I've been selected in a sample for the national statistics office of my country. As researcher working with surveys know, we are annoyed by missing data, inconsistent answers, measurement errors, attrition, specially if they are not random.
My dilemma is the following: if I strategically provide missing data, inconsistent answers, slight miscalculations just as an average dude would do, I reduce selection problems on the above issues. Yet, I'm not contributing to a better dataset. However, if I do my best effort to answer as accurate as possible and never "attrite", then I'm furthering the selection problem, yet improving data quality.
What should I do?
PS: Yes, I know. I'm statistically insignificant :(
