“”Not evil, but longing for that which is better, more often directs the steps of the erring”
Theodore Dreiser, Sister Carrie
neq1
Karma: 380
If someone was locked in to a belief, then they’d use a point mass prior. All other priors express some uncertainty.
A rationalist sits down next to an attractive woman at the bar.
He asks “are you familiar with immediate reward bias?”
“No,” she responds.
“Well, people tend to place irrationally high value on immediate rewards, relative to future rewards. So, for example, they might prefer $50 today over $55 next week. This is a bias that a more rational person can take advantage of in trade negotiations. Unfortunately, I am an impatient person. With that in mind, I have an offer for you. If you agree to have sex with me ONCE tonight, I will agree to have sex with you TWICE next week.”
The body count argument annoys me, and it’s disappointing to see people like Hitchens use it. Whether or not there is evidence-based reasons to believe in god is a separate issue from whether people who do or do not believe in god do other stupid or immoral things. There are atheists, I’m sure, who reject god for completely irrational reasons and are generally irrational themselves. It matters not just what you believe, but why you believe it.
The primary reason SIA is wrong is because it counts you as special only after seeing that you exist (i.e., after peeking at the data)
My detailed explanation is here.
Self-indication assumption is wrong for interesting reasons
fixed. thanks.
If you repeat the experiment, does who you call I stay the same (e.g., they might not get selected at all)? If so, then that person was labeled as special a priori, and if they find themself in a room, then the probability of blue is 0.99.
But.. I’m arguing that anyone who is selected, whether it is one person or 99 people, will all think of themselves as I. When you think of frequentist properties then, you have to think about the label switching each time. That changes everything. The fact that you were selected just means that someone was selected, and that was a probability 1 event. Thus, probability of blue door is .5.
Good point. Thanks
Thanks for directing me to that post.
I think calling it ‘pre-selection bias’ makes sense. Would be good to have a name for it, is it’s an error that is common and easy to miss.
It’s a hidden label switching problem.
If Laura exists, she’ll ask P(blue door | laura exists). Laura=I
If Tom exists, he’ll ask P(blue door | Tom exists). Tom=I
If orthonormal exists, s/he will ask P(blue door | orthonormal exists). orthonormal=I
and so on. Notice how the question we ask depends on the result of the experiment? See how the label switches?
What do Tom, Laura and orthonormal have in common? They are all conscious observers.
So, if orthonormal wakes up in a room, what orthonormal knows is that at least one conscious observer exists. P(blue room | at least one conscious observer exists)=0.5
Yes, I agree. That would be interesting. I’d be more tempted to read a post that I saw had a lot of both up and down votes, than I would be one that had few votes
“History is like the weather. Themes do repeat themselves, but never in the same way. And analogies became rhetorical flourishes and sad ex post facto justifications rather than explanations. In the end, they explain nothing.”
-Errol Morris
Here is what he said prior to making the statement I quoted (to give you some context):
Take historical analogies. I believe that historical analogies are always wrong. This a long discussion, but, to me, the most dangerous thing about Chamberlain’s capitulation to Hitler at Munich is not the fact that Munich happened and it led to further Nazi aggression and so on and so forth, but that the example of Munich has been used to support thousands upon thousands of bad policies and inappropriate decisions. LeMay called JFK’s recommendation for a “quarantine” (that is, a blockade) in the Cuban Missile Crisis “worse than Munich”. Would nuclear war have been a better alternative? But nuclear war was averted by Kennedy’s policies. And thirty years later the Soviet Union collapsed without the need for nuclear war. Was LeMay right? I don’t think so. But again, the example of Munich was invoked to justify the invasion of Iraq. Appeasing Saddam, appeasing Hitler. The use of the Munich analogy does not clarify, it obscures.
The prior probability of 0.13 is wrong. That would be correct if 13% of fires resulting in fatalities of children were intentionally set by the children’s dad.
“I meant just to emphasize that the prior probability for Willingham’s guilt is at least an order of magnitude higher than the prior probability of Knox’s guilt.”
I think you made an interesting observation. Just thought it was worth noting that the prior prob is probably too high
Another aspect of this is whether we are being fully Bayesian or not. If fully Bayesian we’d have a prior distribution for the probability. that prior might have a mean or mode at 14%, but still be pretty flat (reflecting uncertainty)
Ah, I think that’s it (posted to drafts). Thanks. Not sure how I missed that.
Beauty quips, “I’d shut up and multiply!”
Not all of the waking moments have the same probability of occurring. If you estimate the probability of heads by the proportion of waking moments that were preceded by heads, you’d be throwing out information. Again, on a random waking moment, Monday preceded by heads is more likely than Monday preceded by tails.
So why do you still take vitamins? If you look at their Figure 2, there aren’t many studies that ‘favored antioxidants’, and some of those studies had low doses.
“A linear analysis assumes that if 10 milligrams is good for you, then 100 milligrams is ten times as good for you, and 1000 milligrams is one-hundred times as good for you.” That’s only true if the range of data included both 10 milligrams and 1000 milligrams. Linearity is only assumed within the range of data of the data sets.
The hockey stick approach seems too restrictive as well. Just use a p-spline.
There doesn’t appear to be statistician on the paper. This study really needed one. Using meta-regression to estimate a dose effect is challenging, especially when you don’t have access to the original data (just using aggregate, study-level covariates). In fact, the dose effect and the concept of study heterogeneity are conflated here.
I agree with you that it’s unclear what they actually did.