I have asked following questions before

More than one outcome (dependent) variables in ordinal logistic regression

How to handle more than one dependent variable (categorical) in logistic regression?

Using Rasch model to explain relationships between a set of dependent and independent variables

Use of further analysis on factors formed by principal component analysis in regression

answers of the above questions help me to understand what could be done with my data. My data consists of 2 independent variable (income and years- continuous data) with control variables (dichotomous or categorical data) and I have a dependent variable with 6 proxies e.g for dependent variable Business development (latent variable), six questions were asked in 5-point likert scale (categorical-ordinal data) increase in profit, sales, size, marketing, asset and labour (manifest variables).

I want to find out the relationship between dependent and independent variable. I want to use measurement model to find out this. I came to know from http://en.wikipedia.org/wiki/Latent_variable_model

that I could use latent variable model. I am very much sure that my manifest variables are categorical and I suppose my latent variable is also categorical (need to know), therefore Latent Class Analysis seems to be the best choice for my data. My question is that:

  • Am I right that the Latent Class Analysis is best for my data or have to think of any other thing?
  • Could you give an example when latent variable is continuous and manifest variable is categorical? these are the situations when there is need to use latent trait analysis. I just ask this question to clarify my latent variable as continuous or categorical.
  • I know that for latent class analysis LatentGold software and Mplus are best but expensive to buy, Do you think Lem-a free software could serve my purpose, as I want to do factor and regression in Latent class analysis? or what to you think of SATA for it?
  • It is possible in latent class analysis to do cluster method, factor analysis and regression for categorical data. Is it possible to do more than that in it like structural equation modelling and path analysis for categorical data etc?
  • Latent class analysis is mostly used in medicine and psychology and less frequently in development or applied economics. Could I able to justify the use of latent class analysis for research related to business (mix of management and development economics area).

Thanks for the answers in advance.

  • $\begingroup$ You might consider accepting (and upvoting) some of those previous answers to your questions. $\endgroup$
    – Andy W
    Commented Jan 30, 2012 at 13:44
  • $\begingroup$ @Andy W: Thanks for the comment. Actually the previous answers help me to understand which method is best. Now when I selected Latent class analysis, I have some questions on it. I have not asked about Latent class before. $\endgroup$
    – wxa
    Commented Feb 1, 2012 at 0:40
  • 1
    $\begingroup$ the way the site works is if you find answers helpful, you should upvote them, and if they reasonable help solve your problem you should mark the prior question as answered (the tools to do both of these are outside the top left corner of your question). Given all of your questions I have flagged for a moderator to attempt to merge your accounts, and I would suggest you register your account with the site. $\endgroup$
    – Andy W
    Commented Feb 1, 2012 at 13:02
  • 2
    $\begingroup$ I have merged your four accounts. (Thanks to @Andy for noticing that!) Since you have registered this one, you can now use it in the future so that you get system notification, you can vote up and accept responses that you found useful or that did answer your questions--even if they lead to new related questions. $\endgroup$
    – chl
    Commented Feb 1, 2012 at 13:14


Browse other questions tagged or ask your own question.