Last year, a study came out showing that beef jerky and other cured meats, could trigger mania in bipolar disorder (paper, popular article). It was a pretty big deal, getting coverage in the national press and affecting the advice psychiatrists (including me) gave their patients.
The study was pretty simple: psychiatrists at a mental hospital in Baltimore asked new patients if they had ever eaten any of a variety of foods. After getting a few hundred responses, they compared answers to controls and across diagnostic categories. The only hit that came up was that people in the hospital for bipolar mania were more likely to have said they ate dry cured meat like beef jerky (odds ratio 3.49). This survived various statistical comparisons and made some biological sense.
The methodology was a little bit weird, because they only asked if they’d ever had the food, not if they’d eaten a lot of it just before becoming sick. If you had beef jerky once when you were fourteen, and ended up in the psych hospital when you were fifty-five, that counted. Either they were hoping that “ever had beef jerky at all” was a good proxy for “eats a lot of beef jerky right now”, or that past consumption produced lasting changes in gut bacteria. In any case, they found a strong effect even after adjusting for confounders and doing the necessary Bonferroni corrections, so it’s hard to argue with success.
Since the study was so simple, and already starting to guide psychiatric practice, I decided to replicate it with the 2019 Slate Star Codex survey.
In a longer section on psychiatric issues, I asked participants “Have you ever been hospitalized for bipolar mania?”. They could answer “Yes, many times”, “Yes, once”, or “No”. 3040 people answered the question, of whom 26 had been hospitalized once, 13 many times, and 3001 not at all.
I also asked participants “How often do you eat beef jerky, meat sticks, or other similar nitrate-cured meats?”. They could answer “Never”, “less than once a year”, “A few times a year”, “A few times a month”, A few times a week”, or “Daily or almost daily”. 5,334 participants had eaten these at least once, 2,363 participants had never eaten them.
(for the rest of this post, I’ll use “beef jerky” as shorthand for this longer and more complicated question)
Power calculation: the original study found odds ratio of 3.5x; because the percent of my sample who had been hospitalized for mania was so low, OR = RR; I decided to test for an odds ratio of 3. About 1.2% of non-jerky-eaters had been hospitalized for mania, so I used this site to calculate necessary sample size with Group 1 as 1.2%, Group 2 as 3.6% (=1.2×3), enrollment ratio of 0.46 (ratio of the 921 jerky-never-eaters to 2015 jerky eaters), alpha of 0.05, and power of 80%. It recommended a total sample of 1375, well below the 2974 people I had who answered both questions.
Of 932 jerky non-eaters, 11 were hospitalized for mania, or 1.2%. Of 2042 jerky-eaters, 27 were hospitalized for mania, or 1.3%. Odds ratio was 1.12, chi-square statistic was 0.102, p = 0.75. The 95% confidence interval was (.55, 2.23). So there was no significant difference in mania hospitalizations between jerky-eaters and non-eaters.
I also tried to do the opposite comparison, seeing if there was a difference in beef jerky consumption between people with a history of hospitalization for mania and people without such a history. I recoded the “beef jerky” variable to a very rough estimate to how many times per year people ate jerky (“never” = 0, “daily” = 400, etc). The rough estimate wasn’t very principled, but I came up with my unprincipled system before looking at any results. People who had never been hospitalized for mania ate beef jerky an average of 16 times per year; people who had been hospitalized ate it an average of 8 times per year. This is the opposite direction predicted by the original study, and was not significant.
I tried looking at people who had a bipolar diagnosis (which requires at least one episode of mania or hypomania) rather than just people who had been hospitalized for bipolar mania. This gave me four times the sample size of bipolar cases, but there was still no effect. 63% of cases (vs. 69% of controls) had ever eaten jerky, and cases on average ate jerky 15 times a year (compared to 20 times for controls). Neither of these findings was significant.
Why were my survey results so different from the original paper?
My data had some serious limitations. First, I was relying on self-report about mania hospitalization, which is less reliable than catching manic patients in the hospital. Second, I had a much smaller sample size of manic patients (though a larger sample size of controls). Third, I had a different population (SSC readers are probably more homogenous in terms of class, but less homogenous in terms of nationality) than the original study, and did not adjust for confounders.
There were also some strengths to this dataset. I had a finer-grained measure of beef jerky consumption than the original study. I had a larger control group. I was able to be more towards the confirmatory side of confirmatory/exploratory analysis.
Despite the limitations, there was a pretty striking lack of effect for jerky consumption. This is despite the dataset being sufficiently well-powered to confirm other effects that are classically known to exist (for example, people hospitalized by mania had higher self-rated childhood trauma than controls, p < 0.001). This is an important finding and should be easy to test by anyone with access to psychiatric patients or who is surveying a large population. I urge other people (hint to psychiatry residents reading this blog who have to do a research project) to look into this further. I welcome people trying to replicate or expand on these results. All of the data used in this post are freely available and can be downloaded here.
How could we tell that the causal arrow goes from beef jerky-> mania rather than mania -> beef jerky? It seems likely that people who are manic have different cravings. I know personally that mood is linked to wanting different foods?
Suppose mania leads to protein cravings and also to very short time preferences. If someone wants protein and they want it right now, they are likely to to buy jerky or other cured meats. Those who are hospitalized for mania must be exhibiting mania before they are hospitalized and eat jerky while manic, before their mania gets them hospitalized.
The study should never have been taken seriously to begin with. It appears to be a classic case of the Garden of Forking Paths. Add to that the sloppiness of the methodology, and you have another poster child for the replication crisis.
Not forking paths.
They Bonferroni’d the stats, which compensates for multiple comparisons.
I read this as the researchers doing the necessary Beefaroni corrections and was very confused.
If someone is looking into this sort of thing, then why not go to bipolar hospital patients and ask them what they’ve eaten lately?
How can they even publish the original study without running a second data sweep to test only for the “beef jerky” .
How can anyone who’s done science or engineering or even read Feynman take anything they say seriously?
Because if they do a second study without publishing the first, they publish zero studies.
If they publish the first, they publish one study, which is then cited by the many studies that fail to replicate it.
Because the bar for publishing is not “definitely true”, but “possibly interesting to other scientists”.
Certainly you wouldn’t expect the crazies who read your blog to resemble garden-variety crazies in any way, would you? 🙂
Isn’t the study’s initial question (“Have you ever eaten locally procured dry cured meat?”) trying to get at something completely different from what their followup question dealt with? (Not to mention the exciting conclusion of the study.) It sure sounds to me (and perhaps to some of the respondents?) that the “locally procured” part is important. Doesn’t that suggest weird artisan or homemade stuff, like wild game jerky?
…
OK, where are the odds ratios for “other”? As “other” would seem to include bacon and sausage, both very commonly eaten, shouldn’t “other” be a common response?
Your question and the study’s question seem substantially different.
The study:
SSC survey:
To me, “other similar nitrate-cured meats” includes bacon and sausage — both more commonly consumed than jerky. Your question is more inclusive, assuming SSC respondents are similarly aware of their food, a good bet.
Bacon and sausage must be cooked before eating, while jerky and meat sticks are ready to eat. They’re used in completely different ways.
I now some sausages that are only cured and smoked (which get’s about as hot as industrial jerky production).
Those sausages are also eaten straight from packaging.
As long as we’re on this topic, do people have thoughts on the usefulness of hypomania? I read this article a few years ago and realized I occasionally get spurts of energetic, inspired productivity which resemble what the author describes. Are there any downsides to trying to be in that state all the time? It seems like when I’m not in that state, all the things I want to do end up taking way longer.
Other than that mania is associated with poor decision making, which includes poor ability to evaluate whether your decision making is impaired?
Are you asking “would there be downsides to magically staying in that state or forever?” or “What are the downsides of trying to stay in it forever?” Because if it’s the latter, then it really matters what you’re actually doing when you try to get/stay hypomanic.
When I was younger, I depended on my hypomanic episodes to get anything done. I discovered that hypomanic episodes were more likely to begin when I was tired or hungry, so over time I started depriving myself of food and sleep more and more aggressively in hopes of having more hypomania.
That turned out to have considerable downsides.
Unipolar hypomania and unipolar mania are real things. (Hypomania or mania without depression.) They don’t have much coverage in studies because psychology of that general field area focus a lot more on stuff like suicide rates.
Downsides to unipolar hypomania exist and you would need to assess how strong the hypomania is and what downsides it has on that specific person to answer your question. If the hypomania is very weak, has low side effects, and doesn’t vary much over time, then it may be mostly beneficial. I despise typing that out though because any person who has hypomania or mania is automatically going to be overconfident in their ability to control their own actions, underestimate (or forget or ignore) the actual historical variance in their hypomania/mania over time, and underestimate (or forget or ignore) how harmful the side effects they have are. I’ve watched it happen and it isn’t pretty.
Downsides:
Grandiosity (self-focused)
Psychotic symptoms (losing touch with reality in a variety of forms)
Excess energy
Increased Positivity
Increased overall Enthusiasm
Irritability
Lasts longer than hypomanic or manic episodes for bipolar patients
Also, (BIG ONE) hypomania, at least with bipolar people, historically has a risk of turning into full blown mania. Full mania is bad bad bad bad bad.
Symptoms like grandiosity, positivity, and enthusiasm sound fine at first, but have to be considered in terms of both how strong the hypomania is, how long it lasts, and how much variance it has. If the increase in grandiosity, positivity, enthusiasm, etc. is something like 10% above normal, only lasts for one week a year, and only varies between 0% and 10% max, then things are probably going to be perfectly fine for that person. If the increases are 50%, it lasts 3 months, or has a variation patthern even slightly resembling something like 20% -> 50% -> 30% -> 10% -> 50%, then that’s going to be a very unhealthy experience for a person.
Additionally, psychological and emotional stability are very important for interpersonal relationships. If you can’t count on someone to be regularly stable or the same person each time you see them, then that can kill your evaluation of them as a dependable person and someone worth investing in relationship-wise.
Any patience or sympathy I once had for the people writing and publishing these kinds of studies is fully converted to disgust at this point. The replication crisis has gone on too long for this to be at all excusable.
Hypothesis: it’s not jerky that’s weird. Instead, it’s the people who have never eaten jerky who are weird.
Suppose there’s a segment of the population that has a property that would inherently lead to their never eating jerky. Veganism, for example. If a vegan diet protects from bipolar mania—and if this sub-population was a significant segment of the population who have never eaten jerky—then the original study would product the same results.
I find beef jerky to be repellent. Perhaps I am weird, but I don’t think it’s uncommon to dislike the stuff.
I’m sure I had some when I was a child, so I do have some memory of the flavor and odor. I didn’t find it objectionable then, but as an adult, even the memory makes me gag.
And yet in this study–not Scott’s, but the prior one–you’d be in the jerky eater group.
Either I’ve missed something or the original study showed us the population of Baltimore eats significantly more locally procured cured meat than undercooked or raw meat and fish. This does not seem a surprise, but is an odd thing to report in a psychiatric journal rather than in a marketing report to a salami or sushi manufacturer.
Basically, the study asked a possibly representative* sample of Baltimorians (? Baltimorons is probably wrong… Baltimorese?) about their eating habits, and did statistical testing on this, showing the tendency to eat cured meats is significantly more common in Baltimore than other forms of not fully-cooked meat. They did not survey any population that was not in the hospital to establish a baseline for the prevalence of eating these various meats, so therefore did not establish that the consumption of cured meat was actually unusually high for patients with mania compared to the general population. To put it succinctly, unlike Scott, they forewent a control group. Without this we can’t say that the link between eating nitrate-cured meat and mania is significant, because for all we know more people without mania may have been exposed to beed jerky, slim jims et al!
What is worrying is a paper with an apparent methodological hole like this was published. What were the reviewers thinking? Clearly it wasn’t about methodology, which is a problem considering in scientific research method is everything.
To move into speculation: it looks like there was a theory about the possible effects of nitrates on mania, which I’m not able to evaluate. It appears that a test was done to see if there was grounds for this, which there was on an initial level. Then rather than providing a proper control to show this was a significant finding, a paper was rushed out. This doesn’t mean the nitrate-mania hypothesis is wrong: it means some proponents did what looks like bad science** by not considering their population as part of the general population. And ended up doing a marketing report which somehow affected psychiatrists practices…
*These were first-time patients, so the effects of mental health problems on their lives were presumably less marked than in long-term sufferers, thus meaning the cohort was as likely as a cohort of people suffering from mania will be to reflect wider society.
** Good science here might have at least attempted a replication with a different cohort, although Scott’s reports on the bureaucratic difficulties of psychiatric research suggest this might not be that easy.
Well it depends on which controls they compared to, yeah? I figure they were simply checking whether they ate more jerky than the national rate in previously published research. Like, the control was a rate from a national survey maybe?
I can’t tell what you think that they did. Every paragraph seems to accuse them of something different. They certainly had a control group. How good was it? Hard to tell. They adjust for, eg, race because the racial distribution of the controls is different from the racial distribution of the patients.
But there is another group that acts as controls, which is the other patients. If the patients all ate jerky and the controls didn’t, that would be suspicious, probably something about how they got the controls. But its only the mania patients that eat jerky, not the schizophrenics.
Ah. I see the problem. I searched for control (I was scanning through the paper) and missed the earlier references, so only got control mechanisms mentioned. Danger of scan reading whilst distracted epitomised there…
I’m a huge fan of all your work looking at the SSC surveys to replicate other research. Its… a very good use of time.
It benefits substantially from the fact that you make the data public — but it would be even stronger if you made code for the analyses public as well. Given that numerous studies have coding errors that cause results, being able to dissect exactly what you do is helpful — moreover, if I feel like you did more or less the right thing, but I want a daily jerky to be 365 on the year rather than 400, it makes it easy to do that. It also makes your hints towards psych students easier to follow up on — as it becomes much more plug’n’play with the survey data.
I do realize however that it would incur some cost to you with making code that is fully runnable — if you’re even using any stats computing language to begin with, and ensuring that PII doesn’t end up in the code can be a bit of a pain, but I do think it is worth considering.
It’s such a simple analysis. How about you replicate it and publish your code? There is value to people replicating without code.
What were they thinking?
The paper seems to say that the questionnaire had 5 questions, 4 about raw foods and one about beef jerky. It sounds like they thought raw food was relevant and added the beef jerky as a control that didn’t fit their theory.
If it was intended as a control, did they have a plan for when it triggered? It seems like if they had gamed this out ahead of time, the answer wouldn’t be “publish immediately,” but accept that they’re confused and replicate. If they didn’t game it out ahead of time, why the control? One person wanted it as a control and the rest went along with the demand, but not the long-term plan?
(They didn’t simply publish immediately. Instead they did animal experiments. Better than nothing, but I doubt that’s what they’d plan for if asked ahead of time. Also, they modified the questionnaire 80% through the study period. So the last 20% would constitute a replication of registered hypothesis, but I don’t think that they break out the numbers. Instead they break out the results of the finer questionnaire, which is the typical path to not noticing failure to replicate.)
I heard a similar story in which Mendenhall tried to use stylometry to compare Shakespeare and Bacon. He included Marlowe as a control. He distinguished Shakespeare from Bacon, but not Marlowe from Shakespeare. What should he conclude?
In both cases, it may be that they did have a plan for a control, but only in the other direction. If your method identifies Shakespeare and Bacon, you shouldn’t conclude anything until it you checked that it doesn’t identify everyone.
That once again nitrate cured meat stands out?
Looks to me that more likely they were looking for an effect from nitrates as they have something of an explanation for it in the paper.
What’s the explanation? I didn’t see one. Is this in the animal model? I’d think that came later. Don’t they say that the hypothesis is infection? which makes more sense with raw meat than cured.
At this point, I’m curious about what people who’ve had manic episodes think set them off?
My understanding is that it can be a whole lot of things. Lack of sleep is a big one, also lack of food, changes in drug/alcohol use, travel, life changes, general excitement. It’s often seasonal too. Basically it’s predictable enough thar you can *try* to anticipate and control it, but not predictable enough that you can reliably do so.
Just a cycle, for me. A very slow but predictable one in my case. There are triggers that really set me off while manic but nothing induces the mania itself.
Stress, definitely. Diet changes, probably. Menstrual cycles, lately. Sleep, possibly. Change of seasons…? TMI below.
My worst (hypo)manic episode was triggered by the combination of ketogenic diet and family stress. I had been on keto for some time without serious mood issues (though it was a strain and very not good for me), but spiraled out after a thing with a family member went south. Since then, I’ve had similar/worse family stresses without episodes, and I’ve had diet changes including significant weight loss (I’m small and can’t afford to lose much) without episodes, so I think it was specifically that combination. I had also lowered my dose of meds several months before either the diet or the stress, which probably lowered my ability to deal with those triggers.
Lately I’ve had my mood cycling slightly but very noticeably each month. (Example: down starting around March 28th, up at about May 3rd, down again starting June 3rd, etc.) My life has been uniquely calm and stress-free recently, so it isn’t that, and my diet has varied. I don’t have actual periods due to my birth control but I do have cycles, so given the timing, I figure it’s that.
It’s hard for me to tell whether lack of sleep is a trigger or a symptom–I avoid it under the assumption that it’s a trigger, but most often it appears to be a symptom. That is, if I’m suddenly unable to sleep, either, 1. it has no apparent cause and ends up being one of the first signs of where I’m going, or 2. I’ve already arrived.
I suspect I’m significantly more likely to be up in the winter and down in the spring, but it’s difficult to clear out confounding factors enough to decide whether that’s actually the case. If so, that’s about the exact opposite to most people afaik.
The summary is that it’s a lot of guesswork, hindsight, and process of elimination. A lot of warning signs are also potentially normal, like insomnia, feeling creative, and seeking out social interaction. So far it’s not very predictable for me, and so I worry that every fluctuation in mood or behavior might be a Sign of What’s To Come. Depression is bad, but I’m used to it and can live with it; I’m genuinely afraid of the hypomania. It feels awful to be plugged into that much current.
Perhaps the requirement of having been hospitalized for a manic episode gave too small of a sample size? It’s my understanding that a very small number of people who have had hypomania or mania have ever been hospitalized. Maybe a new experiment could compare beef jerky consumption with number of manic episodes per year, without the hospitalization requirement.
Scott said he addressed this in the survey by looking at people with a bipolar diagnosis, but not necessarily hospitalized.
So you and other psychiatrists were seriously counseling bipolar patients that there was a real risk they’d snap after a Slim Jim, ohhh yeah!
How long after? Answer not in evidence.
Does speaking like Macho Man Randy Savage count as having a psychotic episode? Because I start doing that pretty much every time I even think of slim jims!
This sounds like a classic green jelly beans study. Ask patients about dozens of different foods, and it is likely that one of the foods will have a “statistically significant” correlation with disease.
Thankfully, they adjusted for this within the study by raising the bar required to reject the null hypothesis
Wat? I feel horrible for these people. A life without nitrate-cured meats is a life not worth living!
I’m a little confused, because why is beef jerky the canonical example of a nitrate-cured meat? Why not bacon?
I love the understated sarcasm.
Anyway, maybe you could have found something that did have an effect if you included twenty more types of food.
Any issues with reduced accuracy of self-reports while one is in the throes of a manic episode severe enough to require hospitalization? I guess it would be odd for that to create a false correlation with one particular type of food and not any others but still.
Antipsychotic drugs are ruining my life but they’re also like magic for ending psychotic episodes. A shot of Geodon will take someone in full-on psychosis to a sufficiently composed state within like an hour, sooner really. I’ll sure these people were being medicated before doing the questionnaire.
Hey, I’m sorry to hear you’re having to deal with these issues (although I now see posts on the internet about it from a few years ago). Hope things are going reasonably well, friend. One thing that interested me about your comment, even though it isn’t directly relevant to the topic of beef jerky, is why, if a single dose of medication typically relieves the severity of a major manic episode, this group is nonetheless hospitalized notwithstanding the rapidity with which it can be dealt with. Why not just give them a dose of Geodon, wait a couple hours to make sure they’re under control, and send them home?
Edit: Or maybe I was just assuming they were inpatient cases. I guess the post above actually doesn’t say that.
Well you’d give them a shot of Geodon (or Haldol) and then give them oral atypical antipsychotics like Seroquel etc to use long term. Why keep them? Legal issues (if you release them and they harm themselves/others afterwards you’re exposed to legal liability); the desire to get them into therapy; to get them on meds long enough to get them into a rhythm; in extreme cases, I guess, to have them on lithium long enough for it to kick in (3-4 weeks). To do more doctoring.
Heya!
I have bipolar type 1, and historically I’ve stocked up on dried meats, meal squares, and (now) laräbars when pursuing my mania-fueled self-defined super-tasks.
My suspicion is that manic people like instantaneous nutrients more than a causal link
Yeah, the people who did the study duly nodded at the issue of causality. Alone, it’s like the well-known schizophrenia-nicotine and schizophrenia-marijuana correlations. The thing that seems to indicate the existence of causality is that nitrated beef seemed to cause mania in rats.
Then why wouldn’t I have found this?
Social class of who you’re asking, maybe?
I’d imagine your average SSC survey respondent is going to have access to more resources than your average Baltimore mental patient. If mania implants a driving compulsion of “I need to consume some kind of food with a bunch of fat and calories” then what the people enduring that mania actually consume is going to vary wildly. Maybe rho stocks up on laräbars, while the people the study asked just go to town on some Slim Jims.
Presumably there are many such differences between the average Baltimore mental patient and the average SSC reader.
This is perhaps a bit of a prejudiced view of Baltimore. The average Baltimore mental patient is likely more affluent than the national average. There’s a lot of wealth in the state/region. More specifically the study was done at Sheppard Pratt, a historically renowned mental institute that’s actually located in Towson on a large suburban campus it shares with GBMC. It services people from the entire region, not just the inner city. And I’d assume the socioeconomic demographics aren’t wildly different than SSC, tho I’m sure SSC skews much more white male and affluent in general
Hahaha! What?
@Cliff,
Sure SSC is bizarrely white and male, but the income/class background probably isn’t as wildly different as you imagine. If you think all of Baltimore = The Wire you have an aggressively ignorant view of the metro area. Go walk around the Sheppard Pratt campus some day, it’s not at all what you’re imagining.
Where are the survey results? Pretty sure the income/class of SSC is very different from any city in the world.
https://slatestarcodex.com/blog_images/2019%20SSC%20Survey.html
Yeah, so, 50% upper-middle class or rich, 90% middle class or higher, 75% with Bachelor’s degree or higher (while only 70% have completed their education), average IQ of like 135, average income something like $100k…
Which city did you have in mind as not wildly different?
@Cliff
Princeton
Well, if there were some rather more male version of Princeton, anyway.
Meh, hard to tell from that link.
Median household income there is $100k, but I guess it depends on your definition of “wildly different”
Per-capita income includes everyone, not just wage-earners. So it’s dragged down by students, children, and those weird old guys who wander around college campuses and you think might be professors but also might be bums.
Slavoj Žižek?
Isn’t Beef Jerky expensive? Maybe just in Germany, cause it’s a delicatess from the United States.
I actually like it very much, though I do not suffer from mania.
I just want to point out that as a German I had never eaten beef jerky until I visited China. Nor had I ever consciously seen it.
I also wonder how veganism might distort the SSC results.
But I think your country’s many wonderful hams and sausages are often nitrate/ite-cured.
Every Rewe, Real, and Edeka have it on their usual product line, and the discounters usually have it every so often as special offer. (It is stupidly expensive though).
And like Lambert says all the dryed sausages are usually heavly cured.
One possibility: your results could be different due to asking people a long time after they’ve been hospitalized for mania vs. immediately after.
Assuming that people eat more cured meats during periods of mania and that people mostly forget what they’ve eaten after a few months, then if you ask a bunch of people hospitalized for mania (who just ate a whole bunch of jerky) if they’ve ever eaten jerky, they’ll say yes. If you ask the same population six months later, they may not remember all the beef jerky they ate months ago, and say no.
Well, the available evidence suggests I’m fantastically strange, so perhaps I have no insight here. But in the case of food I either optimize for deliciousness, optimize for healthiness, or optimize for speed of delivery. If I’m manic I’ll attempt to optimize speed of delivery while satisficing for health and deliciousness, because I’m at the height of my industry. Beef jerky is particularly suited for this if you need to ingest protein. Beyond the miscellaneous preservatives, there’s nothing really objectionable about beef jerky as well. Larabars simplify the nutritional calculus because they are transparent about their ingredients and typically use 5 or less substances. Mealsquares were useful back when I didn’t understand nutrition as well, but the cost to calorie ratio was too high for their flavor profile so i stopped buying them.
To return to the topic, I think it’s generally applicable that manic people would seek efficiency first when it comes to eating. When I’m manic I’m at an extreme of goal-oriented behavior and hunger is a mere distraction to be rid of swiftly, but perhaps this is a feature of me alone? Also, not many people can do (or even bother to attempt really) the instantaneous and granular assessment of their nutritional needs and the food at hand that I can so maybe they would arrive at a different answer when trying to eat fast?
As someone who has experienced drug-induced mania, I too immediately thought of this. I don’t think your story is abnormal in the least – the desire to eat something as conveniently and quickly as possible (to disappear the hunger) follows the rest of the logic of that state.
I think a necessary adjacent study would be to compare rates of beef jerky consumption in bipolar mania patients with beef jerky consumption in methamphetamine users. Then cross-check other gas station and packaged meal foods.
This has “multiple comparisons false positive” written all over it. When you say that it “survived various statistical comparisons,” do you mean they accounted for this in a satisfactory way (ideally by confirming the beef jerky correlation as an a priori hypothesis on a separate, independent sample)?
nah, it’s fine. They “adjusted for cofounders.”
I think I’m going to start just throwing that in as a line in any email I send or anything I say that someone might be tempted to doubt or be skeptical of.
“No need to worry, I adjusted for confounders.”
“Furthermore, Carthage must be adjusted for confounders”
– Cato the Statistician
I mean, if they adjusted for multiple comparisons in a fairly conservative way (like a Bonferroni correction), I’d be willing to take them semi-seriously. I’m just worried because the description above sounds a bit like a fishing expedition.
*edit* It looks like they did Bonferroni-correct their results, so I’ll grudgingly admit they might have an actual result on their hands.
Right, I agree with you. I should have included the /sarcasm tag in my original response.
This means that they probably didn’t get the result just due to random variation in their sample, but it doesn’t rule out sampling bias and uncontrolled confounders.
What if your co-founders are not themselves well-adjusted?
In the paper they report a p-value of 10^-7 and that they performed Bonferonni correction for multiple comparisons. Also note that they only asked about 5 different food categories, of which one was cured meat. Within cured meat, it looks like they asked about 3 types of cured meat. So there really isn’t that large of a multiple hypothesis testing problem here, at least assuming they’re reporting all the tests they might have performed.
I just want to note that instead of speculating about whether they do it correctly, one can just read the paper to see.
By “replicate” in the second last sentence, he means for you to eat a lot of beef jerky and see if you get hospitalized for mania.
I’ll post my results.
Possible confounder: No internet access in the institution.
Rho could post every day “I’m not institutionalized today.”
“Psychiatric institutionalization linked to power failures.”
“Psychiatric institutionalization linked to feelings of disempowerment.”
I actually finangled 20-40 minutes of internet access a day the last time i was in the joint, which was exceedingly recently.
So why does the beef jerky/mania correlation make biological sense?
Fudging some here but cured foods tend to have a higher tyramine content => tyramine is a precursor for/has effects on several neurotransmitters => mumblemumble => therefore mania?