Locrian said:
There's no mean, no variance, no statistics in your world, there's just a sample of n=1 (you) and it somehow reaches Bayesian credibility.
Everyone is different. Every situation is different. That's why I tend to talk a lot about my own experiences. I know my life and my situation. I can guess what is your life and what is your situation, but I'm not in a position to talk about your life.
Someone makes a comment about corporations. I mention my personal experience about my corporation. Would you rather me talk about a corporation that I've never worked in, and I know nothing about? So we want to talk about corporations in general. Fine. If we don't talk about specific details, then we just end up talking about stereotypes and preconceptions.
And there is no variance or statistics in my world, because I have only one life. I suppose it's like quantum mechanics. I can say that there is a 25% chance that I get into Harvard, but in the end what does that ***mean***? Either I get in or I don't.
One common question is "what are the chances of X?" and I really think that this is the wrong question, and people should stop thinking in ways that would result in that question being asked.
The way I think about statistics might have something to do with my work. Suppose someone tells you that there is a 0.0001 chance of major housing crash. A lot of people might think, well the probability is low enough so we don't have to worry about that, but people that thought like that ended up bankrupt after the crash. So "what is the chance of a housing crash" is the wrong question. The right question is "if there is a housing crash, what happens next?"
"What are my chances of getting accepted to Harvard?" That's a bad question. I tell you 0%, 20%, 40%, 80%, 100%? How does that change what you are going to do? If you are going to apply whether the answer is 1% or 99% then there is no point in even asking. If you apply, either you get in or you don't. What is your plan if you get in? What is your plan if you don't?
It may be that I don't think in terms of "average cases" because thinking about "average cases" is what got us into this financial mess in the first place. When the government comes and asks you for data, they don't care about the "average case" because the "average case" is meaningless. What they want you do to is to describe the worst case, and see if you can survive that, and if you tell them that you don't have to worry about the worst case because the probability of that happening is "low" then you will get a very dirty look.
People on in my line of work don't think very much about averages or variances. It's probably a result of what happened in 2007. If you create certain financial instruments, then on "average" you make a decent amount of money, but in the worst case scenario, the world blows up. Looking at the variance, you can reassure yourself that the probability of that happening were "low" but it turns out that the probabilities didn't mean a damned thing. Since 2007, people have done a lot of "deep thinking" about what a probability really *means* and I think within the industry people agree that the language of "average" and "variance" is what got us into trouble in the first place.
Someone's chances of getting a university position must be the same as they were for those who went to your school, no matter where they went to school or what field of physics they're in; It's always one in ten.
Would you be happier if someone asks me what the odds are of getting a job in academia, and I reply 23.432432 +/- 0.0005 %
That seems to be a meaningless number, and since we are talking about round numbers, then we can talk about this in terms of orders of magnitude. So let's do this in half orders of magnitude
1 in 1
1 in 10^0.5
1 in 10
1 in 10^1.5
1 in 100
OK, which is the closest to the right number? I pick 1 in 10. That number should be "good enough." I hope we aren't in a situation where the number being 21.5% will cause you to do one thing and 22.5% will cause you to do another.
And as a matter of fact, it does matter which field that you go into. If you get a Ph.D. in finance or economics, you are pretty much guaranteed of getting a job as tenure track faculty. For physics, I can imagine things going to 1 in 3 if all of the stars align. But there is no set of decisions that will cause things to be more than 1 to 1, and then you look at the structure of physics hiring and find out that mathematically this must be the case.
Also the reason I use UT Austin Astronomy is that there is a professor there that keeps track of this (and I can message you her e-mail if you want to chat with her). She keeps a record of what every graduate is doing. The other source of statistics are the AIP surveys.
Now it may be that my school is weird, but I don't have data about other schools. So the most that I can say is here is the data that I have, what can and can't we figure out from it? One thing that I strongly suspect is that universities don't keep this sort of data because the numbers will make them look bad. If you had a school that places 80% of its Ph.D.'s in tenure track jobs, I think they would announce this quite loudly. That's consistent with a 1 in 10 statistic.
I try to talk from data. Sometimes the data is sparse. Sometimes I extrapolate and sometimes I outright guess, but I try to minimize that and when I'm guessing or extrapolating, I try to make it clear what I'm doing. I tell you what I see, and if you see something different then great, we can piece together what is going on.
I don't see why that should be considered weird.