Lies, Damned Lies, and You Know the Rest

Here are three numbers.

• 83
• 64
• 33

They're the percentages of adults in this country who, according to three public opinion polls released this month, support the death penalty.  The first is from Angus Reid.  The second from Gallup.  The third from Lake Research Partners for the Death Penalty Information Center.

Doug Berman finds Angus Reid convincing and doubts the significance of what Lake Research concluded.  Thirty-two minutes after Doug posted about the Lake Research Poll, Bill Otis put up a comment with the declaration that the poll reveals DPIC

exploring new ground in embarrassing itself. The dishonesty and intentional avoidance of the main issue is mind blowing.

Otis and Berman conclude that the numbers are wrong because people were asked which of several sentences they thought appropriate for murder.

Rather, they say, the proper question is which is the maximum sentence that should be available for killers.  Ask that question, they say, and Angus Reid (which didn't ask that question) will be right.  Maybe.

Neither explains precisely why that's the right question, nor why they know for sure that Angus Reid numbers will come out right (though Otis does say it's close to the percentage that thought Tim McVeigh and now think Steven Hayes should die.

We all know what you can do with statistics.  It's how you ask the questions and what questions you ask and which numbers you look at and how you massage them.  None of that's new, and it's probably why I tend to roll my eyes, even when folks pretend that they're just giving real world results.  Sure they are (see here, for instance), but selectivity distorts.

The numbers that count, after all, aren't from the people who answer polls on line or by phone or when people come to the door.  The numbers that matter come from the folks in the real world.

When last year's Gallup numbers came out (not much different from this year's, by the way), I said that the poll that really counted was the one in the jury room and that death sentences were way down.  And Capital Defense Weekly reports that actual executions this year are also down.

[S]ave for the Baze v. Rees related moratorium on executions, 2010 will be the lowest number of executions in a given year since Bill Clinton’s first term as President.

Those are the polls that matter. 

And those tell us that the death penalty may not be about to be killed, but it's slowly dying.