Estimating Smooth Country–Year Panels of Public Opinion

Abstract

At the microlevel, comparative public opinion data are abundant. But at the macrolevel—the level where many prominent hypotheses in political behavior are believed to operate—data are scarce. In response, this paper develops a Bayesian dynamic latent trait modeling framework for measuring smooth country–year panels of public opinion even when data are fragmented across time, space, and survey item. Six models are derived from this framework, applied to opinion data on support for democracy, and validated using tests of internal, external, construct, and convergent validity. The best model is reasonably accurate, with predicted responses that deviate from the true response proportions in a held-out test dataset by 6 percentage points. In addition, the smoothed country–year estimates of support for democracy have both construct and convergent validity, with spatiotemporal patterns and associations with other covariates that are consistent with previous research.

In the Mood for Democracy? Democratic Support as Thermostatic Opinion

“…In other words, this article tests theories of democratic learning against the thermostatic model. To carry out these tests, I make use of new country-by-year measures of democratic mood estimated using the Bayesian latent variable model of Claassen (2019b) years. This dataset, in turn, permits the use of dynamic models, which control for the effects of previous levels of mood, and first difference models, which eliminate the confounding effects of time-invariant, country-specific factors.…”

“…I use Claassen's "model 5," which performed best in validation tests on a held-out portion of the data. AlthoughClaassen (2019b) also applies this model to the topic of democratic support, this article use an expanded dataset of survey responses (see also Claassen 2019a).…”

Modeling Dynamic Comparative Public Opinion

“…Such models have been used to measure many difficult concepts in comparative politics, from democracy (Treier and Jackman 2008;Pemstein, Meserve and Melton 2010) through governance (Arel-Bundock and Mebane 2011) and state capacity (Hanson and Sigman 2019) to judicial independence (Linzer and Staton 2015) and respect for human rights (Fariss 2014). Two recent works, Claassen (2019) and Caughey, O'Grady and Warshaw (2019), have introduced a Bayesian latent variable approach to the study of comparative public opinion that maximizes the information gleaned from available surveys to overcome issues of sparse and incomparable data and allow comparativists to examine the dynamics of public opinion. This paper presents DCPO, a new model and software for estimating Bayesian latent vari-ables of comparative dynamic public opinion from cross-national survey data.…”

“…The paper begins with a detailed description of the DCPO model, building up from the individual level. It continues with a comparison of the performance of the DCPO model to the alternate models of public opinion presented in Claassen (2019) and Caughey, O'Grady and Warshaw (2019).…”

“…This is true both in the internal validation test and on average across the ten folds of the external validation test. Again, this discrepancy is possible due to the greater variance in the survey data when its ordinal nature is preserved, as in the Caughey, O'Grady and Warshaw (2019) model, rather than dichotomized, as inClaassen's (2019) approach. An examination of column 6 reveals that the Caughey, O'Grady and Warshaw (2019) model yielded predictions with credible intervals that are much too narrow, encompassing only about 13% of the actual sample observations, while those for Claassen's (2019) Model 5 were slightly too conservative.…”