Question About Bayesian Analysis

Great Debates
21

It depends on which subgroups Smith and Brown come from (and how they are chosen).

If you pick one Fleeper and one Meeper at random from the entire population, and ask each one whether he’s been tickled, the probability of a FTR (false tickling report) is exactly the same. (Because Fleepers and Meepers are equally honest)

However, if you pick one Fleeper and one Meeper and random from among those people who have made a TA (tickling accusation), the probability is different. (Because background statistics matter).

22

Since the OP has been answered and only debates are left, let’s move this at the request of the OP(and so my head stops hurting).

samclem

23

You know, I think i’ve finally got how this works. Hooray!

24

Yes, but we aren’t picking anyone at random. If we were, we’d be making a prior prediction, just like rolling a dice with 4 red sides and 2 blue and guessing which side is more likely to land up.

If Fleeper Smith called you up and said that he’d been tickled, the chances of him telling you the truth are equivalent to Meeper Brown.

Just like if we rolled the dice and Fleeper Smith told you that it was blue, he’d be just as credible as Meeper Brown saying the same thing.

Get it now?

25

Er, just to expand on that, I think i’ve found the area in which we’re arguing about, and it’s this;

psychloan is assuming the percentage of “likely to lie” is out of the entire population.

I was assuming the percentage of “likely to lie” was out of the accusing population.

So for psychloan’s method, he assumed that percentage was equal for both fleepers and meepers, since there are equal total amounts of fleepers and meepers. For my method, I assumed the percentage was different, since there are different amounts of accusers in each group.

So I think we’re both right; we just started off with different starting assumptions.

26

Actually, not only did he assume that the percentage of false tickling reports were equal for both groups, he also assumed that fleepers and meepers had equivalent numbers in the population, when the hypothetical was intentionally unbalanced in a way that didn’t match the equalities he was giving it. There’s nothing wrong with his math that I see, there’s only something wrong with the method in which he applied it (by making shit up that he needed to actually apply the theory) to the hypothesis that was given. He added assumptions that fit neither the hypothesis nor reality, and then used the math to attempt to demonstrate that it was accurate. If one adds their own parameters, one can come up with any result you want.

27

Then how are Smith and Brown selected?

It depends how they are selected (or how they select themselves).
How are they selected?

28

They select themselves. They make the claim and you are called to assess their credibility.

They call you up on the phone, laughing hysterically, claiming they were tickled mercilessly by a band of Meeper or Fleeper thugs.

29

Google ads:

Ostrich Feather Dusters
from $2 each. 100+ designs. Same day ship. Wholesale & Retail
snerk

30

you with the face, I do understand what psychloan is attempting to do, but since he had to add made up parameters to accomplish it, resulting in an answer that is obviously far removed from reality, it’s got you rather upset, especially since we were attempting to answer a real world question. So, I have a suggestion. I’ll set up a scenario in which he doesn’t have to add any assumptions, and it will be entirely based on real world data where possible.

For the purposes of this hypothetical, we’ll limit our group to females that self-identify as only white or black, and we’ll base it on the North Carolina numbers for these groups, since that was what raised the question in the first place. Keep in mind that even with the actual numbers, this is a thought experiment, and there is no general consensus that Bayesian Inference is going to give us a valid picture. Let’s just give it the proper tools to do any painting at all.

Given:

2,964,000 white females
872,000 black females

1,398 actual white rape victims
790 actual black rape victims

Since there are no generally agreed upon statistics as to the frequency of false rape allegations, we’ll pick a number between the two most frequently cited extremes and go with 25%.

1,864 white women who claimed rape
1,053 black women who claimed rape

466 white women out of 1,864 were lying
263 black women out of 1,053 were lying

psychloan, if you require any other parameters, just ask and I’ll be happy to provide them, using real world equivalents if possible.

If anyone has issue with the parameters laid out, or if anyone else thinks any are missing, let’s iron it out.

31

But how? By what criteria? Why is it that those particular people chose to call and other people didn’t?

32

Why do some women report rape and some don’t?

33

There’s no need. You are assuming that the rate of false reporting is proportional to the rate of rape for each group.

In other words, you are assuming that someone who is more likely to be raped is also more likely to make a false accusation.

In still other words, you are assuming that the group that is victimized more is also less honest.

I think your assumption is silly. Feel free to disagree.

(By the way, I doubt that the statistics you give show the number of inter-racial rapes, which is really what we are talking about here. )

34

And just because I’m masochist, let me continue:

There are three types of people in the total population.

  1. Those who, when tickled, will report being tickled. (Honest)
  2. Those who, when NOT tickled, will not report being tickled. (Honest)
  3. Those who, when NOT tickled, will report being tickled. (Liar)

After a claim is made and guilt or innocence is determined, then you’ll know which category the claimant belongs to. Group 2) people don’t really count though, since if you have a claim, you automatically know they are not in that group.

When an allegation is made, there are only two types of people that are relevant:

  1. Those who were tickled and reported it (Honest)
  2. Those who were NOT tickled, but reported it (Liar)

It goes without saying that if someone is not tickled, they won’t allege an attack if they are honest, and therefore there will be no allegation.

Using psychloan’s assumptions, the same fraction of people in both subpopulations are liars (25%). The remaining 75% of both populations are therefore honest.

We know that the subpopulations have different levels of risk for being attacked, but that doesn’t matter. When confronted with a claim, there’s no reason to factor in the probability that the given complainant was attacked. Only the probability that they are telling the truth. This is identical for both groups.

35

Look, I know you mean well, but I’m sick of the insinuations that I’m upset or crazy or hysterical because stuff goes against my worldview.

I’m not any more upset than anyone else who has taken issue with psychloan’s “math”. I understand perfectly well how he derived his answer, but I take issue with his assumptions. Just as you have done. Just as Scylla has done. Just as Hentor has done. I don’t get why people are accusing me of being emotional, when I’ve projected nothing that no one else has.

Like I said, I know that you mean well. But this “upset” business really has me scratching my head.

36

If you have any cites showing otherwise, feel free to bring them to the table.

This is a thought experiment, so we work with what we have. Again, if you have better information, let’s see it and we can change the parameters accordingly.

I’m just throwing up actual numbers, with the single exception for false rape claims, for which we don’t have actual numbers. I split the difference between two extremes to get them.

Less silly than throwing in made up numbers on top of numbers that were already made up. Here, we took everything possible from the real world.

Unless you can show that people lie about inter-racial rape at a different rate than rape in general (good luck with that), I don’t see why that matters for this.

Willing to do the math?

37

I didn’t assume crazy or hysterical, but I did assume upset. That was likely projection, as I was rather annoyed at the “it happens less often so we shouldn’t believe them equally” crowd. I’m sorry for the mischaracterization.

I still want him to perform this on what are as close to actual numbers as we can find. For some reason, he seems hesitant to do so.

38

Let’s say you work at the sheriff’s office. On Monday, August 14th, Fleeper Smith calls 911 and reports that he was tickled on the corner of 45th and Crenshaw by a gang of Meeper thugs.

Two hours later, Meeper Brown calls 911 and reports that he was tickled as well.

Who is more credible?

39

My assumption is that both groups of women are equally honest.

In other words, I have assumed that women of both groups are equally likely to fabricate a claim of rape.

I believe that this is a natural assumption. Further, I believe that in any discussion of race and behavior, the burden of proof is on those people who claim that races behave differently.

My opinion only.

40

Why would anyone report tickling to the sheriff’s office or by calling 911? The appropriate agency to report tickling to is the Tickling Authority. I will assume that you meant to say that I work for the Tickling Authority and the reports were made to me there.

The answer is that Meeper Brown is more credible. For reasons that have been explained numerous times.