hyunsooseol / snowlatent

Latent Class analysis: This module allows users to conduct LCA, Multiple group LCA, and Multilevel LCA based on glca R package, and provide plot such as Profile plot and Radar chart within module.

snowLatent

This module allows users to conduct LCA, Multiple group LCA, and Multilevel LCA based on glca R package.

snowLatent module can be installed and used in R as a standard R packages.

Installation

At the moment, snowLatent is not in CRAN yet, so you need to install it via devtools

library(jmv)

library(haven)

library(jmvReadWrite)

library(jmvconnect)

devtools::install_github("hyunsooseol/snowLatent")

snowlatent's Issues

Relative model fit 산출 관련 문의 드립니다.

안녕하세요 교수님, 워크샵에 참여했던 학생입니다.
메일 남겨드렸으나 확인하시지 못하는 것 같아 깃허브에서 인사드리게 되었습니다. :)

모듈 작동시에 변수 개수가 늘어날 경우 Relative model fit 이 산출되지 않는 경우가 있습니다.
여러 방법을 테스트해 본 결과, 특정 변수의 문제이거나 문제가 발생하는 변수 개수는 특정되지는 않았습니다.
말 그대로 랜덤하게 결과가 산출되지 않는 것으로 보입니다. 보통 6~7개 사이에서 문제가 발생합니다.

혹시 데이터 상 문제가 있는 것일까요? 혹시 확인 필요하실지 몰라 사용한 데이터는 메일에 첨부드렸습니다..!
감사합니다.

Probabilities not properly assigned to rows if filter is applied

Hi,

LCA in snowLatent works pretty fine, however I ran into one limitation:
If a filter is applied, probabilities like class membership are not properly assigned to the rows.

Concretely, if e.g. 100 of 1.000 cases are selected via a filter and LCA is performed, the probabilities are assigned to the first 100 cases instead of the 100 selected cases.

Is there any workaround at the moment?

Best,
Stephan

Classify membership using multigroup and multivel LCA

Dear Dr. Seol,

Based on my experience with your snowLatent package (Multiple and Multilevel) I realized that there is no option to classify membership, as in a classic LCA, wich means that it is not possible to input the classification of each participant and run additional analyses. Using the classic LCA (first analysis of the package), this is fully possible. Could you please add this option, please?

Another request: In the Multiple and Multilevel analyses, the system shows the "item response probabilities by class" only for the overall sample, but does not separate it for each of the groups. It is certain that this separation will help us better understand the behavior of the classes, as well as provide material for scientific articles. Note that the "class prevalences by group" already appears separated from each group.

In addition, I would like to congratulate you on your package, as it is helping us a lot and making LCA more accessible to researchers.

Best regards,