A Comparison of Multilevel Mediation Modeling Methods: Recommendations for Applied Researchers

Multilevel structural equation modeling (MSEM) has been proposed as a valuable tool for estimating mediation in multilevel data and has known advantages over traditional multilevel modeling, including conflated and unconflated techniques (CMM & UMM). Recent methodological research has focused on comparing the three methods for 2-1-1 designs, but in regards to 1-1-1 mediation designs, there are significant gaps in the published literature that prevent applied researchers from making educated decisions regarding which model to employ in their own specific research design. A Monte Carlo study was performed to compare MSEM, UMM, and CMM on relative bias, confidence interval coverage, Type I Error, and power in a 1-1-1 model with random slopes under varying data conditions. Recommendations for applied researchers are discussed and an empirical example provides context for the three methods.     

A Comparison of ML,WLSMV, and Bayesian Methods for Multilevel Structural Equation Models in Small Samples: A Simulation Study

Multilevel structural equation models are increasingly applied in psychological research. With increasing model complexity, estimation becomes computationally demanding, and small sample sizes pose further challenges on estimation methods relying on asymptotic theory. Recent developments of Bayesian estimation techniques may help to overcome the shortcomings of classical estimation techniques. The use of potentially inaccurate prior information may, however, have detrimental effects, especially in small samples. The present Monte Carlo simulation study compares the statistical performance of classical estimation techniques with Bayesian estimation using different prior specifications for a two-level SEM with either continuous or ordinal indicators. Using two software programs (Mplus and Stan), differential effects of between- and within-level sample sizes on estimation accuracy were investigated. Moreover, it was tested to which extent inaccurate priors may have detrimental effects on parameter estimates in categorical indicator models. For continuous indicators, Bayesian estimation did not show performance advantages over ML. For categorical indicators, Bayesian estimation outperformed WLSMV solely in case of strongly informative accurate priors. Weakly informative inaccurate priors did not deteriorate performance of the Bayesian approach, while strong informative inaccurate priors led to severely biased estimates even with large sample sizes. With diffuse priors, Stan yielded better results than Mplus in terms of parameter estimates.     

儿童攻击和自我感知的关系

使用基于多水平分析技术的元分析方法对有关儿童攻击和自我感知相关的研究结果进行分析,并对计算结果用模拟方法进行模拟。结果显示现有研究中攻击和自感认知能力之间的相关为-0．01,攻击和自我总体感知间地相关为-0．08,相关结果为低度负相关但达到显性水平,攻击和自感社交能力之间的相关关系不显,模拟结果与此类似。现有关于儿童攻击和自我感知关系的研究数量还偏少。     

近10年中国飞行员心理健康状况的元分析

[摘要] 目的：探讨近10年中国飞行员的心理健康状况。方法：搜索2001-2010年10年间所有用SCL-90量表研究中国飞行员心理健康的文献,对符合要求的文献用元分析和横断历史研究的方法进行重新处理和分析。结果：（1）2000年到2001年飞行员SCL-90得分增加,2001-2009年飞行员SCL-90数据呈U型曲线,05年得数最低,09年得分最高,其中人际敏感、抑郁、焦虑、恐怖四个因子的分数在这十年中的变化显著;（2）SCL-90各因子的平均效果量在-0.27到-0.46之间。其中“躯体化”、“抑郁”、“焦虑”“恐怖”四个因子与一般成人常模差异不显著,其余五个因子均与常模差异显著,且分数要低于常模;（3）民航飞行员SCL-90的各个因子得分要低于军事飞行员,但未达到显著性水平。     

A Comparison of Inverse-Wishart Prior Specifications for Covariance Matrices in Multilevel Autoregressive Models

Multilevel autoregressive models are especially suited for modeling between-person differences in within-person processes. Fitting these models with Bayesian techniques requires the specification of prior distributions for all parameters. Often it is desirable to specify prior distributions that have negligible effects on the resulting parameter estimates. However, the conjugate prior distribution for covariance matrices—the Inverse-Wishart distribution—tends to be informative when variances are close to zero. This is problematic for multilevel autoregressive models, because autoregressive parameters are usually small for each individual, so that the variance of these parameters will be small. We performed a simulation study to compare the performance of three Inverse-Wishart prior specifications suggested in the literature, when one or more variances for the random effects in the multilevel autoregressive model are small. Our results show that the prior specification that uses plug-in ML estimates of the variances performs best. We advise to always include a sensitivity analysis for the prior specification for covariance matrices of random parameters, especially in autoregressive models, and to include a data-based prior specification in this analysis. We illustrate such an analysis by means of an empirical application on repeated measures data on worrying and positive affect.     

基于多层结构方程模型的情境效应分析—— 兼与多层线性模型比较

多层(嵌套)数据的变量关系研究, 必须借助多层模型来实现。两层模型中, 层一自变量Xij按组均值中心化, 并将组均值 置于层2截距方程式中, 可将Xij对因变量Yij的效应分解为组间和组内部分, 二者之差被称为情境效应, 称为情境变量。多层结构方程模型(MSEM)将多层线性模型(MLM)和结构方程模型(SEM)相结合, 通过设置潜变量和多指标的方法校正了MLM在情境效应分析中出现的抽样误差和测量误差, 同时解决了数据的多层(嵌套)结构和潜变量的估计问题。除了分析原理的说明, 还以班级平均竞争氛围对学生竞争表现的情境效应为例进行分析方法的示范, 并比较MSEM和MLM的异同, 随后展望了MSEM情境效应模型、情境效应无偏估计方法和情境变量研究的拓展方向。     

A Multilevel Contextual Model of Family Conflict and Deviance

This paper presents a multiple-group multivariate hierarchical specification of family problem behaviors across ethnicities using structural equation modeling techniques which explicitly model the individual-level and family-level covariance matrices in familial problem behavior. Analyses were conducted across White and African American ethnic groups. The sample (N = 1,168; 647 White and 521 African Americans) comprised children and their parents from 392 families. In addition to relations between family conflict and deviant behaviors, covariates were included at each level of analysis: neighborhood desirability at the family (between) level, and age and gender at the individual (within) level. At the between level, neighborhood desirability influenced family conflict and family conflict influenced family levels of deviance. At the within level, conflict was significantly related to individual levels of deviance. Discussion focuses on the substantive results as well as the application of multilevel analyses to contextual influences of family problem behavior.     

儿童攻击和同伴关系的相关:20年研究的元分析

用多水平分析技术对近20年关于儿童攻击和同伴关系相关的研究结果进行的元分析显示,现有研究中关于攻击和同伴关系相关结果的变异程度比较大,总体来说攻击和同伴接受之间有负向相关关系,和同伴拒绝之间有正向相关关系。     

Multilevel Investigation into the Community Context of Child Maltreatment

Official rates of child maltreatment vary considerably from community to community. Extensive research on the role of the community context in maltreatment suggests that the neighborhood in which families live could impact their maltreatment behavior. However, the vast majority of existing studies and literature reviews on the topic have not used statistical methods to appropriately control for the impact of these variables at the individual level. This literature review approaches the question of whether the community context affects maltreatment behavior above and beyond the individual-level variables by critiquing the studies that have successfully measured maltreatment risks at both the individual and community level using multilevel modeling.     

三类多层中介效应分析方法比较

比较了贝叶斯法、Monte Carlo法和参数Bootstrap法在2-1-1多层中介分析中的表现。结果发现：1)有先验信息的贝叶斯法的中介效应点估计和区间估计都最准确;2)无先验信息的贝叶斯法、Monte Carlo法、偏差校正和未校正的参数Bootstrap法的中介效应点估计和区间估计表现相当,但Monte Carlo法在第Ⅰ类错误率和区间宽度指标上表现略优于其他三种方法,偏差校正的Bootstrap法在统计检验力上表现略优于其他三种方法,但在第Ⅰ类错误率上表现最差;结果表明,当有先验信息时,推荐使用贝叶斯法;当先验信息不可得时,推荐使用Monte Carlo法。     

儿童期受虐经历与抑郁的元分析

本研究采用元分析技术探讨儿童期虐待亚类型与抑郁的关系。通过文献搜索与检查获得了31项研究和92个独立效应量,共包含24283名被试。元分析表明儿童期虐待亚类型(心理虐待与忽视、性虐待、身体虐待与忽视)对抑郁障碍存在不同程度的影响,其中心理虐待与忽视影响程度最高;另外,被试的年龄、性别可以调节儿童期虐待与抑郁的关系。这些结果表明了在预防和治疗儿童期不良经历对抑郁影响时,尤其要关注心理虐待与忽视的消极影响,且应给予青少年群体、女性更多的关注和保护。     

The Validity of the Employment Interview: A Meta-Analysis

The interview is the most widely used personnel selection method, but has revealed low reliability and validity compared with other selection methods (Mayfield 1964; Ulrich and Trumbo 1965; Schmitt 1976; Arvey 1979). Thirty-one studies on the validity of the interview were meta-analyzed. The result was an average validity coefficient of .27. The estimated true validity of the interview was calculated to be .38. These values indicated that the interview has moderate validity as a personnel selection device. Six characteristics of the interview were also examined in relation to the validity of the interview: structure of the interview; number of interviewers; length of the interview; gender of the applicant pool; blue-collar/white-collar jobs; and use of college students versus job applicants. The six study characteristics accounted for 30.9% of the variance in the validity of the interview. Structure of the interview appeared to be the only characteristic that moderated the validity of the interview. The relationship of this study to other meta-analyses of the employment interview is discussed.     

Socio-Economic Status and Creativity: A Meta-Analysis

In the present study, we examined the relationship between socio-economic status (SES) and creativity through a meta-analysis of 885 effect sizes from 151 samples in 117 studies. Analyses of a multivariate model with a robust variance estimator showed a small positive correlation between creativity and SES (r = .120, 95% CI [0.086, 0.154]). Further analyses with four groups of moderators (i.e., creativity measure, SES indicator, sample characteristics, and study characteristics) indicated that the mean effect size is significantly higher when the creativity index is flexibility rather than fluency or when SES is measured by parental or one's own level of education and parental occupation rather than the location of residency. In addition, the relationship between creativity and SES dissipated as the students progressed from elementary school through college. The potential role of schooling in the creativity-SES link is discussed.     

A Meta-Analysis of After-School Programs That Seek to Promote Personal and Social Skills in Children and Adolescents

A meta-analysis of after-school programs that seek to enhance the personal and social skills of children and adolescents indicated that, compared to controls, participants demonstrated significant increases in their self-perceptions and bonding to school, positive social behaviors, school grades and levels of academic achievement, and significant reductions in problem behaviors. The presence of four recommended practices associated with previously effective skill training (SAFE: sequenced, active, focused, and explicit) moderated several program outcomes. One important implication of current findings is that ASPs should contain components to foster the personal and social skills of youth because youth can benefit in multiple ways if these components are offered. The second implication is that further research is warranted on identifying program characteristics that can help us understand why some programs are more successful than others.     

A Meta-Analysis of the Relationship between Curiosity and Creativity

Curiosity, the desire to know, may be associated with creativity, which involves generating ideas that are novel and valuable. This meta-analytic investigation consolidated the results of studies of the association between curiosity and creativity. Across 10 studies, which included 2,692 individuals, there was a significant association between more curiosity and greater creativity (weighted effect size r = .41, 95% CI [.27, .54], p = .0001). For studies examining the association of the exploration dimension of curiosity with creativity, the weighted effect size was r = .48, 95% CI [.09, .74], while for studies examining the deprivation sensitivity dimension of curiosity with creativity the weighted effect size was r = .20, 95% CI [.10, .29]. The association of self-report measures of curiosity with self-reports of creativity was r = .52, 95% CI [.40, .62], while the association of self-report measures of curiosity with rated creativity was r = .16, 95% CI [.10, .22]. These meta-analytic results are congruent with some theoretical assumptions regarding curiosity and creativity and can be a foundation for efforts to facilitate creativity.     

Persuasion Knowledge in the Marketplace: A Meta-Analysis

Since the introduction of the persuasion knowledge model more than 25 years ago, many research studies have investigated how consumers’ persuasion knowledge affects their reactions to persuasion attempts. While most results have shown that persuasion knowledge increases coping responses and leads to less favorable evaluations of marketer actions, the findings vary considerably, leaving researchers with a limited understanding of the substance and structure of persuasion knowledge effects and the conditions that explain their variability. To develop a better understanding of persuasion knowledge effects in the marketplace, this study builds on the concept of persuasion to predict responses to marketers’ attempts to persuade consumers with different levels of persuasion knowledge. The study presents a meta-analysis of the findings in 148 papers and 171 distinct data sets. Persuasion knowledge effects can be viewed as substantial compared with persuasion attempts, but persuasion knowledge cannot suppress or eliminate persuasion effects in the marketplace, as it only reaches around 50% of the explanatory power of persuasion. Persuasion knowledge effects on evaluations and coping depend on the characteristics of the persuasion process. All persuasion elements that help consumers identify and better understand benefits not just for themselves, but also for marketers and how marketers realize their benefits—such as the use of personal communication, communication about unfamiliar products or products with experience attributes, and receiver experience—lead to less favorable effects for marketers. This paper’s insights provide a new framework for persuasion knowledge effects in the marketplace, ideas for future research, and implications for researchers, consumers, policymakers, and marketers.     

Double Decomposition of Level-1 Variables in Multilevel Models: An Analysis of the Flynn Effect in the NSLY Data

This paper introduces an extension of cluster mean centering (also called group mean centering) for multilevel models, which we call “double decomposition (DD).” This centering method separates between-level variance, as in cluster mean centering, but also decomposes within-level variance of the same variable. This process retains the benefits of cluster mean centering but allows for context variables derived from lower level variables, other than the cluster mean, to be incorporated into the model. A brief simulation study is presented, demonstrating the potential advantage (or even necessity) for DD in certain circumstances. Several applications to multilevel analysis are discussed. Finally, an empirical demonstration examining the Flynn effect (Flynn, 1987 Flynn, J. R. (1987). Massive IQ gains in 14 nations: What IQ tests really measure. Psychological Bulletin, 101(2), 171. https://doi.org/10.1037/h0090408.[Crossref], [Web of Science ®] , [Google Scholar]), our motivating example, is presented. The use of DD in the analysis provides a novel method to narrow the field of plausible causal hypotheses regarding the Flynn effect, in line with suggestions by a number of researchers (Mingroni, 2014 Mingroni, M. A. (2014). Future efforts in Flynn effect research: Balancing reductionism with holism. Journal of Intelligence, 2(4), 122. https://doi.org/10.3390/jintelligence2040122.[Crossref] , [Google Scholar]; Rodgers, 2015 Rodgers, J. L. (2015). Methodological issues associated with studying the Flynn effect: Exploratory and confirmatory efforts in the past, present, and future. Journal of Intelligence, 3(4), 111. https://doi.org/10.3390/jintelligence3040111. [Crossref] , [Google Scholar]).     

The Relation between Motivation and Goal Attainment: A Correlational Meta-Analysis

This meta-analysis investigated the strengths of the relationship between various types of motivations and accompanying future outcomes that individuals intend to change, based upon 337 effect sizes from 62 studies. Considerable variation exists within and between the effect sizes of the 14 types of motivations, ranging from a small negative effect size, r = −.13, I² = 93.85% (k = 13), to a medium positive effect size, r = .38, I² = 0.0% (k = 3). The following factors moderated some of the 14 summary effect sizes: (a) the type of assessment data (self-report vs. physical data); (b) the type of future outcomes (physical behavior, psychological state, and intellectual ability); (c) the use of a motivational intervention; (d) the use of a longitudinal design; and (e) the time period between the point that measured motivation and future outcomes. The moderating effects suggest that the effect size of motivations may fluctuate across various domains, while future outcomes may be almost unaffected or even affected negatively by particular types of motivations, although certain other types of motivations play positive roles.     

The Country's Crime Rate Moderates the Relation Between Authoritarian Predispositions and the Manifestations of Authoritarianism: A Multilevel,Multinational Study

We performed a multilevel, multinational test of Stenner's model on authoritarianism using the 2008 European Values Survey dataset (N = 55 199, nested in 38 nations). We focussed on the effects exerted on four authoritarian manifestations (racial intolerance, political intolerance, negative attitudes towards immigrants, and moral intolerance) by the cross‐level interaction between participants' authoritarian predispositions (assessed in terms of childrearing values) and their country's crime rate. Associations between authoritarian predispositions and racial intolerance, political intolerance, negative attitudes towards immigrants, and moral intolerance were significantly stronger among participants living in countries characterised by high crime rates than those among participants living in countries with low crime rates. Limitations, implications, and future directions of this study are discussed. Copyright © 2013 European Association of Personality Psychology.     

Multilevel covariance structure analysis of sibling substance use and intrafamily conflict

Conventional covariance structure analysis, such as factor analysis, is often applied to data that are obtained in a hierarchical fashion, such as siblings observed within families. A more appropriate specification is demonstrated which explicitly models the within-level and between-level covariance matrices of sibling substance use and intrafamily conflict. Participants were 267 target adolescents (mean age=13.11 years) and 318 siblings (mean age=15.03 years). The level of homogeneity within sibling clusters, and heterogeneity among families, was sufficient to conduct a multilevel covariance structure analysis (MCA). Parental and family-level variables consisting of marital status, socioeconomic status, marital discord, parent use, and modeling of substances were used to explain heterogeneity among families. Marital discord predicted intrafamily conflict, and parent marital status and modeling of substances predicted sibling substance use. Advantages and uses of hierarchical designs and conventional covariance structure software for multilevel data are discussed.