Hi Abdulah,

How important is the sub-construct? Is it integral to the instrument? If so, you may not be able to drop it without rendering the entire instrument useless. If you can live without the sub-construct, have you tried running your factor analyses without it and reviewing the results? Finally, perhaps you may get somewhat better behaving properties if you utilize item response theory as opposed to factor analysis (probably not, but perhaps worth investigating).

Ariel

Good day Ariel,

thanks a lot for your valuable response.

this is a reflective construct and it is possible to drop it without causing a serious damage to the theory. in other words, i could trade-off between model validity and retaining such fat sub-construct. however, i found that three items may load with another independent construct. instead of deletion, I thought of merging these items with the other independent construct. whereas, the easiest and safest is to drop it.

this scenario (deletion) has met all assumptions and fulfill the validity, liability and the overall model fit. further, the final results did not change in terms of significance and sing for all paths.

regarding (item response theory) i would google about it.

once again I appreciate your help.

Regards.

Hi Abdullah, it sounds like you may have two viable choices here. Do you have some colleagues with expertise in the theory? You may want to show them both versions (with and without the problematic sub-construct) and see if there is a consensus on a particular option.

Hi Ariel,

Sound good idea my friend.

Thanks