1,000
3.00Here is my analysis: https://docs.google.com/document/d/1sR1T2Kb1X1fCLYeEE-U3Rweinn0tKunIP9bKGi6XXa8/edit?usp=sharing
I apologize for taking a long time to post this. At first I intended to include my analysis in this post, but then the post grew too large, so I published it without that section and then forgot about the whole potassium thing.
Ok, I have a plan. It just needs @Aella's help.
https://twitter.com/mold_time/status/1608626779656450049
Surely it would help the state of amateur research for SMTM to respond in good faith to their critics.
The analysis came out (https://slimemoldtimemold.com/2022/12/20/people-took-some-potassium-and-lost-some-weight/) and they concluded a positive result.
The average weight loss among respondents who submitted at least two weight datapoints was ~0.79 lb. Unfortunately, though, this group comprised only about half of participants who sent in any data, which means that selection effects can be responsible. It seems very plausible that people are more likely to report their body weight if it’s going down than if it’s flat or stable! This is not something that was addressed by the SMTM authors.
Moreover, there was no control group. This was not a problem with e.g. their potato diet, because the magnitude of the weight loss (~10 lb over 4 weeks) couldn’t plausibly have been caused by e.g. the Hawthorne effect. But that is not true about the modest weight loss in the potassium trial.
I will do a more quantitative analysis of their results later.
@NataliaMendonca There is also a very basic addition error in their analysis.
They claim that a linear regression model fitted to the potassium trial results (with potassium dose as the predictor variable) can predict the weight loss they found in their potato trial. But when you look at their numbers, that is very clearly wrong — they fitted the model on supplemental potassium doses from the potassium trial (around 2 g/day on average) and then used the model parameters to predict the potato diet’s average weight loss based on the (estimated) total daily potassium intake from it (11 g/day).
Average potassium consumption among adults in the US is about 2 - 3 g/day, so the total potassium dose in this most recent trial was 4 - 5 g/day, not 2 g/day as they consistently imply in the post.
@NataliaMendonca oops -- I meant to say "if it’s flat or going up," not "if it’s flat or stable."
@NataliaMendonca Oops — I meant to say in my first comment that fewer than half of participants who submitted their weight on day 1 submitted their weight on the last day as well.
@NataliaMendonca they've been making weight loss claims throughout, so if they retreat to "well it did stop people feeling hungry but they didn't lose weight" we can just laugh and resolve this ambiguous.
@JoyVoid Because this market is conditional on them concluding a positive result, the less likely you think the potassium hypothesis is to be true, the more likely it is to be easy to find a hole in their conclusions.
@AlexL Agreed. I do think it's pretty likely that potassium is a driving factor in weight loss, I know some people have had positive effect from supplementing it (though I'd be unsure about weight regain after study, and I think there might be a severe selection effect going on).
