Check the following post:

https://www.researchgate.net/post/Did_you_use_single-item_measures_in_your_SEM_models_How_did_it_work_out

and this article:

http://www.palgrave-journals.com/jma/journal/v1/n2/full/jma20137a.html

Regards

Asish

i would recommend conceptualize all your variables as a second order contract  

For this, you have to fix loading and error term (for example 1 and 0.0). Without this, model will not be estimated.

Regard

Imtiaz

The single item will become as observed variable. You can run as usual with other latent variables

Hi Thien,

Basically you need to adjust for measurement error through attenuation.  The below link explains how this is done in AMOS.

Wanous & Reichers (1996) is the one to look at for dealing with single items and measurement error.

Best of luck with your research.

http://spssx-discussion.1045642.n5.nabble.com/Correcting-measurement-error-in-AMOS-td5726574.html

Appreciate for all the responses. I did as recommended By Wan Mohamad and it ran. Cheers 

Hi Chalamalla, that variable was actual literacy score. The rest were Likert-scale items for latent variables. This issue was solved. Cheers