If you were a scientist and you designed a series of experiments with a variety of experimental measures, would you consider it to be dishonest if you were only to report a few of these outcomes? Particularly the ones that show a significant change as opposed to the ones that didn't?
If you set a series of thresholds as to what would be considered a significant change, but then found that level of change was rarely met, would you change the thresholds in the report to make the results sound more significant?
If to secure funding, you were required to publish your experimental protocol beforehand, would you still continue to cherry pick the results that you report and the experimental thresholds? Despite the fact that your bait and switch will be evident in the literature.
If you did all the above, would you expect your research to be published in 'top' journals such as Science, Nature, PNAS, NEJM or The Lancet?
Would you be surprised to learn that the above does occur regularly?