Self-report measures of functional status play an important role in group-level research, as well as individual- and group-level clinical assessment. A myriad of functional status tools exist today. Such measures have been developed with rigorous attention to reliability and validity. However, little investigation has been devoted to item bias or differential item functioning (DIF). A person's response to a functional status item should be a result of his/her ability level (the amount of the latent trait s(he) possesses) and nothing else. An item functions differentially if two individuals with equal ability do not have the same probability of item endorsement. Self-report measures of functional status can fall prey to such systematic bias because human beings interpret such items within the context of culturally- and socially-determined mindsets. The purpose of this application is to employ a multi-method approach to identify DIF in widely-used measures of activities of daily living (ADLs) and instrumental activities of daily living (IADLs). We hypothesize that respondent characteristics will interact the content of ADL and IADL items, thereby producing DIF. We will employ techniques of item response theory and other statistical procedures to identify DIF in eight large datasets. Once DIF has been identified, we will employ qualitative methods to discover the potential sources of the DIF. Once the range of causes of DIF has been identified, we will develop practical recommendation for writing new functional status questions to be free of DIF and correcting DIF in existing functional status items. This application will advance our knowledge base about characteristics of items and populations that cause functional status items to exhibit DIF. This contribution will help to bring about necessary change in how scientists develop self-report measurement tools, whether they are for physical or mental assessment. The health outcomes field is transitioning from classical test theory to increasing use of item response theory. Along with that transition will hopefully be more a priori and rigorous attention paid to DIF in instrument development and pretesting phases, rather than after measures have been in long use. This much-needed attention to DIF will result in assessment and outcome tools that are relevant and fair to members of a multicultural society. Thus, this application has broad-reaching implications for advances in item writing, item analysis, test construction, and test evaluation that go beyond the specific ADLs and IADLs we will study.
| McHorney, Colleen A; Martin-Harris, Bonnie; Robbins, JoAnne et al. (2006) Clinical validity of the SWAL-QOL and SWAL-CARE outcome tools with respect to bolus flow measures. Dysphagia 21:141-8 |
| McHorney, Colleen A; Monahan, Patrick O (2004) Postscript: Applications of Rasch analysis in health care. Med Care 42:I73-8 |
