|
Message/Author |
|
Anne Buu posted on Tuesday, April 15, 2008 - 12:00 pm
|
|
|
Q1: If I use probit link with MLR, are the resulting factor loadings comparable to the factor loadings estimated by WLSMV? Q1: If my observed binary dependent variable x has coding 0/1, does Mplus model Pr(x=0) or Pr(x=1)? Q3: Assuming I have one latent variable z (continuous), and A is the slope (factor loading), B is the threshold corresponding to a binary variable x, tell me if the following equation is true: P(x=1|z)=F(-B+A*z) where F is cdf of normal/logistic distribution. |
|
|
Q1: Yes, they are both probit regression coefficients. Q2: Pr(x=1). Q3: Yes where F is the normal distribution function. |
|
Back to top |
|
|