On Tuesday, David mentioned that in the latest DailyKos/SEIU/PPP poll, 14% of Democrats support impeaching President Obama. David suggested this shows some people don’t know what the term ‘impeachment’ means. Some commenters wondered whether respondents were just messing with the poll.
I dug a little deeper into the numbers and this is what I found: About half of Dems for Impeachment do not approve of Obama. About half of them do approve. We’ll start by looking at some views of those who do not approve:
Only 22 respondents are Democrats who do not approve of Obama and support his impeachment. It is obvious that this group has consistently conservative views, so much so that even with so few respondents we can feel confident that there’s a difference between this group and other Democrats. Only two did not describe Obama as ‘too liberal’ and they both described themselves as Conservative. One of the respondents appears to be a crank – this individual said Obama is too conservative on Question 3, and then pressed ‘1’ for every remaining question.
The question is, do these people really consider themselves Democrats? The answer is ‘probably.’ Some of them may be Dixiecrats – about a third of these people are white Southerners. See digression below for details.
What about those Democrats who do approve of Obama and support impeachment? They are consistently Democratic; 15 of 20 approve of Congressional Democrats and 19 of 20 prefer Democrats on the generic ballot question. So why would they impeach Obama? Perhaps in some of these cases it is the result of not understanding the term ‘impeachment.’ Here is the relationship between income and percent of Obama-approving Dems who either support impeachment or are not sure:
We see most of the Dems who approve of Obama and support impeachment fall into the lowest income category. If we take income as indicating, roughly, education levels (see census numbers), this would be consistent with the idea that these may be people who aren’t familiar with the meaning of the word ‘impeachment.’ (Another survey has shown the number of respondents who answer civics questions correctly increases with both educational level and income, although it did not ask about the term 'impeachment.')
So, Democrats who want to impeach Obama are about half conservative Democrats-in-name-only, and about half people who potentially may not understand the term, with a couple of likely cranks (giving nonsense answers to a pollster) thrown in the mix. Democrats who disapprove of Obama from the left and support impeachment are too few to measure in a poll of 1000, although they certainly exist.
Updated for clarity and with new links.
Digression.
While answers to the opinion questions for the Obama-disapproving impeachment supporters are consistently conservative, the respondents may have accidentally or purposefully pressed the wrong number on the party identification question. But if somebody’s gonna mess with the poll, you might expect them to identify as not just a Democrat, but a liberal Democrat, and only one of these respondents did. You might also expect this type of prankster to be white, and press 3 for African-American instead. But the racial balance is not terribly out of whack either. So while another couple of pranksters may be included in the mix, it seems unlikely they’re the majority.
That still leaves accident – on the face of it, this is more plausible. For instance, about a third of all those who support impeachment are Southern whites, and about a third of Democrats who disapprove of Obama and want to impeach are also Southern whites. Are we just looking at a sample of those who pressed the wrong button? This is highly unlikely. First, we have some information about error. We know the ‘downward slippage’ error rate – those who press 4, 5, or 6 instead of 1, 2, or 3. I’m calling this ‘downward slippage’ because one imagines the respondent’s finger slipping down a row of numbers. It is zero. Out of 9000 some-odd respondents this year, nobody accidentally pressed a number greater than 3 on the party question. (On the first question of the poll, the slippage rate is 0.27%, on the second it is 0.20% - there appears to be a learning curve. By the time the party ID question comes along, people have got this button-pushing thing down pat!) Another type of error would be forgetting which number you’re supposed to press. That error, too, seems unlikely in this case, because of the phrasing of the question (“If you are a Democrat, press 1. If you are….”) which does not give the respondent a chance to forget.
For more analysis of DailyKos/SEIU/PPP polls see the DKos Polls tag.