Although measures are increasingly used in medical research, there is a sparse for measurement bias assessment in practice. Differential item functioning (DIF) analysis is the core of a measurement bias study. Measures that are very likely to have problems with DIF are used without caution, making the comparison of measure scores between different groups questionable. This phenomenon may be partly due to the difficulty of DIF analysis, given the fact that most of current DIF analysis procedures are tedious and formidable for practical medical researchers. The major aims of this project were to: (a) develop an easily implemented DIF analysis method for dichotomized variables; (b) to evaluate the performance of this new method under different scenarios; and (c) to apply this new method on a dataset and to compare the results with those got from applying a standard DIF analysis method on the same dataset. |