Re:Is prison contagious?
by rockpoodle on Thursday June 26, 2014 @11:41PM
(#7952652)
Your summary is completely false. Did you actually read the original paper or just the article in the link? I know not everyone has journal subscriptions to that might be too much to ask, but maybe you should not dismiss it entirely having read only a summary. All of the parameters in the model are taken from real data, there was no tuning. And there was no parameter in the model that determined how sentence length would affect population-level incarceration rates. The tipping point occurring between 14 and 17 months was an emergent property, not an input. You clearly have no understanding of simulation science in general or this study in particular.
Agreed, it looks flawed to me. The assumption being that the transmission rate is constant this will obviously lead to longer sentences = more transmission.
If a %age of your time is spent on each person you know then there will obviously be a maximum point for the likelihood of infection.
Then the modellers have to choose an infection rate, this they took from the data.
The whole model is begging the question.
Re:Is prison contagious?
by jessaustin on Thursday June 26, 2014 @08:14PM
(#7952190)
I don't really disagree with you, but think about it in reverse. If those responsible for sentencing are in a sense racist, wouldn't we expect them to eventually tune their actions to obtain racist outcomes?
"According to our model, the tipping point is somewhere between the 14- and 17-month sentences"
Surprise surprise! You put those numbers in and they came back out!
Or are we to believe that somehow society just randomly happened upon exactly the correct number to tip one race one way and one the other?
Their error check to California later doesn't help because that's part of the data they used to design their simulation in the first place!
And finding that sentence length alone explains everything is actually proof that this simulation is useless!
In the real world there is a huge difference in the number of single parent homes, and those without a good male role model between the races. Those are well known factors for crime, and the fact that there is no excess difference in the model that needs to be attributed to this shows that the model is just parroting their preconceived explanation.
Summary: They carefully tuned their model so that 14 months causes the incarceration seen by whites, and 17 months causes the rate seen by blacks. They did that for a good reason: In order to match the real world. But it's no surprise that those are exactly the numbers their model came back with.