Report all results obtained, including those that do not support the hypothesis.

Modify anomalous data to make them more consistent.

Remove difficult-to-interpret results without reporting their removal.

Present only the results that support the initial hypothesis.