Thread replies: 8
Thread images: 2
Thread images: 2
Hey /sci/,
a friend of mine told me recently that racial profiling leads to more crime in the profiled group. His argument was basically that the crimerate would always be bigger in the smaller group because you're inspecting all elements in this group while you're not inspecting all elements in the other group.
Anyway that got me thinking so I made a rather simply model for racial profiling thinking that I have two groups W and B with [math]E_B = |B|*P(c)*(k_b)[/math] and [math]E_W = |W|*P(c)*P(k_w)[/math] where P(c) is the probability of someone being a criminal and P(k_b) is the probability that someone is being checked by the authorities. E is my expected Value. Thus [math]%_b = \frac{|B|*P(k_b)}{|B|*P(k_b)+|W|*P(k_w)}[/math] becomes the percentage in the crime statics, which does not depends on P(c), because I assume that P(c) is the same for both groups. With some simple math I can reduce this further to [math]%_b = \frac{B}{B+(1-B)*V}[/math] with B being the percentage of people of group be living in an area and V being the proportion that the people in the W are checked less then the B group. The funny thing is you can recreate the crime statistics for all US states and for the US as a whole with this formula by only entering the percentage of blacks living in that state and the ratio that blacks are controlled against whites.
This is pretty neat, huh?
>>
Uh errm, this is missing.
>>
>>8539086
That is neat
>>
>>8539086
Wait so you made an equation that gives you how more likely blacks have to be caught for the same crime ration compared to whites assuming both are equally likely to be criminals?
>>
>>8539109
>blacks have to be caught
not caught, just checked. Let's say B_c and W_c are the number of people you check all you need is that [math]B_c/|B| > W_c/|W|[/math] and blacks will be over represented in the crime statistics, like they are.
>>
>>8539086
P(c) shouldn't be the same for both and it isn't independent of P(k_w), pretty neat that it fits so well, but you can conclude a lot from that bro.
>>
>>8539141
Do you take the ratio of Blacks being checked by police versus White people from some statistic? Or do you just take ratio of criminal blacks to whites and fit [math]B_c[/math] and [math]W_c[/math] to the statistic?
>>
Do you have the data?
Thread posts: 8
Thread images: 2
Thread images: 2