#statstab #427 Incorporating Historical Control Data Into an RCT
Thoughts: In frequentist stats, historical controls are a dangerous proposition. But in bayesian, they may have a more useful purpose.
Historical data (HD) are being used increasingly in Bayesian analyses when it is difficult to randomize enough patients to study effectiveness of a treatment. Such analyses summarize observational studies’ posterior effectiveness distribution (for two-arm HD) or standard-of-care outcome distribution (for one-arm HD) then turn that into a prior distribution for an RCT. The prior distribution is then flattened somewhat to discount the HD. Since Bayesian modeling makes it easy to fit multiple models at once, incorporation of the raw HD into the RCT analysis and discounting HD by explicitly modeling bias is perhaps a more direct approach than lowering the effective sample size of HD. Trust the HD sample size but not what the HD is estimating, and realize several benefits from using raw HD in the RCT analysis instead of relying on HD summaries that may hide uncertainties.
I went to an AI training this morning, and they were showing us how to generate an image on Gemini. One of the prompts was a Fijian man, and I found it interesting that it put up a picture of an Indigenous Fijian man, and didn't choose a Fijian Indian.
Made me think about the time I was in Turkey, on a tour with a bunch of Australians. One of them asked me where I was from, and I said I was a Kiwi. He said that I don't look like a New Zealander. Any of you of you who know me can guess my response - no, I didn't swear. Just said that this was what we look like now, so he better get used to it.
I remembered this because I know that when the prompt for a New Zealander is put in to generate an image, it won't bring up a picture of someone like me.
That realisation hit me, unexpectedly, with some feelings.
Ryan Crosswell: "The most galling part about all of this was that Attorney General Pam Bondi also said, 'No one is above the law.'
Except we know that’s not true. Eric Adams is above the law. Tom Homan is above the law. Apparently anybody in the Epstein files is above the law."
#injustice #bias #incompetence #politics #ethics
Commonly used #chatbots show #gender #bias when users ask for #salary negotiation advice
Surface Fairness, Deep Bias: A Comparative Study of Bias in Language Models
In the era of memory-based #AI assistants, the risk of persona-based #LLM bias becomes fundamental. Therefore, we highlight the need for proper debiasing method development and suggest pay gap as one of reliable measures of bias in LLMs.
https://arxiv.org/abs/2506.10491
KB NL research
Brains Matter
Dr Mircea Zloteanu ☀️ 🌊🌴
Bypass Blues 🌻🇺🇦
anjum
abmoore
RonSupportsYou
Benjamin Carr, Ph.D. 👨🏻💻🧬