Re: Cops 4: The Good, the Bad and the Ugly
You're the one who used the five coin flip false equivalency to start us off so yeah...
Your point on it only being 80 cases is well taken and yes there's probably a bit of variance from the average due to it being a smaller sample and not random. But it's still incredibly far off from the average and knowing how these cases always unfold there's zero reason to assume the most drastic variance in your favor. You're also ignoring the grand jury part, grand jury ---> trial happens the vast majority of the time (something like 99%) and that number is definitely not the same for police.
Ok one more try.
The point about it being only 80 is minor to the problem. I showed how selecting 500,000 coin flips out of a million would still give you a wrong result. But I'll try a different tack: Let's say you know as a fact that all mammals average 8 hours of sleep per day. Using the same reasoning you're using here you commence your own study of giraffes
expecting naturally, that they will sleep about 8 hours a day as they are mammals. Much to your dismay, the giraffe's you study must all be sick because they are only sleeping 3.5 hours a day!
What is wrong here? "Well Wisko, you idiot, I have you this time because I sampled 20,000 giraffes and therefore have a 99% confidence level +- 1! Ha! And they are all mammals and they are all suffering sleep deprivation! Call the ASPCA you goon!"
Except all giraffes do in fact only sleep 3.5 hours a day and they are fine. The EXPECTATION that they would sleep eight is based on the faulty notion that there is a relationship between the average number of hours all mammals sleep, and the number of hours the subset 'giraffes' sleep -- because statistics! But there isn't a relationship like that.* For example the subset 'brown bats' sleeps 20 hours a day. A probability relationship
would exist if you took
a random sample across all mammals, but
not if you just sample giraffes. There is not just "a bit of variance", there is no effective relationship. The correct number of sleep hours for an eastern spotted skunk could be anything!
Here we are taking the subset 'cops' (plus, no doubt, other variables) excluding all others and EXPECTING a result that is equivalent to the average result of all other cases. Why? Is it because that fits in with what we think it should be? That's fine,
but it doesn't come from the numbers.
I suspect the conviction rate for 'alienation of affection' is quite low these days. DUI conviction rates are fairly high. The best we can say right now from the data presented is that a cop has a roughly 1/3 chance of being convicted in a jury trial. And even that is suspect.
*Theoretically, you possibly could come up with a formula that estimates values for all giraffe variables and spits out the generally correct answer. But we're not doing that for giraffes and certainly not for all the variables connected with high profile court cases in front of juries. If we could, the answer for cops may be 20%, (like giraffes) or 90% (like brown bats). We don't know!