I analyze a retrospective data-set where same people did two laboratory tests with different proportion of pathological result (45% vs 17%) - these are paired data. McNemar test can show that the proportion of patients with positive test B is significantly lower than with positive test A. However, to analyze whether patients with positive test B group among patients with positive test A Chi-squared test is needed and would be used if analyzing relationship with gender or some other feature without a problem. However when defining something as a test of same feature it somehow becomes illegal to use it? It seems to me that these two statistical tests investigate different hypotheses (McNemar assesses size of proportions, Chi squared assesses whether do patients with one feature group with patients with other feature of disease). Therefore I question whether is it absolutely illegal to use Chi-squred test in this situation to obtain complementary information?
- Badges
- Science method
- Similar topics
- Methodology
- Research Methodology
- Qualitative Research