A Reflection on the Use of P-Values

A Reflection on the Use of P-values
Student name
Institution
Professor
Course
Date
A Reflection on the Use of P-values
This reflective essay will discuss two papers from the special issue of The American Statistician on Statistical Inference in the 21st Century: A World Beyond p<0.05; the first paper, by Wasserstein and Lazar (2016), provides the context for the rest of the documents in the issue, outlining the American Statistical Association’s (ASA) statement on p-values and their appropriate use. The second paper to base on is by Goodman (2019), titled “why is getting rid of p-values so hard?” The third paper is by Greenland (2019), titled “valid p-values behave exactly as they should: some misleading criticisms of p-values and their resolution with s-values.”
Wasserstein and Lazar (2016) discuss p-value misuse and misinterpretation in statistical inference. They claim that p-values are routinely misused to prove a proposition. Instead, researchers should use p-values to make decisions in the context of the study design and the evidence (Wasserstein & Lazar, 2016). The authors emphasize better-presenting effect sizes and confidence intervals to understand the observed effect's magnitude and uncertainty. Wasserstein and Lazar's (2016) work emphasizes the limitations of p-values as a statistical tool and the need to use them properly. P-values have been used to determine statistical significance (Wasserstein & Lazar, 2016). This determination of the statistical significance has led to an emphasis on getting a p-value below the arbitrary threshold of 0.05 rather than effect size magnitude and direction. This action can misinterpret study results and hide the studied effect. Wasserstein and Lazar propose using p-values as a continuous measure of evidence against the null hypothesis rather than a binary signal of significance (Wasserstein & Lazar, 2016). Thus, p-values should be assessed with effect size, study methodology, and the body of evidence. Effect sizes indicate effect magnitude, whereas confidence intervals indicate estimate precision. These indicators complete the evidence supporting a hypothesis.
Goodman (2019) discusses the difficulty of modifying statistical inference culture, building on Wasserstein and Lazar's work. He argues that the statistical community must work hard to change scientific practice away from p-values. According to Goodman (2019), the null hypothe...