0
$\begingroup$

I am reading (also learning) a paper using structural equation modeling (SEM) as their analysis method. Their analysis was constructed by using a correlation matrix and asymptotic variances. They follow the two-step method (Anderson and Gerbing, 1988) with the baseline model being constructed first to confirm the stability of the latent variables. I am still confused about the term "baseline model" here. The author proposed a causal model (all paths), which is later examined. However, I cannot find they defined "baseline model". What is the baseline model here?

Thanks

Improve this question
$\endgroup$

1 Answer 1

Reset to default
1
$\begingroup$

There are a few definitions of baseline model. Typically, it's also the null model, which contains no parameters (except means and variances). It's (just about) the worst model you could have. But that can get more complicated: https://www.researchgate.net/publication/10764067_On_Specifying_the_Null_Model_for_Incremental_Fit_Indices_in_Structural_Equation_Modeling (you probably don't need to worry about that).

It's a long time since I've read that paper, but IIRC, they might also be talking about the structural baseline model, which is the model with all of the measurement parameters estimated, but the structural parts (relationships between the latents - both regression and correlation) constrained to zero.

That paper is pretty old, and the way that things are thought about has changed somewhat (e.g. people now put measured variables into the structural model, which wasn't possible in 1988, without some hacks).

Perhaps describe what the problem is that you are trying to solve with the baseline model?

Improve this answer
$\endgroup$
5
  • $\begingroup$ Thanks so much for your reply. The paper is Vance, D. E., Roenker, D. L., Cissell, G. M., Edwards, J. D., Wadley, V. G., & Ball, K. K. (2006). Predictors of driving exposure and avoidance in a field study of older drivers from the state of Maryland. Accident Analysis & Prevention, 38(4), 823-831. doi.org/10.1016/j.aap.2006.02.008. $\endgroup$
    – AsleepOrDead
    Commented Jan 7, 2021 at 4:29
  • $\begingroup$ The study examined age, gender, health, cognitive as predictors of driving avoidance, and driving exposure. They used SEM as the analysis model, health, cognitive, and physical are latent variables measured by various tests or questionnaires. They said that they followed the two-step SEM method, started with a baseline model. However, I have no idea what is the baseline model, what the "baseline model" referred to. All they preposed or presented is their casual model. The purpose of the baseline model is to "check the validity of the latent variables", as stated in the paper. $\endgroup$
    – AsleepOrDead
    Commented Jan 7, 2021 at 4:39
  • $\begingroup$ I don't have access to the paper. $\endgroup$
    – Jeremy Miles
    Commented Jan 7, 2021 at 5:25
  • $\begingroup$ Sorry about the paper, but I think I have described clearly. The baseline model is for checking the validity of the latent variables, no specific problem included, it is the first step of the routine two-step SEM (Anderson and Gerbing, 1988). Thanks for your reply. $\endgroup$
    – AsleepOrDead
    Commented Jan 7, 2021 at 22:38
  • $\begingroup$ The first step in the A&G approach is a saturated structural model. This is not (usually) called a baseline model. A baseline model is (usually) a null model, with no parameters. See. e.g. stats.idre.ucla.edu/stata/faq/… This is why I'm confused. $\endgroup$
    – Jeremy Miles
    Commented Jan 7, 2021 at 23:02

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

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