M<i>plus</i> Trees: Structural Equation Model Trees Using M<i>plus</i>

Serang, Sarfaraz; Jacobucci, Ross; Stegmann, Gabriela; Brandmaier, Andreas M.; Culianos, Demi; Grimm, Kevin J.

doi:10.1080/10705511.2020.1726179

Cited by 14 publications

(23 citation statements)

References 26 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…The LIP yields the percent improvement in the −2LL from the baseline model for the more parameterized model. The same metric, but scaled in terms of the proportional improvement in model fit, has been implemented in recursive partitioning algorithms for mixed-effects (Abdolell et al, 2002;Stegmann et al, 2018) and structural equation models (Serang et al, 2020). The main reason to do this is to provide context regarding the relative improvement in model fit.…”

Section: An Effect Size Measure For Model Comparisonmentioning

confidence: 99%

Model Fit and Comparison in Finite Mixture Models: A Review and a Novel Approach

2021

Self Cite

View full text Add to dashboard Cite

One of the greatest challenges in the application of finite mixture models is model comparison. A variety of statistical fit indices exist, including information criteria, approximate likelihood ratio tests, and resampling techniques; however, none of these indices describe the amount of improvement in model fit when a latent class is added to the model. We review these model fit statistics and propose a novel approach, the likelihood increment percentage per parameter (LIPpp), targeting the relative improvement in model fit when a class is added to the model. Simulation work based on two previous simulation studies highlighted the potential for the LIPpp to identify the correct number of classes, and provide context for the magnitude of improvement in model fit. We conclude with recommendations and future research directions.

show abstract

Section: An Effect Size Measure For Model Comparisonmentioning

confidence: 99%

Model Fit and Comparison in Finite Mixture Models: A Review and a Novel Approach

2021

Self Cite

View full text Add to dashboard Cite

show abstract

“…This approach builds on the decision trees popularized by Breiman et al (1984) by fitting SEMs in each node, thereby performing a form of exploratory multiple-group SEM. I will discuss a version of SEM Trees known as Mplus Trees (Serang et al, 2020), a variant where the SEMs are fit in Mplus, but the trees themselves are grown in R. The Mplus Trees algorithm works as follows. First, an LGM (without covariates) is fit to the full dataset.…”

Section: Structural Equation Model Treesmentioning

confidence: 99%

“…(1984) by fitting SEMs in each node, thereby performing a form of exploratory multiple‐group SEM. I will discuss a version of SEM Trees known as M plus Trees (Serang et al., 2020), a variant where the SEMs are fit in M plus , but the trees themselves are grown in R.…”

Section: Structural Equation Model Treesmentioning

confidence: 99%

A comparison of three approaches for identifying correlates of heterogeneity in change

Serang

2021

New Directions for Child and Adolescent Development

Self Cite

View full text Add to dashboard Cite

Longitudinal research is often interested in identifying correlates of heterogeneity in change. This paper compares three approaches for doing so: the mixed-effects model (latent growth curve model), the growth mixture model, and structural equation model trees. Each method is described, with special focus given to how each structures heterogeneity, attributes that heterogeneity to covariates, and the kinds of research questions each can be used to address. Each approach is used to analyze data from the National Longitudinal Survey of Youth to understand the similarities and differences between methods in the context of empirical data. Specifically, changes in weight across adolescence are examined, as well as how differences in these change patterns can be explained by sex, race, and mother's education. Recommendations are provided for how to select which approach is most appropriate for analyzing one's own data. K E Y W O R D S growth mixture model, latent growth curve model, longitudinal, mixed-effects model, structural equation model trees Longitudinal research is undertaken for a number of reasons. Baltes and Nesselroade (1979) highlighted five main rationales for using longitudinal research within developmental psychology: (1) identification of intraindividual change, (2) identification of interindividual differences in intraindividual change, (3) analysis of interrelationships in behavioral change, (4) analysis of causes (determinants) of intraindividual change, and (5) analyses of causes (determinants) of interindividual differences in intraindividual change. This paper will focus primarily on the fifth rationale, comparing three statistical approaches for implementing it.

show abstract

“…There are currently two software packages for the statistical programming language R that allow fitting SEM trees. One is the package ( Brandmaier et al, 2013b ) that has been widely applied in the literature ( Brandmaier et al, 2013a , 2016 , 2017 , 2018 ; Jacobucci et al, 2017 ; Usami et al, 2017 , 2019 ; de Mooij et al, 2018 ; Ammerman et al, 2019 ; Serang et al, 2020 ; Simpson-Kent et al, 2020 ). The other software implementation is the package ( Hothorn and Zeileis, 2015 ).…”

Section: Introductionmentioning

confidence: 99%

“…Recently, Zeileis (2020) demonstrated on his blog how MOB can be coupled with the SEM software ( Rosseel, 2012 ) to estimate SEM trees. Outside of the R ecosystem, SEM trees have also been fitted in M plus ( Serang et al, 2020 ).…”

Section: Introductionmentioning

confidence: 99%

Score-Guided Structural Equation Model Trees

2021

Self Cite

View full text Add to dashboard Cite

Structural equation model (SEM) trees are data-driven tools for finding variables that predict group differences in SEM parameters. SEM trees build upon the decision tree paradigm by growing tree structures that divide a data set recursively into homogeneous subsets. In past research, SEM trees have been estimated predominantly with the R package semtree. The original algorithm in the semtree package selects split variables among covariates by calculating a likelihood ratio for each possible split of each covariate. Obtaining these likelihood ratios is computationally demanding. As a remedy, we propose to guide the construction of SEM trees by a family of score-based tests that have recently been popularized in psychometrics (Merkle and Zeileis, 2013; Merkle et al., 2014). These score-based tests monitor fluctuations in case-wise derivatives of the likelihood function to detect parameter differences between groups. Compared to the likelihood-ratio approach, score-based tests are computationally efficient because they do not require refitting the model for every possible split. In this paper, we introduce score-guided SEM trees, implement them in semtree, and evaluate their performance by means of a Monte Carlo simulation.

show abstract

Mplus Trees: Structural Equation Model Trees Using Mplus

Cited by 14 publications

References 26 publications

Model Fit and Comparison in Finite Mixture Models: A Review and a Novel Approach

Model Fit and Comparison in Finite Mixture Models: A Review and a Novel Approach

A comparison of three approaches for identifying correlates of heterogeneity in change

Score-Guided Structural Equation Model Trees

Contact Info

Product

Resources

About