Page 1 of 1

Excluding Case ID for better AVE values

Posted: Sat Apr 06, 2019 12:05 pm
by saqib.phd@gmail.com
My model have 1 I.V, 1 D.V 2 Mediating and same are being tested for Moderation. I have collected the data of 1050 where number of items in my instrument is 86. After running the model in SmartPLS 3.0 the AVE values of the variables are not satisfactory i.e above 0.05. While reading the discussion in the fourms i came to know that some responses effect the over all results in producing AVE. I am now trying to check the case ID of Latent variables. The values ranges from 0.080 to 5.20.

Can any one guide which values (lower for higher) are effecting by results?

Re: Excluding Case ID for better AVE values

Posted: Sat Apr 13, 2019 3:27 pm
by jmbecker
It is not the observations that you might want to inspect (i.e., the case id values), but the items (i.e., the indicators in your measurement models). You need to see if there are some exceptional low loadings that negatively influence the AVE.