Hello

You may look for a correlation among demographics and not important items and the rates of important items. This may go as t test, chi square or multivariate analysis to see what factors influence not important items. I would not remove them from the analysis, it is better to seek for any meaningful relations at this stage.

If items were classified as not important, this is not equal to zero in the Likert scale. I will not go with the solution of {not losing cases in the sample} adding the not important to analysis. To be with more accurate solution, only analyze those who answered the Likert scale.

Lets remember that the construct validity by Factor Analysis in this case could be preceded by content validity. This will improve the structure of the scale to be run by statistical solution.

I agree with Muayyad. In some scales these option are deleted and answers treated as missing values, since not applicable/not important is not a option in a Likert scale

Thanks for responding. The items that are identified as not important are cooking and other household activities by men and return to work by women. The items are developed based on patient interviews which showed  there's a gender specific difference in activity preference. We wanted our scale to accommodate this difference, but we encountered this issue while analyzing the data.

Hello,

I too agree with Muayyd Ahmed. Go with content validity and if the activities explained by the participants as not important . We can ignore the content from the outcome measures .. I think so .

We cant ignore because activities identified as not important by men are identified as important by women