Sean Mackinnon

@seanpmackinnon.bsky.social

478

Followers

202

Following

1047

Posts

Instructor at Dalhousie University. Personality, statistics, mixed methods

Mastodon Account: @[email protected]

Posts Replies Media

A cool example of a more than 100 year old data visualization from the Romanian census @datavisfriendly.bsky.social you might like this!

Check out this incredibly cool analog mechanical calculator! You flick switches, and turn the crank and the answer pops up at the bottom. Negative numbers cause an error with everything flipping to 9999s

I'm traveling in Romania for a conference and for once, my Canadian dollar is actually worth something. The two of us just ate at a decent restaurant for 30 CAN with a beer. My god, is this what Americans feel like all the time when travelling? I feel so powerful, so flush with cash.

Reposted by

Sean Mackinnon

🚨 New Substack Post - A secret third way: Likelihoodist statistics I provide a quick tutorial on conducting research using the Likelihoodist approach. It's step-by-step + #R code open.substack.com/pub/mzlotean... #substack #statistics #education #guide #science #research #edusky

A secret third way: Likelihoodist statisticsopen.substack.com A tutorial on designing and reporting experiments in the most objective framework

Reposted by

Sean Mackinnon

Preview of an image from the slides for my 2-day workshop on longitudinal data analysis and visualization in beautiful Calgary, Alberta. Only 4 spots left and registration closes soon! haskayne.ucalgary.ca/CCRAM/academ...

Reposted by

Sean Mackinnon

Reply to

Markus Pössel

For all the debates about the demarcation problem, extensive data cleaning is what makes a discipline a science.

Reposted by

Sean Mackinnon

Time for some facts

Every time I read up the basics of #statistics and re-learn something I have this brief moment of panic where I have to go back and double check that I haven't messed up the analysis in a previous paper. I usually haven't, but this reveals why i don't want to be a coauthor on too many papers

Reposted by

Sean Mackinnon

"This paper uses simulation methods to assess the accuracy..., power, and robustness of 10 tests designed to check the significance of the difference between two dependent correlations with overlapping variables (i.e., the correlation between X1 and Y and the correlation between X2 and Y)." #Stats

British Journal of Mathematical and Statistical Psychology | Wiley Online Librarydoi.org The analysis of multiple bivariate correlations is often carried out by conducting simple tests to check whether each of them is significantly different from zero. In addition, pairwise differences a...

Another confusing missing data #stats term to add to the pile. Fraction of Missing Information (FMI). Is it the percentage of missing data? No! It's how much information about an estimand is lost due to missing data after imputation. Statisticians shouldn't be allowed to name things.

Reposted by

Sean Mackinnon

Statisticians! Please use hashtags like #stats, so your work or question or idea appears in the StatsSky feed and can be found by people interested in that topic! And please promote other people's work there too, especially if they aren't typically represented in science feeds.

StatsSky

Feed by @steveharoz.com

Does anyone have a good resource to get a good understanding of survey weights? Like journal article length, not book length. These are super uncommon in psychology, but I'm on a dissertation committee where a student is using Statistics Canada data that has weights #StatsSky

I've been so distracted by other work trying to finish writing a paper over the past month I accidentally wrote the first 3 paragraphs of the discussion twice. These paragraphs are ... totally different, and I have no idea why I scrapped the first draft. Who needs LLMs when I got an absent mind?

I know Ghosts of Mars was a box offoce flop and basically killed John Carpenters career, but it's honestly not that bad if you like campy action movies (which I do). I had a good time, tbh

Reposted by

Sean Mackinnon

SCIENCE CAT (my latest cartoon for New Scientist)

I was about to cite Kaiser (1974) on the KMO index for factor analysis, but decided I had better check the original source. Apparently, the wrong article has been cited for years. It's one of the most cited psychometrika articles, but it's the wrong one! #stats cda.psych.uiuc.edu/psychometrik...

cda.psych.uiuc.edu

Reposted by

Sean Mackinnon

Oh god. Oh fuck.

As an elder millennial, I need to constantly suppress the urge to capitalize Internet

Check out our Quantitative Methods section newsletter for the Canadian Psychological Association! It has our voted best Canadian 2023 stats paper, lifetime achievement award, 1-page tutorials, talk schedule for the convention and some #stats memes! canadianquantpsych.wordpress.com/wp-content/u...

canadianquantpsych.wordpress.com

Reposted by

Sean Mackinnon

“I Will Fucking Piledrive You If You Mention AI Again — Ludicity” ludic.mataroa.blog/blog/i-will-... > This isn't a recipe for disaster, it's a cookbook for someone looking to prepare a twelve course fucking catastrophe. Today’s must-read

I Will Fucking Piledrive You If You Mention AI Again — Ludicityludic.mataroa.blog

All the forced AI integration in new Windows updates seemed crazy to me until I realized that the only data source left to train their models on after scraping the Internet (which is now contaminated with AI material) is private data.

Reposted by

Sean Mackinnon

Reposted by

Sean Mackinnon

Deliberately undermining one of the most powerful tools against the threat of global pandemics from novel emerging infectious diseases is not only evil — it’s stunningly shortsighted zero-sum thinking that puts the US population at risk as well. www.reuters.com/investigates...

Pentagon ran secret anti-vax campaign to incite fear of China vaccineswww.reuters.com The covert effort began under Trump and continued into Biden’s presidency, Reuters found. Health experts say it endangered lives for possible geopolitical gain.

One of my least favorite brands of reviewer comments are "You should have controlled for more confounders. No, I won't tell you what ones but you totally forgot some." I don't mind a reasoned argument to account for confounders, but like, give me a rationale or some actual variables, at least!

Here's a #stats opinion question, brought to you from a journal reviewer comment. Say you have a multiple regression with 10 predictors, and thus 10 p-values, one for each slope. Should one control for familywise error? If so, are these 10 values one "family" for adjusting the p-values?

Reposted by

Sean Mackinnon

#statstab #112 Mixed Models with R Thoughts: A very nice overview of what mixed models are, how to use them, and how to interpret the results (even mentions issues with p-values). #rstats #r #lmer #mixedeffects m-clark.github.io/mixed-models...

Mixed Models | Mixed Models with Rm-clark.github.io This is an introduction to using mixed models in R. It covers the most common techniques employed, with demonstration primarily via the lme4 package. Discussion includes extensions into generalized mi...

Reposted by

Sean Mackinnon

PSA: All #rstats package on #cran will get an official DOI! This will facilitate bibliometrics and giving credit to R package authors. Registering all 20,000+ packages will still take a few more days. But the first couple of thousand are already live. Example:

Answering emails is a sucker's game. The more emails I answer, the more I get! If I answer a question, my reward is more questions. If I don't answer, I get emails asking me why I didn't answer their question yet.