Back to Programme

How to Explain Variation in Institutional Trust in the World, a 4-level Longitudinal Model

Claire Durand (Université de Montréal)
Luis Patricio Pena Ibarra (Université de Montréal)
Anne-Sophie Charest (Université Laval)

Keywords: Comparative research

Abstract

A number of international surveys – the Barometers, the Latin American Public Opinion Project (LAPOP), the International Social Survey Program (ISSC), the World Values Survey (WVS) – include questions pertaining to trust in institutions. However, it is not obvious if and how we can combine these measures and if and how we can analyse them jointly. The paper will present the results of our research project where we combined the data from the Arab, Africa, Asia, East Asia and Latin America Barometers together with LAPOP and are in the process of integrating data from the ISSC and the WVS. As of January 2017, we have combined 635 files for surveys conducted in 98 countries between 1994 and 2015. We have close to 1 million respondents and 10 million answers to measures pertaining to 60 different institutions. In order to analyze these data while taking into account change over time together with the context at the regional and country level as well as the characteristics of the individual respondents and of the institutions for which they are asked about their level of trust, we use a 4-level multilevel longitudinal model of repeated measures. In this model, answers to trust questions are at level 1; they are embedded within individuals (level 2), themselves polled in a given year (level 3) in a given country (level 4). The multilevel model shows that the major part of the variance is at the level of institutions (63%), followed by 27% at the respondent level, 2.6% at the year level and 7.5% at the country level. The analyses confirm our hypothesis that there is no significant change over time in the overall level of trust. Region explains 30% of the variance between countries. However, the variables available at the individual level – age and sex – do not explain much of the variance at that level. Finally, the institutions themselves and the variation in trust in specific institutions over time and between regions explain 7.9% of the between institutions variance. The results presented at the conference will use an enhanced data base completed with new surveys – ISSC and WVS – and with information about the economic and political characteristics of the different countries.