Research Narrative Presents:
The THINKerry
Is Significance Testing Becoming Less Significant?
Statistical Testing
Academic Research
p-Value Hacking
July 18, 2019
A recent article in Nature – and follow-up coverage from the likes of NPR and Vox – recently posited the underlying question: has academia taken significance testing too far? More than 800 academics from around the world recently came together to call for “a stop to the use of p-values in the conventional, dichotomous way — to decide whether a result refutes or supports a scientific hypothesis.” They call this trend “dichotomania” and argue that this binary interpretation of statistics has led to dismissal of crucial – and materially meaningful – effects.
In the world of academic research, p-values are considered the gold standard, and p<.05 has become the baseline requirement for publishing. But is that really good for the future of research and academic progress? Has that mentality bled into market research? We argue that the use of p-values in an exclusively binary “pass/fail” way is not only arbitrary and potentially dangerous, it’s also a misinterpretation of statistics and probability theory as a whole.
In this episode of The THINKerry podcast, we talk about the debate at hand, the true nature of probability, and the benefits and limitations of stat testing. And as an added bonus, we learn about Guinness’s relationship with the history of statistics!
This episode features:
More from The THINKerry…
Like this episode? Share it on social media: