For my master thesis I'm making a structural equation model. However, I do not have much experience with this. So here are some different questions.

  1. I read that one of the assumptions important for SEM is: Linearity: A linear relationship is assumed between endogenous and exogenous variables. However, I see in other papers that this is only considered for the sum score of a latent variable. But not for 'normal/other' variables, such as age. Does this have to be done?

  2. In addition, I never intended to make a sum score of my variables, is this always the case?

  3. I only have 86 respondents, because I have a very specific population and because of time limitation. Does this mean that I can only have 86/10 = 8 variables in total in my SEM model? Or does this mean that I may have 5 variables per latent factor? I find conflicting information.

  4. Is it possible to include 'independent' variables in a SEM model, such as car ownership. So not link this to a latent variable, or is that not possible?

Thank you in advance!


  • $\begingroup$ Linearity just means that the arrows in the path diagram represent linear (as opposed to curvilinear or nonlinear) effects. If age has a curvilinear effect on your latent variable, a standard SEM will not be able to model that, and your model will be misspecified. It has nothing to do with sum scores. $\endgroup$ – Noah Apr 25 '19 at 18:12

welcome to CrossValidated!

  1. I'm not sure quite what you mean by this. You assume linearity, but it's often impossible to test. I don't know what the sum score of a latent variable is.
  2. It's not always the case, but if you're going to make a sum score, why have a latent?
  3. Whoa! That's a really small sample size. It's going to be hard to say anything interesting in SEM with a sample that small, however many variables you have.
  4. I'm not sure what you're getting at here. What is this variable doing if it's not linked to a latent variable? (You can have SEMs with no latent variables though, so I think the answer is yes.)

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.