Bayesian meta-analysis using skewed elliptical distributions

Meta-analysis refers to a quantitative method for combining results from independent studies in order to draw overall conclusions. We consider hierarchical models including selection models under a skewed heavy tailed error distribution proposed originally by Chen, Dey, and Shao [M. H. Chen, D. K. Dey, Q. M. Shao, A new skewed link model for dichotomous quantal response data, J. Amer. Statist. Assoc. 94 (1983), pp. 1172–1186.] and Branco and Dey [D. Branco and D.K. Dey, A general class of multivariate skew-elliptical distributions, J. Multivariate Anal. 79, pp. 99–113.]. These rich classes of models combine the information of independent studies, allowing investigation of variability both between and within studies and incorporating weight functions. We constructed a detailed computational scheme under skewed normal and skewed Student's t distribution using the MCMC method. Bayesian model selection was conducted by Bayes factor under a different skewed error. Finally, we illustrated our methodology using a real data example taken from Johnson [M.F. Johnson, Comparative efficacy of Naf and SMFP dentifrices in caries prevention: a meta-analysis overview, J Eur. Organ. Caries Res. 27 (1993), pp. 328–336.].     

Emotional strain and organizational citizenship behaviours: A meta-analysis and review

This paper provides a qualitative review and quantitative summary of the relationship between emotional strain and organizational citizenship behaviour (OCB), and discusses five potential moderators of the strain-OCB relationship. OCB refers to discretionary behaviours that benefit organizations and their members. Emotional strain is important to consider because it has a broad impact on employee behaviours and is possibly more fundamental than other forms of strain. However, it has received less attention than aspects of job-related strain, such as job dissatisfaction. Based on the results of 29 empirical studies with 52 unique effect sizes, meta-analytic results revealed a negative relationship between strain and OCB (corrected estimate of the population correlation coefficient, ρ=-.16). Furthermore, this relationship is moderated by the type of OCB (OCB directed at the organization vs. that directed at individuals), type of organization (private vs. public), publication status (published vs. unpublished), OCB rating source (self vs. other), and type of sample (full-time employees vs. employed students). We present theoretical and practical implications of these findings, including steps that could be taken by organizations to increase OCB and to reduce emotional strain, and suggest directions for future research.     

Behavior Genetic Research on Gambling and Problem Gambling: A Preliminary Meta-Analysis of Available Data

A meta-analysis of family and twin studies on gambling and problem gambling was initiated in an effort to determine the probable role of genetic factors in high risk wagering. Two twin studies and 17 investigations employing the family history or family study method were included in this meta-analysis. A summation of the 19 studies produced a small but significant overall effect size (weighted = .10, unweighted = .13), with both family (weighted = .12, unweighted = .14) and twin (weighted = .06, unweighted = .05) studies achieving significant individual mean effects. Given the paucity of twin data, further analysis was confined to family studies and revealed a stronger familial effect for the sons of problem gambling fathers than for the daughters of problem gambling mothers and for more severe forms of problem gambling than for less severe forms of problem wagering, and was strongest for high severity problem gambling in males. The implications of these results are discussed.     

Conducting systematic literature review in operations management

Systematic literature review (SLR) is a well-known research method. However, there is a paucity of detailed SLR guidelines in operations management (OM). The recent interest in SLR in OM has not been followed by the same rigour observed in disciplines as medical sciences and public policy. There are no OM-specific SLR protocols, detailed step-by-step methods and reporting procedures. Therefore, this paper provides a step-by-step approach to SLR for OM scholars and an overview of SLR’s evolution as a research method in OM and the resulting progression of themes. The step-by-step approach aims to serve as a guideline sufficiently broad to avoid skipping any significant step, but still being easy to be understood and applied. The paper describes procedures for rigourous SLR, reveals a growing use of literature review in OM, specially for qualitative SLR and traditional narrative reviews, assesses contemporary and emerging themes in OM, and provides a research agenda.     

A Meta-Analytic Review of the Association Between Disgust and Prejudice Toward Gay Men

ABSTRACT

A sizeable number of studies have documented a relationship between heterosexual persons’ experience of disgust (measured as an individual difference variable or induced experimentally) and prejudice toward gay men (i.e., homonegativity). Yet, to date, no one has attempted to meta-analytically review this corpus of research. We address this gap by conducting a meta-analysis of published and unpublished work examining heterosexual men and women’s disgust and their homonegativity toward gay men. Fourteen articles (12 published, two unpublished) containing 17 studies were analyzed (N = 7,322). The average effect size for disgust sensitivity studies was moderate to large (d = 0.64), whereas for disgust induction studies, the effect was large (d = 0.77). No evidence of effect size heterogeneity emerged. Future directions and recommendations for methodological improvements are outlined.     

Editorial note: referees

Abstract

This study investigates three methods to handle dependency among effect size estimates in meta-analysis arising from studies reporting multiple outcome measures taken on the same sample. The three-level approach is compared with the method of robust variance estimation, and with averaging effects within studies. A simulation study is performed, and the fixed and random effect estimates of the three methods are compared with each other. Both the robust variance estimation and three-level approach result in unbiased estimates of the fixed effects, corresponding standard errors and variances. Averaging effect sizes results in overestimated standard errors when the effect sizes within studies are truly independent. Although the robust variance and three-level approach are more complicated to use, they have the advantage that they do not require an estimate of the correlation between outcomes, and they still result in unbiased parameter estimates.     

Rating Methodological Quality: Toward Improved Assessment and Investigation

Assessing methodological quality is considered essential in deciding what investigations to include in research syntheses and in detecting potential sources of bias in meta-analytic results. Quality assessment is also useful in characterizing the strengths and limitations of the research in an area of study. Although numerous instruments to measure research quality have been developed, they have lacked empirically-supported components. In addition, different summary quality scales have yielded different findings when they were used to weight treatment effect estimates for the same body of research. Suggestions for developing improved quality instruments include: distinguishing distinct domains of quality, such as internal validity, external validity, the completeness of the study report, and adherence to ethical practices; focusing on individual aspects, rather than domains of quality; and focusing on empirically-verified criteria. Other ways to facilitate the constructive use of quality assessment are to improve and standardize the reporting of research investigations, so that the quality of studies can be more equitably and thoroughly compared, and to identify optimal methods for incorporating study quality ratings into meta-analyses.     

Organizational constraints: a meta-analysis of a major stressor

The stressor organizational constraints, reflecting aspects of the work environment that inhibit or interfere with an individual's performance of job tasks, is prevalent in the workplace. This meta-analysis summarizes findings concerning the relationships of organizational constraints with other variables. Using five search methods, we located 84 research reports that provided data from 119 independent samples that contained 33,998 employed individuals. Fitting a stressor–strain framework, constraints had significant relationships with behavioural (counterproductive work behaviour), physical (somatic symptoms), and psychological (e.g. emotions and job dissatisfaction) strains and with well-being variables. Relationships were also found with all other work environment variables, and most personal characteristics (demographics and personality) except for gender. Moreover, a series of regression and relative weights analyses based on meta-analytic correlations showed that constraints had a unique contribution to strain beyond other stressors. The current meta-analysis shows that many commonly studied variables are related to organizational constraints, and it should more often be the centre of empirical research and theoretical development.     

家庭社会经济地位与儿童汉语阅读能力关系的元分析

众多实证研究表明,家庭社会经济地位是影响儿童汉语阅读能力发展的重要因素,然而已有研究结论存在分歧。采用元分析的方法,筛选出发表2002-2022 的 8篇实证研究进效应统计效应验结表家庭社会经济地位综合指标及其各单项指标与儿童汉语阅读能力均显著相关。学段的调节效应检验结果表明,与小学阶段相比,学前阶段儿童父母的职业和家庭收人与儿童汉语阅读能力的相关关系更强;汉语阅读能力测量类型的调节效应结果表明,汉语阅读能力测量类型不同,父母职业和家庭收人与儿童汉语阅读能力的相关关系存在差异。上述结果表明。家庭社会经济地位与儿童汉语阅读能力关系密切,同时还应考量儿童学段和汉语阅读能力测量类型在二者关系中的作用。