Evaluation of Analysis Approaches for Latent Class Analysis with Auxiliary Linear Growth Model

Abstract

This study investigated the performance of three selected approaches to estimating a two-phase mixture model, where the first phase was a two-class latent class analysis model and the second phase was a linear growth model with four time points. The three evaluated methods were (a) one-step approach, (b) three-step approach, and (c) case-weight approach. As a result, some important results were demonstrated. First, the case-weight and three-step approaches demonstrated higher convergence rate than the one-step approach. Second, it was revealed that case-weight and three-step approaches generally did better in correct model selection than the one-step approach. Third, it was revealed that parameters were similarly recovered well by all three approaches for the larger class. However, parameter recovery for the smaller class differed between the three approaches. For example, the case-weight approach produced constantly lower empirical standard errors. However, the estimated standard errors were substantially underestimated by the case-weight and three-step approaches when class separation was low. Also, bias was substantially higher for the case-weight approach than the other two approaches.

High‐Need Phenotypes in Medicare Beneficiaries: Drivers of Variation in Utilization and Outcomes

“…20 However, these methods were tested using simulation data with the acknowledgment that practical application of these methods can lead to model convergence issues and potential changes in classification of persons into latent groups. 20,21 Therefore, in this study, we descriptively compared outcomes across HN phenotypes rather than using regression analysis, which can result in biased estimates of outcome effect sizes and standard errors. 21 Moreover, the assumption of conditional independence stipulates that predictor variables within a latent class should be independent of one another, or uncorrelated.…”

“…20,21 Therefore, in this study, we descriptively compared outcomes across HN phenotypes rather than using regression analysis, which can result in biased estimates of outcome effect sizes and standard errors. 21 Moreover, the assumption of conditional independence stipulates that predictor variables within a latent class should be independent of one another, or uncorrelated. However, this assumption was difficult to test formally in our data because statistical tests of association are systematically significant in large samples, regardless of effect size.…”

Distinct age‐related patterns of overweight development to guide school healthcare interventions

“…To account for the uncertainty of class assignment, we used the caseweight approach, which we considered to be the most appropriate because of the relative entropy being less than 0.80 and the presence of small latent classes. 32 We treated children as fractional members of all identified latent classes by allowing each boy and girl to be included in each of the sex-specific classes, when accounting for the corresponding posterior probabilities of class membership by means of weighting. We then calculated the effective sample size and weighted statistics of the key attributes for each class as well as performed statistical comparisons of these attributes.…”

Ideological profiles of US adults and their support for youth mentoring

“…One approach involves classifying people according to some measurement indicators, and then independently analyzing a prediction model using some set of independent variables. This approach is referred to as the “classify‐and‐analyze,” or hard partitioning 2‐step, approach (Kamata, Kara, Patarapichavatham, & Lan, ). Implicit in this approach is the assumption that class specification is independent of prediction and that classes are errorless during prediction (i.e., the model excludes measurement error), which is not the case in reality.…”

“…Moreover, the 1-step approach has been found to improve class separation (Clark & Muthén, 2009). There are acceptable alternatives to the 1-step model, including variants on a 3-step model (see Kamata, et. al, 2018); however, the 1-step approach performs as well or better than the 3-step approach when using large sample sizes, and is recommended when entropy is less than 0.80, as in our case.…”