Not Trolling Zaku-Illegal Immigration

[Docfeelgood]

Plus they have lower rates of criminality than do native born Americans

If you live in Alaska

[Darius871]

Plus they have lower rates of criminality than do native born Americans

I wouldn't find that particularly surprising if true, but what's your source on this? It strikes me as something that'd be difficult to quantify.

[KrazyHorse]

It's not particularly difficult to quantify incarceration rates. Criminality rates are a little more difficult, but the gap in incarceration rates is so big that sampling bias etc. would be hard to offer as an explanation.

For all immigrants (legal and illegal) the incarceration rate was 4-5 times smaller than among native-born Americans (controlling for age and sex). Comparing only Hispanics to Hispanics yields incarceration rates for native-born Americans ~8-9 times larger than for immigrants.

These numbers are robust and are based solely on US Census data. There are a number of studies which attempt to determine actual criminality rates, but they are less trustworthy.

You should note that illegals are ~1/3 of the immigrant population to the US and these are overwhelmingly Hispanic, so there's no realistic way to claim that the distinction between illegal and legal matters very much when you're comparing criminality rates to those of native-born Americans.

[Drake Tungsten]

When did geology become an "advanced science job"?

[Darius871]

It's not particularly difficult to quantify incarceration rates. Criminality rates are a little more difficult, but the gap in incarceration rates is so big that sampling bias etc. would be hard to offer as an explanation.

For all immigrants (legal and illegal) the incarceration rate was 4-5 times smaller than among native-born Americans (controlling for age and sex). Comparing only Hispanics to Hispanics yields incarceration rates for native-born Americans ~8-9 times larger than for immigrants.

These numbers are robust and are based solely on US Census data. There are a number of studies which attempt to determine actual criminality rates, but they are less trustworthy.

That's more or less my point; I just wanted to make sure you weren't conflating incarceration rates with criminality rates. Because illegal immigrants are far less likely to have a permanent address, documented job, documented relatives, or any other sort of reliable documentation that police could use for that matter, and because they would tend to hide in a relatively insular community less likely to cooperate with police, and because they would probably be involved in more low-level street crime well below skilled detectives' radar, among other things, I'd imagine differences in the diligence and efficacy of police investigation would create an enormous sampling bias. I don't disregard incarceration rates entirely, but this is one particular area where they are especially useless.

Just as importantly, I'm surprised "legal and illegal" wasn't a dispositive problem for you, when legal immigrants (with so much to lose) would understandably be very law-abiding, and when there are approximately 37 million legal immigrants already here versus a mere 12 to 20 million legal immigrants. I highly doubt that "Hispanic" is a very accurate way to cleave those two categories apart, given that nearly half of even legal immigrants are Hispanic. [EDIT: Even after your DanS this point still stands.]

[KrazyHorse]

More importantly, I'm surprised "legal and illegal" wasn't a dispositive problem for you, when there are approximately 37 million legal immigrants already here versus a mere 12 to 20 million legal immigrants. I highly doubt that "Hispanic" is a very accurate way to cleave those two categories apart, given that nearly half of even legal immigrants are Hispanic.

37 mill is total immigrant pop. Not legal immigrant pop. 25 is legal immigrant pop.

[Darius871]

37 mill is total immigrant pop. Not legal immigrant pop. 25 is legal immigrant pop.

I just relied on wiki's "[l]egal immigrants to the United States now are at their highest level ever at over 37,000,000 legal immigrants" for that, but perhaps they included in that figure legal immigrants who were later naturalized. Even if the census' definition of "legal immigrant" didn't include the naturalized and your figure's correct, it still means a sizable majority of your sample is not comprised of the category in question. And it doesn't change the fact that nearly half of legal immigrants are Hispanic, which make it an unreliable factor to separate legal from illegal.

[KrazyHorse]

The point of me saying that it wasn't particularly important for attempting to compare is that even if you assume that all(!) crimes by immigrants are committed by illegals you still get that the incarceration rates for illegals are lower than for native-born Americans.

When you break it down by race, the the picture becomes even clearer; if you assume that all crimes by Hispanic immigrants are committed by illegals (and that all illegals are Hispanic, which is actually a pretty good assumption) then you get a better upper bound for incarceration rates of illegal immigrants, and this improved upper bound is significantly lower than the incarceration rates for native-born Americans.

[KrazyHorse]

http://www.census.gov/population/www/socdemo/foreign/acst2.html

lists foreign-born pop (should be both illegal and legal) of US in 2003 at 33.5 mill

[Darius871]

The point of me saying that it wasn't particularly important for attempting to compare is that even if you assume that all(!) crimes by immigrants are committed by illegals you still get that the incarceration rates for illegals are lower than for native-born Americans.

When you break it down by race, the the picture becomes even clearer; if you assume that all crimes by Hispanic immigrants are committed by illegals (and that all illegals are Hispanic, which is actually a pretty good assumption) then you get a better upper bound for incarceration rates of illegal immigrants, and this improved upper bound is significantly lower than the incarceration rates for native-born Americans.

This still leaves out the blatant sampling bias. I don't see any reason to think that police would be anywhere near as diligent or effective at catching undocumented illegals for petty thefts and assaults as they would be at catching citizens who have accumulated mountains of documentation instrumental to police finding them. I'm saying this as someone who has spent years trying to track people down on a daily basis and who would be an utterly hopeless babe in the woods when dealing with someone who was born in rural Mexico, lives in a basement, and works for cash. He is essentially a ghost, which would explain why he'd be tougher to "incarcerate" than citizens, regardless of whether or not he has engaged in criminal acts or how many. As a practical matter that much should be obvious.

[KrazyHorse]

http://www.census.gov/population/www/socdemo/foreign/faq.html

says that illegals are part of foreign-born

http://www.census.gov/population/socdemo/foreign/ST023/tab3-1a.xls

gives total number of foreign born from Latin America as 17.5 mill.

This actually means that Hispanic immigrants are more likely than not to be illegal (~2-1 margin) which means that our upper bound from Hispanic incarceration rates is actaully going to be better than I thought previously (I was basing it on 1-1 illegal/legal)

[KrazyHorse]

This still leaves out the obvious sampling bias. I don't see any reason to think that police would be anywhere near as diligent or effective at catching undocumented illegals for petty thefts and assaults as they would be at catching citizens who have accumulated mountains of documentation instrumental to police finding them.

I don't disagree with you that there is a sampling bias present in incarceration rates. I do question whether or not that bias is going to give you the factor of 3 or more that you need to explain incarceration rates without admitting a differential in criminality rates.

[KrazyHorse]

By the way, arrests for theft and other fairly trivial crimes are overwhelmingly made at or near the scene of the crime within a couple of hours of the crime occurring.

[Darius871]

This still leaves out the obvious sampling bias. I don't see any reason to think that police would be anywhere near as diligent or effective at catching undocumented illegals for petty thefts and assaults as they would be at catching citizens who have accumulated mountains of documentation instrumental to police finding them.

I don't disagree with you that there is a sampling bias present in incarceration rates. I do question whether or not that bias is going to give you the factor of 3 or more that you need to explain incarceration rates without admitting a differential in criminality rates.

Having experience in these matters as a former debt collector and current investigator for a federal agency, I would posit that a factor of 3 would be surprisingly low. Obviously it'd be incredibly high when comparing blacks to whites, for example, but in the peculiar case of undocumented(!) suspects who have faded into a sea of people separated by language, frankly it's amazing police could manage to incarcerate any that aren't right at the scene, let alone proportionally as many as citizen suspects. It's not like we'll ever know for sure, though.

[KrazyHorse]

Darius, your experience is with the people who the police:

a) Did NOT catch on the streets
b) DID collect a significant amount of accurate, identifying information about

Your experience with sampling bias is hugely biased.

Most simple assault and theft cases in which arrests are made do NOT fall into those categories. Instead, the suspect is caught because the cops find him within a few blocks of where the crime occurred, and he's identified by a victim or witness.