Abstract: Inference and Interval Estimation for Indirect Effects With Latent Variable Models

Models specifying indirect effects (or mediation) and structural equation modeling are both popular in the social sciences. Yet relatively little research has compared methods that test for indirect effects among latent variables and provided precise estimates of the effectiveness of different methods.

This simulation study provides an extensive comparison of methods for constructing confidence intervals and for making inferences about indirect effects with latent variables. We compared the percentile (PC) bootstrap, bias-corrected (BC) bootstrap, bias-corrected accelerated (BC _a ) bootstrap, likelihood-based confidence intervals (Neale & Miller, 1997), partial posterior predictive (Biesanz, Falk, and Savalei, 2010), and joint significance tests based on Wald tests or likelihood ratio tests. All models included three reflective latent variables representing the independent, dependent, and mediating variables. The design included the following fully crossed conditions: (a) sample size: 100, 200, and 500; (b) number of indicators per latent variable: 3 versus 5; (c) reliability per set of indicators: .7 versus .9; (d) and 16 different path combinations for the indirect effect (α = 0, .14, .39, or .59; and β = 0, .14, .39, or .59). Simulations were performed using a WestGrid cluster of 1680 3.06GHz Intel Xeon processors running R and OpenMx.

Results based on 1,000 replications per cell and 2,000 resamples per bootstrap method indicated that the BC and BC _a bootstrap methods have inflated Type I error rates. Likelihood-based confidence intervals and the PC bootstrap emerged as methods that adequately control Type I error and have good coverage rates.     

Leisure activities,caregiving demands and catecholamine levels in dementia caregivers

This study examined whether satisfaction from leisure activities moderates the relationship between caregiving demands (i.e., hours per day spent caring for a spouse with dementia) and resting levels of the catecholamines norepinephrine (NE) and epinephrine (EPI). Spousal caregivers (n?=?107; mean age?=?73.95?±?8.12 years) were assessed in home for plasma levels of NE and EPI, amount of care provided, and leisure satisfaction. Regression was used to determine whether leisure satisfaction moderated the relationship between hours providing care per day and catecholamine levels. A significant interaction was found between hours caregiving and leisure satisfaction for NE, but not for EPI. Post hoc regressions were conducted for both NE and EPI. At low leisure satisfaction, time spent caring for a spouse was positively associated with plasma NE (β?=?0.41; p?=?0.005) and EPI (β?=?0.44; p?=?0.003). In contrast, at high levels of satisfaction, time caregiving was not significantly associated with plasma NE (β?=?–0.08; p?=?0.57) or EPI (β?=?0.23; p?=?0.12). These findings suggest that leisure satisfaction may protect caregivers from increases in catecholamines, which have been implicated in cardiovascular risk. Further support for these findings may impact psychological treatments for distressed caregivers.     

Self-Regulation and Performance in High-Fidelity Simulations: An Extension of Ego-Depletion Research

This article extends the research literature related to “ego-depletion.” Although numerous studies have focused on the self-regulatory failure associated with ego-depletion, the extant literature is generally characterized by relatively simple behavioral manipulations and dependent measures. Two studies are described that extend previous ego-depletion findings by employing a high-fidelity, customer service simulation as an ego-depleting manipulation (Study 1) and by using performance on a cognitively demanding naval combat simulator as a dependent measure (Study 2). Results of both studies show the generalizability of the effects of ego-depletion; the implications for self-regulatory failure in more naturalistic settings are discussed.     

The Effects of Cognitive Ability and Conscientiousness on Performance Over Time: A Censored Latent Growth Model

This study examined the extent to which cognitive ability and conscientiousness predicted initial academic performance (i.e., grade point average) and changes in performance over the course of college students' careers. A censored latent growth model that controlled for initial performance was used. Results showed positive effects of cognitive ability and conscientiousness on initial performance, but only positive effects of conscientiousness on performance change. However, because initial performance negatively affected performance change, cognitive ability and conscientiousness negatively indirectly affected performance change through initial performance. In addition, results revealed cognitive ability accounted for more variance in initial performance than conscientiousness, and conscientiousness accounted for more variance in performance change than cognitive ability. Finally, results show that, beyond the third semester, conscientiousness is a better predictor of student performance than cognitive ability.     

Considering the Knowledge You Have: Effects on Realism in Confidence Judgements

In three experiments, the hypothesis was tested that the realism of subjects' item-specific confidence judgements of their own answers to general knowledge questions would be improved by making the subjects heed content which might be expected to lower their confidence ratings. In each experiment, all subjects answered knowledge questions and then rated their confidence in their chosen answer. Before each question, the subjects in the experimental condition assessed the extent of their knowledge in a knowledge-area encompassing the knowledge question. Experiments 1 and 2 differed only in the design of the knowledge-area assessment scale. In Experiment 3, when making knowledge-area assessments, the subjects were asked to find examples of knowledge they lacked in the knowledge-area. Both Experiments 2 and 3 showed significantly improved realism in the subjects' confidence judgements. An aggregate assessment of the total number of questions believed to be answered correctly, given by the subjects at the end of Experiments 2 and 3, showed underconfidence in the experimental condition, particularly in Experiment 3. The results support the idea that the content active in subjects' memory at the time when confidence judgements are made affects their realism. Furthermore, our results show that the realism of subjects' item-specific confidence judgements can be improved without giving feedback.     

Motor Development During the First Year: A Comparative Study

This investigation of the motor development characteristics of Brazilian infants during the 1st year of postnatal life compared the findings with those of a widely used U.S. sample (N. Bayley, 1993). Analysis of the 12 monthly assessments indicated that Brazilian infants' scores increased with age, with the greatest increase occurring over the first 8 months. As a general comparison, the results are similar to the U.S. patterns of development. However, the Brazilian sample exhibited mean scores significantly lower than the U.S. sample's in the 3rd, 4th, and 5th months. More specific analyses indicated that the difference within these months was on tasks of sitting and grasping. Possible explanations for the differences focus on variations in child-rearing practices and the influence of biological maturation in early movement behaviors.     

The Research Use of Projective Techniques: A Seventeen Year Survey

The past 17 volumes (from 1947 to 1064) of the Journal of Projective Techniques and Personality Assessment were surveyed with regard to the degree of research usage of the various projective techniques. The most commonly used techniques in research have remained quite unchanged over the past 17 years, i.e., the Rorschach, the thematic tests, human figure drawings, and sentence completions. An analysis and comparison of the first five years covered by the survey with the last five years suggests that the Szondi, the Mosaic, Finger Painting Analysis, and the Make-A-Picture show relative decline in research usage; the Blacky, the Holtzman Inkblots, and the Hand Test appear to be increasing. The correlation between the degree of clinical usage reported by Sundberg (1961) and the degree of research usage found in this study was .60, suggesting a high relationship between the two measures. However, it was suggested that an increase or decline in the research usage of an instrument may precede a corresponding increase or decline in its clinical usage.     

Time Doesn’t Change Everything: The Longitudinal Course of Distress Tolerance and its Relationship with Externalizing and Internalizing Symptoms During Early Adolescence

Although distress tolerance is an emerging construct of empirical interest, we know little about its temporal change, developmental trajectory, and prospective relationships with maladaptive behaviors. The current study examined the developmental trajectory (mean- and individual-level change, and rank-order stability) of distress tolerance in an adolescent sample of boys and girls (N?=?277) followed over a four-year period. Next we examined if distress tolerance influenced change in Externalizing (EXT) and Internalizing (INT) symptoms, and if EXT and INT symptoms in turn influenced change in distress tolerance. Finally, we examined if any of these trends differed by gender. Results indicated that distress tolerance is temporally stable, with little mean- or individual-level change. Latent growth models reported that level of distress tolerance is cross-sectionally associated with both EXT and INT symptoms, yet longitudinally, only associated with EXT symptoms. These results suggest that distress tolerance should be a focus of research on etiology and intervention.     

At the interface of social cognition and psychometrics: Manipulating the sex of the reference class modulates sex differences in personality traits

Psychometric surveys suggest that sex differences in personality are minimal. Herein, we argue that (a) the mind is likely biased toward assessing oneself relative to same-sex others, and (b) this bias may affect the measurement of sex differences in personality. In support of this, an experiment demonstrates modulation of sex differences on the HEXACO facets by manipulating the sex of the “reference class”—the group of people subjects compare themselves to when making self-assessments on survey items. Although patterns varied across traits, sex differences were relatively small in the “unspecified” and “same-sex” reference class conditions—but substantially larger in the “opposite-sex” condition. These findings point to a same-sex comparison bias that may impact the measurement of sex differences in personality.