Portsmouth Republican statistical misdirection

Click to embiggen.

The Portsmouth Republicans have had a blaring yellow insert in our weekly Portsmouth Times for the past month, and while I've previously covered their gibberish, it didn't seem worthwhile to go after their cherry-picked statistics. But this week's installment featured such a shameless whopper that as a graduate of Edward Tufte's workshop, I feel obligated to point it out.

Here's their statement:

DID YOU KNOW...that voter turnout in Portsmouth increased from 54.4% in 2010 to 65.5% in 2012? Keep up the good work! See you at the polls.

Let's start with the less obvious problem. The Republicans appear to be doing their percentages against the entire Portsmouth population (17,343) rather than the eligible voter universe (12,327). That's the only way I can get their numbers to work, since in the the 2012 election, Portsmouth voters cast at least 9,182 votes for President, which works out to 74% turnout, rather than the 54% they computed, apparently using the census number.

But that's merely the little chocolate squiggle atop this confection of innumeracy.

Let's leave percentages aside for just a moment and look at the raw numbers. Here are the vote totals for the past five elections, using the Portsmouth Town Clerk (the race with the the single most local votes) and the "top of the ticket" contest to represent total Portsmouth turnout.

  Clerk Top of ticket
2012 6,733 9,182
2010 5,806 7,051
2008 7,096 9,740
2006 5,744 8,069
2004 7,918* 9,205
*contested race

Source: RI Board of Elections

Notice something about the numbers — kind of, like, a pattern? The numbers seem to fluctuate regularly. What could possibly be driving that? Oh, yeah, right, 2012, 2008, and 2004 were Presidential elections.

If you pick a local minimum (2010) and only use one other data point (2012) you can make the data say something quite dishonest, because comparing total votes across the range doesn't support the inference the Republicans want to make. As Tufte drilled into us, always ask, "Compared to what?"

And if you're willing to distort the data to support a relatively minor point like turnout, clearly, you cannot be trusted to play fair when there are issues of significance on the table.

Editorial note: I have Tufte's three books within arms' reach, and I'm not afraid to use them.