The relevance of contextual generalised trust in explaining individual immigration sentiments

ABSTRACT

This paper measures both individual- and contextual effects of generalised trust on Europeans’ attitudes towards immigration. Our data come from round 7 of the European Social Survey (ESS) with which it is possible to measure generalised trust also at the subnational level (NUTS levels). This enables us to capture evident variation in generalised trust within countries. Our main contribution is to test whether two persons who have the same level of generalised trust, but who live in regions differing in mean generalised trust, have different opinions about immigrants. Our results show, first of all, that people with high generalised trust have positive immigration attitudes. Second, living in a high-trusting region does not seem to generally encourage even more positive attitudes towards immigrants. Third, there is, however, another type of contextual effect that moderates the relationship between individual-level generalised trust and pro-immigration attitudes. A high-trusting regional context encourages high-trusting people to develop even more positive attitudes towards immigrants.

KEYWORDS:

• Generalised trust
• immigration attitudes
• social cohesion
• contextual effects
• multilevel model

Disclosure statement

No potential conflict of interest was reported by the author(s).

Notes on contributors

Josefina Sipinen is doctoral student at the Faculty of Management and Business, Tampere University. In her doctoral thesis she studies candidates of foreign origin in the Finnish municipal elections.

Peter Söderlund is a senior lecturer in political science at Åbo Akademi University. His main research interests include political attitudes and voting behaviour.

Maria Bäck is Ph.D. and Senior Lecturer at the Faculty of Social Sciences (politics and organisations) at University of Helsinki. Her research interests cover such topics as political trust, social capital, political participation and elections.

Notes

1 Our main conclusions are robust to the exclusion of respondents with any missing values on the immigration attitudes index. N would drop from 30,062 to 29,915 by excluding respondents with one or two missing values. The coefficient for individual generalised trust would remain 0.19, while the coefficient for regional generalised trust would decrease from 0.20 to 0.17 (see Model 2 in Table 1). The coefficient for the interaction variable (individual trust × regional trust) would also be 0.02 (as in Model 3 in Table 1).

2 The unexpected positive effect with respect to non-European immigrants could be a result of not measuring a change in status quo. If the regional share of non-European immigrants has remained stable, the context is likely not to appear threatening.

3 Another way to demonstrate that there is no contextual effect is to run a regression model where the raw score for the generalised trust index (0–10) is included instead of the deviation from the regional mean. In such a model, the coefficient for the raw score for the trust index (0.19) is identical to the coefficient for within-region variable in Model 3, Table 2. In contrast, the coefficient for the regional mean coefficient is 0.03 and statistically insignificant (p = 0.72).