|
Message/Author |
|
Sanjoy posted on Sunday, July 10, 2005 - 6:35 pm
|
|
|
Dear Professor/s ...I have a three equation simultaneous system, like the following ANALYSIS: ESTIMATOR=WLSMV; PARAMETERIZATION=THETA; MODEL:! model 1 R on B X1; B on R X2; U on R B X3; (R, B's are five-point categorical, U is 0/1, Xi's are the vector and they share some common elements) ...as we can see the dependent variables have single indicators. Following your previous suggestion ...I tried this alteration MODEL: !model 2 f1 by B; f1@1 !setting factor variance equal to 1 f2 by R; f2@1 f3 by U; f3@1 f1 on f2 X1; f2 on f2 X2; f3 on f1 f2 X3; Q1. This model converges nicely like our first one, giving me equivalent but NOT same result ...is there any way to get the exact result that we got in our first model Q1. How could I get reliability estimate ...I mean where to look at in MPlus output ...I'm thinking to fix the error variance @(1-reliability), instead of fixing factor variance thanks and regards |
|
bmuthen posted on Monday, July 11, 2005 - 2:27 pm
|
|
|
Q1. I don't recall what your first model was, but the difference might lie in the metrics of the solutions. Please email me the output for the 2 models so I can tell what you mean. Q2. You find the reliability as the R**2 in the output when requesting Standardized in the Output. |
|
Sanjoy posted on Tuesday, July 12, 2005 - 6:06 am
|
|
|
Thank you Professor...I'm sending you a mail regards |
|
Back to top |
|
|