Generalized Trust, Cultural Diversity and Institutions
Project Research Design and Methodology
H1. Institutions and context matter, in particular, school, prison, volunteer associations, shape trust:

H1a. Generalized trust increases when the subject has a new group of peers with higher level of trust.

H1b. Generalized trust increases when the subject moves to a locality with higher level of trust.

H2. Exposure to acts of corruption and dishonesty has a negative effect on generalized trust.

H3. Both perceptions of income inequality at society level and income inequality at mezzo level have a negative effect on generalized trust.

H4. Diverse and strong social ties reduce the effect on generalized trust of the exposure to a social context, such as peer group, neighborhood and region. In particular, students who have many friends outside class, are active members in groups, as well as prison inmates who have frequent visits and who work outside prison, are less influenced by their colleagues.

H5. Certain types of social relationship magnify the effect of cultural diversity. In particular, cultural diversity has a more positive effect on those with stronger ties to their family.

A Longitudinal and Experimental Study

General Hypotheses

E1. Analysis of secondary data of two youth panel surveys (the 2010-2011 Soros survey and the 2006-2009 CSD survey), on samples that are representative for the Romanian high school students. Both studies include measures of generalized trust, exposure to social and cultural diversity, unfairness and corruption, and migration.

E2. Conducting 40 semi-structured interviews with students and 40 semi-structured interviews with prison inmates. Half of them will be conducted before the quantitative part and half after.

E3. Conducting a panel survey on Romanian college students. The respondents will be selected from four universities that differ according to two criteria: (1) location (regions with high v. low levels of generalized trust), and (2) diversity (the university attracts students from the same region v. various regions of Romania). Additionally, in order to be able to assess the effects of contextual variables by multilevel models, we will include entire classes of students in the sample (N=1,200). The two waves will be conducted 9 months apart.

E4. Conducting a panel survey on prison inmates (N=900), with 9 months between the two waves.

E5. Conducting experiments on 200 subjects from a subsample of students and on a 200 subsample of inmates. We use an innovative approach that consists of combining two different types of experiments (Mosteanu 2011). The first experiment builds on Berg, Dickhaut and McCabe?s (1995) experimental trust games. In order to corroborate the findings of the student panel survey and those of the first experiment, the second experiment will use measures of implicit cognition (Nosek et al. 2002).

This research design will contribute to addressing the problems of measurement, endogeneity and model underspecification that undermined previous studies on generalized trust. In conjunction with the other three components, E5 will allow us to improve the validity and reliability of the measurement of generalized trust (O1). E3, E4 and E5 components ensure the longitudinal character of the research.

The panel surveys of E3, with 12 months between waves, combined with results of E1, which are based on panels with a longer span of time between the waves, will enable us to evaluate the stability of trust over time and across diverse settings (O2). At the same time, the longitudinal and experimental components are necessary for establishing the causal direction between generalized trust and some of the key independent variables. Multilevel analyses on E3 and E4 data will allow us to include mezzo-level attributes in the explanatory models, in addition to the standard individual and regional level variables. The longitudinal and multilevel approaches are necessary in order to test the H1 to H5 hypotheses by assessing the effects of social and cultural diversity (O3) and institutions (O4) on trust.

The Structure of the Analysis

Research Design

The project benefits from a complex analytical strategy that consists of a quantitative component, including longitudinal, multi-level and experimental aspects, and of a qualitative component.