Null hypothesis significance testing is not dead, but in this webinar, Ken Rothman shares why it should be.
Although seemingly quantitative, statistical significance testing degrades quantitative information into a dichotomy, based on the P-value. This statistical analysis process fosters misinterpretation of results stemming from the mixing of two essential data descriptors - effect size and precision - into a single number or label.
The American Statistical Association has issued a warning against statistical significance testing. There is a growing drumbeat to ditch this flawed data analytics method and replace it with more insightful and reliable approaches to quantitative data analysis in epidemiological research.
Join us for this webinar where you’ll learn:
- To articulate why a P-value is a “confounded” measure, mixing effect size with precision.
- To illustrate why, in hypothesis testing, a P-value is not equal to the probability that the null hypothesis is true.
- To broaden the interpretation of a P-value into a P-value function, which displays both effect size and precision.
- To describe the connection between a confidence interval and a P-value function.
Kenneth J. Rothman, DrPH
Distinguished Fellow, Vice President of Epidemiology Research
Heather Danysh, PhD