If you look into the articles which used SMART-PLS the items with loading between 0.4 to 0.7 may be considered for deletion only if it improves the AVE and CR. You can refer the article by Ringle,  Hair etc.,. Strictly speaking loadings less than 0.30 should be deleted. 

I've read that you delete items below 0.70, not 0.50. Yo only keep items close to 0.70 (i.e., 0.67/0.68...) if you think the item is important and AVE doesn't decrease.

So, in Smart PLS you examine outer loadings and keep >0.70

I hope this will help.

Regards

If you look into the articles which used SMART-PLS the items with loading between 0.4 to 0.7 may be considered for deletion only if it improves the AVE and CR. You can refer the article by Ringle,  Hair etc.,. Strictly speaking loadings less than 0.30 should be deleted. 

Alejandro and RajKumar Thank you for your answers. But recently i came to run one analysis in Smart PLS-3.2 and found that all the outer -loading values below 0.7 becomes red in color.  Which shows that all values below 0.7 are not anticipated but when you modify items to suite in your context research area than 0.4 or even 0.5 is acceptable when supported by some relevant studies. 

Well, generally you delete items below 0.70 because it doesn't affect AVE.

However, some researchers keeps items above 0.40 or 0.50. Is this what you mean?

I think this is because of the nature of the study. Exploratory studies are less conservative and consider that item loadings above 0.40 are enought. However, as I said, general approach is to keep only > 0.70.

They are shown in red in SmartPLS because SmarPLS only recommend to keep items > 0.70.

I hope this will help.

Just a remark, the evaluation of the measurement model is actually only useful if you use PLSc. Otherwise your loading estimates are biased in case of an underlying common factor model. 

Best regards,

Florian

You have to take out, because more thn 50% of the variance is explain by error term..

.......and this is why the hypothesized model → Re-specified model → Generated Model.

We have just published an article about scale purification, i.e. the process of eliminating items from multi-item scales. We have used the example of SCM, but our framework can be applied to any other discipline. Download: https://doi.org/10.1108/SCM-07-2016-0230 (or request via my ResearchGate page).

Dear Andread, hello. Could I please have your articles on scale purification please? thanking you. Regards, Tasmiha