Toward a Unified Framework for the Study of Between-Person and Within-Person Structures: Building a Bridge Between Two Research Paradigms

Manuel C Voelkle,Annette Brose,Florian Schmiedek,Ulman Lindenberger

doi:10.1080/00273171.2014.889593

Abstract

The vast majority of empirical research in the behavioral sciences is based on the analysis of between-person variation. In contrast, much of applied psychology is concerned with the analysis of variation within individuals. Furthermore, the mechanisms specified by psychological theories generally operate within, rather than across, individuals. This disconnect between research practice, applied demands, and psychological theories constitutes a major threat to the conceptual integrity of the field. Following groundbreaking earlier work, we propose a conceptual framework that distinguishes within-person (WP) and between-person (BP) sources of variation in psychological constructs. By simultaneously considering both sources of variation, it is shown how to identify possible reasons for nonequivalence of BP and WP structures as well as establishing areas of convergence. For this purpose, we first introduce the concept of conditional equivalence as a way to study partial structural equivalence of BP and WP structures in the presence of unconditional nonequivalence. Second, we demonstrate the construction of likelihood planes to explore the causes of structural nonequivalence. Third, we examine 4 common causes for unconditional nonequivalence—autoregression, subgroup differences, linear trends, and cyclic trends—and demonstrate how to account for them. Fourth, we provide an empirical example on BP and WP differences in attentiveness.

Highlights

View supplementary material Submit your article to this journal View related articles Citing articles: 61 View citing articles
As apparent from the three-dimensional likelihood plane in Figure 5, with two exceptions, not a single WP structure is unconditionally equivalent to a BP structure at any occasion t. This is exactly what we would expect for most psychological constructs and is further empirical evidence that a simple generalization from a BP structure to a WP structure and vice versa is not justified (Borsboom, Mellenbergh, & van Heerden, 2003; Molenaar, 2004; Molenaar & Campbell, 2009)
Controlling for some very basic factors like mean differences, autoregression, and gender differences, the BP factorial structure of attentiveness turned out to be indistinguishable from the WP structure in about 50% of all possible comparisons

Summary

A One-Factor Model

The data (yit ) for each person and occasion combination (each “cell”) were independently generated in the baseline condition (according to the true model), the results of the T · N likelihood ratio tests in Figure 2 are not independent of each other This is because we do not compare single cells. Ing for conditional equivalence—that is, when controlling for the autoregressive effect (β) at the within level—the number of significant likelihood ratio tests reduces to 0.76% (at an alpha level of 1%) and results in a likelihood plane (right side of Figure 3A) that is similar to the likelihood plane in the baseline condition (Figure 2) Another reason for nonequivalence of BP and WP structures may be that individuals are grouped in a meaningful way. We used R Version 2.15.0 (R Core Team, 2012) and mkfm (Dolan, 2010) for all statistical analyses

Results

Discussion of the Empirical Example

OVERALL DISCUSSION AND CONCLUSIONS