The other angle on this I’ve seen multiple times is comparisons of crime rate per capita for populations segmented by migrant status like this, where the usual conclusion is “A person is less likely to commit crime, given that they are a migrant”. It seems pretty believable to me that this reflects a raised background level of caution around crime caused by migrant status.
This does depend on the methodology stated by the institution collating the data. The most recent methodological statement I can find from the FBI’s stats reporting indicates it uses locality population estimates derived from the federal census - which in turn aim to count not the people each locality has registered as a resident, but the number of people who usually live and sleep in a locality regardless of documentation status.
Obviously, undocumented migrants may consciously attempt to be undercounted in that divisor but they don’t necessarily disappear entirely.
The other angle on this I’ve seen multiple times is comparisons of crime rate per capita for populations segmented by migrant status like this, where the usual conclusion is “A person is less likely to commit crime, given that they are a migrant”. It seems pretty believable to me that this reflects a raised background level of caution around crime caused by migrant status.