratings of a typical German. Adults who had spent a much longer part of their lives behind the iron curtain and in a joint state of Czechoslovakia, during which the attention was focused on ‘brotherly Slavic nations’, agreed with younger participants less on Slavic stereotypes than on stereotypes of Germanic nations. An obvious limitation of this finding is the presence of two age groups only in one of the five countries under study.
The comparison between national autostereotypes rated in 2003 and 2008 showed a relatively high temporal stability in Poland, the Czech Republic, Germany and Slovakia. Previous cross‐cultural studies within the five‐factor model also reported that Polish, Estonian (Realo et al., 2009) and Slovak (Kouřilová & Hřebíčková, 2011) autostereotypes did not dramatically change over 5 years' time. Research suggests that even important socio‐political events leave national stereotypes relatively unchanged. Perception of a typical American by respondents from different countries, including Lebanon, was similar before and after the Iraq invasion (Terracciano & McCrae, 2007). However, other studies brought evidence that, over longer periods, the content of stereotypical beliefs changes as a result of shifts in social norms (Ferjenčík, 2006; Madon et al., 2001). Ferjenčík (2006) found differences in an ingroup perception of typical Slovaks and outgroup perceptions of typical Czechs, Poles, Germans and Hungarians when comparing ratings from 1994 and 2004. In a similar manner, Madon et al. (2001), comparing content of national stereotypes in four studies across 60 years, concluded that stereotypical beliefs change in time. The representations of the studied nationalities were, in both quoted researches, more positive and less ethnocentric in the later period as compared with the previous data sampling.
It seems that content of national stereotypes changes in time—simply, the period over which stereotypes are measured must be long enough to capture the shift in stereotypical beliefs. Five years or even less used as a time frame for the assessment of temporal stability in the earlier cited studies (Kouřilová & Hřebíčková, 2011; Realo et al., 2009; Terracciano & McCrae, 2007) and our research is presumably not long enough for a change to take place. Furthermore, both Madon et al., (2001) and Ferjenčík (2006) included a broader variety of characteristics pertaining to national stereotypes, not just personality traits. As Madon et al. (2001, Study 1) showed, the putative stability of stereotypes can be caused by a limited choice of stereotype attributes. Thus, the stability found in our and previous studies that employed five‐factor model can be further bolstered by the fact that participants could rate only personality traits. However, stereotypes are also composed of other characteristic, such as behaviours or preferences, that can be more sensitive measures of shifts in stereotypical beliefs as compared with personality descriptors.
Agreement between national stereotypes
The previous section discussed a high level of agreement in stereotypical ratings within countries, indicating that stereotypes are not ad hoc estimates but shared beliefs. In order to determine whether sharing of national stereotypes crosses country boundaries, we compared ratings of national heterostereotypes, ratings of national autostereotypes with their corresponding heterostereotypes and, lastly, national autostereotypes rated by participants living in various parts of central Europe.
The comparison of national heterostereotypes showed a high level of agreement among participants from different countries. However, the level of agreement depended on the target country. A typical German was perceived virtually identical by participants from all countries under study, whereas a lower level of agreement was found in case of the other typical country representatives. Our findings also indicated a role of participants' language and cultural background with a higher level of agreement between participants belonging to the same language group. Austrian and German ratings of Czech, Polish and Slovak heterostereotypes were more similar than Czech, Polish and Slovak stereotype ratings of these Slavic countries. In a similar vein, when rating a typical Austrian, participants speaking Slavic languages agreed more as compared with ratings by German participants. People speaking Slavic languages perceived their Germanic neighbours similarly, and vice versa; people speaking Germanic languages shared their views on Slavic nations in the central European region. Thus, the sharing of national stereotypes extends beyond boundaries of one particular country. However, it cannot escape a broader cultural background defined by language similarity.
Comparison of national autostereotypes and their corresponding heterostereotypes suggested other factors that influence sharing of stereotypical beliefs. We found a convergence between an autostereotype and heterostereotypes only in ratings of a typical German. In other countries, no agreement between ratings of national stereotypes from an ingroup and outgroup perspective was found. This finding indicates that it is the size of a target country and its political and economic significance that play a role in intergroup sharing of its national stereotypes. Similarly, Boster and Maltseva (2006), in their study of European national stereotypes, found the highest level of agreement concerning British, Italian and German autostereotypes and heterostereotypes. In a different study, people from 49 cultures around the world rated a typical American similarly to Americans themselves (ICC = .71) (Terracciano & McCrae, 2007). A moderate level of agreement was also recorded among ingroup and outgroup perceptions of a typical Russian in the Baltic Sea Project (Realo et al., 2009). Thus, it seems that national heterostereotypes resemble autostereotypes in case of world powers with economic and political influence, such as USA, Germany or Russia. On the contrary, when members of smaller or less significant countries are rated, the outgroup and ingroup perspectives differ. International media report more often on world powers and their inhabitants than about less influential countries. As a result, stereotypical beliefs disseminated this way can be shared more broadly across the ingroup and different outgroups, resulting in a convergence between national autostereotypes and heterostereotypes.
Besides the convergence between national autostereotypes and heterostereotypes, we also examined the similarity of national autostereotypes rated by participants from different countries. Generally, we found a low level of agreement in the ratings of one's own country representatives. The convergence of autostereotypes followed the division into the two language groups—Czech autostereotype resembled the Polish, and German autostereotype resembled the Austrian. Austrian participants differentiated their autostereotype from the other countries the least—it also converged with Czech and Polish ingroup ratings. The only country with no agreement with other central European autostereotypes was Slovakia. However, the disagreement between national autostereotypes was not based on existing differences between personality traits of real people. The ratings of the inhabitants of central Europe converged much more than views on stereotypical characteristics of the given countries representatives. When rating national stereotypes, participants tended to emphasize the distinctiveness of their own country in that they overrated the differences between countries. Our findings correspond to the predictions of social identity theory (Tajfel & Turner, 1979), positing that individuals strive to achieve a distinct social identity. Social identity can be defined as a part of self‐concept that derives from membership in social groups. Citizenship is one of the possible categories with which individuals identify. In the situation of intergroup comparison represented by ratings of national autostereotypes and heterostereotypes in our research, participants could exaggerate differences between the countries under study in order to satiate their need for distinctiveness.
Limitations
The main strength of our study—the large number of participants from five different countries—also turned out to be its main limitation. In order to examine the effect of age and different measuring instruments on the agreement between ratings of national stereotypes and real people in the whole sample, we would need to include another age group and NCS for self‐report in each of the four remaining countries, not only in the Czech Republic. Another limitation pertains to the use of different norms for standardization of stereotypical and real people ratings. The fact that there are no international norms available for self‐reports on NEO‐PI‐R and only separate international norms for men and women in two different age groups for observer ratings complicated a fair estimate of convergence between national stereotypes and real people. Although previous research reported similar results obtained from analysis of data standardized along different norms (McCrae et al., 2005), the effect of standardization on subsequent correspondence between two profiles deserves rigorous examination in future studies.
The estimate of agreement between stereotypical and real people traits can be further confounded by low correspondence between the two methods employed for self‐reports, the NCS and NEO‐PI‐R. As illustrated by previous research, NCS and NEO‐PI‐R do not highly correlate even if they are used for ratings of the same targets (Konstabel et al., 2012). This way, our conclusion that national stereotypes do not reflect personality traits of real people can partially mirror the differences between the two methods used for both types of ratings. All in all, future research needs to focus on the effect of different/same method(s) used for ratings of targets under comparison, the effect of different/same raters providing the assessment, the effect of different/same norms used for standardization and the effect of ICC/Pearson's r on the resulting level of agreement between two profiles that are being compared.
ACKNOWLEDGEMENTS
This research was supported by a grant P407/10/2394 ‘Intergroup attitudes and contact in five Central European countries’ by the Czech Science Foundation and by RVO: 68081740 of the Institute of Psychology, Academy of Sciences of the Czech Republic. The authors wish to thank Annu Realo and René Mõttus for their insightful comments on the earlier version of the manuscript and Martin Jelínek for statistical consultations.
REFERENCES
Citing Literature
Number of times cited according to CrossRef: 17
- Anu Realo, Jüri Allik, National Character, Encyclopedia of Personality and Individual Differences, 10.1007/978-3-319-24612-3, (3099-3101), (2020).
- Robert R. McCrae, The Five-Factor Model of Personality: Consensus and Controversy, The Cambridge Handbook of Personality Psychology, 10.1017/9781108264822, (129-141), (2020).
- , The Cambridge Handbook of Personality Psychology, 10.1017/9781108264822, (2020).
- Luc van Doorslaer, Embedding imagology in Translation Studies, Slovo.ru: Baltic accent, 10.5922/2225-5346-2019-3-4, 10, 3, (56-68), (2019).
- Samuel Piha, Terhi Pohjanheimo, Anu Lähteenmäki-Uutela, Zuzana Křečková, Tobias Otterbring, The effects of consumer knowledge on the willingness to buy insect food: An exploratory cross-regional study in Northern and Central Europe, Food Quality and Preference, 10.1016/j.foodqual.2016.12.006, 70, (1-10), (2018).
- Martina Hřebíčková, René Mõttus, Sylvie Graf, Martin Jelínek, Anu Realo, How Accurate Are National Stereotypes? A Test of Different Methodological Approaches, European Journal of Personality, 10.1002/per.2146, 32, 2, (87-99), (2018).
- Bryor Snefjella, Daniel Schmidtke, Victor Kuperman, National character stereotypes mirror language use: A study of Canadian and American tweets, PLOS ONE, 10.1371/journal.pone.0206188, 13, 11, (e0206188), (2018).
- Florian Klapproth, Henrike Kärchner, Sabine Glock, Do Students' Religion and School Absences Moderate the Effect of Ethnic Stereotypes on School-Placement Recommendations?, The Journal of Experimental Education, 10.1080/00220973.2017.1293602, 86, 2, (173-194), (2017).
- Anu Realo, Jüri Allik, National Character, Encyclopedia of Personality and Individual Differences, 10.1007/978-3-319-28099-8, (1-3), (2017).
- Martina Hřebíčková, Sylvie Graf, Tamás Tegdes, Ivan Brezina, We are the opposite of you! Mirroring of national, regional and ethnic stereotypes, The Journal of Social Psychology, 10.1080/00224545.2017.1284738, 157, 6, (703-719), (2017).
- Gerry Larsson, Carina Brandow, Maria Fors Brandebo, Alicia Ohlsson, Gunnar Åselius, Swedish military officers through other nations’ eyes, International Journal of Organizational Analysis, 10.1108/IJOA-11-2014-0817, 24, 4, (615-633), (2016).
- A Timothy Church, Personality traits across cultures, Current Opinion in Psychology, 10.1016/j.copsyc.2015.09.014, 8, (22-30), (2016).
- Barbara Lášticová, Andrej Findor, Developing explicit measures of stereotypes and anti-Roma prejudice in Slovakia: Conceptual and methodological challenges, Human Affairs, 10.1515/humaff-2016-0022, 26, 3, (2016).
- Esther Guillaume, Erica Baranski, Elysia Todd, Brock Bastian, Igor Bronin, Christina Ivanova, Joey T. Cheng, François S. de Kock, Jaap J. A. Denissen, David Gallardo‐Pujol, Peter Halama, Gyuseog Q. Han, Jaechang Bae, Jungsoon Moon, Ryan Y. Hong, Martina Hřebíčková, Sylvie Graf, Paweł Izdebski, Lars Lundmann, Lars Penke, Marco Perugini, Giulio Costantini, John Rauthmann, Matthias Ziegler, Anu Realo, Liisalotte Elme, Tatsuya Sato, Shizuka Kawamoto, Piotr Szarota, Jessica L. Tracy, Marcel A. G. van Aken, Yu Yang, David C. Funder, The World at 7:00: Comparing the Experience of Situations Across 20 Countries, Journal of Personality, 10.1111/jopy.12176, 84, 4, (493-509), (2015).
- Michael W. Morris, Chi-yue Chiu, Zhi Liu, Polycultural Psychology, Annual Review of Psychology, 10.1146/annurev-psych-010814-015001, 66, 1, (631-659), (2015).
- Jüri Allik, Ekaterina Alyamkina, Boris Meshcheryakov, The Personality Stereotypes of Three Cohabiting Ethnic Groups, Cross-Cultural Research, 10.1177/1069397114540861, 49, 2, (111-134), (2014).
- Sylvie Graf, Stefania Paolini, Mark Rubin, Negative intergroup contact is more influential, but positive intergroup contact is more common: Assessing contact prominence and contact prevalence in five Central European countries, European Journal of Social Psychology, 10.1002/ejsp.2052, 44, 6, (536-547), (2014).
