实证研究中的因果推理与分析

心理学期刊中的实证研究论文,很多时候都在检验变量之间的因果关系,但学界对因果研究存在一些不同的看法。本文试图回答下面问题:(1)实验中不能操纵的变量是否可以作为原因?(2)非实验研究能不能检验因果关系?(3)因果分析(尤其是中介分析)是否一定要使用追踪数据?追踪设计的主要目的是什么?通过因果要素和因果推理逻辑的辨析,对前面两个问题都给出了肯定的回答,并讨论了判明变量先后顺序、统计控制无关变量的方法;为了回答问题3,厘清了追踪设计在因果分析中的作用:区分变量的先后顺序、有效获得历时性的影响结果,但即时性的因果影响采用追踪设计可能是不合适的。     

国内追踪数据分析方法研究与模型发展

追踪研究因其可以得到比横断研究更有说服力的变量关系论证, 在心理学等科学中具有重要地位。梳理国内以心理学为主的相关领域中追踪数据分析方法研究的发表现状、主要解决的研究问题和模型发展。追踪研究可以进行均值差异比较、分析多变量相互影响、描述总体发展趋势及差异和探究心理动态变化过程。近20年的研究热点和发展思路也集中在上述研究问题当中, 特别是总体发展趋势及差异、多变量相互影响、总体发展趋势与多变量相互影响的融合、追踪研究设计、缺失数据等议题上。最后, 比较国内外研究的差异, 并结合交叉学科对国内追踪研究未来发展做出展望。     

因果影响与函数式依赖:从逻辑的视角分析

近年来,关于如何刻画系统中变量间的相互依赖和相互影响在计算机以及哲学研究中激起了广泛讨论。其中非常经典的工作包括Pearl、Gales以及Halpern等逻辑学家和计算机科学家提出的因果模型和基于因果模型的因果逻辑。而最新Baltag与van Benthem的工作又提出了通过函数式依赖这一概念分析变量间影响的模型。本文将介绍并探讨这两种路径之间的关系,并且提出,在对Halpern等人的因果模型和逻辑做认知方面的扩充之后,我们能在这两种路径中找到更多共通之处。     

纵向关系的探究：基于交叉滞后结构的追踪模型

基于交叉滞后结构的追踪模型对于揭示变量间纵向关系具有重要作用,也为因果关系的验证奠定了基础。交叉滞后面板模型在一定条件下可转换为其他形式的模型,如何选择适当的模型是重要的议题。研究对各模型进行概述,并从模型结构、预设轨迹、时间点要求等方面进行比较,最后通过一个实例说明如何选择适当的模型。结果表明,不同模型在变量关系的判断上可能给出很不同的结果,实际运用中应当有模型选择和模型比较的意识。     

密集追踪数据分析：模型及其应用

在心理学、教育学和临床医学等领域, 越来越多的研究者开始关注个体内部的行为、心理、临床效果等随时间而产生的动态变化, 重视针对个体的差异化建模。密集追踪是一种在短时间内对个体进行多个时间节点密集追踪测量的方法, 更适合用于研究个体内部心理过程等的动态变化及其作用机制。近年来, 密集追踪成为心理学研究的一大热点, 但许多密集追踪的研究分析仍停留在较为传统的方法。方法学领域已涌现出较多用于密集追踪数据分析的模型方法, 较为主流的模型包括以动态结构方程模型(Dynamic Structural Equation Model, DSEM)为代表的自上而下的建模方法, 以及以组迭代多模型估计(Group Iterative Multiple Model Estimation, GIMME)为代表的自下而上的建模方法。二者均可以方便地对密集追踪数据中的自回归及交叉滞后效应进行建模。     

因果模型构建的特点和新思路

因果模型构建作为心理学研究的一项中心任务,主要包括因果关系理论模型的构建和验证、因果关系数据模型的构建和验证,理论模型和数据模型之间进行匹配的因果关系推论过程,文章讨论了因果模型构建的基本条件。因果模型构建时所面临的问题及解决的策略,包括如何确定因果模型的类型,如何在不同层次进行建模、如何进行因果模型的验证。另外,文章还讨论了一种因果建模的新思路及发展趋势。     

二分数据的多层线性模型:原理与应用

分类数据的多层线性模型在我国的心理学研究中鲜有使用。本研究旨在将这种模型引入到我国心理学研究之中。论文首先介绍了二分数据的多层线性模型的原理和假设条件、参数估计和假设检验,然后以6187名小学生为被试,采用二分变量的多层线性模型,说明了个体因素和学校因素对儿童攻击行为的影响,并对分析结果进行了解释。     

发展心理学研究中个体定向的理论与方法

长期以来, 发展心理学研究主要采取变量定向的方法。近年来, 随着发展心理学理论的发展和人类发展个体差异的科学事实的积累, 个体定向的方法得以发展并且成为发展心理学研究的主要方法思路之一。个体定向的方法关注作为整体的个体, 旨在确定有意义的同质性个体类别或亚组。个体定向的方法包括一系列的基本理论原则, 具体数据分析方法包括传统聚类分析、基于模型的聚类方法以及配置频次分析等。发展心理学研究需综合运用个体定向的方法与变量定向的方法。     

贝叶斯结构方程模型及其研究现状

在心理学研究中结构方程模型(Structural Equation Modeling, SEM)被广泛用于检验潜变量间的因果效应, 其估计方法有频率学方法(如, 极大似然估计)和贝叶斯方法两类。近年来由于贝叶斯统计的流行及其在结构方程建模中易于处理小样本、缺失数据及复杂模型等方面的优势, 贝叶斯结构方程模型发展迅速, 但其在国内心理学领域的应用不足。主要介绍了贝叶斯结构方程模型的方法基础和优良特性, 及几类常用的贝叶斯结构方程模型及其应用现状, 旨在为应用研究者介绍新的研究工具。     

组织承诺研究综述

该文对国内外组织承诺的研究进行了综述。包括组织承诺的内涵, 组织承诺形成发展过程模型,以及组织承诺因果模型中与其它变量之间的关系。最后分析了以往研究中存在的问题以及今后研究努力的方向     

Model Implied Instrumental Variables (MIIVs): An Alternative Orientation to Structural Equation Modeling

Few dispute that our models are approximations to reality. Yet when it comes to structural equation models (SEMs), we use estimators that assume true models (e.g. maximum likelihood) and that can create biased estimates when the model is inexact. This article presents an overview of the Model Implied Instrumental Variable (MIIV) approach to SEMs from Bollen (1996). The MIIV estimator using Two Stage Least Squares (2SLS), MIIV-2SLS, has greater robustness to structural misspecifications than system wide estimators. In addition, the MIIV-2SLS estimator is asymptotically distribution free. Furthermore, MIIV-2SLS has equation-based overidentification tests that can help pinpoint misspecifications. Beyond these features, the MIIV approach has other desirable qualities. MIIV methods apply to higher order factor analyses, categorical measures, growth curve models, dynamic factor analysis, and nonlinear latent variables. Finally, MIIV-2SLS permits researchers to estimate and test only the latent variable model or any other subset of equations. In addition, other MIIV estimators beyond 2SLS are available. Despite these promising features, research is needed to better understand its performance under a variety of conditions that represent empirical applications. Empirical and simulation examples in the article illustrate the MIIV orientation to SEMs and highlight an R package MIIVsem that implements MIIV-2SLS.     

The longitudinal and bidirectional association between parenting stress and couples' relationship quality

This study examined the longitudinal and bidirectional association between parenting stress and relationship quality, and whether the associations differed between married and cohabiting couples, using cross‐lagged panel models (CLPM) and random‐intercept cross‐lagged panel models (RI‐CLPM). Using data from the Fragile Families and Child Well‐being Study, the findings indicated that the RI‐CLPM model provided a better fit to the data compared to the CLPM model for mothers and fathers. The RI‐CLPM results showed high levels of stability in parenting stress and relationship quality over time, that parenting stress was associated with higher levels of relationship quality for both parents, and that the between‐person effect was similar for married and cohabiting couples. This study highlighted the importance of accounting for between‐ and within‐person variability.     

多变量追踪研究的模型整合与拓展：考察往复式影响与增长趋势

追踪研究当中, 交叉滞后模型可以探究多变量之间往复式影响, 潜增长模型可以探究个体增长趋势。对两类模型进行整合, 例如同时关注往复式影响与个体增长趋势, 同时可以定义测量误差、随机截距等变异成分, 衍生出随机截距交叉滞后模型、特质-状态-误差模型、自回归潜增长模型、结构化残差潜增长模型等。以交叉滞后模型和潜增长模型分别作为基础模型, 从个体间/个体内变异分解的角度对上述各类模型梳理, 整合出此类模型的分析框架, 并拓展建立“因子结构化潜增长模型(factor latent curve model with structured reciprocals)”作为统合框架。通过实证研究(早期儿童的追踪研究-幼儿园版, ECLS-K), 建立21049名儿童的阅读和数学能力的往复式影响与增长趋势。研究发现, 分离了稳定特质的模型拟合最优。研究也对模型建模思路和模型选择提供了建议。     

Does high self-esteem foster narcissism? Testing the bidirectional relationships between self-esteem,narcissistic admiration and rivalry

We examined the longitudinal associations between self-esteem and narcissism in a three-wave panel study (N = 557). In a standard cross-lagged panel model, self-esteem had a positive bidirectional relationship with narcissistic admiration. Narcissistic rivalry predicted increases in narcissistic admiration, but the corresponding reciprocal cross-lagged effect was not significant, nor were the cross-lagged associations between self-esteem and narcissistic rivalry. However, a random-intercept cross-lagged panel model (which partitions between- and within-person variance) failed to identify significant cross-lagged relationships between self-esteem and admiration or rivalry. Rather, self-esteem correlated positively with narcissistic admiration (but not rivalry) only at the trait level. Furthermore, we observed positive bidirectional associations between admiration and rivalry, suggesting that the within-person fluctuations in these two sub-dimensions of narcissism mutually reinforce each other.     

Data Analytic Advances That Bridge the Scientist-Practitioner Gap

In this commentary, I advance the view that the scientist-practitioner gap is partly due to the research designs commonly used in psychotherapy research. Specifically, I believe that randomized controlled trials, which are important for establishing treatment efficacy and as leverage when making the case for the value of psychotherapy in relation to various stakeholders, are limited for further development of clinical theories. Instead, I find recent advances in cross-lagged panel modeling to be both clinically intuitive and stronger for causal inference than most other nonexperimental designs. In addition to discussing causal inference and clinical interpretation of cross-lagged panel models, I discuss the potential of improving mediation analysis, personalization of research, and studying issues of clinical timing. Finally, I briefly discuss some limitations of cross-lagged panel models. It is my belief that the use of these data analytic advances can make empirical research better live up to the innovations in Beck’s work.     

Positive gain spirals at work: From job resources to work engagement, personal initiative and work-unit innovativeness

The present cross-lagged panel study aimed to investigate the energizing power of job resources and related gain spirals. Drawing on Hobfoll’s Conservation of Resources (COR) theory’s rarely tested assumptions of cumulative resource gains and gain spirals a reciprocal process was expected: (1) job resources lead to work engagement and work engagement leads to personal initiative (PI), which, in turn, has a positive impact on work-unit innovativeness, and (2) work-unit innovativeness leads to PI, which has a positive impact on work engagement, which finally predicts future job resources. The study was based on a two-wave 3-year panel design among 2555 Finnish dentists. Structural equation modeling was employed to study cross-lagged associations. The results mainly confirmed our hypotheses: positive and reciprocal cross-lagged associations were found between job resources and work engagement and between work engagement and PI. In addition, PI had a positive impact on work-unit innovativeness over time.     

School System Evaluation by Value Added Analysis Under Endogeneity

Value added is a common tool in educational research on effectiveness. It is often modeled as a (prediction of a) random effect in a specific hierarchical linear model. This paper shows that this modeling strategy is not valid when endogeneity is present. Endogeneity stems, for instance, from a correlation between the random effect in the hierarchical model and some of its covariates. This paper shows that this phenomenon is far from exceptional and can even be a generic problem when the covariates contain the prior score attainments, a typical situation in value added modeling. Starting from a general, model-free definition of value added, the paper derives an explicit expression of the value added in an endogeneous hierarchical linear Gaussian model. Inference on value added is proposed using an instrumental variable approach. The impact of endogeneity on the value added and the estimated value added is calculated accurately. This is also illustrated on a large data set of individual scores of about 200,000 students in Chile.     

The role of gratitude in the development of social support, stress, and depression: Two longitudinal studies

In two longitudinal studies, the authors examined the direction of the relationships between trait gratitude, perceived social support, stress, and depression during a life transition. Both studies used a full cross-lagged panel design, with participants completing all measures at the start and end of their first semester at college. Structural equation modeling was used to compare models of direct, reverse, and reciprocal models of directionality. Both studies supported a direct model whereby gratitude led to higher levels of perceived social support, and lower levels of stress and depression. In contrast, no variable led to gratitude, and most models of mediation were discounted. Study 2 additionally showed that gratitude leads to the other variables independently of the Big Five factors of personality. Overall gratitude seems to directly foster social support, and to protect people from stress and depression, which has implications for clinical interventions.     

Cross-Lagged Analyses of Prolonged Grief and Depression Symptoms With Insomnia Symptoms

Prolonged grief disorder, characterized by severe, persistent and disabling grief, has recently been added to the DSM-5-TR and ICD-11. Treatment for prolonged grief symptoms shows limited effectiveness. It has been suggested that prolonged grief symptoms exacerbate insomnia symptoms, whereas insomnia symptoms, in turn, may fuel prolonged grief symptoms. To help clarify if treating sleep disturbances may be a viable treatment option for prolonged grief disorder, we examined the proposed reciprocal relationship between symptoms of prolonged grief and insomnia. On three time points across 6-month intervals, 343 bereaved adults (88% female) completed questionnaires to assess prolonged grief, depression, and insomnia symptoms. We applied random intercept cross-lagged panel models (RICLPMs) to assess reciprocal within-person effects between prolonged grief and insomnia symptoms and, as a secondary aim, between depression and insomnia symptoms. Changes in insomnia symptoms predicted changes in prolonged grief symptoms but not vice versa. Additionally, changes in depression and insomnia symptoms showed a reciprocal relationship. Our results suggest that targeting insomnia symptoms after bereavement is a viable option for improving current treatments for prolonged grief disorder.     

Reciprocal Relations of Worry,Rumination, and Psychopathology Symptoms After Loss: A Prospective Cohort Study

Bereavement can precipitate symptoms of depression, prolonged grief disorder, and posttraumatic stress disorder. Targeting repetitive negative thought (i.e., worry, rumination) in treatment may help reduce post-loss psychopathology. Yet, evidence on longitudinal associations of depressive rumination and worry with post-loss psychopathology symptoms has been mixed and the directions of effects are still unclear. Recently bereaved adults (78% female) completed questionnaires assessing depressive rumination (brooding), worry, and depression, prolonged grief and posttraumatic stress symptoms 11 times in 1.5 month intervals. We applied random-intercept cross-lagged panel models (RICLPMs) to examine reciprocal within-person associations between worry and psychopathology symptoms, between rumination and these symptoms, and between worry and rumination. Main findings were that worry showed reciprocal relationships with psychopathology symptoms (although worry did not consistently predict prolonged grief symptoms). Depressive rumination was predicted by psychopathology symptoms, but not vice versa. Worry showed reciprocal relations with depressive rumination. Findings suggest that worry may be part of a downward spiral, enhancing psychopathology symptoms following loss, whereas depressive rumination is solely a consequence of such symptoms.