有调节的中介模型检验方法：竞争还是替补?

在心理和其他社科研究领域, 经常遇到中介和调节变量。模型的变量多于3个时, 可能同时包含中介和调节变量, 一种常见的模型是有调节的中介模型。本文检视文献上各种检验有调节的中介模型的方法, 理清方法之间是竞争关系(分清优劣)还是替补关系(分清先后), 在此基础上总结出检验有调节的中介模型的步骤, 并用一个实例进行演示。文中也讨论了有调节的中介模型与有中介的调节模型的联系与区别。     

有调节的多层中介效应分析

多层中介和多层调节效应分析在社科领域已常有应用,但如果将多层中介和调节整合在一起,可以产生2（多层中介类型）×2（调节变量的层次）×3（调节的中介路径）共12种有调节的多层中介模型。面对有调节的多层中介效应分析,研究者往往束手无策。详述了基于多层线性模型的12种有调节的多层中介的分析方法和基于多层结构方程模型的4类有调节的多层中介分析方法,包括正交分割法,随机系数预测法,潜调节结构方程法和贝叶斯合理值法。这四类方法的核心议题在于如何处理潜调节项。当样本量足够大时,建议选择潜调节结构方程法;当样本量不足时,建议选择贝叶斯合理值法。随后用一个实际例子演示如何进行有调节的多层中介效应分析并有相应的Mplus程序。最后展望了有调节的多层中介效应分析研究的拓展方向。     

Causal graphical views of fixed effects and random effects models

Despite the long-standing discussion on fixed effects (FE) and random effects (RE) models, how and under what conditions both methods can eliminate unmeasured confounding bias has not yet been widely understood in practice. Using a simple pretest–posttest design in a linear setting, this paper translates the conventional algebraic formalization of FE and RE models into causal graphs and provides intuitively accessible graphical explanations about their data-generating and bias-removing processes. The proposed causal graphs highlight that FE and RE models consider different data-generating models. RE models presume a data-generating model that is identical to a randomized controlled trial, while FE models allow for unobserved time-invariant treatment–outcome confounding. Augmenting regular causal graphs that describe data-generating processes by adding the computational structures of FE and RE estimators, the paper visualizes how FE estimators (gain score and deviation score estimators) and RE estimators (quasi-deviation score estimators) offset unmeasured confounding bias. In contrast to standard regression or matching estimators that reduce confounding bias by blocking non-causal paths via conditioning, FE and RE estimators offset confounding bias by deliberately creating new non-causal paths and associations of opposite sign. Though FE and RE estimators are similar in their bias-offsetting mechanisms, the augmented graphs reveal their subtle differences that can result in different biases in observational studies.     

Bayesian Causal Mediation Analysis for Group Randomized Designs with Homogeneous and Heterogeneous Effects: Simulation and Case Study

A fully Bayesian approach to causal mediation analysis for group-randomized designs is presented. A unique contribution of this article is the combination of Bayesian inferential methods with G-computation to address the problem of heterogeneous treatment or mediator effects. A detailed simulation study shows that this approach has excellent frequentist properties, particularly in the case of small sample sizes with accurate informative priors. The simulation study also demonstrates that the proposed approach can take into account heterogeneous treatment or mediator effects without bias. A case study using data from a school-based randomized intervention designed to increase parent social capital leading to improved behavioral and academic outcomes in children is offered to illustrate the Bayesian approach to causal mediation in group-randomized designs.     

基于不对称区间估计的有调节的中介模型检验

有调节的中介模型是中介过程受到调节变量影响的模型。评介了基于Bootstrap不对称置信区间和贝叶斯不对称可靠区间进行有调节的中介模型检验的3种方法, 包括亚组分析法、差异分析法和系数乘积法。模拟研究发现, 偏差校正的百分位Bootstrap置信区间和无先验信息的贝叶斯可靠区间在有调节的中介模型检验中表现相当, 都优于百分位Bootstrap置信区间的表现。建议使用系数乘积法进行第一阶段或第二阶段被调节的中介模型检验, 使用差异分析法进行两阶段被调节的中介模型检验, 并用一个实际例子演示如何用不对称区间估计检验有调节的中介模型。随后评述了3种有调节的中介模型检验方法在国内心理学的应用现状, 并展望了检验的拓展方向。     

基于结构方程模型的有调节的中介效应分析

有调节的中介模型是中介过程受到调节变量影响的模型。指出了目前有调节的中介效应分析普遍存在的问题：当前有调节的中介效应检验大多使用多元线性回归分析,忽略了测量误差;而基于结构方程模型(SEM)的有调节的中介效应分析需要产生乘积指标,又会面临乘积指标生成和乘积项非正态分布的问题。在简介潜调节结构方程(LMS)方法后,建议使用LMS方法得到偏差校正的bootstrap置信区间来进行基于SEM的有调节的中介效应分析。总结出一个有调节的中介SEM分析流程,并有示例和相应的Mplus程序。文末展望了LMS和有调节的中介模型的发展方向。     

Methods for integrating moderation and mediation: a general analytical framework using moderated path analysis

Studies that combine moderation and mediation are prevalent in basic and applied psychology research. Typically, these studies are framed in terms of moderated mediation or mediated moderation, both of which involve similar analytical approaches. Unfortunately, these approaches have important shortcomings that conceal the nature of the moderated and the mediated effects under investigation. This article presents a general analytical framework for combining moderation and mediation that integrates moderated regression analysis and path analysis. This framework clarifies how moderator variables influence the paths that constitute the direct, indirect, and total effects of mediated models. The authors empirically illustrate this framework and give step-by-step instructions for estimation and interpretation. They summarize the advantages of their framework over current approaches, explain how it subsumes moderated mediation and mediated moderation, and describe how it can accommodate additional moderator and mediator variables, curvilinear relationships, and structural equation models with latent variables.     

日常环境中的暴力暴露对攻击行为的长期影响:一个有调节的中介模型

以850名大学生为被试,探讨日常环境中的暴力暴露与攻击行为的关系,并在此基础上提出一个有调节的中介模型,考察攻击性信念的中介作用和人际信任的调节作用。结果发现:(1)日常环境中的暴力暴露对攻击行为有显著的正向预测作用;(2)攻击性信念在日常环境中的暴力暴露与攻击行为的关系中起部分中介作用;(3)攻击性信念的中介作用会受到人际信任的调节。对于低人际信任个体,攻击性信念起部分中介作用;对于高人际信任个体,攻击性信念的中介效应不显著,日常环境中的暴力暴露对攻击行为只有直接作用。     

More Precise Estimation of Lower-Level Interaction Effects in Multilevel Models

In hierarchical data, the effect of a lower-level predictor on a lower-level outcome may often be confounded by an (un)measured upper-level factor. When such confounding is left unaddressed, the effect of the lower-level predictor is estimated with bias. Separating this effect into a within- and between-component removes such bias in a linear random intercept model under a specific set of assumptions for the confounder. When the effect of the lower-level predictor is additionally moderated by another lower-level predictor, an interaction between both lower-level predictors is included into the model. To address unmeasured upper-level confounding, this interaction term ought to be decomposed into a within- and between-component as well. This can be achieved by first multiplying both predictors and centering that product term next, or vice versa. We show that while both approaches, on average, yield the same estimates of the interaction effect in linear models, the former decomposition is much more precise and robust against misspecification of the effects of cross-level and upper-level terms, compared to the latter.     

When moderation is mediated and mediation is moderated

Procedures for examining whether treatment effects on an outcome are mediated and/or moderated have been well developed and are routinely applied. The mediation question focuses on the intervening mechanism that produces the treatment effect. The moderation question focuses on factors that affect the magnitude of the treatment effect. It is important to note that these two processes may be combined in informative ways, such that moderation is mediated or mediation is moderated. Although some prior literature has discussed these possibilities, their exact definitions and analytic procedures have not been completely articulated. The purpose of this article is to define precisely both mediated moderation and moderated mediation and provide analytic strategies for assessing each.     

Social capital,migration stress,depression and sexual risk behaviors among rural-to-urban migrants in China: a moderated mediation modeling analysis

Background: Few studies have examined the complex relationship of migration stress and depression with sexual risk behaviors among migrants. The relationship between migration stress and sexual risk behaviors may be mediated by depression, and the mediation process may be modified by social capital. The study aims to investigate this moderated mediation mechanism among rural-to-urban migrants.

Methods: Data were collected from rural-to-urban migrants in China. Migration stress, depression, and social capital were measured with validated scales and used as predictor, mediator and moderator, respectively, to predict the likelihood of having sex with risk partners. Mediation and moderated mediation models were used to analyze the data.

Results: Depression significantly mediated the migration stress–sex with risk partner relationship for males (the indirect effect [95%CI]?=?0.36 [0.08, 0.66]); the mediation effect was not significant for females (0.31 [?0.82, 0.16]). Among males, social capital significantly moderated the depression-sex with risk partner relation with moderation effect ?0.12 [?0.21, ?0.04], ?0.21 [?0.41, ?0.01] and ?0.17 [?0.30, ?0.05] for total, bonding and bridging capital respectively.

Conclusion: Social capital may weaken the association between migration stress and sexual risk behavior by buffering the depression-sexual risk behaviors association for males. Additional research is needed to examine this issue among females.     

What mediation analysis can (not) do

The present article is concerned with a common misunderstanding in the interpretation of statistical mediation analyses. These procedures can be sensibly used to examine the degree to which a third variable (Z) accounts for the influence of an independent (X) on a dependent variable (Y) conditional on the assumption that Z actually is a mediator. However, conversely, a significant mediation analysis result does not prove that Z is a mediator. This obvious but often neglected insight is substantiated in a simulation study. Using different causal models for generating Z (genuine mediator, spurious mediator, correlate of the dependent measure, manipulation check) it is shown that significant mediation tests do not allow researchers to identify unique mediators, or to distinguish between alternative causal models. This basic insight, although well understood by experts in statistics, is persistently ignored in the empirical literature and in the reviewing process of even the most selective journals.     

The Combined Effects of Measurement Error and Omitting Confounders in the Single-Mediator Model

Mediation analysis requires a number of strong assumptions be met in order to make valid causal inferences. Failing to account for violations of these assumptions, such as not modeling measurement error or omitting a common cause of the effects in the model, can bias the parameter estimates of the mediated effect. When the independent variable is perfectly reliable, for example when participants are randomly assigned to levels of treatment, measurement error in the mediator tends to underestimate the mediated effect, while the omission of a confounding variable of the mediator-to-outcome relation tends to overestimate the mediated effect. Violations of these two assumptions often co-occur, however, in which case the mediated effect could be overestimated, underestimated, or even, in very rare circumstances, unbiased. To explore the combined effect of measurement error and omitted confounders in the same model, the effect of each violation on the single-mediator model is first examined individually. Then the combined effect of having measurement error and omitted confounders in the same model is discussed. Throughout, an empirical example is provided to illustrate the effect of violating these assumptions on the mediated effect.     

Effect size measures for mediation models: quantitative strategies for communicating indirect effects

The statistical analysis of mediation effects has become an indispensable tool for helping scientists investigate processes thought to be causal. Yet, in spite of many recent advances in the estimation and testing of mediation effects, little attention has been given to methods for communicating effect size and the practical importance of those effect sizes. Our goals in this article are to (a) outline some general desiderata for effect size measures, (b) describe current methods of expressing effect size and practical importance for mediation, (c) use the desiderata to evaluate these methods, and (d) develop new methods to communicate effect size in the context of mediation analysis. The first new effect size index we describe is a residual-based index that quantifies the amount of variance explained in both the mediator and the outcome. The second new effect size index quantifies the indirect effect as the proportion of the maximum possible indirect effect that could have been obtained, given the scales of the variables involved. We supplement our discussion by offering easy-to-use R tools for the numerical and visual communication of effect size for mediation effects.     

Estimation of Controlled Direct Effects in Longitudinal Mediation Analyses with Latent Variables in Randomized Studies

Abstract

In a randomized study with longitudinal data on a mediator and outcome, estimating the direct effect of treatment on the outcome at a particular time requires adjusting for confounding of the association between the outcome and all preceding instances of the mediator. When the confounders are themselves affected by treatment, standard regression adjustment is prone to severe bias. In contrast, G-estimation requires less stringent assumptions than path analysis using SEM to unbiasedly estimate the direct effect even in linear settings. In this article, we propose a G-estimation method to estimate the controlled direct effect of treatment on the outcome, by adapting existing G-estimation methods for time-varying treatments without mediators. The proposed method can accommodate continuous and noncontinuous mediators, and requires no models for the confounders. Unbiased estimation only requires correctly specifying a mean model for either the mediator or the outcome. The method is further extended to settings where the mediator or outcome, or both, are latent, and generalizes existing methods for single measurement occasions of the mediator and outcome to longitudinal data on the mediator and outcome. The methods are utilized to assess the effects of an intervention on physical activity that is possibly mediated by motivation to exercise in a randomized study.     

Translating intentions into nutrition behaviors via planning requires self-efficacy: evidence from Thailand and Germany

A first step towards the improvement of daily dietary behaviors is forming an intention to change one's nutrition. However, an intention by itself is not sufficient for successful action. Rather, to translate intentions into behavior, careful planning is recommended. Thus, planning constitutes a mediator between the intention and the behavior. However, if a person lacks self-efficacy, this mediation might fail. Previous research in Costa Rica and South Korea has identified perceived self-efficacy as a moderator of the intention-planning-behavior relationship. To examine further the moderator role of self-efficacy, two additional studies were designed in Thailand and Germany. Study 1 surveyed 1718 Thai university students in terms of a low-fat diet; Study 2 surveyed 1140 German internet users in terms of their fruit and vegetable consumption at two measurement points in time, 6 months apart. Intentions served as predictor, planning as mediator, self-efficacy as moderator, and behaviors as outcomes. First, intentions were translated into nutrition behaviors by planning. Second, self-efficacy moderated this mediation in both studies: The strength of the mediated effect increased along with levels of self-efficacy, even when accounting for baseline behaviors. For planning to mediate the intention-behavior relation, people must not harbor self-doubts. If they lack self-efficacy, intentions are not well translated into nutrition behavior through planning.     

Evaluating mediation and moderation effects in school psychology: A presentation of methods and review of current practice

Third variable effects elucidate the relation between two other variables, and can describe why they are related or under what conditions they are related. This article demonstrates methods to analyze two third-variable effects: moderation and mediation. The utility of examining moderation and mediation effects in school psychology is described and current use of the analyses in applied school psychology research is reviewed and evaluated. Proper statistical methods to test the effects are presented, and different effect size measures for the models are provided. Extensions of the basic moderator and mediator models are also described.     

Low self‐esteem,entrapment, and reason for living: A moderated mediation model of suicidal ideation

Adolescent suicidal ideation has become a top public health concern. It is thus significant to explore both risk and protective factors of adolescent suicidal ideation. The present study tested a moderated mediation model of suicidal ideation in a sample of Chinese adolescents. Chinese adolescents of 1074 (54.2% females, aged between 11 and 18 years) completed questionnaires assessing self‐esteem, entrapment, reason for living, and suicidal ideation. We found that entrapment mediated the association between low self‐esteem and suicidal ideation. The association between entrapment and suicidal ideation was moderated by reason for living. Findings of this study may expand our understanding of the development of suicidal ideation, and facilitate future research exploring the interplay of risk and protective factors of suicidality. Clinical implications of these findings were also discussed.     

Thinking meta-theoretically about the role of internalization in the development of body dissatisfaction and body change behaviors

Internalization of societal standards of physical attractiveness (i.e., internalization of the thin ideal for women and internalization of the mesomorphic ideal for men) is a widely studied and robust risk factor for body dissatisfaction and maladaptive body change behaviors. Substantial empirical research supports internalization as both a mediator and a moderator of the relation between societal influences and body dissatisfaction. In this paper, a primer on mediation and moderation is followed by a review of literature and discussion of the extent to which internalization can theoretically fulfill the roles of both mediation and moderation. The literature review revealed a stark contrast in research design (experimental versus non-experimental design) when alternate conceptualizations of internalization are adopted. A meta-theoretical, moderated mediation model is presented. This model integrates previous research and can inform future empirical and clinical endeavors.     

A general approach to causal mediation analysis

Traditionally in the social sciences, causal mediation analysis has been formulated, understood, and implemented within the framework of linear structural equation models. We argue and demonstrate that this is problematic for 3 reasons: the lack of a general definition of causal mediation effects independent of a particular statistical model, the inability to specify the key identification assumption, and the difficulty of extending the framework to nonlinear models. In this article, we propose an alternative approach that overcomes these limitations. Our approach is general because it offers the definition, identification, estimation, and sensitivity analysis of causal mediation effects without reference to any specific statistical model. Further, our approach explicitly links these 4 elements closely together within a single framework. As a result, the proposed framework can accommodate linear and nonlinear relationships, parametric and nonparametric models, continuous and discrete mediators, and various types of outcome variables. The general definition and identification result also allow us to develop sensitivity analysis in the context of commonly used models, which enables applied researchers to formally assess the robustness of their empirical conclusions to violations of the key assumption. We illustrate our approach by applying it to the Job Search Intervention Study. We also offer easy-to-use software that implements all our proposed methods.