基于多元回归的调节效应分析

在心理学和其他社科研究领域,大量实证研究建立调节模型,以分析自变量对因变量关系的影响机制,但在基于多元回归的调节效应分析实践中仍存在不足。我们回顾了均值中心化在基于多元回归的调节效应分析中的作用,均值中心化不影响乘积项(即调节效应)的检验,仅对一阶项(即主效应)的检验有影响。讨论了简单斜率的检验方法,建议在调节变量为连续变量时,使用Johnson-Neyman法进行简单斜率检验;在调节变量为类别变量或研究者对某个调节变量值感兴趣时,使用选点法。并用一个实际例子演示如何进行调节效应分析。随后展望了调节效应检验的拓展方向。     

Modelling determinants of walking and cycling adoption: A stage-of-change perspective

Shifting travel away from cars and towards more active modes has proven a formidable policy challenge. This study aims to uncover the determinants of walking and cycling adoption by applying a stage-of-change framework. Drawing on the Transtheoretical Model, this framework models the adoption of active modes as a series of stages from pre-contemplation to maintenance. Ordinal logit models applied to US data (n = 914) illustrate the importance of both observable demographic-personal and perceptual-attitudinal variables for determining stage-of-change membership. Comparing walking and cycling, the model reveals both shared variables (vehicle ownership, self-identity) and differing factors (gender, environmental spatial ability) distinguishing among adoption stages, which has significant implications for transport policy. Results indicate that a model combining both demographic-personal and perceptual-attitudinal factors has the best fit and validity, suggesting that travel behavior interventions would benefit from multivariate segmentation methods that consider an array of individual and group characteristics. This research also gives evidence of different determinants motivating change processes for cycling versus walking. Taken together, results suggest a need for tailored policy interventions to promote behavioural adoption based not merely on the specific mode selection, but also on the usage stage under consideration.     

Intellectual, behavioral, and personality correlates of violent vs. non-violent juvenile offenders

The overall aim of this study was to examine the relationship between offender status (violent vs. nonviolent) and selected predictor variables from personality, behavioral, and intellectual domains. The two main sub goals were (a) to determine which variables from these domains were most closely associated with offender status, and (b) to construct a stepwise logistic regression model that could help identify which juveniles were more likely to be incarcerated for violent vs. nonviolent offenses. The participants for this investigation were juvenile offenders referred to the Juvenile Court Assessment Center by the Juvenile Justice Division of the Eleventh Judicial Circuit. The court-ordered assessment included the following measures: (a) The Millon Adolescent Clinical Inventory (MACI), (b) the Behavior Assessment System for Children (BASC), (c) the Peabody Picture Vocabulary Test-Third Edition (PPVT-III), (d) the Wide Range Achievement Test-Third Edition (WRAT-III), (e) the Kaufman Brief Intelligence Test (K-BIT), and (f) records of school achievement. The ten variables that had the strongest association with offender status were entered into the stepwise logistic regression analysis; five of these strategically chosen predictor variables accurately differentiated violent from nonviolent offenders 86.3% of the time. Reading Percentile (β=-.051), PPVT-III (β=-.059), MACI-Inhibition (β=-.033), MACI-Eating Dysfunction (β=.051), and BASC-Sense of Inadequacy (β=-.072). Gender differences were explored.     

Simple and multiple P‐splines regression with shape constraints

In many research areas, especially within social and behavioural sciences, the relationship between predictor and criterion variables is often assumed to have a particular shape, such as monotone, single‐peaked or U‐shaped. Such assumptions can be transformed into (local or global) constraints on the sign of the nth‐order derivative of the functional form. To check for such assumptions, we present a non‐parametric regression method, P‐splines regression, with additional asymmetric discrete penalties enforcing the constraints. We show that the corresponding loss function is convex and present a Newton–Raphson algorithm to optimize. Constrained P‐splines are illustrated with an application on monotonicity‐constrained regression with both one and two predictor variables, using data from research on the cognitive development of children.     

Regression to the mean and football wagers

Football performances are an imperfect measure of abilities, and consequently exaggerate differences in abilities. The skills of those football teams that perform the best and the worst are not really that far from average; thus their future performances regress to the mean. Betting data indicate that gamblers do not fully account for this regression. Copyright © 2002 John Wiley & Sons, Ltd.     

Robust regression: Testing global hypotheses about the slopes when there is multicollinearity or heteroscedasticity

A well-known concern regarding the usual linear regression model is multicollinearity. As the strength of the association among the independent variables increases, the squared standard error of regression estimators tends to increase, which can seriously impact power. This paper examines heteroscedastic methods for dealing with this issue when testing the hypothesis that all of the slope parameters are equal to zero via a robust ridge estimator that guards against outliers among the dependent variable. Included are results related to leverage points, meaning outliers among the independent variables. In various situations, the proposed method increases power substantially.     

Returning to the present moment: Thinking about one’s childhood increases focus on the hedonistic present

Children seem to live more in the present moment than adults. But is it possible to make the present time perspective more available among adults? Four experiments demonstrated that activating childhood selves can lead adults to be more embedded in the hedonistic present. Studies 1 and 2 showed that recalling memories from childhood, either in an open-ended or a structured form, increased participants’ focus on the hedonistic present. Study 3 showed that this effect also occurred after an implicit childhood manipulation. Study 4 revealed that taking the perspective of a child significantly increased orientation to the hedonistic present, compared to taking the perspective of an adult. The effects of activation of childhood selves were not mediated by mood, nostalgia, impatience, concentration, difficulty of the task, enjoyment connected with the task, or involvement in the task. Evaluation of one’s childhood and tendency to relive childhood memories did not affect the results.     

What's Behind the “Nones‐sense”? Change Over Time in Factors Predicting Likelihood of Religious Nonaffiliation in the United States

The proportion of people in the United States who identify as unaffiliated with any religious tradition (Nones) has risen steadily since the 1990s. Empirical investigations have examined this phenomenon, and point to a range of sociodemographic and associational variables as significant predictors of religious nonaffiliation. To build on these, the research reported here uses nearly five decades of General Social Survey data and binary logistic regression to examine change over time in the direction and size of effect on the likelihood that various factors predict religious nonaffiliation. While some factors (like age and political orientation) behave as expected over time, other factors decrease in their effect on likelihood (e.g., residence in the Far West), lose effect on likelihood (e.g., being college‐educated), or never showed likelihood of effect in the first place (e.g., residence in New England).     

The interactive effect of neuroticism and extraversion on the daily variability of affective states

The interaction between neuroticism and extraversion is thought to predict affective variability. In this study, neuroticism and extraversion were assessed with questionnaires, and affects were measured by experience sampling, with five daily assessments over 2 weeks. Affective variability was studied within a three-dimensional affective space whose three axes were oriented along the main affective dimensions: positive affect, negative affect, and activation. Quantile regression mixed-effects models allowed predicting zones in which affective states were most likely to occur according to personality. Beyond the well-known effect of personality on affect level, high neuroticism and/or high extraversion were accompanied by heightened affective variability. Results were interpreted as potentially reflecting positive feedback loops oriented toward negative affect for neuroticism, and toward positive affect and activation for extraversion.