Reporting on reported results may increase confusion

I had decided not to comment on political polling during this election season. Although this post concerns the election, it isn’t about polling. It isn’t directly about research, but it does show the problems that can happen when numeric results are reported imperfectly, and, even more important, when reporting on the report confuses instead of clarifying.

Today (10/7/08), I read a snippet from the Puget Sound Business Journal “Washington State Voter Registration hits 3.5 million” The article stated that Voter registration totals in Washington state hit 3,515,393. Reading further, I see that the previous record was 3,514,078 in 2004, and that number was reduced “significantly” after duplicate registrations and deceased voters were removed.

What do these statements mean in absolute terms? And what do they mean relative to the changing state population?

The original release from the Washington Secretary of State has the same figures – naturally. But there we find that the 3,515,393 number is merely the latest tally, and that it is expected to grow as applications continue to be processed. There is also some more information about the total of voters registered in 2004, and the processes involved in reducing the numbers. I find the writing somewhat unclear, but it appears that the earlier number of 3,514,078 was the high water mark reached during the previous registration process – which wasn’t clear from the Puget Sound Business Journal.

I’m not going to comment on why the Secretary of State decided to publish a release today (the application deadline was last Saturday), but let’s compare the tallies from 2008 and 2004. The current registration count is 0.04% more than last time. Given that the Washington State population increased by 8.5% from 2000 to 2006 according to the US Census Bureau, the 0.04% seems like a so-what. It may very well be that the final tally increases by 5% or more (even before trimming), but why not wait and report on more solid numbers? And (for the Business Journal), why not add some clarity about the process and the meaning?

For more information about reporting on reporting and the way that mistakes have a way of taking on a life of their own, I recommend Damned Lies and Statistics (untangling numbers from the media, politicians, and activists) by Joel Best. It’s a great way to learn to read critically, whether or not you are a researcher. And, of course, to learn what you can do to prevent your own results being abused (answer – not much).

Idiosyncratically,
Mike Pritchard