Before posting, check our FAQ to see if your question is already covered.
Post Reply
PLS Junior User
Posts: 1
Joined: Mon Jan 08, 2024 11:14 am
Real name and title: Mohammed Hassanein-PhD Student


Post by hassaneinmohd »


I am working on a research project to identify patient profiles of treatment acceptability to different modalities. Since we do not have a variable that label or assign each participant to a group, we run a FIMIX-PLS analysis to look for unobserved classes within the sample. Acceptability was measured using a scale of 13 items all on the same Likert scale. After I run the measurement model, identified number of segments, and each respondent was assigned to a group. I created a new data file. My questions are:

1- Is this FIMIX-PLS same as the "bias-adjusted three step approach"?

2- How to look at the explanatory variable that derived the segmentation of the sample. I found some used cross tabs and looked at the closet distribution between the original sample and the resulted segments? other looked at the discrete segment assignment correlations and indicators with high correlation were used to generate new groups? I am not sure which is the right method here in FIMIX-PLS?

3-How to define the latent classes (is it based on the correlation of each FIMIX segment with its indicators or any other method?)

4-Can i used the generated categorical classes as an outcome variable to build a supervised classification tree to look at the most important predictors of each generated group and in which I aim to add covariates (demographics and other related variables?) is this a right method?

Thank you very much

Best regards,
Post Reply