I'm not sure whether the McNemar test is an appropriate choice in the situation below. I'd like to see whether an intervention had an effect on screening adherence, which can be done only once your life. For instance, before the intervention the proportion was 15% (15/100). After the intervention, this proportion (in the same people) increased to 45% (45/100).
Is the McNemar test appropriate here (even if one of the four cells is forced to have zero counts)? Indeed, the proportion may only increase (or remain unchanged). For the same reason, if this test can be applied, should I use the one-tailed p-value?
