As Not Related! fans will know well, I haven't made an episode in several months! This is for a variety of reasons, but I've decided to switch it up for an episode tomorrow by recording LIVE, streaming on YouTube.
The link will be HERE. The chatroom will be open until then. I might stop in every so often in the wait. I've scheduled it for 2PM New York time tomorrow, but I might change that by an hour or two if I need to. Check back at that link for updates.
Here's how it's going to work:
- Not Related! episodes I typical arrange as 30 minutes of depthy content, followed by a brief break and donation reading, followed by another 30 minutes of content. The beginning of the stream will work like this.
- The podcast is still audio and will still be released as such. I might occasionally use a minor visual aid, but nothing that will make consuming the content via audio diminished. I'll probably just have the topic and a webcam in the video.
- Chat will be live, but I will not be actively reading it until the break and it will not appear on the video (it will, as all YouTube streams, appear on the edge).
- Related/serious donation will be answered in that in-between break time. Non-serious/meme/joke or just unrelated donations will be answered at the very end (this will not be included in the audio version of the podcast, unless I deem it high-quality discussion).
In this arrangement, we can have all the typical glory of a Not Related! episode while still "devolving" into a general purpose stream as time goes on. Obviously it can be interactive at the same time.
The starting topic of the episode/stream:
This episode/stream will be on the absolute(ly terrible) state of "statistics" as it is used in the "soft" sciences: psychology, neuroscience, medicine, econometrics, here in linguistics and many others. All of these disciplines are in hot water. A lot of people are aware that there is flimsy knowledge of statistics in a lot of non-specialist fields, but in this episode, I go to an even deeper problem: the "Null Ritual" (Gerd Gigerenzer's term), a statistical analysis procedure that is essentially required for publication, but is also a historical misunderstanding and mistake that has produced entire fields of people running statistical tests of unclear scientific and epistemological meaning, while also abiding by totally arbitrary conventions (p < 0.05 and more).
I'll go into the original history of modern statistics, as expounded by Ronald Fisher and his "adversaries" Neyman and Pearson. Their competing theories were motivated in entirely different ways, but have been distorted and confused over the decades, resulting in the hybridized "null ritual" we see today.
I'll also talk about how this contributed to the "replication problem", reminding one of Ioannidis's now well-known 2005 paper: "Why Most Published Research Findings Are False."
Do not read Wikipedia or even a textbook on anything related to this before the podcast! You will be misinformed! Refer to Gigerenzer's 2004 article "Mindless Statistics" and the ensuing literature, but I should say that there is a very big gap between this specialist literature and the mostly cargo-cult understanding of the statistical methods that filter into textbooks and then into public repositories like Wikipedia, etc. This is why I think this topic is highly important to get out there!