Cross-Temporal Meta-Analysis: A Conceptual and Empirical Critique
- PDF / 874,481 Bytes
- 18 Pages / 595.276 x 790.866 pts Page_size
- 10 Downloads / 234 Views
ORIGINAL PAPER
Cross-Temporal Meta-Analysis: A Conceptual and Empirical Critique Cort W. Rudolph 1
&
David P. Costanza 2 & Charlotte Wright 2 & Hannes Zacher 3
# Springer Science+Business Media, LLC, part of Springer Nature 2019
Abstract The proper estimation of age, period, and cohort (APC) effects is a pervasive concern for the study of a variety of psychological and social phenomena, inside and outside of organizations. One analytic technique that has been used to estimate APC effects is cross-temporal meta-analysis (CTMA). Although CTMA has some appealing qualities (e.g., ease of interpretability), it has also been criticized on theoretical and methodological grounds. Furthermore, CTMA makes strong assumptions about the nature and operation of cohort effects relative to age and period effects that have not been empirically tested. Accordingly, the goal of this paper is to explore CTMA, its history, and these assumptions. Using a Monte Carlo study, we demonstrate that, in many cases, cohort effects are misestimated (i.e., systematically over- or underestimated) by CTMA. This work provides further evidence that APC effects pose intractable problems for research questions where APC effects are of interest. Keywords Age . Cohorts . Cross-temporal meta-analysis . Generations . Lifespan
Across literatures, a great deal of effort has been invested to properly estimate the independent influence of age, period, and cohort (APC) effects (e.g., Baltes, 1968; Costanza & Finkelstein, 2015; Costanza, Darrow, Yost, & Severt, 2017; Hofer & Sliwinski, 2006; Koslowski, 1986; Schaie, 1986; Yang, 2008; Yang & Land, 2013). For example, research has looked at the effect of age as a marker of individual development in the work context (e.g., Rudolph & Baltes, 2016), of major historical events on various populations (e.g., Elder, 1974; Elder & Liker, 1982), and on differences among cohorts (e.g., Gerstorf, Ram, Hoppmann, Willis, & Schaie, 2011; Schaie, 2013). However, in most of these studies, age, period, and cohort are not independent of each other and, hence, separating out their relative effects is challenging, if not impossible.
A pre-print version of this work can be found here: https://psyarxiv.com/ exskp/. Code to replicate the simulations presented here can be found at: https://osf.io/mak6y/. A Shiny web-app is also available here: https:// cortrudolph.shinyapps.io/CTMA_Simulation. * Cort W. Rudolph [email protected] 1
Department of Psychology, Saint Louis University, St. Louis, MO, USA
2
The George Washington University, Washington, DC, USA
3
Institute of Psychology, Leipzig University, Leipzig, Germany
It is understood that, statistically, the separation of age, period, and cohort effects generally represents an intractable problem (Glenn, 1976; Glenn, 2005; Bell & Jones, 2013, 2014). This problem becomes intractable when these three effects are each defined by non-independent temporal variables. Although this has long been acknowledged by statisticians, there is no shortage of theories that beg for t
Data Loading...
