February 07, 2008

What Immigration Did For the Romney Campaign—The Numbers

By “An Economist”

[VDARE.COM NOTE: Mitt Romney has withdrawn from the GOP presidential race, and John McCain appears to be winning. The usual suspects at the Wall Street Journal and other immigration enthusiast publications can be expected to claim that this because of the fact that Romney displayed the normal American attitude towards illegal immigration, which is that it's illegal. This is an analysis of how voters responded on this one issue.]

A number of Republican primaries have now been completed. Based on the CNN exit polls from these primaries it appears that in all but one case, Romney gained strongly from his position on illegal immigration. The average gain for 11 states (AL, AZ, CA, FL, GA, IL, MA, MI, NH, NY, OK) was 5.5% using one method and 7.21% using another. These are actual percentage gains. For example, in New Hampshire Romney gained either 9.66% or 12.73% depending on what method was used. This was roughly 1/3rd of his total vote in New Hampshire. Conversely, Romney appears to have lost Latino votes in a few states. Only in Florida were his losses substantial, but the reasons are unclear. See the chart with the raw exit polling data and the calculations.

All these calculations are based on heroic assumptions. However, they are not unreasonable. To calculate Romney's gains from the illegal immigration issue, I compared his results with immigration voters versus all voters and non-immigration voters. Immigration voters are defined as those who stated that illegal immigration was their most important issue. All voters are defined as everyone who voted in the Republican primary. Non-immigration voters are everyone who ranked Iraq, the Economy, or Terrorism as their most important issue.

An example may help here. In New Hampshire, 23% of voters ranked illegal immigration as their number one issue. They gave Romney a 37% edge versus McCain. By contrast, overall Romney lost by 5%. This means that Romney gained 42% versus McCain on the immigration issue. Multiplied times 23% that is an overall gain of 9.66%.

However, this may not be the complete story. If you look at only non-immigration voters (Iraq, Economy, Terrorism), McCain beats Romney by 18.36% in New Hampshire. Using non-immigration voters as the starting point, Romney actually gained 55.36% versus McCain on the immigration issue. Multiplied times 23% that is an overall gain of 12.73%.

Of course, just because a voter listed illegal immigration as his most important issue, doesn't mean that he favors Romney's approach. However, given that Romney did much better among these voters that is a reasonable conclusion. Note that in general, voters who ranked Iraq as their most important issue favored McCain.

The reverse question is how much Romney lost among Latino voters. To calculate Romney's Latino losses, I compared his results with White voters versus Latino voters. Once again I am making a key assumption—namely, that any shift against Romney among Latino voters is entirely a consequence of his stand on illegal immigration. For a number of reasons this is probably not correct. Latino Republicans are much more liberal than general Republicans, so Romney could be expected to lose Latino votes in any case. Nonetheless to be conservative, this is the approach I used.

An example may help here. In California, Latino voters were 14% of Republican voters (highest of any state I checked). White voters were 74% of Republican voters. Romney lost Latino voters by 16% versus McCain. However, he also lost White voters by 5%. This means that Romney might have lost 11% of Latino voters because of his position on illegal immigration. Multiplied by 14% gives a net loss of 1.54% among Latino voters. By contrast, Romney gained 8.12% or 12.32% based on his stand on immigration overall.

The one case where Romney may have net lost because of immigration is Florida. He gained either 3.68% or 4.2% because of his position on illegal immigration. This number is partially smaller than other states because only 16% of Florida voters ranked illegal immigration as their number one issue. Even Florida voters who ranked illegal immigration as their number one priority shifted less to Romney than in other states. Conversely, using the standard approach described earlier, Romney lost 4.92% among Latinos because of immigration.

This sounds bad, but the details are murky. It turns out that 11-12% of all Florida Republican voters are Latino. These voters are 7% Cuban and 4% non-Cuban. Logically, one might expect non-Cuban Latino voters to be much more interested in immigration than Cuban Latino voters. After all, Cubans have automatic legal status when they arrive in the US. However, this was manifestly not the case. Cuban voters gave McCain a 45% edge over Romney (Romney got 9% of the Cuban vote). By contrast, non-Cuban Latino voters favored McCain by 32% (Romney got 21% of the non-Cuban Latino vote).

Why were Cubans so much more favorable to McCain? Are they more interested in Amnesty? More concerned about Iraq? More generally liberal? Or are they more Spanish oriented and less interested in assimilation? Given that Miami is a non-English speaking city, there may be some truth to this. However, any real answer can not be determined from the available exit polling data.

Overall, it appears that Romney did quite well via the immigration issue. Indeed, it may have been a strong winner in every state including Florida. Obviously the small Latino share in the Republican electorate was a major factor. However, even in Arizona and California Romney enjoyed strong net gains. Since the Latino Republican share in California and Arizona is much greater than the national Latino share for all voters (both parties), it looks like illegal immigration would be a strong issue for any Republican in a national election.

Note that in several cases, the gain to McCain from Latinos is shown as zero. This may not be correct. However, the exit polling data did not break down the Latino vote in these cases. Given that the Latino vote was small in these states, this doesn't materially change the results.