Jouni Kuha

We propose a general procedure for estimating the variance–covariance matrix of two‐step estimates of structural parameters in latent variable models. The method is partially simulation‐based, in that it includes drawing simulated values of the measurement parameters of the model from their sampling distribution obtained from the first step of two‐step estimation, and using them to quantify part of the variability in the parameter estimates from the second step. This is asymptotically equal to the standard closed‐form estimate of the variance–covariance matrix of two‐step estimates, but it avoids the need to evaluate a cross‐derivative matrix which is the most inconvenient element of the standard estimate. The method can be applied to any type of latent variable models. We present it in more detail in the context of two common models where the measurement items are categorical: latent‐class models with categorical latent variables and latent trait models with continuous latent variables. The good performance of the proposed procedure is demonstrated with simulation studies and illustrated with two applied examples.

We assess how key features of the survey experience and characteristics of respondents are related to survey enjoyment, with a particular focus on questionnaire topic and the direct and moderating effects of respondent personality. Beyond the intrinsic value of promoting surveys that are more enjoyable for the public to participate in, understanding what shapes whether and how respondents enjoy completing surveys matters instrumentally for minimising nonresponse and maximising data quality. To address these questions, we model respondent self-reports of survey enjoyment over repeated waves of an online probability panel (OPP). Our findings show that the topics included in the questionnaire are a key determinant, with the probability of enjoyment varying substantially across topic areas. Enjoyment of the survey is also influenced by respondent personality, survey mode, and how interested respondents are in these topics ex ante.

Social desirability bias (SDB) occurs when respondents systematically distortsurvey responses through self-deception and impression management. While there isa large literature on interviewer-induced SDB, less is known about how and why thisvaries between different respondents across demographic groups. We present newevidence on between-group heterogeneity in SDB, with a specific focus on respond-ent sex, from a randomised mode experiment where respondents were randomlyassigned to either computer-assisted telephone interviewing (CATI) or ComputerAssisted Web Interview (CAWI). Allied with the availability of rich covariates onboth respondents and nonrespondents, this design enables us to estimate the causaleffect of interview mode on SDB, separately for men and women, across a large setof both positively and negatively valenced attitudes and behaviours. Our results re-veal large mode effects on SDB: CATI administration inflates reporting of positivelyvalenced items while suppressing reports of negatively valenced items compared toCAWI. These differences were largest and most consistent for positively valenceditems. Women showed significantly greater interviewer-induced SDB than men onthese outcomes. Our findings point to the superiority of online self-completion formeasuring normatively valenced attitudes and behaviours. They also suggest thatinterviewer-administration can systematically distort between-group comparisonson normatively sensitive questions.

We consider estimation of two-level latent class models for clustered data, when the measurement model for the observed measurement items includes non-equivalence of measurement with respect to some observed covariates. The parameters of interest are coefficients in structural models for the latent classes given covariates. We propose a two-step method of estimation. This extends previously proposed methods of two-step estimation for models without non-equivalence of measurement by specifying the model used in the first step in such a way that it correctly accounts for non-equivalence. The properties of these two-step estimators are examined using simulation studies and an applied example.

Latent class (LC) analysis is a model-based clustering approach for categorical data, with a wide range of applications in the social sciences and beyond. When the data have a hierarchical structure, the multilevel LC model can be used to account for higher-level dependencies between the units by means of a further categorical LC variable at the group level. The research interest of LC analysis typically lies in the relationship between the LCs and external covariates, or predictors. To estimate LC models with covariates, researchers can use the one-step approach, or the generally recommended stepwise estimators, which separate the estimation of the clustering model from the subsequent estimation of the regression model. The package multilevLCA has the most comprehensive set of model specifications and estimation approaches for this family of models in the open-source domain, estimating single- and multilevel LC models, with and without covariates, using the one-step and stepwise approaches.

This contribution presents a guide to the R package multilevLCA, which offers a complete and innovative set of technical tools for the latent class analysis of single-level and multilevel categorical data. We describe the available model specifications, mainly falling within the fixed-effect or random-effect approaches. Maximum likelihood estimation of the model parameters, enhanced by a refined initialization strategy, is implemented either simultaneously, i.e., in one-step, or by means of the more advantageous two-step estimator. The package features i) semi-automatic model selection when a priori information on the number of classes is lacking, ii) predictors of class membership, and iii) output visualization tools for any of the available model specifications. All functionalities are illustrated by means of a real application on citizenship norms data, which are available in the package.

We propose a two-step estimator for multilevel latent class analysis (LCA) with covariates. The measurement model for observed items is estimated in its first step, and in the second step covariates are added in the model, keeping the measurement model parameters fixed. We discuss model identification, and derive an Expectation Maximization algorithm for efficient implementation of the estimator. By means of an extensive simulation study we show that (1) this approach performs similarly to existing stepwise estimators for multilevel LCA but with much reduced computing time, and (2) it yields approximately unbiased parameter estimates with a negligible loss of efficiency compared to the one-step estimator. The proposal is illustrated with a cross-national analysis of predictors of citizenship norms.

In the digital age, understanding people’s online behaviours is vital. Digital trace data has emerged as a popular alternative to surveys, many times hailed as the gold standard. This study critically assesses the use of web tracking data to study online media exposure. Specifically, we focus on a critical error source of this type of data, tracking undercoverage: researchers’ failure to capture data from all the devices and browsers that individuals utilize to go online. Using data from Spain, Portugal, and Italy, we explore undercoverage in commercial online panels and simulate biases in online media exposure estimates. The paper shows that tracking undercoverage is highly prevalent when using commercial panels, with more than 70% of participants affected. In addition, the primary determinant of undercoverage is the type and number of devices employed for internet access, rather than individual characteristics and attitudes. Additionally, through a simulation study, it demonstrates that web tracking estimates, both univariate and multivariate, are often substantially biased due to tracking undercoverage. This represent the first empirical evidence demonstrating that web tracking data is, effectively, biased. Methodologically, the paper showcases how survey questions can be used as auxiliary information to identify and simulate web tracking errors.

We define a model for the joint distribution of multiple continuous latent variables which includes a model for how their correlations depend on explanatory variables. This is motivated by and applied to social scientific research questions in the analysis of intergenerational help and support within families, where the correlations describe reciprocity of help between generations and complementarity of different kinds of help. We propose an MCMC procedure for estimating the model which maintains the positive definiteness of the implied correlation matrices, and describe theoretical results which justify this approach and facilitate efficient implementation of it. The model is applied to data from the UK Household Longitudinal Study to analyse exchanges of practical and financial support between adult individuals and their non-coresident parents.