Many states disenfranchise prisoners but prisoners are counted by the census as "residents" of the place in which they are incarcerated. This practice is particularly notable when it comes to congressional apportionment.
Because Congressional apportionment is done by population, the more residents you have the more congressional districts, and representation in Congress, that your state is likely to get. States with large prison populations get to count the prisoners as residents even though the prisoners are not allowed to vote. The upshot of this practice is the state to provide more representation for its voting citizens by using the prisoners who of course get no representation. The practice is even more pernicious in states that permanently disenfranchise ex-felons.
It seems that the Census Bureau is taking some steps to at least make this issue more transparent. See here.