Ferris Bueller

Another thing, and while certainly not condoning any kind of behavior, there are exceptions to the rule, especially with sex offenders. One of my best friends growing up is a convicted sex offender. While the label on him earns him a huge degree of dirty looks and stigma, he doesnt deserve it any more than any of us do.

Why? Simple: how many of you had sex before you were 18 with someone who was also under 18? I had an encounter or two with girls I was in high school with. So did he. Difference is that he got caught by the girl's father, who pressed statutory rape charges. Needless to say, my friend lost the case and is now on the list of sex offenders. Interestingly enough, he married the girl he was sleeping with, and they're still married to this day, 8 years later.

So yes, while some of those people deserve to be jeered at and regarded harshly, be careful of whom you judge and for what reason.

__________________
|TG-6th|Ferris Bueller

Kerostasis

@ Tybalt:
While Bayes' theorem probably is applicable to this scenario, you need to clarify your use of "99% accurate" more before it can be used. Do you mean that of all people looked at by the courts, 99% of them will have the correct verdict issued? Thats the scenario that would produce your 33% statistic, but I don't think thats what you intended when say there is a "1% false conviction rate".

The courts accuracy needs to be described by 2 numbers to be complete -- the % chance that an innocent man will be convicted, and the % chance that a guilty man will be set free. These numbers are probably different. And assigning a chance of 1% for each innocent man to be convicted, in the way you used it statistically, refers to 1% of ALL US Citizens, not just people brought in front of the court.

I couldn't say for sure that this is incorrect, but you are definately not giving the court the benefit of the doubt with that prediction -- thats an extremely high error rate. It might be accurate to say that 1% of innocent men actually tried in court are falsely convicted, but that gives an entirely different statistical result because now you can't use your "1 in 200 citizens are child molestors" as the other half of the Bayes theorem. In this scenario we dont care how much of the general population is a child molestor, we just care about what portion of the population of citizens on trial for child molesting are child molestors. And that will be quite a bit higher than 1 in 200.

So using some new stats: Lets suppose that 99% of innocent men are set free, and 99% of guilty men are convicted, and 50% of those accused of child molesting are actually guilty. Now a convicted offender is actually 99% likely to be guilty, rather than 33%. Marked improvement in accuracy, just by using the correct stats.

Now lets try some more cynical stats: 99% of innocent men are set free, 10% of guilty men are set free, and 40% of those accused are guilty. A convicted man is still 98.4% likely to actually be guilty.

__________________

xTYBALTx

Twelve year old registered sex offenders?

http://www.reason.com/blog/show/117140.html

__________________
Current Tybalt theme song: Mason v. Princess - Perfect

Rincewind

I did not mean that your example shows poor understanding of Bayesian math (it's a "copy paste" example from wikipedia). I meant that your understanding of how that example relates to finding a person guilty of a crime is flawed; it doesn't. Bayesian probability is relevant to prosecuting and defending a crime, but not in the manner you're referencing.

This only due to a lack of understanding in how bayesian mathematics should, or should not be used when looking at rates of criminality.

__________________
~~ Veritas simplex oratio est ~~
No matter how far a wizard goes, he will always come back for his hat. --T. Pratchett

<---- You know you're getting old when you rely on your forum meta-data to remind you how old you are.

Required Reading for all TG sandboxers

xTYBALTx

@ Kero - What impact would "false negatives" have on the 99%/1% number? I explicitly restricted my example to talk of "false positives."

__________________
Current Tybalt theme song: Mason v. Princess - Perfect

xTYBALTx

@ Rincewind - So where is my reasoning flawed?

__________________
Current Tybalt theme song: Mason v. Princess - Perfect

Rincewind

That sure sucks. But I don't see how this is relevant to your arguments or this thread in general.

__________________
~~ Veritas simplex oratio est ~~
No matter how far a wizard goes, he will always come back for his hat. --T. Pratchett

<---- You know you're getting old when you rely on your forum meta-data to remind you how old you are.

Required Reading for all TG sandboxers

Kerostasis

Sorry, I have a habit of continuing to edit my posts for clarity for a minute or two after I post it. Its more complete now.

For extra clarity: The key failing in your example was applying the 1% false judgment rate to all citizens instead of just those citizens actually accused of a crime. If you actually believe we pass false judgement on 1% of our entire population, then ok, but you expressed your estimate as a generous one rather than a cynical one:
1% false judgement on our entire population is a stretch allright, but its a stretch in the other direction.

Also, the Bayes Theorem requires use of False Negatives. You can't use it without statistics on the errors in both directions. So I'd say that evidences your mis-use of the theorem even more than your previous post.

__________________

Rincewind

Where is my assertion that you misunderstand incorrect?

And Kerostasis is providing some good info.

__________________
~~ Veritas simplex oratio est ~~
No matter how far a wizard goes, he will always come back for his hat. --T. Pratchett

<---- You know you're getting old when you rely on your forum meta-data to remind you how old you are.

Required Reading for all TG sandboxers

CingularDuality

Really? I'd like the link to his profile on the National Registry, please. Juvenile crimes aren't supposed to be registered.

Now, we can argue about 19 year olds with 16 year olds, and you'll probably find that I'm on the same side of the argument as you, but the fact is that we have to draw the line somewhere and hope that the justice system uses discretion in meting out punishment. All convictions for sex crimes do NOT automatically result in registration as sex offenders.

__________________

Become a supporting member! -- Shop at the Tactical Market!

Take the world's smallest political quiz! "I was touched by His Noodly Appendage."
TacticalGamer TX LAN/BBQ Veteran:

Rincewind

@ty

Your misunderstanding of how Bayes Theorem applies is evident (to me) from your use of the drug testing example off of wikipedia to promote your idea that some (exaggerated) percentage of pedophiles are falsely convicted (for arguments sake). This idea somehow fits with a bias you hold so you like it and run with it.

Kero is explaining nicely how Bayesian probability should be approached when looking at the probability of a false conviction. I've been told I'm only capable of explaining things meanly, so I'm lucky he's around

__________________
~~ Veritas simplex oratio est ~~
No matter how far a wizard goes, he will always come back for his hat. --T. Pratchett

<---- You know you're getting old when you rely on your forum meta-data to remind you how old you are.

Required Reading for all TG sandboxers

Arf

That's not as uncommon as you might think, around here we call that a shotgun wedding.

__________________

xTYBALTx

__________________
Current Tybalt theme song: Mason v. Princess - Perfect

xTYBALTx

Kafka is alive and well!

http://www.slate.com/id/2184798/

__________________
Current Tybalt theme song: Mason v. Princess - Perfect

Augustus Gloop

I just want to state for the record, that child molestation is a banning offense on the 2142 server.

__________________