If you were to take the latest Quinnipiac University poll at its face value (as most pundits, media outlets and commentators have) you’d come away with a data-supported opinion of the political moment. And in mid-July of a presidential election year, that’s about as good as it gets. But if you were looking for specific trends, say of the Latino electorate – the poll found that Latino support for Preasident Obama has eroded to 59% while Mitt Romney’s had increased to 30% – you’d more than likely come away with a false idea of that same political moment.
Here’s why: most national polls treat the Latino electorate as an “also asked” afterthought. And that afterthought comes at the tail end of the increasingly difficult task of finding Americans to poll. Bear with me while I cite a couple of reasons.
In May of this year the Pew Research Center published a paper in which they explained how difficult the business of opinion polling had become. Over the years the work of contacting people and getting them to respond to polls has gotten harder:
The percentage of households in a sample that are successfully interviewed – the response rate – has fallen dramatically. At Pew Research, the response rate of a typical telephone survey was 36% in 1997 and is just 9% today.
The reason given is the ubiquity of cell phones and caller ID. This difficulty is addressed, assessed and factored into the final polling results, somehow, so that the numbers are ultimately weighted. But the fact that the universe of poll respondents has diminished is important.
And that brings me to a second point.
The Latino slice of the Quinnipiac poll consisted of 143 persons, that’s according to NBC Latino reporting. So what Quinnipiac did – which is what almost all of the national polling organizations do – is they took those respondents in their poll (a total of 2722) who self identified as Latino, bunched them together, and called that a national Latino poll. NBC Latino puts the margin of error at about 8. Which means that the pro-Romney approval among Latinos, according to that poll, is anywhere between 22 and 38 percent.
I’m interested in a national poll that targets Latino voters across the country – and only Latino voters, not a sub-strata of a larger universe of 9% respondents. I’d like to see a polling organization go into the Latino community across the USA, take a scientific sample of the people who live there, ask them what they feel about a variety of issues and publish the results. Not as an afterthought of a larger poll.
Anything less than that, as far as I can see, is misleading.
[Photo by DonkeyHotey]