Try again. Although non-compliance may increase the risks to research subjects and affect the scientific validity of your research findings, you should not exclude someone based on assumptions. Rather, you should try to understand why they have been non-compliant previously – were their treatment not working (in which case they might benefit from an experimental treatment)? Or were they simply forgetful (in which case you can give them more reminders)?

Try again. Hard-to-reach groups could be unfairly excluded from the benefits of research because of inconvenience. When less research is done on them, the health services they receive will also be less evidence-based.  

Correct. You do not have to include someone in a study that will do them more harm than good.

Try again. It is not acceptable to manipulate your choice of subjects in order to present your research findings in the best possible light. This is a form of selection bias and could reduce the validity of your research findings.