Polls fail on measuring racism

As the pollster for Florida’s first elected Hispanic governor (Bob Martinez, elected in 1986) and first Hispanic U.S. senator (Mel Martinez, elected in 2004), I have been called upon to try and measure the electoral impact of race through opinion surveys. 

In those and other campaigns in which my client faced a black or Hispanic opponent, it was supposed that polling could reveal the hand of racism in shaping votes. 

My experience leaves me skeptical that polls do a very good job of gauging complex attitudinal constructs like racism and sexism.  

{mosads}Sure, I believe that racist beliefs and attitudes exist, but I’m not sure that fixed-choice poll questions can figure them out.

An old professor of mine once suggested that the best way to measure any opinion is simply to ask a direct question. 

If you want to know whether someone will vote in the upcoming election, just ask them, “Are you going to vote or not?” 

Well, most of us in the polling trade just don’t subscribe to such elegant simplicity. 

We don’t believe that a poll respondent would tell the truth, especially about a sensitive topic like race and racism. So while we could ask, “Are you racist?” we try and come at it indirectly — to “trick” the respondent into revealing his or her real opinions. 

In the 1980s, I would not ask Floridians whether they would vote against Bob Martinez “because he is Hispanic.” I assumed they wouldn’t come clean. I thought I was clever in asking voters whether they knew others that would vote against him on account of his being Hispanic. 

The fact that many Floridians readily admitted to knowing someone so prejudiced seemed to validate both the existence of racism and the indirect approach to measuring it. 

But, to be honest, I didn’t really find the exercise wholly satisfactory.

Looking at polls about racism conducted a year ago, when the George Zimmerman case first gained national attention, and recently, during and after the trial, I am left with a similar sense that racism is not easily or satisfactorily measured with polls. 

In particular, I worry that polling about race topics creates opinions that didn’t exist before the questions were asked. 

The assumption of polls is that we are measuring opinions that existed before the respondent took the pollster’s call or put hands to keyboard to do an online survey. 

My own sense of this topic is that the vast majority of Americans does not sit around thinking about race all the time and consequently do not have a well-formed belief system when it comes to racial topics. 

Now, I know serious professionals who disagree. A well-known expert on opinion formation who lives in Atlanta has tried to convince me that residents of his metro area think about race all the time, and that advocates must always acknowledge this reality in campaign communications. 

Perhaps. But I am not completely convinced. Are there racists? Sure. But are they pervasive? No.

Lately, I am interested in going beyond polling to look at online discussions to identify and understand racist attitudes. 

Generally, online discussants in public forums believe they are anonymous, and thus presumably say what they really think. And, typically, no one has solicited their participation. They chose to join in the discussion voluntarily. 

The liabilities to this approach are that the locus of the forum affects results. 

I have read discussions of the Zimmerman verdict on a major daily newspaper site, a conservative media outlet site and a sports site, and they were understandably all different. 

But read enough of these focus groups online, and you start to see things. While discussion boards fail to provide insight on non-posters’ opinions, they might be the best way to understand the chattering class.

David Hill is a pollster that has worked for Republican candidates and causes since 1984.{mosads}