MULTIPLE IMPUTATION OF INCOMPLETE CATEGORICAL DATA USING LATENT CLASS ANALYSIS

We propose using latent class analysis as an alternative to log-linear analysis for the multiple imputation of incomplete categorical data. Similar to log-linear models, latent class models can be used to describe complex association structures between the variables used in the imputation model. However, unlike log-linear models, latent class models can be used to build large imputation models containing more than a few categorical variables. To obtain imputations reflecting uncertainty about the unknown model parameters, we use a nonparametric bootstrap procedure as an alternative to the more common full Bayesian approach. The proposed multiple imputation method, which is implemented in Latent GOLD software for latent class analysis, is illustrated with two examples. In a simulated data example, we compare the new method to well-established methods such as maximum likelihood estimation with incomplete data and multiple imputation using a saturated log-linear model. This example shows that the proposed method yields unbiased parameter estimates and standard errors. The second example concerns an application using a typical social sciences data set. It contains 79 variables that are all included in the imputation model. The proposed method is especially useful for such large data sets because standard methods for dealing with missing data in categorical variables break down when the number of variables is so large.     

The evaluation and selection of adequate causal models: a compensatory education example

Procedures for ascertaining relative model adequacy in latent variable structural relations models are discussed. Under diverse methods of estimation, this determination may be assessed using the chi square goodness of fit statistic, incremental fit indices for covariance structure models, and latent variable coefficients of determination. An example from evaluation research is taken (cf. Magidson, 1977; Bentler & Woodward, 1978). Numerical sensitivity of parameter estimates under alternative model specifications is demonstrated. Interpretive implications based on these procedures are discussed in terms of parameter sensitivity to alternative model specifications.     

Regression Analysis of Multivariate Binary Response Variables Using Rasch-Type Models and Finite-Mixture Methods

A model is considered for the regression analysis of multivariate binary data such as repeated-measures data (for example, panel data) or multiple-indicators with measures of some underlying characteristic such as attitude or ability (for example, surveys or tests). The model is related to the usual Rasch model, the usual latent-class model, and other familiar models such as logistic regression. In addition to a regression specification, the model includes parameters that describe heterogeneity not accounted for by the predictors. In contrast to most other approaches, a nonparametric specification of the latent mixing distribution is used, leading to a formulation based on scaled latent classes. We examine the relationship between this model and several other models, give a tractable formulation of the likelihood function and likelihood equations, present an algorithm for maximum-likelihood estimation, and analyze marginal and conditional latent structures. The approach is illustrated with longitudinal data from the German Socioeconomic Panel.     

The Natsal-SF Measure of Sexual Function: Comparison of Three Scoring Methods

The Natsal-SF is a psychometrically validated measure of sexual function for use in community health surveys, derived from 17 questions reflecting three components of sexual function. Scoring requires knowledge of complex statistical modeling and, given the methodological complexities, we assessed the validity of two simplified scoring methods calculated using the factor loadings produced when originally modeling the Natsal-SF items. Method 1 uses these factor loadings to three decimal places, while method 2 assigns whole numbers to each item based on the factor loadings. Scores from these simplified methods are compared to the original score using correlation coefficients, by comparing the distributions and the scores of each method in a linear regression model with key variables. We found scores from the simplified methods both correlate highly with the original score, and the distributions of scores closely match. The simplified methods result in different regression coefficients for gender and relationship context but estimate the coefficients of all other variables similarly to the original method. While the Natsal-SF should ideally be scored using latent variable modeling, the simplified methods perform well so can be used in similar contexts, increasing the utility of the Natsal-SF and enabling future studies to measure sexual function more comprehensively.     

Dimensional Latent Structure of Relationship Quality: Results of Three Representative Population Samples

A fundamental question facing every latent construct (e.g., relationship quality) is whether the construct is categorical or dimensional in nature. The authors analyzed the latent status of relationship quality in 3 large general German population samples using the Partnership Questionnaire (PFB), the Quality of Relationship Inventory, and the short form of the PFB (PFB‐K; N₁ = 1,330, N₂ = 1,494, N₃ = 1,390). They applied 3 widely used taxometric methods: (a) MAXEIG, (b) MAMBAC, and (c) L‐Mode. Simulation data were created to evaluate the comparison curve fit index values, which were below 0.5 across the 3 German population samples, supporting a dimensional solution. Hence, in the current study the latent structure of relationship quality encompassed differences in degree rather than kind. Implications of the dimensional latent status of relationship quality, as well as causes of the discrepant findings with previous taxometric studies, are discussed.     

Latent Class Marginal Models for Cross-Classifications of Counts

The standard latent class model is a finite mixture of indirectly observed multinomial distributions, each of which is assumed to exhibit statistical independence. Latent class analysis has been applied in a wide variety of research contexts, including studies of mobility, educational attainment, agreement, and diagnostic accuracy, and as measurement error models in social research. One of the attractive features of the latent class model in these settings is that the parameters defining the individual multinomials are readily interpretable marginal probabilities, conditional on the unobserved latent variable(s), that are often of substantive interest. There are, however, settings where the local-independence axiom is not supported, and hence it is useful to consider some form of local dependence. In this paper we consider a family of models defined in terms of finite mixtures of multinomial models where the multinomials are parameterized in terms of a set of models for the univariate marginal distributions and for marginal associations. Local dependence is introduced through the models for marginal associations, and the standard latent class model obtains as a special case. Three examples are analyzed with the models to illustrate their utility in analyzing complex cross-classifications.     

The Measurement of Values in Surveys: A Comparison of Ratings and Rankings

Social values are most commonly measured using ranking techniques,but there is a scarcity of systematic comparisons between rankingsand other approaches to measuring values in survey research.On the basis of data from the 1980 General Social Survey, thisarticle evaluates the comparability of results obtained usingrankings and ratings of valued qualities. The comparison focuseson (1) the ordering of aggregate value preferences and (2) themeasurement of individual differences in latent value preferences.The two methods are judged to be similar with respect to orderingthe aggregate preferences of the sample, but dissimilar withregard to the latent variable structure underlying the measures     

Approaches to blockmodeling dynamic networks: A Monte Carlo simulation study

Blockmodeling refers to a variety of statistical methods for reducing and simplifying large and complex networks. While methods for blockmodeling networks observed at one time point are well established, it is only recently that researchers have proposed several methods for analysing dynamic networks (i.e., networks observed at multiple time points). The considered approaches are based on k-means or stochastic blockmodeling, with different ways being used to model time dependency among time points. Their novelty means they have yet to be extensively compared and evaluated and the paper therefore aims to compare and evaluate them using Monte Carlo simulations. Different network characteristics are considered, including whether tie formation is random or governed by local network mechanisms. The results show the Dynamic Stochastic Blockmodel (Matias and Miele 2017) performs best if the blockmodel does not change; otherwise, the Stochastic Blockmodel for Multipartite Networks (Bar-Hen et al. 2020) does.     

Work orders: Analysing employment histories using sequence data

The analysis of employment histories has been facilitated recently by advances in survey methodology, statistical processes and computing power. While much work has focused on transitions between states and time spent in one state, the potential of analysing a series of states (i.e. careers) has largely been ignored. A concentration on movement between two states, whilst allowing relevant contextual covariates to be controlled for, often ignores valuable data both prior to and after the episode in question. Analysis of extended sequences of employment states is better able to describe employment trajectories. Furthermore, comparison of sequences permits either allocation to theoretical categories, or the identification of latent groupings using cluster analysis. The resulting typology of careers can then be used in inferentially based analyses. This paper explores sequence analysis using Optimal Matching Analysis (OMA). OMA is explained in relation to one broad substantive issue: the relationship between employment trajectories and gender.     

Ein Mixed-Methods-Evaluations-Ansatz zur Untersuchung von Makro-Mikro-Interaktionen

Especially in the case of longitudinal evaluations, mixed-methodological approaches appear to be useful, since the studies are basically suitable for evaluating evaluation objects of a high degree of complexity, such as, for example, dynamic processes in educational structures. Basing on the longitudinal development of professions-related beliefs of mathematics student teachers, this article describes how macro- and microstructures can be taken into account by integrating results obtained through different research methods. With the help of latent growth curve models the development of beliefs of 235 student teachers at five universities could be modeled longitudinally. On the basis of 19 interviews, afterwards interrelationships between institutional frameworks and the individual development of beliefs could be identified on the basis of case studies. Amongst others, a high degree of agreement to constructivist-oriented forms of teaching and learning could be revealed, which can be partly attributed to the experience of university teaching.     

Die Integration qualitativer und quantitativer Forschung – theoretische Grundlagen von „Mixed Methods“

Mixed Methods have been used in social research for quite a long time, as can be seen by looking on influential empirical studies from the first half of the 20^th century, where the combined use of qualitative and quantitative methods was justified in most cases only pragmatically (if at all). Unfortunately, the discussions led from the 1970s onwards about theoretical justifications of mixed methods designs often suffered from a focus limited on formal and epistemological issues. This holds true for the debates about the notion of “triangulation” as well as for the controversies about the paradigms presumably underlying quantitative, qualitative or mixed methods research. By contrasting “positivism”, “constructivism” and “pragmatism” outdated frontlines were often preserved and complex philosophical problems oversimplified. Furthermore, the focus on epistemology led to a neglect of substantial issues relating to the research field (which are always critical for the use of methods).The necessity for using (and combining) quantitative and qualitative methods follows from the nature of social structures relevant for the concrete field of investigation which may differ widely according to their stability and flexibility. Thereby structures with limited scope or “contingent structures” create specific problems for qualitative and quantitative methods which can be detected and dealt with by drawing on the strengths of the respective other methodological tradition in mixed methods designs.     

Conditionally exponential random models for individual properties and network structures: Method and application

Exponential random models have been widely adopted as a general probabilistic framework for complex networks and recently extended to embrace broader statistical settings such as dynamic networks, valued networks or two-mode networks. Our aim is to provide a further step into the generalization of this class of models by considering sample spaces which involve both families of networks and nodal properties verifying combinatorial constraints. We propose a class of probabilistic models for the joint distribution of nodal properties (demographic and behavioral characteristics) and network structures (friendship and professional partnership). It results in a general and flexible modeling framework to account for homophily in social structures. We present a Bayesian estimation method based on the full characterization of their sample spaces by systems of linear constraints. This provides an exact simulation scheme to sample from the likelihood, based on linear programming techniques. After a detailed analysis of the proposed statistical methodology, we illustrate our approach with an empirical analysis of co-authorship of journal articles in the field of neuroscience between 2009 and 2013.     

Kulturen und Spiralen des „Eigentlich“. Hierarchisch gesteuerte Veränderungsprozesse am Beispiel nachgelagerter Dienststellen des österreichischen Bundesministeriums für Bildung

Amidst the current debates on education, in Austria the former academies of the compulsory and vocational teachers’ training shall be transformed to educational “colleges” or even “universities”. The Federal Ministry is driving this process top-down, first and foremost tackling the manifest structures. This gradually unsheathes a latent “culture of eigentlich” within the affected educational institutions. It turns out that this culture, which is functional for the system in dealing with ambiguous, contradictory, or even unattainable requirements, does not disappear, but that it rather absorbs and transforms the manifest demands “from top” in practice “at the bottom”. This again provokes hierarchical interventions top-down, and eventually leads to self-enforcing dynamics between manifest and latent structures, fuelling the spiralling process.     

THE METHODOLOGICAL INCLINATIONS OF GENDER SCHOLARSHIP IN MAINSTREAM SOCIOLOGY JOURNALS

This article examines the types of research methods sociologists studying women and gender use in articles published in mainstream journals. The research is based on an analysis of 1,826 gender-content articles published between 1984 and 1993 in 15 major sociology journals. Specifically, we explore whether feminist-oriented articles use different types of data, data collection methods, and statistical techniques than other articles addressing women and gender. The relationship between the rank of an academic journal, author's sex, and type of research method is also examined. We find that a large majority of gender-content articles are based on secondary, quantitative data. When the first author is a man, articles are somewhat more likely to be based on secondary data and to use quantitative statistical techniques. In general, we find that feminist research in the discipline is based on a melange of methodological approaches, reflecting the contested nature of feminist epistemologies.     

Evaluation of strategies to mitigate contagion spread using social network characteristics

Computer simulation is an effective tool for assessing mitigation strategies, with recent trends concentrating on agent-based techniques. These methods require high computational efforts in order to simulate enough scenarios for statistical significance. The population individuals and their contacts determined by agent-based simulations form a social network. For some network structures it is possible to gain high accuracy estimates of contagion spread based on the connection structure of the network, an idea that is utilized in this work. A representative social network constructed from the 2006 census of the Greater Toronto Area (Ontario, Canada) of 5 million individuals in 1.8 million households is used to demonstrate the efficacy of our approach. We examine the effects of six mitigation strategies with respect to their ability to contain disease spread as indicated by pre- and post-vaccination reproduction numbers, mean local clustering coefficients and degree distributions. One outcome of the analysis provides evidence supporting the design of mitigation strategies that aim to fragment the population into similarly sized components. While our analysis is framed in the context of pandemic disease spread, the approach is applicable to any contagion such as computer viruses, rumours, social trends, and so on.     

A SNP,Gene, and Polygenic Risk Score Approach of Oxytocin‐Vasopressin Genes in Adolescents’ Loneliness

Not much is known regarding underlying biological pathways to adolescents’ loneliness. Insight in underlying molecular mechanisms could inform intervention efforts aimed at reducing loneliness. Using latent growth curve modeling, baseline levels and development of loneliness were studied in two longitudinal adolescent samples. Genes (OXTR, OXT, AVPR1A, AVPR1B) were examined using SNP‐based, gene‐based, and polygenic risk score (PRS) approaches. In both samples, SNP‐ and gene‐based tests showed involvement of the OXTR gene in development of loneliness, though, significance levels did not survive correction for multiple testing. The PRS approach provided no evidence for relations with loneliness. We recommend alternative phenotyping methods, including environmental factors, to consider epigenetic studies, and to examine possible endophenotypes in relation to adolescents’ loneliness.     

FOUR USEFUL FINITE MIXTURE MODELS FOR REGRESSION ANALYSES OF PANEL DATA WITH A CATEGORICAL DEPENDENT VARIABLE

This paper describes and contrasts two useful ways to employ a latent class variable as a mixture variable in regression analyses of panel data with a categorical dependent variable. One way is to model unobserved heterogeneity in the trajectory, or change in the distribution, of the dependent variable. Two models that accomplish this are the latent trajectory model and latent growth curve model for a categorical dependent variable having ordered categories. Each latent class here represents a distinct trajectory of the dependent variable. The latent trajectory model introduces covariate effects on the composition of latent classes, while the latent growth curve model introduces covariate effects on both the "intercept" and the "slope" of growth in logit, which may vary among latent classes.
The other useful way is to model unobserved heterogeneity in the state dependence of the dependent variable. Two models that accomplish this are introduced for a simultaneous analysis of response probability and response stability, and the latent class variable is employed to distinguish two latent populations that differ in the stability of responses over time. One of them is the switching multinomial logit model with a time-lagged dependent variable as its separation indicator, and the other is the mover-stayer regression model.
By applying these four models to empirical data, this paper demonstrates the usefulness of these models for panel-data analyses. Example programs for specifying these models based on the LEM program are also provided.     

Multivariate methods in rehabilitation

The complexity of relationships between variables in a variety of rehabilitation settings can be appropriately addressed by the use of multivariate methods of statistical analysis. The purpose of this article is to describe (a) some basic concepts and principles of multivariate methods as well as (b) common misconceptions in the application of such methods. Specifically, the authors focus on the use of multivariate analysis of variance (MANOVA) in rehabilitation research and some pitfalls that rehabilitation researchers face when choosing between MANOVA for a set of outcome variables versus separate analyses of variance (ANOVAs) for each outcome variable. Examples of the use of MANOVA in rehabilitation research, as well as references to other multivariate methods, are also provided.     

The evolution of a project involving an integrated analysis of structured qualitative and quantitative data: From N3 to NVivo

Developments in Qualitative Solutions and Research (QSR software) from N3 through to NVivo have impacted not only on the convenience of using the program but also on the capacity of the researcher to undertake interpretive analysis. Perhaps reflective of changes in the broader context of social research, developments through succeeding generations of this software have also increased its capacity to contribute to integrated mixed methods (qualitative and quantitative) analysis. New tools in the software have facilitated bringing quantitative (demographic, categorical and other numeric) data into a qualitative project, primarily to use for comparison of texts across subgroups. In addition, with N4 it has became possible to display tabulated data showing either the presence or absence of coding in particular documents or matrix displays (cross-tabulations) of grouped data. NVivo allows for both greater flexibility and greater specificity of quantitative output. Quantitative outputs from the qualitative database allow for graphical display or statistical manipulation of the data. Numeric output might be combined with existing statistical data to expand analytic possibilities. Interpretation of numeric data is supported throughout by the availability of the text on which such data are based, while patterns in the text can be elucidated through numeric displays. This ability to treat data both qualitatively and quantitatively, using software, can move mixed methods research beyond comparative, convergent or sequential designs toward truly integrated analysis.