When embarking on a research project, it is important to have a little foresight into the possibility of encountering opportunities for compromise in relation to scientific integrity.  There are many such opportunities and the temptation to treat them lightly is often fueled by the desire to ‘get published’ and to please a more senior colleague.  Having trained as a pure mathematician, I am passionate about rigour in the choice of statistical methodology and the interpretation of findings. It is critical to appreciate that statistical that the statistical models supporting statistical hypothesis tests are only appropriate for use where your data meet the assumptions set by that model.  There are many such examples provided in later pages of StatsforMedics, including at the simplest level, choice of appropriate summary measures (or, descriptive statistics) and correlation coefficients. If you use a statistical procedure where the underlying  assumptions have not been tested, you run the risk of generating erroneous results and drawing  misleading conclusions.
This is one source of compromise in scientific integrity. There are many others.* To support you in developing good standards of statistical practise for your research, I recommend that you refer to the article Ten simple rules for effective statistical practice.
Also, take a look at the site Retraction Watch to gain an insight into the fact that retraction of journal articles is not unusual in some fields!
Scientific integrity by Margaret MacDougall is licensed under a Creative Commons Attribution-NonCommercial-NoDerivatives 4.0 International License.
