Is factor analysis here can extract the most important questions which reflects the anemia status???

Probably factor analysis although PCA may give you similar results.

If you can assume that there are latent variables (factors) that can can explain the answers to the questions of your anemia questionnaire (I would assume this) you should use factor analysis (not PCA) with oblique rotation (e.g. promax) of the factors. To decide on the number of factors I would advice parallel analysis. A good article explaining the decisions to be made is: Preacher & McCallum (2003). Repairing Tom Swift's Electric Factor Analysis Machine. Understanding Statistics, 2, 13-43 (https://quantpsy.org/pubs/preacher_maccallum_2003.pdf).

Dirk Enzmann & Christian Geiser thanks for your response

Actually I have a variable for anemia status but the questionnaire not primarily intended to mesure the anemia although some questions may reflect the anemia status so I need to run a statistical test to extract questions which may reflect the anemia. I don't knownif my question is clear now or not. But the anemia is observed here not a latent factor.

Thanks

Mariam Taher Amin : But you seem to have data (answers to more than two questions/items in you questionnaire) that you assume to be indicators of anemia? You mentioned that there are 15 items. If you believe that they are indicators of anemia that is not directly measured (but you want to use these items to measure anemia) then anemia is an unobserved variable (= latent factor) and you could use factor analysis to investigate whether the items are homogeneous enough to form a scale of anemia (it may be that there is more than one factor -- in this case the items can be grouped into sub-dimensions of anemia that may be correlated).

If the factor analysis seems to indicate that there is only one factor, additionally you should determine the reliability of your scale, e.g. by calculating Cronbach's Alpha or McDonald's Omega (and if there are more than one factor you should do this for these other factors, as well).

However, whether this factor actually is measuring anemia has to be decided based on face validity or on theoretical grounds. I can't imagine any statistical method to establish the validity of your items (as measuring anemia) if you don't have any other measure in your questionnaire to can be used to demonstrate convergent or discriminant validity of the latent factor anemia.

If there are only two items (or if there is only one) then factor analysis cannot be used.

Dirk Enzmann Thanks for your explanation. Yes I have a 15 Items from questionnaire that measure quality of life but it seems that some questions may reflect the anemic status (or it is my research question). I need to extract those questions to be used as screening tool in clinics to decide if this patient need further investigation for anemia or not??

I agree with you in factor analysis point but I am just thinking is there other ways to answer my question or not .... for example can I use regression here?? As anemic status is binary dependent variable and to find the associated questions from the questionnaire.... again it is just brainstorming ideas.

Thanks again for your helpful answer and discussion.

I am not an expert on anemia or anemic status. It seems to me that your question is foremost a theoretical (or conceptual) question that needs to be clarified before deciding on the appropriate statistical method. There should be a theoretical reason why quality of life is part of the questionnaire. As this is not my field of expertise I can't advise your as to the question how this is related to anemia or anemic status.