Consider the following hypothetical study: a study of 20 patients needed knee replacement surgery was conducted. Ten patients were randomized to have the surgery performed using the standard surgical anchors used for knee replacements and ten patients were randomized to have the surgery performed using a new surgical anchor.
The primary outcome for the study was the 6 month change in the Oxford Knee Score (a scale that can be scored between 0 and 48 - with low scores indicating bad outcomes for the knee and high scores indicating healthy outcomes for the knee).
The researchers found that the Oxford Knee Score improved by 9 units in the patients with the new anchor and 3 units with the standard of care anchor, but the p-value for the test of the difference was not statistically significant. In addition, the 95% confidence interval comparing the two groups for the difference in Oxford Knee Scores was 6 +/- 8 (i.e., confidence interval from -2 to +14).
The minimally clinically important change (difference) in a Oxford Knee Score is 5 units based on previous research.
Investigator one says that this study shows that the new anchor is better than the old anchor because the observed difference between groups is 6 units and that is larger than the minimally clinically significant difference of 5 units.
Investigator two says that this study has not shown that the new anchor is better than the old anchor because the p-value for the comparison is not significant and the confidence interval includes 0.
Investigator three says that this is an under-powered study and therefore we cannot tell if Investigator 1 or Investigator 2 is correct,
Based on your understanding of statistical power which Investigator 1, 2, or 3 is correct - Explain your reasoning.
What additional information might you request to better determine whether Investigator 3's claim concerning the study being under-powered is correct?