Each questionnaire which we would like to use in our research, it should be standardized on the society that we want study it. for example if your questionnaire has made in the USA and you want to study on a sample in Germany you should standardized it in Germany.

As you mentioned your research's questionnaire has Norm that it is a sign of standardization.

Hi Elina, that is a very good, and also common, question. I work in psychometrics, and work with more experienced psychometricians, on this very point - regardless of type of measure. Cut-offs, or cut points, are almost never valid. That is, they are rarely based on solid and valid evidence and almost never take into account the different impacts of the several questions/items in the measure. What you have, with most cut-off labels, is a rough estimate of where people stand on the construct.

So, what can we do? For those like you, who want to divide participants/respondents into two extreme groups, you can divide the sample by high (this is subjective, but +1 or 2 standard deviations above the mean or above some validated level) and low (-1 or 2 SDs from the mean), then take out the people in the middle. That way you are comparing those high and low on the construct, which removes much of the ambiguity and error of those in the middle - but also takes away a large segment of your sample and some real-world validity in the spectrum of the construct.

The best choice is to keep your measure continuous, and not group people, only conduct analyses with the full data. However, as you know, for clinical and other practical reasons, that is not ideal or useful.

Good luck! It sounds like an interesting and useful study.

Keith

Thank you so much for taking time to provide me your insights! It was valuable! Thank you!