Dear all
I am using CFA for validating the attitudes toward statistics questionnaire SATS36, when I use the items of subscales it yielded a poor fit, while when use total score for subscales as an observed values it yielded good fit. I am wondering if that is right or not?
any help would be appreciated
Rita,
It is possible, given that the assumptions of the CFA are met with the new variables you create. Also, I think CFA is a good idea for this data.
Hope this helps. Email if you have other questions I can answer!
Marcie
I Think the subscales scores should be latent variables not observed variables.
Erturgul Marcie and Mohmoud thank you all for responding!
I try to combine items in to parcels according to mean, skewness, SD, and kurtosis even though many researchers claimed that parcels seems tricky way to prove the nature of the data. Using parcels model have moderate fit for CFI, GFI, and AGFI but chi square value was significant and RMSEA equals 0.14
Normality was checked, there were outliers but when I omit them the model fit get worse so I think that those respondents were offer some important information.
the model of the attitudes toward statistics as a latent variable and the total scores of sub scales yielded a high fit.
- Badges
- Science topic
- Similar topics
- Mathematics
- Statistics
- Applied Statistics
- Survey Design
- Questionnaire