I am trying to examine the impact of cognitive and psychological factors, socio-demographic factors, Experimental processing factors and socio-cultural and institutional factors on farmers risk perception of floods. Based on the assumption of causality among these observed variables, i am running the following model. Cognitive and psychological variable (latent) has been measured by five indicator variables, these indicator variables are ordinal variables (1 to 4 point scale). Socio-demographic observed variables such as education, income and gender are continuous, continuous and dichotomous variables respectively. Experimental processing factor is a latent variable measured by four indicator variables (continuous and dummy). Socio-cultural and institutional factors is also a latent variable measured by five indicators variables (which are dummy and continuous variables). Risk perception of the flood, which is an endogenous latent variable measured by four ordinal indicator variables.
I run the model, but the model is not good fit (see the attachment). My concern is: can i measure the latent variable where observed variables are dummy and continuous? I am also trying to model by using reflective and formative latent variables in AMOS, but it did not run the model.
Any suggestion on the attachment.
Dear Nanda,
First of all, whether the EFA results were promising? Moreover, if poor cross loading and cronbach value leads to poor CFA analysis reuslting in bad fit. Also, if using reflective you can use SEM. But your constructs are formative which is also a concern.
Refer:
http://statwiki.kolobkreations.com/index.php?title=Main_Page
- Badges
- Science topic
- Similar topics
- Developmental Psychology
- Attachment