transcript of episode 35: SEASON THREE PREVIEW, 1st December 2023

[🎶 INTRO: "Spring Swing" by Dee Yan-Kee 🎶]

welcome to the error bar: returning after the longest three months in history

my name's nick holmes and this is the error bar, a podcast by a middle age man who likes statistics, for middle aged men who like statistics

here is the brain news on the 1st December 2023:

[🎶 BIG BEN BONGS 🎶]

SEASON THREE RETURNS (probably)

sources


SEASON THREE RETURNS

yesterday i listened back to episode #34, in which i promised:

<>the error bar would return in "spring 2023" <>

that, dear listener, was an error. 2023 has been one of the longest years on record. it took me six months to get a new computer in my new office in my new job & i wasted most of rest of the year writing my career-largest research paper. as that eight-experiment, twenty-thousand word paper is nearly finished, i can now return to my side-hustle of polishing glasses that no-one drinks from in a bar that no-one visits, creating a podcast that, well, no-one listens to.

sometime between episode #34 & now, i came to a few decisions about the bar. season 3 will change in several ways.

first, there will be only one episode per month. i have more papers i need to write & this won't happen if i spend two or three days a month reading the daily mail's eclectic science output.

second, i will no longer cover science stories in the daily mail, other tabloid newspapers or any news media outside the UK. this should help with my first aim, to cut the time i spend on producing the podcast.

third, there will be no interviews or special discussion episodes. there are many interesting people to talk to & many discussions to have about science. i've learnt a great deal from my guests. but each of those discussion episodes took a full week's work to produce.

fourth, the error bar will no longer make cheap jokes at other scientists' expense. i have found it difficult to know which side of decency the podcast landed. sometimes i got that wrong. my justification for satirical or sneering criticism was that i focussed on scientists whose work was amplified - unjustly - by the news media. if scientists play the game of hyping their science to the daily mail, then they deserve the criticism. publish, promote, pillory. but that was too harsh. often the scientists have no choice or control over how the media reports their work.

fifth, i am going to try not to mention a study's sample-size or the statistical values that it generates. so much of the discussion around scientific reform, reproducibility & the replication crisis over the last few years has been narrowly focussed on a small set of metrics for judging the quality of science. some discussions of meta-science - that's the scientific study of scientists & their practices - particularly in some corners of twitter, have focussed almost exclusively on the statistical distributions of these metrics, rather than what's arguably more important, on what the scientists are studying, how their doing it & why. just for example, some corners of meta-science-twitter have been counting how many p-values a scientist produced between the values of 0.01 & 0.05. if the numbers were too high, the scientist is bad, if they were low, they're good. it's a farce. but other corners of meta-science twitter are thriving, focussing on what scientists do & how they do it. the error bar will try to move into those corners.

sixth, the new season of the error bar will accompany my departure from twitter & my arrival in the bright blue skies of an alternative social medium. more than half of my interactions on twitter are spam. bye bye elon, hello bluesky.

seventh - we're nearly there, i promise - each episode will contain one new study, one news story & one statistical essay (i.e., rant), by me. the episodes will be titled according to the statistical essay rather than the study or the news story. ideally, the essay will be related to the study & to its news story, but it might just be whatever i want to talk about.

the reason for this seventh change is that i have things to say that i only say to reviewers, authors or editors during opaque manuscript review processes, or in the pub after too many alcohols. there will be, for example, statistical rants - sorry, essays - on researchers complaining that a study sample is 'too small'; on researchers claiming that you can't interpret an interaction when there's a significant main effect (or is it the other way around - who cares?); on why researchers do not need to test their data for 'normality', & when they find it's not normal, why they don't need to change all of their statistical tests. in part, i can use these episodes in my new teaching materials on statistics. in another part, i just want to be david runciman on past, present, future.

i've been piloting these rants - sorry, essays - during years of relatively pointless interactions in peer review. so far, i've not been argued against, so i'm going to try them out in public. picture me as both the barman refusing to serve & the drunk at the bar slurring "& another thing..." as he's shuffled towards the door. if you want to pick an error bar fight with me over something i say, then that's what i'm here for. write to talk@theerrorbar.com & i will respond.

season three of the error bar will begin in january 2024. have a very happy winterval.

[🎶 OUTRO: "Cosmopolitan - Margarita - Bellini"by Dee Yan-Kee 🎶]

it's closing time at the error bar, but do drop in next time for more brain news, fact-checking & neuro-opinions. take care.

the error bar was devised & produced by Dr Nick Holmes from University of Birmingham's School of Sport, Exercise and Rehabilitation Sciences. the music by Dee Yan-Kee is available from the free music archive. find us at the error bar dot com, on twitter at bar error, or email talk at the error bar dot com