In a clinical trial, the p-value for the primary endpoint is 0.03. The investigators conclude this provides 'strong evidence against the null hypothesis'. However, the sample size was 12 patients per group. Which concept BEST challenges the validity of this conclusion?
- A The false positive report probability (FPRP) is high because the prior probability of a true effect is low in a small exploratory study with low statistical power ✓
- B Type I error (alpha inflation due to multiple comparisons)
- C Confidence intervals are preferred over p-values; no other concern applies
- D The p-value threshold should be 0.01 for clinical trials, making p=0.03 non-significant
Explanation
Ioannidis's framework shows that in small studies (low statistical power), even statistically significant findings have a high false positive report probability (FPRP). With low prior probability of a true effect AND low power, the positive predictive value of a significant p-value is low — many 'significant' results are false positives. This is distinct from Type I error (alpha) which is fixed at 0.05. Small sample sizes create not only underpowered studies (inflated Type II error) but also unreliable significant findings when they occur by chance. The scientific recommendation is to report effect sizes, confidence intervals, and pre-registration, not just p-values.
Reference: Park's Textbook of Preventive and Social Medicine, 27th ed.
High-yield for: NEET PGINI-CETNExTFMGEUSMLEPLABMRCP
Written and medically reviewed by the StethoPrep medical team.