I mean, I’m of the Hegelian perspective that nobody even chose to be born so, like, wtf.
If that’s not Hegel a universe brain can correct me. I ain’t googling it right now.
Also, I’m so anti-fat-shaming that, what with all the food we throw away making me weep, I consider people who eat “too much” to be doing the Lord’s work.
Like, I’m imagining two cows in cow heaven talking and one says, “Well, at least they ate every part of me, even the buuhole, what about you?” To which the other replies, “Omg those skinny mfs threw half of me in the garbage are you shitting me right now?!?”
It says that if the results occurred by simple chance, the observation would be at least this extreme 7% of the time. The way you phrased it is a common misunderstanding.
Similarly, a 95% confidence interval doesn’t say that the mean lies within the interval 95% of the time. This is why I’m not a fan of these Frequentist stats. They don’t actually answer the questions one wants to answer.
For all the complexity, the fundamental difference can be expressed rather simply. In frequentist statistics, the parameters are assumed to be fixed. In Bayesian statistics, the parameters themselves are treated as random variables. The parameters aren’t literally considered random though. It’s a philosophical approach to our knowledge and beliefs about the parameter.
That’s why there’s nittery involved in expressing what these things actually say. So for example, if the parameter value (which is always unknown) is fixed, there can’t be a “95% chance it lies within the confidence interval.” It either does or doesn’t, because it can’t change, and what shifts are the endpoints of the confidence intervals based on the data observed in repeated experiments.
Bayesian methods produce a credible interval that has fixed endpoints. Here the parameter is assumed to be random such that you could correctly say “there’s a 95% chance the parameter lies in the credible interval.” It’s basically a philosophical difference that matters in the way questions are formed and answered.
This is where people usually say, “Wait a second. I thought Bayesian statistics was all about using priors.” And yeah, you must specify a prior distribution to generate a posterior, but this is a technical point. It’s perfectly fine (and perhaps correct in most cases) to use objective or “uninformative” priors that don’t tilt the scales. Using an objective prior, the Bayesian maximum a priori (MAP) estimator converges to the maximum likelihood estimator (MLE) asymptotically. In other words, Bayesian and frequentist methods become the same value as sample size approaches infinity if you’re agnostic about the prior values. Surely you can see some ways me might weakly inform a prior, such as left-bounding at zero if the parameter can’t possibly take negative values.
This paper is short and runs through an example where the frequentist p-value actually turns out to be 0.07 and the Bayesian method gives something very different.
I’ve already articulated exactly what CICO is and it has nothing to do with doctors. If you want to know what doctors think CICO means, you should ask doctors. I’d bet most say something different than you since you’ve already conceded that other CICO advocates are using different definitions. In fact, this is really starting to shape up like all of the ancap / “libertarian” debates. At first they claim to be for the same thing, but by the end, after being asked to explain inconsistencies in the logic or evidence that doesn’t square with reality, what we see are as many definitions of the thing as there are people supporting it.
In Professor Spector’s study, while one person had a fat or sugar spike after eating one type of food, another got the same response from a different type. So when you learn a friend lost two stone on a diet yet you struggled to lose a few pounds, it may be because they cut out ingredients that caused the spike for them but you’re still eating the ones that cause it for you – you’re metabolising the same food very differently. The time people ate also affected their response.
If you think someone who has a BMI in the ‘healthy range’ must have a faster metabolism than someone categorised as being ‘overweight’ or ‘obese’, and that if you lose weight your metabolism might pick up too, you’re wrong.
It’s true that if you have a larger build, Dr Yeo says it’s likely your metabolism will be higher than someone who has a slimmer build because “you have a bigger engine to run”. But here’s the kicker: if you lose weight your metabolic rate decreases.
“So, I’m 75kg”, says Dr Yeo. “Imagine I have a twin who used to be 85kg, but has lost 10kg so is now 75kg too. But I’ve never been 85kg. I will always be able to eat more than my twin… because he’s defending a higher weight than me.
“As you lose this weight, your brain hates it”, he continues, “because it considers it as reducing your likelihood of surviving. Keep in mind we’ve evolved from a time when there was not enough food. So whenever you lose weight a red flag goes off in your brain and two things happen: it makes you more hungry and it lowers your metabolism slightly to try and drag your weight back up.”
Dr Yeo says this will be the case whether you’ve reduced your weight gently over time or via an extreme diet.
“When you lose weight it doesn’t drop by that much, but you have to remember that 8 to 10 to 20 calories every single day, that kind of difference is enough to make a difference in 5 kilos of body weight over a few years.”