Latent variable GIMME using model implied instrumental variables (MIIVs).

Abstract

Researchers across many domains of psychology increasingly wish to arrive at personalized and generalizable dynamic models of individuals' processes. This is seen in psychophysiological, behavioral, and emotional research paradigms, across a range of data types. Errors of measurement are inherent in most data. For this reason, researchers typically gather multiple indicators of the same latent construct and use methods, such as factor analysis, to arrive at scores from these indices. In addition to accurately measuring individuals, researchers also need to find the model that best describes the relations among the latent constructs. Most currently available data-driven searches do not include latent variables. We present an approach that builds from the strong foundations of Group Iterative Multiple Model Estimation (GIMME), the idiographic filter, and model implied instrumental variables with two-stage least squares estimation (MIIV-2SLS) to provide researchers with the option to include latent variables in their data-driven model searches. The resulting approach is called Latent Variable GIMME (LV-GIMME). GIMME is utilized for the data-driven search for relations that exist among latent variables. Unlike other approaches such as the idiographic filter, LV-GIMME does not require that the latent variable model to be constant across individuals. This requirement is loosened by utilizing MIIV-2SLS for estimation. Simulated data studies demonstrate that the method can reliably detect relations among latent constructs, and that latent constructs provide more power to detect effects than using observed variables directly. We use empirical data examples drawn from functional MRI and daily self-report data.

Integrating a functional view on suicide risk into idiographic statistical models

“…Moreover, the GIMME models used here contain measurement error in their estimates, which can bias effect estimates and standard errors. GIMME now has the ability to incorporate latent variable measurement models (Gates, Fisher, & Bollen, 2020). However, it has not yet been widely employed in empirical data.…”

“How Well Does Your Structural Equation Model Fit Your Data?”: Is Marcoulides and Yuan’s Equivalence Test the Answer?

“…Such specification searches have long been known to produce results that subsequent research typically fails to replicate (see MacCallum, 1986;MacCallum et al, 1992). A recent increase in dedicated software programs that automate specification searches (e.g., Brandmaier et al, 2016;Marcoulides and Falk, 2018;Gates et al, 2019) for the purpose of recommending additional model specification changes that would enhance fit exacerbates the concern that researchers are engaging in "HARKing" (hypothesizing after results are known), theoretically 6 Specifically, for the misspecified model shown in Figure 1, both Mplus and AMOS define a null model's df as the differences in df between the alternative (H A :) and null (H 0 :) baseline models as follows. In Mplus, the H A : baseline model has df values that are the sum of: 1) four variances, four means, and six covariances (14 total) among the response variables (i.e., Video Viewing, MLQ Self-Efficacy Posttest, MLQ Task Value Posttest, and Lab Report: Discussion), plus 2) all possible covariances between MLQ Self-Efficacy Posttest, MLQ Task Value Posttest, and Lab Report: Discussion with Video Viewing (six) plus all possible covariances between MLQ Self-Efficacy pretest, MLQ Task Value pretest, and Lawson's test of Scientific Reasoning with Video Viewing (six; 12 total) for an H A : baseline model total of df = (14 + 12) = 26.…”

BRAIN Initiative: Cutting-Edge Tools and Resources for the Community

“…and colleagues are introducing a set of tools that build from group iterative multiple model estimation (GIMME; Gates and Molenaar, 2012), a model search procedure that reliably recovers sparse patterns of relations among brain areas in functional MRI data collected during the resting state and during task performance. The two primary developments are as follows: latent variable GIMME (LV-GIMME; Gates et al, 2019) and subgrouping GIMME (S-GIMME; Gates et al, 2017). Both are freely available within the R package gimme (Lane et al, 2019) with documentation available on-line (http://gimme.web.unc.edu/63-2/).…”