Three Challenges to Assumptions about ILSA Data Analysis: The Case of Attitudinal Precursors to Adult Populism among Adolescents

• In Event: Citizenship and democratic education using large-scale assessments: Person-centered approaches.

Proposal

This presentation draws upon a review of a wide range of prior secondary analyses of IEA’s civic education studies (Knowles, Torney-Purta & Barber, 2018) and presents a recent secondary analysis of attitudes germane to populism. An underlying premise of the paper is that the research questions chosen should shape the selection of a data set and mode of analysis. With that in mind, the presentation challenges three common assumptions about ILSAs.
The first assumption to be challenged is that the most recent iteration of a cross-sectional large-scale study (e.g., in civics, the 2016 ICCS Study) always provides preferred data when the researchers’ goal is understanding contemporary attitudes and engagement. This assumption should be challenged especially when there is interest in the prevalence of attitudes in a previous historical period to inform the present-day situation. Consider the following: is it possible to identify precursors of anti-immigrant attitudes (and other attitudes associated with populism) among the generation of fourteen-year-olds in 1999 (who currently are adults in their mid-thirties)? To look for adolescent precursors of the current wave of anti-immigrant and populist attitudes among adults, we explore data from IEA’s 1999 CIVED study. The second assumption is that variable-centered analysis (usually regression-type analysis) is the best practice. The review of secondary analyses revealed this as the most frequent type of analysis employed in 100 published civic-related articles. This assumption will be challenged by presenting a person-centered analysis (latent class analysis) of CIVED attitudinal data. An advantage of this approach for many research questions is that it is possible to estimate the percentages of students who hold particular patterns of attitudes. Here that includes anti-immigrant feeling, negative attitudes toward ethnic groups, and low trust in government/media. Results from person-centered analysis can capture interest among the public and policymakers because researchers delineate profiles of students holding specific attitudes (and estimate prevalence) rather than presenting coefficients with little meaning to non-statisticians. The third assumption is that including data from all countries in a given study result in more valuable ILSA analyses. Populism as an attribute of political culture has grown differentially and has salience in specific countries that participated in the civic education studies. This argues for analyzing a selected subset of countries.
These assumptions will be addressed by presenting a latent-class analysis of attitudes often linked with populism in three countries (the United States, Australia, and England) using data collected from 14-year-olds in CIVED 1999. Results show substantial proportions of students with profiles indicating relatively extreme negative attitudes toward rights for immigrants and toward ethnic group members. These attitudinal profiles included low trust in government and the media. Other substantial groups had profiles indicating indifference toward democratic values. Few had activist participatory profiles. In summary, the presentation will suggest the value of an analysis that tailors the choice of countries, of variables and of methods to address research questions of relevance in current political climates. This presentation is appropriate for the ILSA Interest Group, which considers questions of data use across studies.

Authors

• Judith Torney-Purta, University of Maryland

• Frank Reichert, The University of Hong Kong