P-values: Context, Applications and Limitations

# p-values: context, applications and limitations

A p-value is a probability that measures the likelihood of obtaining a result as extreme or more extreme than the observed result, assuming that the null hypothesis is true.
In clinical studies, the null hypothesis is typically a statement that there is no significant difference between two treatment effects and formulated before start of the study. In exploratory studies p-values enable to understand statistically meaningful findings. However, a statistically “significant effect” is not automatically statistically meaningful. The use of p-values has been subject to much debate, particularly in the context of exploratory analysis. Effect sizes and confidence intervals expand the approach to statistical thinking and improve understanding of data. Confidence intervals, i.e. the range in which the true value lies with a certain degree of probability, provide additional information about the strength and direction of the effect of interest. They allow more informative insights and conclusions about statistical plausibility and clinical relevance of study findings.
Furthermore, p-values are often misinterpreted or misused, leading to erroneous conclusions. Understanding p-values is essential for any researcher seeking to interpret and communicate statistical results effectively. This talk presents context, applications, and limitations of p-values as well as alternative approaches.

Presenter:
Aslihan Gerhold-Ay, PhD
Senior Principal Statistical Analyst
Healthcare | Global R & D | Clinical Measurement Sciences | Biostatistics Oncology
Merck Healthcare KGaA