1. The views expressed herein are those of the authors and are not necessarily those of the U. S. Department of State.Google Scholar

2. Among the most numerous in the “other” ethnic group are Ukrainians, Kazakhs, Tajiks, and Uighurs.Google Scholar

3. For an in-depth discussion, see Richard Dobson, “Kyrgyzstan in a Time of Change,” Central Asia Monitor, No. 1, 1995, pp. 19–27. See also Shirin Akiner, Central Asia: Conflict or Stability and Development (London: Minority Rights Group, 1997), pp. 23–28.Google Scholar

4. Richard Dobson, ibid. Google Scholar

5. For details on this important research conducted in Bishkek, see A. Elebayeva, N. Omuraliev, and R. Abazov, “The Shifting Identities and Loyalties in Kyrgyzstan: The Evidence from the Field,” Nationalities Papers, Vol. 28, No. 2, 2000, pp. 343–349.Google Scholar

6. Our research reflects this process as well. Specifically, we found that when we asked respondents how they first and foremost thought of their own identity, 69% said they were above all “Kyrgyzstani.” Far fewer thought of themselves as representatives of their ethnicity (13%), inhabitants of the region in which they grew up (7%) or in which they now live (4%), broadly defined Central Asians (4%), or representatives of their faith (2%).Google Scholar

7. See Social Realities of South Kyrgyzstan (Bishkek: Institute of Regional Studies, 1999).Google Scholar

8. The survey is based on 1,000 face-to-face interviews of the adult population (age 18 and older) with a sample drawn using a national multi-stage random probability design with quota controls introduced at the level of the final clusters in the design. The State Department's Office of Research commissioned the London-based Russian Research, Ltd. firm to conduct the survey. Russian Research, Ltd. coordinated with local survey research organizations: the Kazakhstani firm BRIF and the Kyrgyzstani firm, The Center for Social and Marketing Research. The Office of Research staff prepared survey questions with input from individuals and groups in the foreign affairs community. Questionnaires were translated by RRL, BRIF and The Center for Social and Marketing Research into Kyrgyz, Russian, and Uzbek and were reviewed and approved by the Office of Research staff. The local polling organization conducted the interviews (fieldwork) in May through June of 2001 and performed all the attendant work under the general supervision of the Office of Research.Google Scholar

Theoretically, 19 times out of 20, results from samples like those used by the Office of Research will differ by no more than about four percentage points in either direction from what would be found if it were possible to interview every adult in the country. Sampling error is larger for subgroups within the population (e.g. ethnic groups, regional subgroups). In addition to sampling error, the practical difficulties of conducting surveys of public opinion may introduce other sources of error (non-sampling error) into the results.Google Scholar

9. In our view, these prejudices regarding ethnic outgroups may be subtle or blatant prejudices depending on the specific measurements used within a given societal context. For example, Professors Pettigrew and Meertens argue that “blatant prejudice involves opposition to intimate contact with the outgroup. This anti-intimacy component focuses upon an emotional resistance against any intergroup sexual contact or intermarriage.” See T. F. Pettigrew and R. W. Meertens, “Subtle and Blatant Prejudice in Western Europe,” European Journal of Social Psychology, Vol. 25, 1995, pp. 57–75. However, their conceptual viewpoint focuses on the negative notion of opposition rather than the positive notion of acceptance. In the latter case, it seems to us that failure to accept or to endorse a close relationship is perhaps a more subtle form of prejudice.Google Scholar

10. For a detailed discussion of how we derived the social distance scale, see Appendix A.Google Scholar

11. There may be several dimensions of or some flexibility in how Kyrgyzstanis understand the term “ethnicity.” For instance, individuals could have more than one ethnic heritage. Or they might find it either more socially or culturally advantageous to say they are a certain ethnicity. One indicator of the flexibility of ethnicity is language use. Since respondents were able to choose the language in which the interview would be conducted—Russian, Kyrgyz, or Uzbek—we can see that some who say they are one ethnicity actually used the language of another. Of the entire sample, 57% used Kyrgyz, 33% used Russian, and 10% used Uzbek. Because of the widespread use of the Russian language, it is not surprising that 21% of Uzbeks and 8% of Kyrgyz used Russian in the interview (100% of Russians used Russian). At the same time, 16% of Uzbeks used Kyrgyz, and 2% of Kyrgyz used Uzbek—all of them in rural areas.Google Scholar

Anticipating this flexibility, when we asked about respondents' nationality, we asked them to tell us what they considered themselves to be—not necessarily what was indicated on their passports. The implication for us is that it does not matter what ethnicity respondents “really are.” What is important is that self-identification has clear implications in terms of perceptions of “others.” For instance, using the social distance scale, we show that those who call themselves “Kyrgyz” are systematically distant from those whom they call “Uzbeks.”Google Scholar

12. The questions we use for this scale were framed so that a specific ethnic group was named for each series of social relationships; consequently, any ambiguity regarding the actual referent for the outgroup for a given ethnic group was minimized. (The exact wordings for the questions are presented in Appendix A.) Furthermore, the notion of “friend” was refined to mean “close friend” to minimize the possibility that a relationship with “friend” might mean something less intimate than a relationship with a neighbor. The impetus for this refinement came from a personal discussion with Ainura Elebayeva, whose recent work on interethnic relations in Kyrgyzstan provided us with suggestive evidence on the importance of this sort of distinction.Google Scholar

13. We have detailed tabulations of the Russians' social distance from Uzbeks, but the small subgroups created by a tabular approach constrain our analysis of these data. For an illustration, see Table B5 in Appendix B.Google Scholar

14. The urban-rural dynamic in Kyrgyzstan is possibly more important in shaping views of social status than in influencing ethnic relations. As Eugene Huskey noted, “Potentially the most serious division among the Kyrgyz runs along urban-rural lines. The urbanized quarter of the ethnic Kyrgyz population has appropriated the language and at least some of the cultural values of the Russians,” while among the rural Kyrgyz there has been a rise of atavistic nationalism and pride in traditional Kyrgyz values. See E. Huskey, “Kyrgyzstan: The Politics of Economic and Demographic Frustration,” in I. Bremmer and R. Taras, eds, Nations and Politics in the Soviet Successor States (New York: Cambridge University Press, 1992). However, in our research, urbanicity has no systematic, statistically significant impact on how Kyrgyz view Uzbeks.Google Scholar

15. It also seems that Russians living in the South are more inclined than Russians elsewhere in the country to accept Uzbeks in close relationships. However, because of the small size of the Russian subgroup in the South, it is difficult to make a statistically valid comparison.Google Scholar

16. As Mary McIntosh et al. suggest, there are competing arguments about the link between ethnic composition of a community and ethnic relations within that community. One camp argues essentially that familiarity breeds contempt, or “the greater the density of a minority population in a certain area, the higher the degree of prejudice.” The other camp posits that close proximity of ethnic groups fosters understanding among those groups. In other words, “majority group members who live in close proximity … with minority group members are less prejudiced than those who are without this kind of contact.” Initially, it seemed the latter case held truer in Kyrgyzstan. For a more detailed discussion, see Mary E. McIntosh, Martha Abele MacIver, Daniel G. Abele, and David B. Nolle, “Minority Rights and Majority Rule: Ethnic Tolerance in Romania and Bulgaria,” in Leokadia Drobizheva, Rose Gottemoeller, Catherine McArdle Kelleher, and Lee Walker, eds, Ethnic Conflict in the Post-Soviet World: Case Studies and Analysis (Armonk, NY: M. E. Sharpe, 1996), pp. 36–66.Google Scholar

17. The level of concentration is estimated from our survey data by the percentage of adult Uzbeks for each settlement within our sample. The lower category—average or below—involves those settlements where Uzbeks constitute 12.9% or less of the settlement. The higher category—above average—incorporates those settlements where Uzbeks constitute more than 12.9% of the settlement. This cutpoint of 12.9% has been chosen because the national percentage of adult Uzbeks in Kyrgyzstan is 12.9%, according to the 1999 census.Google Scholar

18. A Kyrgyz respondent's score on this scale is derived from gender-specific differential scoring of a simultaneous latent class analysis of four Muslim religious rites: prayer, fasting, desire for a pilgrimage, and attendance at a place of worship (because of the Kyrgyz-language wording a respondent might interpret “place of worship” to mean not only mosque but any nearby mazar). The interrelationships among these indicators of religious observances are analyzed by gender for all Kyrgyzstanis indicating that they are Muslims. The scoring is based on a statistical model that has all indicators except attendance at a place of worship constrained to be equal between males and females. This model allows attendance to vary between Muslim males and females because attendance is substantially more important in differentiating levels of religious observance among males than among females. Although this model provides an excellent fit to the data (chi-square = 15.038 on 18 degrees of freedom with a probability value >0.50), this model seems to generate better scalability for males than females. Nevertheless, 87.3% of the Kyrgyzstani Muslims were correctly assigned to their proper level of religious practice by our scaling using this model. Details on the model are available from the authors upon request. References on latent class analysis are noted in Appendix A. For insights on the role of Islam in the daily life of Kyrgyzstanis, see A. Tabyshalieva, “The Kyrgyz and the Spiritual Dimensions of Daily Life,” in R. Sagdeev and S. Eisenhower, eds, Islam and Central Asia: An Enduring Legacy or an Evolving Threat? (Washington: Center for Political and Strategic Studies, 2000), pp. 27–38.0.50),+this+model+seems+to+generate+better+scalability+for+males+than+females.+Nevertheless,+87.3%+of+the+Kyrgyzstani+Muslims+were+correctly+assigned+to+their+proper+level+of+religious+practice+by+our+scaling+using+this+model.+Details+on+the+model+are+available+from+the+authors+upon+request.+References+on+latent+class+analysis+are+noted+in+Appendix+A.+For+insights+on+the+role+of+Islam+in+the+daily+life+of+Kyrgyzstanis,+see+A.+Tabyshalieva,+“The+Kyrgyz+and+the+Spiritual+Dimensions+of+Daily+Life,”+in+R.+Sagdeev+and+S.+Eisenhower,+eds,+Islam+and+Central+Asia:+An+Enduring+Legacy+or+an+Evolving+Threat?+(Washington:+Center+for+Political+and+Strategic+Studies,+2000),+pp.+27–38.>Google Scholar

19. Martha Brill Olcott, “Revisiting the Twelve Myths of Central Asia,” Working Paper of the Carnegie Endowment for International Peace Russian and Eurasian Program, No. 23, September 2001.Google Scholar

20. For details, see J. Billiet, P. Scheepers, and R. Eisinga, “Ethnocentrism in the Low Countries: Opinions on ‘Own’ and ‘Other’ People in The Netherlands and Flanders”, paper presented at the Second Conference of the European Sociological Association, Budapest, 31 August, 1995.Google Scholar

21. For an empirical overview of the situation in the Baltic states in the mid-1990s, see Richard Rose and William Maley, Nationalities in the Baltic States: A Survey Study (Glasgow: Center for the Study of Public Policy, University of Strathclyde, 1994).Google Scholar

22. E. S. Bogardus, Social Distance (Yellow Springs, OH: Antioch Press, 1959).Google Scholar

23. For a recent review of the copious literature on this scale, see J. P. Robinson, P. R. Shaver, and L. S. Wrightsman, eds, Measures of Political Attitudes (Vol. 2 of Measures of Social Psychological Attitudes) (New York: Academic Press, 1999), pp. 335–343.Google Scholar

24. A. L. McCutcheon, Latent Class Analysis (Newbury Park, CA: Sage, 1987), p. 45.CrossRefGoogle Scholar

25. All of the Guttman scale models used in this paper were derived from latent class analyses estimated by using C. C. Clogg's computer program called Maximum Likelihood Latent Structure Analysis (MLLSA), which was incorporated into the Categorical Data Analysis System (Version 3.50) developed by S. R. Eliason in 1990. All of the computations were based on a series of cross-tabulations involving the weighted frequencies in the eight cells of a three-way cross-tabulation of the three items (marriage, neighbor, friend) referring to a specific ethnic outgroup (for example, Uzbeks) within each ethnic group (for example, Kyrgyz). The statistical basis for the development of the models was taken from two seminal articles: C. C. Clogg and D. O. Sawyer, “A Comparison of Alternative Models for Analyzing the Scalability of Response Patterns,” in S. Leinhardt, ed., Sociological Methodology 1981 (San Franciso: Jossey-Bass, 1981), pp. 240–280 and C. C. Clogg and L. A. Goodman, “On Scaling Models Applied to Data from Several Groups,” Psychometrika, Vol. 51, No. 1, 1986, pp. 123–135.Google Scholar

26. The chi-square statistics were calculated under the assumption of simple random sampling; therefore, these statistics are likely to be somewhat inflated by the sampling design effect that naturally arises from a multi-stage cluster design such as the one used in this survey. In this context, these statistics probably provide a “liberal” estimate of the deviation between the models and the data.Google Scholar

27. M. B. Brewer's early work on social distance in Africa provides the best example of the need to check the scalability of various points on a social distance scale when one applies this sort of scale in a different cultural context. Her initial assumption that sharing a meal with a member of an ethnic outgroup was a more intimate social relationship than having an outgroup member as a neighbor was completely undermined by a crude Guttman scale analysis which revealed the exact opposite. She concluded that this result was not surprising “if it is realized that, traditionally, visiting strangers were treated to meals and that such hospitality was an important factor in a tribe's prestige.” For details, see M. B. Brewer, “Determinants of Social Distance among East African Tribal Groups,” Journal of Personality and Social Psychology, Vol. 10, No. 3, 1968, pp. 279–289.CrossRefGoogle Scholar