Analyses of Unbalanced Groups-Versus-Individual Research Designs Using Three Alternative Approximate Degrees of Freedom Tests: Test Development and Type I Error Rates

Three approximate degrees of freedom quasi-F tests of treatment effectiveness were developed for use in research designs when one treatment is individually delivered and the other is delivered to individuals nested in groups of unequal size. Imbalance in the data was studied from the prospective of subject attrition. The results indicated the test that best controls the Type I error rate depends on the number of groups in the group-administered treatment but does not depend on the subject attrition rates included in the study