Unfortunately, this sensible and important step is often skipped in haste to produce results. Some researchers will not even look at the means and standard deviations (another place that would provide a clue that something has gone awry), but they will proceed straight to the t-test, look at the p-value and conclude that one group has significantly higher blood pressure than the other just because they see that p < 0.05 (without noticing that they included nonsense data). This also happens when proceeding to regression modelling without looking at descriptive statistics beforehand or regression diagnostics afterwards. Just checking the p-value has led many researchers astray.
展开▼