Short Answer
The p-value is the probability of obtaining, by chance, a result as extreme as or more extreme than the one you observed, assuming the null hypothesis (that there is truly no difference) is true; p<0.05 is a common but arbitrary threshold at which a result is considered "significant," and it does not reflect any biological reality. You use the p-value in hypothesis testing to assess whether an observed difference can be explained by chance. The most common mistake is to treat the p-value as an effect size or as "the probability that the effect is real": in a very large sample, even a clinically trivial difference can yield p<0.001, while p=0.06 does not mean "no difference" but "could not be demonstrated with sufficient power." For correct interpretation, you should always report the p-value together with the confidence interval and effect size, and avoid chasing the threshold through p-hacking.
Serteser Consulting is run by a biomedical engineer (BME MSc) with peer-reviewed publications and PROSPERO-registered systematic reviews; it designs and conducts thesis, manuscript, and clinical research statistics, including p-value, power analysis, and effect size reporting, using SPSS, R, and Python, in a manuscript-ready form that can be defended before a jury or reviewer.
"P came out at 0.03, significant!" or "P=0.06, unfortunately not significant." These are two of the most frequently heard sentences in medical research. However, the p-value is a far more nuanced statistical tool than most researchers assume. Misinterpreting it leads to both scientific error and reviewer rejection.
What Does the P-Value Mean?
The p-value is the probability that a result as extreme as or more extreme than the one you observed arises by chance, under the assumption that the null hypothesis is true.
Put more simply: "If there were truly no difference, what is the probability that I would obtain this result by chance?"
P=0.03 means: "If there were truly no difference, the probability of obtaining a difference this large at random is 3%."
P=0.06 means: "If there were truly no difference, the probability of obtaining a difference this large at random is 6%."
Why Is P<0.05 the Threshold Value?
The 0.05 threshold does not reflect any biological or mathematical reality. It is an arbitrary threshold proposed by the statistician Ronald Fisher in the 1920s that became standard over time.
Many high-impact journals and statistical bodies now criticize the use of this threshold as the sole criterion. In Nature's 2019 article, more than 800 scientists proposed abolishing the concept of "statistical significance."
Despite this, p<0.05 is still used as a widespread standard. In practice, the great majority of journals expect this threshold.
The Most Common Misconceptions About the P-Value
Misconception 1: "If p<0.05, the effect is large"
No. The p-value does not measure effect size. In a very large sample, even a clinically trivial difference can yield p<0.001. For effect size, Cohen's d, OR, RR, or the absolute risk difference should be reported.
Misconception 2: "If p=0.06, there is no difference at all"
No. P=0.06 means "it did not pass the 0.05 threshold." A truly significant difference may exist but have gone undetected due to insufficient sample size (Type II error). In that case, it is more accurate to report it as "not found to be statistically significant, but a difference that could be clinically significant was observed."
Misconception 3: "The p-value shows the probability of a real effect"
No. The p-value is calculated under the assumption that the null hypothesis is true. An interpretation such as "there is a 97% probability that the effect exists" is definitely wrong.
Misconception 4: "Statistical significance is clinical significance"
This is the most dangerous misconception. In a large RCT, it can be shown with p<0.001 that a treatment lowers systolic blood pressure by 2 mmHg. But 2 mmHg means nothing in clinical practice.
How to Report Correctly?
Modern statistical reporting standards present the p-value not on its own, but together with the confidence interval and effect size.
Weak reporting: "A significant difference was found between the groups (p=0.03)."
Strong reporting: "The VAS pain score was found to be 2.3 points lower in the treatment group (95% CI: 0.8-3.8, p=0.03, Cohen's d=0.54)."
What Should You Do If the P-Value Comes Out Borderline?
What to do with borderline values such as P=0.06 or P=0.07 is a frequently asked question.
What must not be done: Continuing to add more data until p<0.05 is obtained. This is called p-hacking and is a serious ethical violation.
What should be done: Report the result as it is. Present the confidence interval, state the effect size, and perform a power analysis. The statement "the study was not sufficiently powered to detect this difference" is a legitimate and accepted conclusion.
To get support with p-value interpretation and reporting, request a free consultation.
Where Do People Most Often Get Stuck in This Analysis?
- P came out at 0.06 and you wrote "not significant," but the reviewer says "discuss the clinical significance."
- You are making multiple comparisons, but after the Bonferroni correction nothing comes out significant.
- P came out at < 0.001, but the effect size is very small, and you cannot decide whether this finding is really important.