Philippe Rast

An empirical comparison of the therapeutic benefits of physical exercise and cognitive training on the executive functions of older adults: a meta-analysis of controlled trials.

UNLABELLED A robust body of aging-related research has established benefits of both physical exercise (PE) and cognitive training (CT) on executive functions related to the activities of daily living of older adults; however, no meta-analysis has compared these treatments. OBJECTIVE The current quantitative review involved a comparison of the overall effect sizes of PE and CT interventions on executive functions (Morris, 2008; pre-post-controlled effect size: d(ppc)), while also exploring contextual moderators of treatment outcomes. METHOD A systematic review identified 46 studies (23 PE, 21 CT, and 2 both) meeting inclusion criteria (i.e., controlled interventions, executive-related outcomes, mean ages 65+, information to calculate d(ppc)). RESULTS The weighted mean dppc values came to 0.12 (p < .01) for PE and 0.24 (p < .01) for CT. Treatment effects differed based on executive constructs for CT, with problem solving presenting the highest d(ppc) (0.47, p < .01). Notably, PE produced similar effect sizes across distinct executive functions. Treatment characteristics (e.g., session length/frequency) did not predict effect sizes. CT had a significant benefit on healthy participants (0.26, p < .01), but cognitively impaired samples did not experience a significant effect. CONCLUSIONS Both treatments improved executive functions, but CT presented a potential advantage at improving executive functions. Improvements in executive functions differed depending on construct for CT, whereas each construct produced similar, modest effect sizes for PE. Publication bias and study quality variability potentially bias these conclusions, as lower quality studies likely produced inflated effect sizes.

Longitudinal design considerations to optimize power to detect variances and covariances among rates of change: Simulation results based on actual longitudinal studies

We investigated the power to detect variances and covariances in rates of change in the context of existing longitudinal studies using linear bivariate growth curve models. Power was estimated by means of Monte Carlo simulations. Our findings show that typical longitudinal study designs have substantial power to detect both variances and covariances among rates of change in a variety of cognitive, physical functioning, and mental health outcomes. We performed simulations to investigate the interplay among number and spacing of occasions, total duration of the study, effect size, and error variance on power and required sample size. The relation between growth rate reliability (GRR) and effect size to the sample size required to detect power greater than or equal to .80 was nonlinear, with rapidly decreasing sample sizes needed as GRR increases. The results presented here stand in contrast to previous simulation results and recommendations (Hertzog, Lindenberger, Ghisletta, & von Oertzen, 2006; Hertzog, von Oertzen, Ghisletta, & Lindenberger, 2008; von Oertzen, Ghisletta, & Lindenberger, 2010), which are limited due to confounds between study length and number of waves, error variance with growth curve reliability, and parameter values that are largely out of bounds of actual study values. Power to detect change is generally low in the early phases (i.e., first years) of longitudinal studies but can substantially increase if the design is optimized. We recommend additional assessments, including embedded intensive measurement designs, to improve power in the early phases of long-term longitudinal studies.

Modeling Individual Differences in Within-Person Variation of Negative and Positive Affect in a Mixed Effects Location Scale Model Using BUGS/JAGS

A mixed effects location scale model was used to model and explain individual differences in within-person variability of negative and positive affect across 7 days (N=178) within a measurement burst design. The data come from undergraduate university students and are pooled from a study that was repeated at two consecutive years. Individual differences in level and change in mood was modeled with a random intercept and random slope where the residual within-person variability was allowed to vary across participants. Additionally changes in within-person variability were explained by the inclusion of a time-varying predictor indicating the severity of daily stressors. This model accounted for 2 location and 2 scale effects and provided evidence that individuals who reported higher severity in daily stressors also exhibited greater variability in affect—but only for participants who showed low overall affect variability and who reported low average negative affect. Those who were more variable in their affect reports overall were less reactive to daily stressors in the sense that their high levels of affect variability remained high. We describe the utility of this model for further research on individual variation and change.

Verbal Knowledge, Working Memory, and Processing Speed as Predictors of Verbal Learning in Older Adults.

The present study aimed at modeling individual differences in a verbal learning task by means of a latent structured growth curve approach based on an exponential function that yielded 3 parameters: initial recall, learning rate, and asymptotic performance. Three cognitive variables-speed of information processing, verbal knowledge, working memory-and the participants age were included in the model in order to explain individual differences in the learning parameters. The data come from the second wave of the Zurich Longitudinal Study on Cognitive Aging (D. Zimprich, Martin, et al., 2008) comprising 334 participants ranging in age from 66 to 81 years (M = 74.43, SD = 4.41). Among the logistic, the Gompertz, and the hyperbolic function, the exponential function described the data best. Reliable individual differences were found in all 3 learning parameters. The cognitive predictor variables affected the verbal learning parameters differentially: All 3 predictors affected positively initial recall, the asymptotic performance increased with better working memory and faster processing speed, and the learning rate was positively associated with verbal knowledge only. Age did not affect the learning parameters but correlated negatively with working memory and processing speed. The finding of large and reliable individual differences in learning is seen as evidence that the potential for positive change, or plasticity in adulthood is maintained and that it is worthwhile to enhance the determinants of learning or learning itself.

The Identification of Regions of Significance in the Effect of Multimorbidity on Depressive Symptoms using Longitudinal Data: An Application of the Johnson-Neyman Technique

Background: The investigation of multimorbidity and aging is complex and highly intertwined with aging-related changes in physical and cognitive capabilities, and mental health and is known to affect psychological distress and quality of life. Under these circumstances it is important to understand how the effects of chronic conditions evolve over time relative to aging-related and end-of-life changes. The identification of periods in time where multimorbidity impacts particular outcomes such as depressive symptoms, versus periods of time where this is not the case, reduces the complexity of the phenomenon. Objective: We present the Johnson-Neyman (J-N) technique in the context of a curvilinear longitudinal model with higher-order terms to probe moderators and to identify regions of statistical significance. In essence, the J-N technique allows one to identify conditions under which moderators impact an outcome from conditions where these effects are not significant. Methods: To illustrate the use of the J-N technique in a longitudinal sample, we used data from the Health and Retirement Study. Analyses were based on time-to-death models including participants who died within the study duration of 12 years. Results: Multimorbidity differentially affects rates of change in depression. For some periods in time the effects are statistically significant while in other periods the same effects are not statistically different from zero. Conclusion: The J-N technique is useful to continuously probe moderating effects and to identify particular interactions with the model for time when certain effects are or are not statistically significant. In the context of multimorbidity this method is particularly useful for interpreting the complex interactions with differential change over time.

The Unity and Diversity of Executive Functions: A Systematic Review and Re-Analysis of Latent Variable Studies

Confirmatory factor analysis (CFA) has been frequently applied to executive function measurement since first used to identify a three-factor model of inhibition, updating, and shifting; however, subsequent CFAs have supported inconsistent models across the life span, ranging from unidimensional to nested-factor models (i.e., bifactor without inhibition). This systematic review summarized CFAs on performance-based tests of executive functions and reanalyzed summary data to identify best-fitting models. Eligible CFAs involved 46 samples (N = 9,756). The most frequently accepted models varied by age (i.e., preschool = one/two-factor; school-age = three-factor; adolescent/adult = three/nested-factor; older adult = two/three-factor), and most often included updating/working memory, inhibition, and shifting factors. A bootstrap reanalysis simulated 5,000 samples from 21 correlation matrices (11 child/adolescent; 10 adult) from studies including the three most common factors, fitting seven competing models. Model results were summarized as the mean percent accepted (i.e., average rate at which models converged and met fit thresholds: CFI ≥ .90/RMSEA ⩽ .08) and mean percent selected (i.e., average rate at which a model showed superior fit to other models: &Dgr;CFI ≥ .005/.010/&Dgr;RMSEA ⩽ −.010/−.015). No model consistently converged and met fit criteria in all samples. Among adult samples, the nested-factor was accepted (41–42%) and selected (8–30%) most often. Among child/adolescent samples, the unidimensional model was accepted (32–36%) and selected (21–53%) most often, with some support for two-factor models without a differentiated shifting factor. Results show some evidence for greater unidimensionality of executive function among child/adolescent samples and both unity and diversity among adult samples. However, low rates of model acceptance/selection suggest possible bias toward the publication of well-fitting but potentially nonreplicable models with underpowered samples.

Abstract: Power to Detect Within- and Between-Person Effects: A Comparison of Multilevel SEM and Unit-Weighted Scale Scores

Power to Detect Withinand Between-Person Effects: A Comparison of Multilevel SEM and Unit-Weighted Scale Scores Jonathan Rush, Philippe Rast, and Scott M. Hofer University of Victoria Intensive repeated measurement designs (e.g., daily diary) are frequently used to investigate within-person variation over relatively brief intervals of time. These designs allow variance to be partitioned into within-person (WP) and between-person (BP) sources of variability, enabling differential effects and factor structures to be estimated at the WP and BP level of analysis. The majority of research utilizing these designs relies on unit-weighted scale scores (Figure 1a), which assumes that the constructs are measured without error. Failing to account for such error has the potential to bias estimates and may decrease the sensitivity to detect WP and BP effects. An alternative approach makes use of multilevel SEM (Figure 1b), which permits the specification of latent variables at both WP and BP levels. These models disattenuate measurement error from systematic variance and should produce less biased WP and BP estimates and larger effects. However, factor models often result in poorer precision (increased standard errors) than observed score models, which can diminish power, despite the larger effect sizes. Differences in power, precision, and bias between multilevel unit-weighted and factor models were compared through a series of Monte Carlo simulations. First, an actual data set of 147 participants measured daily on a 10-item scale over 14 days with a single WP and BP covariate was used to derive population parameters. Second, hypothetical data were generated to examine the models under less desirable conditions (i.e., poor reliability, heterogeneous factor loadings, and fewer items). Both the unit-weighted and factor models performed comparably in power to detect WP and BP effects when population parameters were generated from actual data. Results based on simulated data revealed that precision was consistently poorer in the factor models than the unit-weighted models, particularly when reliability was Correspondence concerning this abstract should be addressed to Jonathan Rush, Department of Psychology, University of Victoria, P.O. Box 1700 STN CSC, Victoria, BC V8W 2Y2, Canada. E-mail: jrush@uvic.ca FIGURE 1 (a) Unit-weighted multilevel model with within-person and between-person covariate; (b) Multilevel factor model with within-person and between-person covariate. low. However, the degree of bias was considerably greater in the unit-weighted model than the factor model. Despite the unit-weighted model consistently underestimating the effect of a covariate, it still generally produced higher power than the factor models, due to the greater precision.