In recent years, both experimental and survey-based research have turned out an impressive amount of
evidence on the possible determinants of trust. However, a recent review of these works by Hardin (2006: 74) concludes that “there is relatively little to learn about trust from these two massive research programs.” He is echoed by (Nannestad 2008), who finds very few results about generalized trust and its correlates upon which scholars agree on. This lack of consensus may be the result of (1) differences in conceptualization and measurement, (2) endogeneity problems, and (3) omitting relevant variables.
Our proposed research aims to address the triple problem of measurement, endogeneity and model underspecification through a complex research design.
First, we will measure generalized trust using both explicit and implicit measures. Second, the research, based on both panel survey data and experimental data, seeks to triangulate the data according to social science best practices which will allow us to establish causal direction between trust and its potential determinants. Third, the research will capture detailed information both at the individual and intermediate (mezzo) levels, such as colleagues, friends, locality and region.
Moreover, we focus on two contexts of socialization that are rarely researched: higher education and prison. We selected these contexts for several main reasons: (1) they are expected to be at opposite ends of the spectrum in terms of the expected direction of change for trust and therefore will provide good settings for testing robustness of findings, (2) they are among the very few contexts where subjects spend large amount of time and where both the subjects themselves and their colleagues can be traced over several months, (3) they are ideal settings as mezzo contexts, forming their own set of relatively closed social relationships, and (4) the subjects are relatively young and, since youth socialization is faster and deeper at this point of human development, changes in level of trust is expected to be easier to observe.