Conditional probabilities and polling analysis

Downscale voters like Clinton.
Latinos like Clinton.
How much more do downscale Latinos like Clinton than downscale Anglos?
And why do political analysts never seem to get beyond simple cross-tabs?

We know a few things clearly about the voter split in the Clinton-Obama race:

* Obama has a big edge among African-Americans.

* HRC has a smaller edge among women.

* HRC does very well among Latinos.

* Obama has a big edge among the 18-29s; HRC has a big edge among the over-65s.

* For education, income, or political awareness, there’s a break-point above with Obama wins and below which HRC wins.

But there’s what the statisticians call an identification problem. People with more education also tend to have higher income and greater political awareness. Latinos tend to have less education, less money, and less political awareness than Anglos. So which factor is really doing the work here?

We’d like to look at the off-diagonal cases: people with low income and education but high political awareness, for example. If they vote the same way as their low-political-awareness neighbors, then the natural heightening of political awareness as the campaign season progresses won’t help Obama in that demographic. On the other hand, if he does better among high-awareness than low-awareness voters even controlling for income and education, then you might expect his downscale disadvantage to shrink over time. [It’s easy to forget now, but as of December HRC’s downscale advantage extended to African-American voters.]

The same is true for the Latino vote (even accepting for the moment that it makes sense to lump Puerto Ricans in the Bronx with Mexican-Americans in East LA into a single voting bloc). It tends to be downscale and low-awareness. If we take a downscale, low-awareness voter, how much more likely is that person to vote for HRC if we then add the fact that he or she is Latino? How did Obama do among high-political-awareness Latinos in California yesterday?

Or look on the other side. Huckabee gets low-income, poorly-educated, rural, conservative, high-religiosity voters. It’s easy to lump that all together into “the Evangelical vote,” but how much better does Huckabee do among rural high-school graduates with no college who are Evangelicals than among rural high-school graduates with no college who are not Evangelicals?

I understand this stuff is too complicated to put in newspaper stories, but I’ve never even heard about it in backstage gossip. Surely those multi-million-dollar pollsters who lead our politicians around like tame dogs have heard of factor analysis and multiple regression. Or at least conditional probability.

Haven’t they?

Author: Mark Kleiman

Professor of Public Policy at the NYU Marron Institute for Urban Management and editor of the Journal of Drug Policy Analysis. Teaches about the methods of policy analysis about drug abuse control and crime control policy, working out the implications of two principles: that swift and certain sanctions don't have to be severe to be effective, and that well-designed threats usually don't have to be carried out. Books: Drugs and Drug Policy: What Everyone Needs to Know (with Jonathan Caulkins and Angela Hawken) When Brute Force Fails: How to Have Less Crime and Less Punishment (Princeton, 2009; named one of the "books of the year" by The Economist Against Excess: Drug Policy for Results (Basic, 1993) Marijuana: Costs of Abuse, Costs of Control (Greenwood, 1989) UCLA Homepage Curriculum Vitae Contact: