Keywords: Challenges of comparative research and International Survey Projects, cross-cultural concerns in data collection and measurement issues

In an analysis of 635 polls conducted in 98 countries of Latin America, Asia, Sub-Saharan Africa and West Asia and North Africa from 1995 to 2014, Durand, Pena Ibarra, Charest and Pelletier (2017) have shown that close to 10 percent of the variance in institutional trust occurs between countries and that more than 40 percent of this variance can be explained by region.
This paper aims at understanding the differences between regions and between countries within regions. In doing so, we face two challenges. First, the characteristics of interest change over time. It therefore appears more relevant to characterize countries according their trajectories rather than to their situation at one point in time. The second challenge is the fact that many indices of the situation of the different countries are available only for countries of the “Western World”. It is difficult to find information that is equally relevant and available in the different regions of the world and for all the countries.
In this paper, we examine how it is possible to cluster the countries into distinct groups. We perform two series of cluster analyses using K-means longitudinal analysis. The first one aims at clustering countries based on change in trust in different institutions over two decades (1995-2015). It uses a combined dataset of close to 700 polls conducted by the Barometer projects, LAPOP and the World Values survey. The second series of analyses aims at clustering these same countries over the same period on covariates that may/should be related to institutional trust (see Zmerli and Van der Meer, 2017). These covariates include indicators of the economic situation, like the PIB and the GINI index, the political situation, like the Governance indexes (WGI, QOG indicators), and the social situation (access to education, the situation of women, population diversity, etc.) that are available for all the countries. We present our different attempts and our criteria to select the most relevant and productive clustering in terms of our capacity to understand the variation in the situation of different countries and regions of the world, and particularly the level of trust and its variation over time for different institutions.