机器学习在解决过拟合现象中的作用

过拟合现象是心理学走向预测科学的重要阻碍。文章综述了机器学习在解决过拟合现象中的价值和实现途径：（1）介绍了过拟合的两种表现形式和现状;（2）分析过拟合的根因,即“高解释力≠高预测力”;（3）厘清机器学习的建模逻辑与核心技术在解决过拟合中的作用;（4）利用样例数据和代码说明机器学习统计思想在模型拟合中的具体应用过程。文章指出心理学应从解决实际问题的角度出发,借鉴机器学习的分析思想,避免过拟合,进而提供更准确更稳定的结论和预测模型。     

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。     

四参数Logistic模型和传统模型对被试作答拟合能力的比较研究

针对测验中高能力被试答错容易试题的睡眠现象,可使用四参数Logistic模型分析数据。研究选取了来自心理测验和成就测验的实际数据,分别采用传统模型和四参数Logistic模型进行拟合,对不同模型的拟合指标及参数估计结果进行比较。结果表明,四参数Logistic模型能够提高拟合程度,增强估计结果的准确性,有效纠正高能力被试能力被低估的现象。建议在必要时使用四参数Logistic模型进行数据分析。     

大数据时代心理学文本分析技术——“主题模型”的应用

主题模型(Topic Model)作为一种计算机化文本分析方法,目前被研究人员应用于心理学文本分析领域的诸多研究中。在心理咨询研究领域中,主题模型可以用于探索咨询过程中咨询师和当事人之间谈话的主题,比较不同治疗类别的相似性,以及进行行为编码;在社交媒体与心理健康上,利用主题模型可识别和预测各种心理障碍以及进行人格计算。未来需要关注主题模型在算法和操作上的改善,并将其应用到中文语言环境之中,探索中文文本的心理学内涵。     

心智社会:作为分布计算的科学

史学家、哲学家、心理学家以及社会学家采用截然不同的方式来研究科学。不仅不同领域的研究者们所用的方法相去甚远,而且他们所关注的科学侧面也大相径庭。最终我们会看到,一方面有些哲学家试图对科学知识进行逻辑分析,另一方面有些社会学家坚持认为,知识的构造具有社会性。本文试图从一种综合性的观点出发看待科学史、科学哲学、科学心理学和科学社会学。 不同领域的研究者们已经开始明确地或不明确地运用模型来探讨科学研究的不同方法间的关系。鉴于后面将要阐述的理由,我考虑的重点将放在科学心理学与科学社会学之间的关系上。还原论模型要么坚持说社会学可以还原到心理学中更具解释性的基础领域,要么认为心理学可以还原到社会学。“补遗”模型不这样极端,按照这一模型,解释的主要任务可以由心理学或哲学或社会学来承担,余下的解释工作则需借助别的领域来完成。其他模型并未主张以某一学科为主,它们认为,科学研究的不同领域是在通过相互协作或相互竞争来解释科学本质的各个侧面,相对于其他领域而言,这些领域各自都具有一定的自主性。     

社会能力:概念分析与模型建构

社会能力是心理学领域中一个经典的研究问题,但是至今仍没有统一的概念和理论模型.总结以往的有关研究,可以归纳出社会能力概念的几个取向,如技能取向、结果取向、综合取向等.在这些概念取向尤其是综合取向的指导之下,研究者也提出了一些比较有代表性的理论模型,如成分模型、过程模型和棱柱模型等.该文对这些概念取向和理论模型进行了综述并尝试提出了社会能力的整合观点,但关于社会能力的概念和理论模型尚需进一步完善,其文化差异问题也尚需未来研究的关注.     

主观阶层研究取向的贡献与弊端

近年来社会阶层心理学研究领域对于主观阶层的研究取向高度重视, 突出表现为强调社会阶层概念中的主观阶层成分, 在理论阐述、概念测量、研究设计和研究问题选择上都倾向于关注主观阶层。这一取向为社会阶层心理学的发展奠定了基础, 也扩展了该领域的研究方法及关注问题的视野。但它也造成了部分研究在理论层面与社会现实有所脱节、在概念层面过于忽视客观阶层、在方法层面过于依赖操纵主观阶层, 以及在问题层面只关注高低二元阶层划分、对现实阶层多样性关注不足等弊端。未来应进一步深入开展基础研究以厘清客观阶层和主观阶层的关系及其不同的预测效应, 在研究设计时对于客观阶层和主观阶层进行更细化的区分, 同时可以更多关注于现实中的具体社会问题。     

积极心理学视野下的工作压力

工作压力是职业健康心理学研究的重要领域。随着经济发展,工作压力问题日益突出,它降低了生产率,使员工的健康受损,增加了成本消耗。但是以往的研究主要关注压力的消极面,干预方法也是以预防和治疗为主。然而积极心理学运动的兴起改变了传统的工作压力研究方法,提出积极压力的概念,倡导以健康、整合的方式建立一个完整的工作压力模型。并提出在预防和治疗消极压力的同时,也要重视激发积极压力。     

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。     

预测视角下双因子模型与高阶模型的模拟比较

双因子模型和高阶因子模型,作为既有全局因子又有局部因子的两个竞争模型,在研究中得到了广泛应用。本文采用Monte Carlo模拟方法,在模型拟合比较的基础上,比较了效标分别为外显变量和内潜变量时,两个模型在各种负荷水平下预测准确度的差异。结果发现,两种模型在拟合效果方面无显著差异;但在预测效度方面,当效标为显变量时,两个模型的结构系数估计值皆为无偏估计;而效标为潜变量时,高阶因子模型表现优于双因子模型:高阶因子模型的结构系数为无偏估计,双因子模型的结构系数估计值则在50%左右的情况下存在偏差。     

Multicollinearity and the use of regression analyses in discrimination litigation

Multicollinearity is a problem that can adversely affect the estimation of coefficients in regression equations. The types of regression models used as evidence in employment discrimination cases may be particularly susceptible to estimation problems resulting from multicollinearity, yet courts have in most instances failed to address this difficulty. This paper discusses the problems associated with multi- collinearity and suggests possible sources of the multicolli- nearity, particularly in the context of employment discrimination litigation. An academic example demonstrates how multicollinearity can be identified and corrected. Other possible techniques for curing the multicollinearity or reducing its effects are also presented. The related problem of “tainted” variables, often discussed by courts in employment discrimination litigation, is differentiated from the multicollinearity problem. For regression equations to be reliable and for inferences regarding discrimination to be accurately made, courts should concern themselves with multicollinearity, and not so-called “tainting.”     

On the Misconception of Multicollinearity in Detection of Moderating Effects: Multicollinearity Is Not Always Detrimental

Due to its extensive applicability and computational ease, moderated multiple regression (MMR) has been widely employed to analyze interaction effects between 2 continuous predictor variables. Accordingly, considerable attention has been drawn toward the supposed multicollinearity problem between predictor variables and their cross-product term. This article attempts to clarify the misconception of multicollinearity in MMR studies. The counterintuitive yet beneficial effects of multicollinearity on the ability to detect moderator relationships are explored. Comprehensive treatments and numerical investigations are presented for the simplest interaction model and more complex three-predictor setting. The results provide critical insight that both helps avoid misleading interpretations and yields better understanding for the impact of intercorrelation among predictor variables in MMR analyses.     

Extracting the Variance Inflation Factor and Other Multicollinearity Diagnostics from Typical Regression Results

Multicollinearity is a potential problem in all regression analyses. However, the examination of multicollinearity is rarely reported in primary studies. In this article we discuss and show several post hoc methods for assessing multicollinearity. One such multicollinearity diagnostic is the variance inflation factor. We outline the post hoc variance inflation factor method, which computes the variance inflation factor from the standardized regression coefficient and semi-partial correlation, both of which can be calculated from commonly reported regression results. Three examples of computing multicollinearity diagnostics using data from published studies are shown. We conclude with a discussion and practical implications.     

Clarifying the role of mean centring in multicollinearity of interaction effects

Moderated multiple regression (MMR) is frequently employed to analyse interaction effects between continuous predictor variables. The procedure of mean centring is commonly recommended to mitigate the potential threat of multicollinearity between predictor variables and the constructed cross-product term. Also, centring does typically provide more straightforward interpretation of the lower-order terms. This paper attempts to clarify two methodological issues of potential confusion. First, the positive and negative effects of mean centring on multicollinearity diagnostics are explored. It is illustrated that the mean centring method is, depending on the characteristics of the data, capable of either increasing or decreasing various measures of multicollinearity. Second, the exact reason why mean centring does not affect the detection of interaction effects is given. The explication shows the symmetrical influence of mean centring on the corrected sum of squares and variance inflation factor of the product variable while maintaining the equivalence between the two residual sums of squares for the regression of the product term on the two predictor variables. Thus the resulting test statistic remains unchanged regardless of the obvious modification of multicollinearity with mean centring. These findings provide a clear understanding and demonstration on the diverse impact of mean centring in MMR applications.     

The Detection and Interpretation of Interaction Effects Between Continuous Variables in Multiple Regression

issues in the detection and interpretation of interaction effects between quantitative variables in multiple regression analysis are discussed. Recent articles by Cronbach (1987) and Dunlap and Kemery (1987) suggested the use of two transformations to reduce "problems" of multicollinearity. These transformations are discussed in the context of the conditional nature of multiple regression with product terms. It is argued that although additive transformations do not affect the overall test of statistical interaction, they do affect the interpretational value of regression coefficients. Factors other than multicollinearity that may account for failures to observe interaction effects are noted.     

Testing Shneidman’s model of suicidality in incarcerated offenders and in undergraduates

Suicidality constitutes a major health issue for society in general and incarcerated populations in particular. Shneidman’s model of suicide proposes that psychache (i.e., intense psychological pain) is the pre-eminent psychological cause of suicidality and mediates the influence of all other psychological correlates such as depression or hopelessness. The current research evaluates Shneidman’s model for samples of 73 male federal offenders, 80 male undergraduates, and 80 female undergraduates. Consistent with Shneidman’s theory, psychache was both a significant and a more important statistical predictor of reported self-harming ideation and action than was either depression or hopelessness. Further, this relatively greater importance of psychache for the statistical prediction of suicidality was not moderated either by offender status or by sex. Overall, findings support the applicability of Shneidman’s model of suicidality to incarcerated individuals and indicate the model’s generalizability across sex and offender status.     

人事测评中的概化理论应用

概化理论是现代心理与教育测量理论之一,可应用在各种人事测评中,如表现性评价、多源评估、心理测验、结构化面试、水平测试、工作分析、评价中心等.与经典测量理论相比,概化理论应用于人事测评,表现出较强的优势,能够同时考察多种因素、确定多个维度权重等,其应用对象主要包括两大类,即企业和机构.概化理论应用于人事测评,存在应用领域、样本数据、测评效度和微观分析等问题.     

心理传记学的研究方法思考

摘 要 当代心理学由于“本体与方法”的割裂形成科学心理学与人文心理学双峰对峙的局面,科学心理学受自然科学方法的影响而难以直面人性关怀问题;以人文心理学为研究传统的心理传记学,通过对悬疑性问题的逐层解读,直面人类心理动力系统,从而能够直面人性关怀问题。为进一步促进心理传记学的发展,对当前心理传记研究的资料筛选、研究模式、理论运用改进,以期研究者能够更好地通过史料分析进入传主生命事件的历史现场,对传主生命历程进行恰当的理论解释,从而进一步激发中国本土心理传记学发展的动力。     

接纳承诺疗法的作用机制——基于元分析结构方程模型

接纳承诺疗法(Acceptance and Commitment Therapy, ACT)被认为是行为治疗“第三浪潮”的重要代表。本研究使用元分析结构方程模型, 考察ACT的作用机制。通过数据库检索与筛选, 最终纳入文献50篇。结果发现: ACT所假设的心理灵活性、接纳、此时此刻、价值的中介作用都达到统计显著, 认知解离这一中介变量并不显著; 中介机制在网络化干预中仍然得到检验; 相较之传统CBT, ACT在所假设的机制上有其区别于CBT的优势。后续临床研究应更全面地测量6大核心机制, 关注对美好生活提升的影响, 采用多点瞬时评价法, 并尽可能使用更高级、更先进的统计方法检验其作用机制。     

The cultural shaping of depression: somatic symptoms in China, psychological symptoms in North America?

The expectation that Chinese people present distress somatically is a central prediction of cultural psychopathology and has been the subject of considerable theoretical speculation. At the same time, empirical studies have been infrequent and have yielded mixed results. The authors examined symptom presentation in Chinese (n=175) and Euro-Canadian (n=107) outpatients, using spontaneous problem report, structured clinical interview, and symptom questionnaire methods. All 3 methods yielded cross-culturally equivalent somatic and psychological symptom subscales. Chinese outpatients reported more somatic symptoms on spontaneous problem report and structured clinical interview compared with Euro-Canadians, who in turn reported more psychological symptoms on all 3 methods. The relation between culture and somatic symptom presentation was mediated by a tendency toward externally oriented thinking. Difficulties with identifying emotions or describing them to others did not differ significantly across cultures, supporting a nonpathological interpretation of observed differences. Psychological symptom effects were larger and more consistent than somatic symptom effects; because other studies have confirmed the ubiquity of somatic presentations worldwide, these results suggest that Western psychologization may be more culturally specific than is Chinese somatization.