Bayesian estimation and testing of structural equation models

The Gibbs sampler can be used to obtain samples of arbitrary size from the posterior distribution over the parameters of a structural equation model (SEM) given covariance data and a prior distribution over the parameters. Point estimates, standard deviations and interval estimates for the parameters can be computed from these samples. If the prior distribution over the parameters is uninformative, the posterior is proportional to the likelihood, and asymptotically the inferences based on the Gibbs sample are the same as those based on the maximum likelihood solution, for example, output from LISREL or EQS. In small samples, however, the likelihood surface is not Gaussian and in some cases contains local maxima. Nevertheless, the Gibbs sample comes from the correct posterior distribution over the parameters regardless of the sample size and the shape of the likelihood surface. With an informative prior distribution over the parameters, the posterior can be used to make inferences about the parameters underidentified models, as we illustrate on a simple errors-in-variables model.We thank David Spiegelhalter for suggesting applying the Gibbs sampler to structural equation models to the first author at a 1994 workshop in Wiesbaden. We thank Ulf Böckenholt, Chris Meek, Marijtje van Duijn, Clark Glymour, Ivo Molenaar, Steve Klepper, Thomas Richardson, Teddy Seidenfeld, and Tom Snijders for helpful discussions, mathematical advice, and critiques of earlier drafts of this paper.     

Latent variables should remain as such: Evidence from a Monte Carlo study

Use of subject scores as manifest variables to assess the relationship between latent variables produces attenuated estimates. This has been demonstrated for raw scores from classical test theory (CTT) and factor scores derived from factor analysis. Conclusions on scores have not been sufficiently extended to item response theory (IRT) theta estimates, which are still recommended for estimation of relationships between latent variables. This is because IRT estimates appear to have preferable properties compared to CTT, while structural equation modeling (SEM) is often advised as an alternative to scores for estimation of the relationship between latent variables. The present research evaluates the consequences of using subject scores as manifest variables in regression models to test the relationship between latent variables. Raw scores and three methods for obtaining theta estimates were used and compared to latent variable SEM modeling. A Monte Carlo study was designed by manipulating sample size, number of items, type of test, and magnitude of the correlation between latent variables. Results show that, despite the advantage of IRT models in other areas, estimates of the relationship between latent variables are always more accurate when SEM models are used. Recommendations are offered for applied researchers.     

无均值结构的潜变量交互效应模型的标准化估计

潜变量交互效应建模研究近年来有两项重要进展, 一是提出了潜变量交互效应模型的标准化估计及其计算公式; 二是发现无均值结构模型可以取代传统的有均值结构模型, 建模大为简化。但标准化估计是在传统的有均值结构模型中建立的, 在简化的模型中同样适用吗？本文在无均值结构模型的框架内, 给出了潜变量交互效应模型的标准化形式、计算公式和建模步骤, 并通过模拟研究比较了极大似然和广义最小二乘两种估计方法、配对乘积指标和全部乘积指标两种指标类型, 结果表明, 在计算交互效应的标准化估计时, 应当使用配对乘积指标建模, 并且首选极大似然估计。     

新生市民主动发展倾向对城市生活适应性的影响:社会支持的中介效应

为探究主动发展倾向和社会支持对新生市民城市生活适应性的作用机制,本研究构建了结构方程模型.采用两个自编量表和一个现成量表,运用两阶段抽样调查法,在作为统筹城乡发展试验区的重庆市范围内,对已经取得城镇户口的新生市民进行问卷调查,获得有效样本322人.对采集的数据运用统计分析软件SPSS17.0和AMOS18.0进行数据分析和模型检验.结果表明,城市化新生市民的主动发展倾向和社会支持均对新生市民的城市生活适应具有显著影响,其中,社会支持在主动发展倾向对城市生活适应影响中具有部分中介作用.     

Evaluating the Factor Structure of the MIDI Personality Scale Using Exploratory Structural Equation Modeling

This study investigated the factor structure of the 26‐item Midlife Development Inventory (MIDI) Personality Scale in a sample of 2,720 Americans. It was found that whereas confirmatory factor analysis (CFA) did not provide an acceptable fit to the data, exploratory structural equation modeling (ESEM) provided an acceptable fit. The results of ESEM revealed that the a priori five‐factor structure of personality was generally consistent with the data, and all items had salient loadings on their target factors. ESEM also revealed that some of the items contributed significantly to more than one personality factor. The results are in line with previous research, and indicate that ESEM is more suitable than CFA for the study of personality traits.     

An all‐or‐nothing thinking turns into darkness: Relations between dichotomous thinking and personality disorders1

Dichotomous thinking is an individual's propensity to think in terms of binary opposition. While this thinking style may be useful for quick decision‐making, some clinical psychologists have indicated that such a style is related to personality disorders. Oshio (2009 ) revealed that the Dichotomous Thinking Inventory (DTI) has significant positive correlations with borderline personality and narcissism. This study examines whether dichotomous thinking is associated with cluster A, B, and C personality disorders. The participants comprised 152 Japanese undergraduates (84 male and 68 female). They completed the DTI and Ten Personality Styles scale (10 PesT; Nakazawa (2006 )), developed from the definitions of personality disorders in the DSM‐IV. Structural equation modeling supported the hierarchical factor structure of personality disorders. The dichotomous thinking effects “cluster” level of disorders more than other levels. The results indicated that thinking dichotomously may lead to wide‐ranging personality disorders.     

Stability and Invariance of Psychopathic Traits from Late Adolescence to Young Adulthood

The current study examined the longitudinal stability and invariance of psychopathic traits in a large community sample of male twins from ages 17 to 23. Participants were assessed across 6 years to gauge the stability and measurement invariance of the Minnesota Temperament Inventory (MTI), a Cleckley-based measure of psychopathic personality traits, and how family functioning and externalizing behavior were linked to these traits. A latent variable approach was used to model the structure of the MTI and provide a statistical test of measurement invariance across time. The results revealed support for invariance and moderate to strong stability of the MTI factors, which showed significant associations with the external correlates in late adolescence but not early adulthood.     

Sex differences in latent cognitive abilities ages 5 to 17: Evidence from the Differential Ability Scales—Second Edition

Sex differences in the latent general and broad cognitive abilities underlying the Differential Ability Scales, Second Edition were investigated for children and youth ages 5 through 17. Multi-group mean and covariance structural equation modeling was used to investigate sex differences in latent cognitive abilities as well as changes in these differences across age. Most broad abilities showed mean differences across the sexes, and all such differences varied across ages. Girls showed an advantage on the processing speed (Gs) first-order residual factor. Girls also showed advantages at some ages on free-recall memory, a narrow ability of long term retrieval (Glr). Boys showed a developmentally-related advantage on a visual-spatial ability (Gv) first-order residual factor, depending on age. Younger girls showed an advantage on short-term memory (Gsm). No statistically significant sex differences were shown on the latent comprehension-knowledge (Gc) factor, or on a second-order, latent g factor. Boys showed larger variances for several broad abilities, some substantial, but those differences were not statistically significant.     

I know how you felt last night,or do I? Self- and external ratings of emotions in REM sleep dreams

We investigated whether inconsistencies in previous studies regarding emotional experiences in dreams derive from whether dream emotions are self-rated or externally evaluated. Seventeen subjects were monitored with polysomnography in the sleep laboratory and awakened from every rapid eye movement (REM) sleep stage 5 min after the onset of the stage. Upon awakening, participants gave an oral dream report and rated their dream emotions using the modified Differential Emotions Scale, whereas external judges rated the participants’ emotions expressed in the dream reports, using the same scale. The two approaches produced diverging results. Self-ratings, as compared to external ratings, resulted in greater estimates of (a) emotional dreams; (b) positively valenced dreams; (c) positive and negative emotions per dream; and (d) various discrete emotions represented in dreams. The results suggest that this is mostly due to the underrepresentation of positive emotions in dream reports. Possible reasons for this discrepancy are discussed.