摘要
ABSTRACTThe article studies the social relationships of natives in the destination countries of migrants. Specifically, the article estimates the share of non-natives in natives' friendship opportunity pools (FOP) and its correlation with natives only having native friends. Danish data is used due to a unique possibility of linking register information, allowing the establishment of an objective measurement of FOP at various levels, and survey information, allowing the measurement of natives' friendship composition, at the individual level. In our sample, 55 per cent of the natives have a migrant friend, and it is found to be strongly correlated with FOP. In line with Blau's macro-sociological framework and assimilation theory, we thereby find support for a transformation of the 'weak' friendship ties of natives in destination countries. The FOP effects are found to be stronger for older respondents and those with negative attitudes toward immigration. The effects are found across measurement at the neighbourhood-, parish – and municipal level and after control for general attitudes toward immigration. In models with three levels simultaneously included, the effects are present at the neighbourhood and municipal level, but absent at the parish level.KEYWORDS: Friendship homogeneitynativesopportunity poolintegrationassimilation Disclosure statementNo potential conflict of interest was reported by the author(s).Notes1 Some studies find strong homophily in friendship but they are typically based on the General Social Survey instrument asking respondents to name those people with whom they discussed important matters within the last six months; typically three names are mentioned (Marsden Citation1987). Though these core discussion networks are highly exclusive, general friendship networks are likely to be more inclusive.2 In this definition, 'natives' also includes persons where one parent is unknown (most often the father) to keep the measure conservative.3 The DCS has a panel structure but we have only included respondents, which participated in their first wave of the survey.4 As discussed in note 1, social networks are operationalized in various ways in the literature. While the item in the General Social Survey on intimate discussion persons is to the intimate side, our measure is to the inclusive side, i.e. getting closer to including acquaintances. In a Danish survey from 2022 (that cannot be linked to register data), we asked about 'close friends' instead of 'friends'. The share having at least one 'close friend' with immigrant background was 45 per cent, compared to the average of 55 per cent in the sample used in this article. Thus, a rough indication is that 'friends' instead of 'close friends' leads to a ten-percentage point increase.5 The survey sample is drawn from the Central Person Register (CPR). We, therefore, have a unique (anonymised) ID on each respondent in the survey that can be linked to the micro-level register information on basic demographic information, disposable income, highest attained education, and place of residence from Statistics Denmark's restricted-use data. Hence, the only self-reported measured from the sample used in the analysis is the dependent variable (on friendship) and the control variable on 'attitudes toward immigrants'.6 We do not build multi-level models as we do not have theoretical expectations to the relative importance of geographical level. Furthermore, it is methodologically difficult to study, as FOPs at the three levels are highly correlated (between 0.71 and 0.80). We leave it to future research to potentially disentangle the relative effects of embedded FOPs.Additional informationFundingThis work was supported by independent research fund Denmark.