Hi, I think, if you have a single-item measure, you cannot calculate composite reliability as well as average variance extracted because both measures need more than a one variable to be calculated. In simplified terms, both measures need at least one correlation between two indicators. 

I don´t know ASV and MSV. 

Hope that helps.

Rafael 

Thanks Rafael. BTW .. MSV and ASV are maximum shared variance and average shared variance respectively .. both are used to check discriminant validity. I have received feedback from James Gaskin on how to deal with this issue, ie to include the single item only in the causal model. Cheers

I would recommend you to do your CFA with SmartPLS (free software, comparable to AMOS, but more convenient): http://www.smartpls.de/

This software gives you direct information about AVE, Composite Reliabilities, Cronbach Alpha's etc. I would then also suggest you to have a look inside the book "A primer on partial least squares structural equation modeling (PLS-SEM)" by Hair et al. It is not only a theoretical guide to issues as yours but also an awesome "tutorial" on how to use the software. 

In case you still want to stick with AMOS, I recommend you the youtube tutorials by James Gaskin: https://www.youtube.com/user/Gaskination/videos

Hope that helps :)

I disagree with Charlott Menke. The use of SmartPls and AMOS is not discretionary, rather it depends  upon the nature of research (qualitative vs quantitative ).

Yeap .. the choice of stat software should b based on nature of research. Having said that both amos and pls have their pros and cons. Thank you.

and SmartPLS is not for free any more... :(

I think you should see this: http://saeedsharif.com/2013/08/22/construct-validity/

Based on Fornell and Larcker’s (1981) validity determination criteria,

1) For convergent validity, CR for the construct should exceed 0.70 and AVE > 0.50

2) For discriminant validity, AVE > MSV, AVE > ASV, the AVE of a latent variable should be higher than the squared correlations between the latent variable and all other variables.

Where:

CR = Composite Reliability

AVE = Average Variance Extracted

MSV = Maximum Shared Variance

ASV = Average Squared Shared Variance

Source:

Claes Fornell and David F. Larcker (1981). Evaluating Structural Equation Models with Unobservable Variables and Measurement Error, Journal of Marketing Research, Vol. 18, No. 1, pp. 39-50