Psychometric evaluation of the Interpersonal Needs Questionnaire (INQ) using item analysis according to the Rasch model
Research output: Contribution to journal › Journal article › Research › peer-review
Final published version, 1.24 MB, PDF document
The Interpersonal Needs Questionnaire (INQ) assesses Thwarted Belongingness (TB) and Perceived Burdensomeness (PB), two predictors of suicidal thoughts. Up to now, the use of item response theory (IRT) for the evaluation of the INQ has been restricted to a single study with clinically depressed and suicidal youth. Therefore, the psychometric properties of the two INQ-15-subscales TB and PB were now evaluated in a general population sample (N = 2508) and a clinical adult population sample (N = 185) using IRT, specifically the Rasch model (RM) and the graphical log-linear Rasch model (GLLRM). Of special interest was whether the INQ-subscales displayed differential item functioning (DIF) across the two different samples and how well the subscales were targeted to the two sample populations. For the clinical sample, fit to a GLLRM could be established for the PB-subscale and fit to a RM was established for a five-item version of the TB-subscale. In contrast, for the general population sample fit to a GLLRM could only be achieved for the PB-subscale. Overall, there was strong evidence of local dependence (LD) across items and of some age- and gender-related DIF. Both subscales exhibited massive DIF related to the sample, indicating that they don't work the same across the general population and clinical sample. As expected, targeting of both INQ-subscales was much better for the clinical population. Further investigations of the INQ-15 under the Rasch approach in a large clinical population are recommended to determine and optimize the scale performance.
|Number of pages||21|
|Publication status||Published - 3 Aug 2020|
- Faculty of Social Sciences - Suicide, Age groups, Depression, Psychometrics, Mood disorders, Questionnaires, Number theory, Social theory
Number of downloads are based on statistics from Google Scholar and www.ku.dk