I was reading this post about the possible strengths and weaknesses of Clinton and Obama among different demographic groups, and I grew very annoyed.
Not at Digby, but the whole debate. What's really frustrating about most voting demographic stories (besides the obvious, which is that they don't have much to do with actual governing) is that I have no way to evaluate the claims made. Sometimes there are bar graphs that show how one group compares to another (blacks vs. whites, old vs. young, etc.). On very rare occasion, there is a two-by-two table, but that's still not enough. If news organizations are serious about 'transparency' and 'openness', then release the data. After all, in many cases, these organizations paid for it: they can do whatever they want with it.
There are quite a few of us who, given a spreadsheet where each row is a voter and each column some information (e.g., age, gender, income, religion), could do a lot with this. We could actually evaluate the data ourselves. After all, you never hear about the relative importance of various factors--for instance, is race more important than economics? You also never hear about interaction terms--that is, how important are combinations of various factors (e.g., how different are poor white Protestants from middle class white Protestants).
So why don't they release the data?
- Log in to post comments
You make the mistake of assuming that there are things called "news organizations" that are interested in telling the news. In fact the main business of modern media is to sell advertising space and any means that efficiently does this is employed. If this happens to result in actual news being transmitted then this is more accident than design. This is why celebrity crimes are given such high prominence and why the threat of crime against the person are over emphasized compared to the actual statistics. If alien abduction stories resulted in an sustainable increase in their viewing figures then we'd no doubt have our nightly news regularly filled with tales of Billy-Joe-Bob and his night of anal-probing terror.
I work at a call centre where we conduct these sorts of polls. Most of our clients are businesses wanting to gather information on the success of their advertising, opinions on future advertising, or what people think of their competitors. Every so often, however, we do get work from news and government agencies looking to gather this sort of stuff.
I must say that most of the questions are very loaded and it usually becomes fairly obvious to me who is sponsoring the poll. Worse than that, most of these polls are just advertisements in disguise. The way that the questions are worded makes it almost impossible for the respondent to do anything but agree. The idea behind it is either that if you repeat something to someone enough times, they start to believe it, or that if you present poll results that show that 90% of people agree with X, people who don't agree with X start to feel as though they are missing something and become more open to the possibility of agreeing with X.
The people who write these polls can make the results say whatever they want them to, even on the level of the data. So my point in all of this is just to say that the interpretation of the data you are getting from the news agencies is probably about as accurate as the data itself. What you want is the data AS WELL as the scripts that were used to gather it.
And even if the poll is being done by a company in the business of collecting more general data in a less biased way and then mining the results to sell slices of it -- you think they'll release it? You might also want to start asking the big oil companies to release the core drilling samples they have from all over the world, in the interest of science.
I'm pretty sure that most professional polling organizations will give you this information if you pay for it.
I'm not sure that's true for the news orgs, but the independent polling groups, if you look carefully, often openly note in their poll results that multi-tiered polling data is accessable for paying customers. And if there's some data that's held back even from Rassmussen's "Premium Members" or whoever, I imagine even that data is accessible to people (say, the RNC) who might have specific need for that data and approach Rassmussen or CBS or whoever about paying for access to it.
If this is the case then one assumes the reason they do not release the polling data to everyone, is because then they will not be able to make people pay for it
Mike,
For many polls, they do release the data, after a few months they are archived at the Roper Center at the Univ. of Connecticut or at ICPSR at the University of Michigan. In some cases, if you contact the polling director, as an academic researcher, they will give you access to the data for analysis for a study etc. Though they are unlikely to release the polls to bloggers, just like scientists dont normally release their data to bloggers who can twist and distort the implications etc.
Pew also releases their data, downloadable at the their Web site, six months after the survey is released.
I dunno....when they started identifying the knuckleheads as Republican pollsters and Democratic pollsters...wasn't that enough evidence for you that it was a complete joke?