Using respondent level data means we can create aggregated indicators representing different groups in society. For example, we can compare the attitudes of different income, education, or age groups. It is also possible to create indicators representing multiple cross sections. For example, we can contrast the views of high income earners with tertiary education against the views of low income earners with primary education. Of course, using respondent level data also means we can conduct micro level analysis controlling for individual level factors.

By combining so many data sources, we can obtain more observations over time. For many countries, this enables monthly or quarterly measures of key attitude indicators. These more granular levels of analysis enable us to evaluate the effects of elections, policy changes, scandals, recessions, and other events. In addition, using data from so many sources lets us investigate the attitudes of numerically small groups. For example, we can pool data from many sources to contrast the preferences of small minority party supporters.

Note that the data we use belongs to the sources and projects that commission or conduct the surveys. We collaborate with others to undertake specific research projects, but cannot share or redistribute the data. However, all the data we use is freely and publicly available from the original sources.

Analyze 19,838,430 survey respondents across 183 countries from 1948 onwards to conduct unprecedented original research


We have formatted thousands of source variables and merged them to create the harmonized variables outlined below. This formatting and harmonization process still leaves survey design and sampling methods to contend with, but creating these variables are an important prerequisite for conducting comparative cross-survey research. The following is a partial list of some of the variables we have harmonized.