I'm conducting an N=1 self-experiment, starting today, to try to gain evidence for or against the idea that taking a multivitamin causes me to gain weight. The experiment will run for about 6 months and will alternate between 30 day phases where I take a multivitamin and where I'm not taking the vitamin. For each phase, I will record my weight on the morning starting the phase, and then record the delta from that first point for each subsequent day.
To determine how to resolve this market, I will use R to fit linear models to both the three vitamin taking phases and the three non vitamin taking phases. I will use R to compute the 95% confidence interval for the coefficient on the day variable. If the lower bound of the vitamin confidence interval is greater than the upper bound of the non-vitamin confidence interval and greater than zero, then I will resolve this market as YES. Otherwise, it will resolve as NO.
The multivitamin I'm taking contains the following:
Vitamin A: 180mcg
Vitamin D: 50mcg
Vitamin E: 6.7mg
Folate: 200mcg
Vitamin B12: 8mcg
Magnesium: 30mg
Zinc: 2.4mg
Omega-3 DHA: 333mg
Boron: 0.7mg
Vitamin K2: 90mcg
The experiment should run for about six months. I have some travel planned during that time. I'm not planning on packing a scale when I go, so I will not record weights during travel and for two days afterwards. If I travel during a vitamin phase, I will keep taking the vitamin during the trip. If a phase ends during a trip, I will just count the data collected before the trip started, and I will delay the start of the next phase until two days after the trip ends. I've set the end date for this market based on the schedule if I don't have to delay any phase starts. If I do delay a phase start, I'll adjust the end date accordingly.
While I'm attempting to make this as scientific as possible, N=1 studies are inherently limited. Obviously the experiment is not blinded, and I have a pretty strong prior that the multivitamin does cause weight gain, so the setup is prone to see weight gain during the vitamin phases caused by a placebo effect.
Today (2025-04-25) I started a vitamin phase, and in 30 days I'll end that phase and begin a non-vitamin phase. I previously was taking this multivitamin for a couple years, but I tapered off and stopped taking them during the second half of last year. Until this morning, I had not taken the multivitamin this year.
Update 2025-04-30 (PST) (AI summary of creator comment): The creator has updated the resolution criteria slightly. The market will now resolve YES only if both of the following conditions are met:
The lower bound of the 95% confidence interval for the daily weight change coefficient during the vitamin-taking phases is greater than the upper bound of the 95% confidence interval for the daily weight change coefficient during the non-vitamin-taking phases.
The lower bound of the 95% confidence interval for the daily weight change coefficient during the vitamin-taking phases is also greater than zero.
Otherwise, the market will resolve NO. This change ensures that the market resolves YES only if there is statistical evidence of actual weight gain (a positive trend) during the vitamin phases, rather than just a slowing of weight loss compared to non-vitamin phases.
1,000I made a slight change to my resolution criteria. I added the clause "and greater than zero," talking about the lower bound of the confidence interval for the coefficient during the vitamin taking phases.
Previously, I only talked about it relative to the non-vitamin phases, which meant there was a possibility that I could lose weight both taking the vitamins or not, but lost weight more slowly while taking vitamins, and therefore would have resolved this saying that the vitamins cause weight gain, rather than just slowed weight loss.
If anyone feels strongly that this change invalidates the experiment and market, I can go back to the original criteria.
I wanted to give some more data to explain my prior that I mentioned in the description. I've used a smart scale to track my weight since early 2016, which means there are a lot of trends I can see. Here's a summary of my change in weight over time:
It's pretty obvious that something significant happened in 2021. For most of the years on record my weight has been pretty flat, with maybe a small downward trend. In 2021, I gained almost 20 pounds, and gained a significant amount the two years afterwards. In 2022, I did the Slime Mold Time Mold potato experiment and lost 16 pounds during that time; so if we were to add that back into the 2022 change, that would be close to 20 pounds gained then too.
In 2024, I came across some blog posts mentioning that their might be a link between vitamin fortification and obesity. I looked at my order history in my email and realized I had started taking a multivitamin in 2021, and on closer inspection I noticed the date I had started taking the supplement corresponded very closely to when I started gaining weight. See the annotated screen shot below:
Based on this, I decided to stop taking the multivitamin and see what changes happened. While I didn't see prodigious weight loss, it's been long enough since I've stopped to see that the gain has mostly stopped and I perhaps even have a downward trend going.
So these data all seem to point in that direction, but these things are notoriously hard to figure out. I had a three month supply of the multivitamins left from when I quit taking them, so I've decided to use them up while trying to gain some more conclusive data. (Speaking of which, that reminds me of another potential confounding factor, which is that the expiration dates on these vitamins are coming up. The bottle that I take for the second vitamin phase will have expired about a month before starting that phase.)