This paper acknowledged the need to examine psychometric properties of the Presence-Severity (P-S) format in questionnaires, and suggested using the nominal response model (NRM) to analyze such data. An empirical example using the short form of the Memorial Symptom Assessment (MSAS-SF) was investigated. Results showed that category usage is not even in many items where the relatively high usage of the highest category and lowest categories were found. Also, items of the psychological subscale fitted well and performed ordinally, whereas those of the physical subscale were problematic. In addition, the authors found that when the Severity items were ignored, it might reduce the accuracy of the measurement. Overall, too many categories are not needed when some MSAS-SF questions are administered, and the P-S format in the MSAS-SF may not be efficiently designed. Here are my comments:
First, this paper highlighted the importance to examine the psychometric properties of a format used in questionnaires, which has been ignored in the literature. It punched users of surveys or questionnaires and reminded us pay attention not only to the content and wording of a questionnaire but also to the used format.
Second, Since the P-S format in the MSAS-SF demonstrates ordinarily, the GPCM or PCM models might be more appropriate for the MSAS-SF data, instead of the NRM. I believe that the usage of the NRM is more flexible than the GPCM or PCM when categories represent nominal information, but not ordinal. Thus, it would be better that the authors selected different dataset to support their suggestions.
Third, can we use the NRM to analyze data from questionnaires asking both frequency and severity (e.g., the bullying scale)?