Partisan effects of immigration reform

by Andrew Gelman on April 20, 2013 · 9 comments

in Immigration,Methodology

Alex Engler writes:

I just finished an article for the Georgetown Public Policy Review that you might be interested in. I took a thorough look at the influence of the Hispanic vote in congressional races in 2012, and how the party balance in the House changes under different levels of Hispanic party support. I used this to gain some insight into the prospects for immigration reform, and the results are really interesting. There’s also a few nice graphs and maps.

From the article: “the Republican Party has a great deal to gain from successful bipartisan immigration reform, House Democrats face little benefit and even, paradoxically, the possibility of significant losses.”

I have to say that I’m suspicious of analyses where the rebound is bigger than the main effect, but that’s just a hunch on my part, not a serious analysis.

{ 9 comments }

KJ April 20, 2013 at 10:17 pm

I’m skeptical of the article because it assumes that passing immigration reform could increase Hispanic support for the GOP, and I’ve simply not seen any evidence that suggests that would happen.

Nadia Hassan April 21, 2013 at 4:14 pm

Matt Bareto’s survey provides evidence.

Phil April 21, 2013 at 7:13 am

I agree with KJ. Why do (some) people keep on acting as if all it takes for the GOP to win over Hispanics is to be nicer to them on immigration (believing Reagan’s adage about Hispanics being Republicans who just haven’t realized it yet)? Hispanics are liberals when it comes to Obamacare, the economy, cutting the deficit (i.e. they want tax hikes for the rich) and helping disadvantaged people. All Republicans will be doing by supporting CIR is to increase the size of the Democratic electorate in the nearterm future. Maybe – and that’s a big maybe IMO – Republicans will reap the benefits 2-3 generations down the line when Hispanics have become more integrated into US society. Protestantism is far more widespread among 3rd generation immigrants than it is among first generation Hispanics for example. They are also far more likely to support a hands off government. But then again a lot can happen over the next 40-60 years anyway.

DavidT April 21, 2013 at 12:45 pm

“All Republicans will be doing by supporting CIR is to increase the size of the Democratic electorate in the nearterm future.”

Under the Gang of Eight proposal, the next presidential election the “legalized” immigrants would be able to vote in would be in 2028.

Nameless April 21, 2013 at 4:40 pm

The core finding here is that heavily Hispanic districts, for the most part, vote Democratic already, which means that, among these, there are many more vulnerable Democratic districts than vulnerable Republican districts.

That part is plausible. The suspicious part is that, according to their article (at least as far as I understand it), even if Hispanic turnout was increased in 2012 to match the average American level (for starters, that would involve legalizing all illegal immigrants and giving citizenship to all Hispanics), the number of seats gained by Democrats would have been … zero.

Naturally, if the main effect is zero, second order effects could go in any direction.

Current white turnout is at least 30% higher than Hispanic turnout and Hispanics vote Democratic 3:1. The idea that, in the entire country, there isn’t a single district where increasing Hispanic turnout by 30% would not flip the seat, seems incredible and makes me want to look at their data.

I hope it’s not another case of an Excel coding error.

Steve Sailer April 22, 2013 at 8:14 pm

Under the Voting Rights Act of 1982, numerous House districts are gerrymandered to be majority minority, which means that most Hispanics and blacks are rounded up into super-Democratic districts that are so liberal they’ll elect a minority Democrat. The Republicans like this because it helps Republicans win the more marginal districts, and it encourages nonwhite Democratic politicians to start their careers as race men, more Bobby Rush than Barack Obama, who thus later have a hard time winning state-wide office.

However, that’s all small potatoes compared to the GOP losing Texas’s Electoral Votes.

Nameless April 22, 2013 at 9:41 pm

Geographical distribution of voters in the U.S. is inherently skewed, with far more super-Democratic neighborhoods than super-Republican neighborhoods.

Here’s precinct-level data from Nov 2012 elections in one random county (Clark County, Nevada – Las Vegas and surrounding areas):

http://redrock.clarkcountynv.gov/electionresults/sov/12G/PRESIDENT_AND_VICE-PRESIDENT.txt

Excluding precincts with fewer than 100 voters and ignoring third party votes, the “reddest” precinct went for Romney 86%/14%, next three reddest went for Romney 84%/16%, 81%/19% and 81%/19%.

At the opposite end of the spectrum, we have not one but two precincts where Obama got 99%(!) of the vote, and 31 precincts more Democratic than the most Republican precinct in the county.

Even though Clark County leans Democratic overall, this pattern holds more generally.

One consequence is that, in the absence of gerrymandering, there is an inherent (slight) Republican bias in the geography.

Steve Sailer April 23, 2013 at 2:14 am

Judging from this precinct data, white voters tend to be more open-minded and nonconformist than nonwhite voters.

Nameless April 23, 2013 at 12:42 pm

I wouldn’t necessarily want to make value judgments about races based on this data. White voters can be very partisan too. (Precincts in UC San Diego went 85% to 90% for Obama, and they are full of white and some Asian students.) I’m just making an observation about geography.

Comments on this entry are closed.

Previous post:

Next post: