Dear all,
Is there any way of comparing two proportion calculated from two samples that are not complete independent or dependent? The problem is that we have two samples were 60% of the individuals are paired, or the same in both samples, and we whant to compare results between the two samples. So we can't use a test considering that the samples are independent nor a paired sample (Mcnemar test) because we can only use 60% of the samples and we will introduce a bias.
My question: is there any test for partial independent samples? Or should I put my data in the garbage?
Thanks to all in advance.
Kind regards,
Baltazar Nunes
Observatório Nacional de Saúde
Instituto Nacional de Saúde Dr. Ricardo Jorge
Av. Padre Cruz 1649-016 Lisboa
Telf. +351217526453 Fax. +351217526499
Url: www.onsa.pt
|