This morning I came across this TED talk by Peter Donnelly. It was was originally posted way back in 2006, so this isn’t something new. However, it is worth a listen for anyone thinking about data science and its applications. The moral of the story is that statistics are often counter-intuitive, and poor applications of statistical methods or invalid assumptions can lead to dramatically incorrect conclusions.
[iframe width=“560” height=“315” src=“http://embed.
Are P values the “gold standard” for statistically significant findings? From an article today that appeared today in Nature:
"The irony is that when UK statistician Ronald Fisher introduced the P value in the 1920s, he did not mean it to be a definitive test. He intended it simply as an informal way to judge whether evidence was significant in the old-fashioned sense: worthy of a second look." In the era of “big data” and automated algorithms that can test thousands of correlations, we are essentially guaranteed to find statistical significant P values in our analyses.