Case of the Month
Jorge is a young faculty member collaborating with Saanvi, a more senior member of his department. In analyzing their data, Jorge and Saanvi discover that the effect they are looking for is not statistically significant, and therefore likely not publishable. Saanvi proposes that can probably find statistical significance by switching to a different test. Jorge is concerned that switching from the test they initially planned could be seen as falsification under the federal definition of research misconduct.
- Is switching to a different statistical test acceptable statistically and, if so, under what circumstances?
- Is switching to a different statistical test misleading? Research misconduct?
- What should Jorge do?