Anees Pari posted on Thursday, April 07, 2016 - 7:09 am
Hello there. I am exploring whether generic preference based health related quality of measures are appropriate to use in conjunction with disease specific scales in severe mental illness. I have three individual scales that measure depression, mania and generic quality of life respectively in a cohort of patients.
I would like to know whether these instruments measure a common construct/s. All the three scales are unidirectional.To begin with, I ran EFA and found out there a three factor model is appropriate based on parallel plot and loading.
However, I notice that some of the items had negative loading on factors. This was the same case for ESEM and CFA. My questions are:
a) How to interpret negative loadings on factors b)My sample has got repeated measurements on same individuals (there is no particular periodicity or pattern to their responses; some have sent their responses weekly, some monthly). I would like to use clustered standard errors to adjust for this clustering. In total , i have around 1400 observations on the three scales completed by 100 patients. Is there any way to do this in MPLUS?
I am new to MPLUS and factor analyses, and would be grateful for your advice.
MODEL: difficulty BY item1 item2 item3; laziness BY item4 item5 item6 ;
OUTPUT: SAMPSTAT MODINDICES(10.00) STAND RESIDUAL TECH2; PLOT: TYPE = PLOT1 PLOT2 PLOT3;
*** ERROR in VARIABLE command TYPE=COMPLEX requires a cluster variable, a stratification variable or replicate weights. Use the CLUSTER, STRATIFICATION or REPWEIGHTS options to specify one of the requirements for TYPE=COMPLEX.