Although political trust is a long-standing interdisciplinary topic, the lack of comparable cross-national time-series data has limited scholars’ ability to analyze its determinants and consequences and to generalize findings across countries and over time. To address this gap, this paper introduces the Trust in National Government (TrustGov) Dataset—a cross-national time-series resource covering 115 countries and territories from 1973 to 2020, harmonizing 1,545 country-year observations from 189 national and cross-national surveys using a Bayesian latent variable model.
This article focuses on a preliminary step in any ex-post data harmonization project—wrangling the pre-harmonized data—and suggests a practical routine for helping researchers reduce human errors in this often-tedious work. The routine includes three steps: (1) Team-based concept construct and data selection; (2) Data entry automation; and (3) “Second-order” opening—a “Tao” of data wrangling. We illustrate the routine with the examples of pre-harmonizing procedures used to produce the Standardized World Income Inequality Database (SWIID), a widely used database that uses Gini indices from multiple sources to create comparable estimates, and the Dynamic Comparative Public Opinion (DCPO) project, which creates a workflow for harmonizing aggregate public opinion data.