Multivariate Behavioral Research最新文献

Student evaluation of teaching (SET) questionnaires are ubiquitously applied in higher education institutions in North America for both formative and summative purposes. Data collected from SET questionnaires are usually item-level data with cross-classified structure, which are characterized by multivariate categorical outcomes (i.e., multiple Likert-type items in the questionnaires) and cross-classified structure (i.e., non-nested students and instructors). Recently, a new approach, namely the cross-classified IRT model, was proposed for appropriately handling SET data. To inform researchers in higher education, in this article, the cross-classified IRT model, along with three existing approaches applied in SET studies, including the cross-classified random effects model (CCREM), the multilevel item response theory (MLIRT) model, and a two-step integrated strategy, was reviewed. The strengths and weaknesses of each of the four approaches were also discussed. Additionally, the new and existing approaches were compared through an empirical data analysis and a preliminary simulation study. This article concluded by providing general suggestions to researchers for analyzing SET data and discussing limitations and future research directions.

With clustered data, such as where students are nested within schools or employees are nested within organizations, it is often of interest to estimate and compare associations among variables separately for each level. While researchers routinely estimate between-cluster effects using the sample cluster means of a predictor, previous research has shown that such practice leads to biased estimates of coefficients at the between level, and recent research has recommended the use of latent cluster means with the multilevel structural equation modeling framework. However, the latent cluster mean approach may not always be the best choice as it (a) relies on the assumption that the population cluster sizes are close to infinite, (b) requires a relatively large number of clusters, and (c) is currently only implemented in specialized software such as Mplus. In this paper, we show how using empirical Bayes estimates of the cluster means can also lead to consistent estimates of between-level coefficients, and illustrate how the empirical Bayes estimate can incorporate finite population corrections when information on population cluster sizes is available. Through a series of Monte Carlo simulation studies, we show that the empirical Bayes cluster-mean approach performs similarly to the latent cluster mean approach for estimating the between-cluster coefficients in most conditions when the infinite-population assumption holds, and applying the finite population correction provides reasonable point and interval estimates when the population is finite. The performance of EBM can be further improved with restricted maximum likelihood estimation and likelihood-based confidence intervals. We also provide an R function that implements the empirical Bayes cluster-mean approach, and illustrate it using data from the classic High School and Beyond Study.

Accelerated longitudinal designs allow researchers to efficiently collect longitudinal data covering a time span much longer than the study duration. One important assumption of these designs is that each cohort (a group defined by their age of entry into the study) shares the same longitudinal trajectory. Although previous research has examined the impact of violating this assumption when each cohort is defined by a single age of entry, it is possible that each cohort is instead defined by a range of ages, such as groups that experience a particular historical event. In this paper we examined how including cohort membership in linear and quadratic multilevel models performed in detecting and controlling for cohort effects in this scenario. Using a Monte Carlo simulation study, we assessed the performance of this approach under conditions related to the number of cohorts, the overlap between cohorts, the strength of the cohort effect, the number of affected parameters, and the sample size. Our results indicate that models including a proxy variable for cohort membership based on age at study entry performed comparably to using true cohort membership in detecting cohort effects accurately and returning unbiased parameter estimates. This indicates that researchers can control for cohort effects even when true cohort membership is unknown.

Recent shifts to prioritize prediction, rather than explanation, in psychological science have increased applications of predictive modeling methods. However, composite predictors, such as sum scores, are still commonly used in practice. The motivations behind composite test scores are largely intertwined with reducing the influence of measurement error in answering explanatory questions. But this may be detrimental for predictive aims. The present paper examines the impact of utilizing composite or item-level predictors in linear regression. A mathematical examination of the bias-variance decomposition of prediction error in the presence of measurement error is provided. It is shown that prediction bias, which may be exacerbated by composite scoring, drives prediction error for linear regression. This may be particularly salient when composite scores are comprised of heterogeneous items such as in clinical scales where items correspond to symptoms. With sufficiently large training samples, the increased prediction variance associated with item scores becomes negligible even when composite scores are sufficient. Practical implications of predictor scoring are examined in an empirical example predicting suicidal ideation from various depression scales. Results show that item scores can markedly improve prediction particularly for symptom-based scales. Cross-validation methods can be used to empirically justify predictor scoring decisions.

Network analysis has gained popularity as an approach to investigate psychological constructs. However, there are currently no guidelines for applied researchers when encountering missing values. In this simulation study, we compared the performance of a two-step EM algorithm with separated steps for missing handling and regularization, a combined direct EM algorithm, and pairwise deletion. We investigated conditions with varying network sizes, numbers of observations, missing data mechanisms, and percentages of missing values. These approaches are evaluated with regard to recovering population networks in terms of loss in the precision matrix, edge set identification and network statistics. The simulation showed adequate performance only in conditions with large samples ($n\ge 500$) or small networks (p = 10). Comparing the missing data approaches, the direct EM appears to be more sensitive and superior in nearly all chosen conditions. The two-step EM yields better results when the ratio of n/p is very large - being less sensitive but more specific. Pairwise deletion failed to converge across numerous conditions and yielded inferior results overall. Overall, direct EM is recommended in most cases, as it is able to mitigate the impact of missing data quite well, while modifications to two-step EM could improve its performance.

The graded forced-choice (FC) format has recently emerged as an alternative that may preserve the advantages and overcome the issues of the dichotomous FC measures. The current study presented the first large-scale evaluation of the performance of three types of FC measures (FC2, FC4 and FC5 with 2, 4 and 5 response options, respectively) and compared their performance to their Likert (LK) counterparts (LK2, LK4, and LK5) on (1) psychometric properties, (2) respondent reactions, and (3) susceptibility to response styles. Results showed that, compared to LK measures with the same number of response options, the three FC scales provided better support for the hypothesized factor structure, were perceived as more faking-resistant and cognitive demanding, and were less susceptible to response styles. FC4/5 and LK4/5 demonstrated similarly good reliability, while LK2 provided more reliable scores than FC2. When compared across the three FC measures, FC4 and FC5 displayed comparable psychometric performance and respondent reactions. FC4 exhibited a moderate presence of extreme response style, while FC5 had a weak presence of both extreme and middle response styles. Based on these findings, the study recommends the use of graded FC over dichotomous FC and LK, particularly FC5 when extreme response style is a concern.