No statistical test produces valid results if the sampling was done incorrectly. Since all of the patients were recruited in the same manner (all through a convenience sample) the inferences are still valid. The results are still valid it the data is analyzed with an alternative analytic tool. Since the lead author was not responsible for monitoring data collection activities, the inferences remain valid.