Immigrants use less welfare than natives, holding income constant.  Immigrants are far less likely to be in jail than natives, holding high school graduation constant.*  On the surface, these seem like striking results.  But I’ve heard a couple of smart people demur with an old statistics joke: “Controlling for
barometric pressure, Mount Everest has the same altitude as the Dead Sea.”  Sometimes controls conceal the truth rather than laying it bare.

Who’s right?  Does adjusted quality matter?  Or is it just a bait-and-switch?

It all depends on what your audience takes for granted.  If listeners falsely assume immigrants are just as welfare-dependent and criminally-inclined as comparable natives, the adjusted results provide new and valuable information.  If reasonable, they may not become pro-immigrant, but at least they should become less anti-immigrant. 

This point is even stronger, of course, if listeners falsely assume immigrants are more welfare-dependent and criminally-inclined than comparable natives.  As far as I can tell, 90% of native-born Americans angrily believe both negative generalizations.  If they would scrupulously face facts – adjusted facts – much of their anger and desire to “do something about immigrants” would dissolve.

Pointing out that immigrants are better than comparable natives is directly analogous to, say, pointing out that the much-maligned Ford Pinto was not unsafe for a compact car.  In both cases, false beliefs lead to foolish actions – scapegoating immigrants and Pintos when they’re at least as good as comparable natives and comparable cars. 

* There is also good evidence that immigrants commit less crime, making no statistical adjustments at all.  But that’s a separate point.