Analysis of survey data - How to create latent variables that can be used for regression?

[Denise Lue]

Hey Stata Users!

I am currently writing my master thesis and I am stuck with my analysis. I have a set of survey data and I need to make a Factor analysis.
Respondents provided the perceived integration levels of different functions in companies. I need to have a latent variable that measures the overall integration level.

My teacher said that I have to use a factor analysis so I googled and read a lot. I think that I have to do a confirmatory factor analysis because I already know what my factors should measure.
I found two different methods that both provide slightly different results. What is the difference between them?

1) sem (Integrationlvl -> extentofintegration, ) (Integrationlvl -> O, ) (Integrationlvl -> P, ) (Integrationlvl -> Q, ) (Integrationlvl -> R, ) (Integrationlvl -> S, ), method(mlmv) standardized latent(Integrationlvl ) nocapslatent

2) sem (INTEGRATIONLVL -> extentofintegration o p q r s)


After completing this step I received the output of the factor analysis, which is totally fine (even though the values of 1) and 2) differ slightly), but now I still don't have a variable which I can use for further analysis. The latent variable "integration level " is the dependent variable in my research model. How can I add it to my data set?

Thanks for your help!

Denise

[William Lisowski]

Welcome to Statalist.

You likely want to use the predict command following your sem - see the output of help sem predict for documentation.