This may lead to a false negative or a false positive on the conclusion. This is because the statistic being tested is not truly indicative of the two samples, which could end up tainting the final statistic. Having a statistic that is incorrect can lead to both Type I and Type II errors.