Anton K. Formann

Linear Logistic Latent Class Analysis for Polytomous Data

Abstract For latent class analysis, a widely known statistical method for the unmixing of an observed frequency table into several unobservable ones, a flexible model is presented in order to restrain the unknown class sizes (mixing weights) and the unknown latent response probabilities. Two systems of basic equations are stated such that they simultaneously allow parameter fixations, the equality of certain parameters as well as linear logistic constraints of each of the original parameters. The maximum likelihood equations for the parameters of this “linear logistic latent class analysis” are given, and their estimation by means of the EM algorithm is described. Further, the criteria for their local identifiability and statistical tests (Pearson- and likelihood-ratio-χ 2) for goodness of fit are outlined. The practical applicability of linear logistic latent class analysis is demonstrated by three examples: mixed logistic regression, a mixed Bradley-Terry model for paired comparisons with ties, and a lo...

Latent class analysis in medical research

In the introduction we give a brief characterization of the usual measures for indicating the quality of diagnostic procedures (sensitivity, specificity and predictive value) and we refer to their relationship to parameters of the latent class model. Different variants of latent class analysis (LCA) for dichotomous data are described in the following: the basic (unconstrained) model, models with parameters fixed to given values and with equality constraints on parameters, multigroup LCA including mixed-group vali dation, and linear logistic LCA including its relationship to the Rasch model and to the measurement of change in latent subgroups. The problem with the identifiability of latent class models and the possi bilities for statistically testing their fit are outlined. The second part refers to latent class models for polytomous data. Special attention is paid to simple variants having fixed and/or equated parameters and to log-linear extension of LCA with its possibility for including interactions on the latent level. Several examples are presented to illustrate typical applications of the model. The paper ends with some warnings that should be taken into consideration by potential users of LCA.

Some Applications of Logistic Latent Trait Models with Linear Constraints on the Parameters

The linear logistic test model (LLTM), a Rasch model with linear constraints on the item parameters, is described. Three methods of parameter estimation are dealt with, giving special consideration to the conditional maximum likelihood approach, which provides a basis for the testing of structural hypotheses regarding item difficulty. Standard areas of application of the LLTM are surveyed, including many references to empirical studies in item analysis, item bias, and test construction; and a novel type of application to response-contingent dynamic processes is presented. Finally, the linear logistic model with relaxed assumptions (LLRA) for measuring change is introduced as a special case of an LLTM; it allows the characterization of individuals in a multidimensional latent space and the testing of hypotheses regarding effects of treatments.

Maternal bonding behaviour in schizophrenia and schizoaffective disorder, considering premorbid personality traits

Objective: Bonding between mother and child is described as a complex two-way process ensuring the needs of the child for nurture and protection. As such, it is dependent on the contribution of mother and child [1–3] whereby characteristics of personality of the child may have consequences on maternal bonding behaviour. In the current study the perception of maternal behaviour, premorbid personality traits and relationships between maternal behaviour and personality traits were investigated in schizophrenic and schizoaffective patients and their same-sex, healthy siblings. Methods: We recruited 36 schizophrenic and schizoaffective patients and their same-sex healthy siblings. Information about maternal bonding behaviour was assessed by the Parental Bonding Instrument, information about premorbid personality traits was obtained from their mothers using the ‘Gießen-Test’. Results: Compared to their siblings, patients showed less social resonance, more permeability, less social competence and a more depressed and anxious mood. Furthermore, patients described their mothers to be less caring and to be more overprotective than their siblings described them. But there were strong associations between maternal bonding behaviour and premorbid personality traits. These findings were supported by missing significant differences in maternal care behaviour between patients and siblings when using premorbid characteristics as covariates. Significant high maternal overprotection perceived by patients with schizophrenia and schizaffective disorders still remained after correcting for the influence of premorbid personality traits. Conclusion: The results suggest that premorbid personality traits should be considered not only in analyses of maternal care behaviour in schizophrenic and schizoaffective patients but also when studying other psychiatric patient groups.

Latent class model diagnostics: a review and some proposals

This contribution has two purposes. First, to review established methods for latent class model diagnostics and, second, to present and discuss less well-known methods. To the first group belong: (a) overall goodness-of-fit tests using, for large samples, the asymptotic chi-square approximation (Pearson, likelihood ratio, Read-Cressie power divergence statistic) and, for sparse data, resampling techniques such as parametric bootstrap; (b) residual analysis of the pattern frequencies for which also bootstrapping their reference distributions in the presence of smaller samples is recommended. Less well-known are (a) a chi-square goodness-of-fit test not being based on the full information provided by the pattern frequencies but rather using partial information available from lower-order marginals of the contingency table formed by the items, (b) a simple test for the odds ratios of pairs of items, (c) the Rudas-Clogg-Lindsay index of lack of fit of which (d) a new variant is considered, too. The methods are applied to both generated and empirical data to demonstrate their usefulness in latent class model diagnosis.

Health-related quality of life in Austrian elementary school children.

Background:Health-related quality of life (HRQL) is frequently used as an outcome criterion to evaluate the quality and effect of different therapies. However, little is known about the HRQL of the general population, the prevalence of specific HRQL problems and about which factors have an impact on HRQL assessments. Objective:To examine childrens HRQL from their own and their parents’ perspectives. The study sample consisted of children attending the third and fourth grades of elementary school in the capital of Austria, Europe. Methods:One thousand four hundred twelve children and 1185 parents completed child- and parent-versions of the Pediatric Quality of Life Inventory™ (PedsQL™). In addition to the PedsQL questions, children and parents were asked a number of questions with regard to sociodemographic information and specific problems that were hypothesized to be associated with the childrens HRQL. Results:Altogether, the children demonstrated a good HRQL and their PedsQL scores were similar to those of children from other developed countries. About 15% of children showed a notably low HRQL, and two-thirds of these children were from financially less privileged families. Multivariate regression analyses identified the following factors associated with the childrens HRQL: the familys perceived financial situation, parents’ quality of life, childrens school performance, medical and/or psychologic problems (eg, dyslexia, recurrent stomachache or headaches), chronic disease, a recent life-event (eg, divorce/separation of parents), and parents’ satisfaction with school. Conclusions:Assessing childrens HRQL may be helpful to take preventive action and to identify those who are in urgent need of special services.

Latent class model diagnosis from a frequentist point of view.

This is in response to Garrett and Zeger (2000, Biometrics 56, 1055-1067) who, within the Bayesian framework, developed mainly graphical methods for latent class model diagnosis. Possible problems with this approach, and with its application to both generated and empirical data, are pointed out. The impact of the proposed tools cannot be understood by their reader, as no comparisons are made to results obtainable using established methods for latent class model diagnosis; this applies especially to overall goodness-of-fit tests, for which alternatives (bootstrap, Rudas-Clogg-Lindsay index of fit) are mentioned. Further, in one case of generated data, the methods proposed by Garrett and Zeger seem to give problematic results as to identifiability; in the case of the empirical data on major depression, they lead to accepting a suboptimal three-class model. In the latter case, one can be rather sure that an identifiable, well-fitting latent class model could have been identified--if Garrett and Zeger had also considered restricted latent class models.

Neurodevelopmental Schizophrenia: Obstetric Complications, Birth Weight, Premorbid Social Withdrawal and Learning Disabilities

Neurodevelopmental schizophrenia seems to be caused by impaired cerebral development and is supposed to be associated with obstetric complications (OCs), poor premorbid adjustment, schizotypal or schizoid personality traits and negative symptoms. In the present study, 36 schizophrenic and schizoaffective patients and their same-sex, healthy siblings were recruited. They were diagnosed according to DSM-III-R, using structured psychiatric interviews and a consensus of 2 psychiatrists. Information on OCs, birth weight, premorbid social and learning functioning was obtained from their mothers. The main results show significant differences in OCs, birth weight, premorbid social and learning functioning between patients and their same-sex, healthy siblings. Using multivariate analyses, both premorbid variables were again identified to discriminate well between affected and unaffected siblings. Our findings seem to confirm the concept of schizophrenia as a neurodevelopmental process.

Pervasiveness of the IQ Rise: A Cross-Temporal Meta-Analysis

Background Generational IQ gains in the general population (termed the Flynn effect) show an erratic pattern across different nations as well as across different domains of intelligence (fluid vs crystallized). Gains of fluid intelligence in different countries have been subject to extensive research, but less attention was directed towards gains of crystallized intelligence, probably due to evidence from the Anglo-American sphere suggesting only slight gains on this measure. In the present study, development of crystallized intelligence in the German speaking general population is assessed. Methodology/Principal Findings To investigate whether IQ gains for crystallized intelligence are in progress in German-speaking countries, two independent meta-analyses were performed. By means of a cited reference search in ISI Web of Science, all studies citing test manuals and review articles of two widely-used salient measures of crystallized intelligence were obtained. Additionally, the electronic database for German academic theses was searched to identify unpublished studies employing these tests. All studies reporting participants mean IQ or raw scores of at least one of the two measures were included in the present analyses, yielding over 500 studies (>1,000 samples; >45,000 individuals). We found a significant positive association between years of test performance and intelligence (1971–2007) amounting to about 3.5 IQ points per decade. Conclusions/Significance This study clearly demonstrates that crystallized IQ gains are substantial and of comparable strength as Flynn effects typically observed for measures of fluid intelligence in Central Europe. Since mean IQ was assessed in a large number of small, non-representative samples, our evidence suggests a remarkable robustness of these gains. Moreover, in both meta-analyses strength of gains was virtually identical. On the whole, results of the present study demonstrate a pervasive and generalizing Flynn effect in German-speaking countries.

Variation in European suicide rates is better accounted for by latitude and longitude than by national percentage of Finno-Ugrians and Type O blood: a rebuttal of Lester and Kondrichin (2004).

The variation in contemporary suicide rates for men and women across 20 European nations is, in terms of shared variance, better accounted for by latitude and longitude than by national percentage of Finno-Ugrians and Type O blood, thus supporting the conclusion of Voracek, et al. in 2003 and refuting the supposition of Lester and Kondrichin of 2004. Discussion focuses on possible bases of this finding and on research problems pertaining to relations between suicide and blood type.