To do so I examined several sources of data. For statistics relating to the general population, including income by race, population by race, and the general population total I used census statistics from Census.gov. For statistics relating to incarceration and the prison population including race, gender, and income I used statistics from the Bureau of Justice Statistics. All statistics are from 2014 and since the census and bureau of justice statistics used different income partitions data within each group was considered uniform for interpolation. Using these statistics I calculated the expected value for racial incarceration using income as the predicting variable and the expected values for income based incarceration using race as the predicting variable. The full results can be found on the following sheet.
Upon reading the above paragraph one might ask themselves why go through all this math. The theory behind the calculations are as follows. If class trumps race then the estimation of incarceration rates by race (which were made using class) will be more accurate than the estimation of incarceration rates by class (which were made using race). If race trumps class the opposite would be true. This works because if race trumps class then it should have a greater effect on the prison system and thus be able to more accurately predict aspects of said prison system. To measure the accuracy of a measurement I calculated the percent error using the actual BJS numbers as the expected values. I summed the absolute value of the percent errors and used that to compare the accuracy of each method. So without further adieu I can report that race trumps class 119-191. Further data can be seen below.
No comments:
Post a Comment