Swing State Project

« OH-Gov: Coleman Drops Out of Dem Primary | Main | Raise a Ruckus: Bush Fundraising for Steele »

Tuesday, November 29, 2005

RON Revisited: I Want Answers

Posted by DavidNYC

Just came across this piece in Editor & Publisher about the Columbus Dispatch mail poll which I mentioned in the post just below. A syndicated columnist for the Dispatch, Robert Koehler, took the paper to task a few days ago for, allegedly, not critically investigating the mind-boggling failure of its poll regarding the RON ballot measures. Koehler seems to believe that there was nothing wrong with the poll itself but rather everything wrong with the election. Though he doesn't use the word, he's pointing a finger at fraud.

Let me say quite clearly: Without further evidence, I disagree strongly with any such conclusion. But I'm also deeply, deeply unsatisfied with the paper's lazy explanations about why their poll was so awful. The paper's Public Affairs Editor, Darrell Rowland (who was also interviewed by Koehler), only offered this to E&P:

Koehler said the Ohio paper -- which he described as "a member in good standing of the mainstream media" -- instead blamed its inaccurate poll on "the notorious volatility of statewide referendum issues."

Rowland said that may have been one reason for the inaccuracies in the Dispatch poll, which was mailed to about 12,600 Ohioans - of which about 1,900 responded. He added that the ballot issues were "incredibly confusing," and that "a lot of last-minute money" was spent to defeat four of them. "It's easier to cause people to vote no than to vote yes," Rowland said.

Two (possible) problems: First, the Dispatch poll was largely accurate on the one non-RON ballot measure (a bond issue). The poll said 53% yes; the actual results were 54% yes. The only other "yes" vote it came close on was Issue 4 (redistricting). The nays were off by enormous margins on every single issue. (Poll | Results)

Second, according to a "voting-rights activist" cited by Koehler, the pre-election ad blitz mostly came from the RON forces, not the anti-RON side. I assume Ohioans in the audience here can either confirm or dispute this.

The Dispatch wants its poll to be taken seriously. The powers that be there are probably keeping their fingers crossed that the next one will be more accurate. If it's not, then perhaps we can conclude that the entire polling methodology is flawed and needs to be re-worked. However, if the next Dispatch mail-in poll is on target, that will raise many more questions about why this one was so skewed. And if the Dispatch wants its reporting to be taken seriously, then it better start looking for answers.

Posted at 09:01 PM in 2005 Elections, Ohio | Technorati

Trackback Pings

TrackBack URL for this entry:
http://www.swingstateproject.com/mt/mt-track-ssp.cgi/1998

Comments

First, the Dispatch poll was largely accurate on the one non-RON ballot measure (a bond issue). The poll said 53% yes; the actual results were 54% yes.

David, you're forgetting that the undecided sample for that question was 20%.

So, for the poll to have been accurate, 95% of undecided voters would have voted no on Question 1.

Posted by: RBH at November 29, 2005 10:38 PM | Permalink | Edit Comment | Delete Comment

Actually, I noted at the end of that paragraphs, "The nays were off by enormous margins on every single issue." I also characterized this as only a "possible" problem for explaining the Dispatch poll's difficulties.

Also, it's usually the norm for undecideds to swing firmly in one direction. Typically, it's to challengers in regular elections. (The big exceptions in recent years would be Wayne Allard in CO-Sen in 2002 and, of course, George Bush last year.) Does it make sense for the mass of undecideds to swing heavily toward one particular side of a ballot issue? I dunno, but I think it's possible.

In other words, the polling on issue one (and issue four) might have been, in some views, fairly reasonable. (At least they got the outcomes right.) The fact still remains that I'm really not satisfied with the Dispatch's explanations. I'd like to see something a lot more rigorous.

Posted by: DavidNYC at November 29, 2005 11:16 PM | Permalink | Edit Comment | Delete Comment

The polling on referendums is shaky by nature, mainly because there's what's on the ballot, and what people think is on the ballot (from the ads).

Posted by: RBH at November 29, 2005 11:45 PM | Permalink | Edit Comment | Delete Comment

David, if you look at especially the recent history of ballot issues/referendums, typically undecided voters break heavily to voting no, particularly when a ballot issue becomes controversial, and even moreso when one party is lined up on one side of it, and the other party on the other side. If people don't understand something (or, as RBH says, misleading ads grossly mischaracterize what something will do), a "No" vote seems like a much safer vote than a "Yes" one. I remember hearing an interview with someone in Colorado in '04 a week or so before the election, when they were voting on whether to assign their electoral votes by Congressional districts -- he said that the "Yes" side (i.e. the side that would have helped Democrats) felt they needed to be polling at around 60% or so yes before the election to have a shot at winning -- they weren't, and the referendum was soundly defeated. I just think these things have gotten increasingly difficult to pass once one political party or well-funded interest group opposes them -- because once confusion is created, a "No" vote appears to be the safer course to take for most voters.

Posted by: IndianaProgressive at November 30, 2005 08:04 AM | Permalink | Edit Comment | Delete Comment

RBH: Most of the polling on the CA ballot measures was pretty accurate.

IP: Agreed. Which makes the Issues 1 and 4 polls somewhat reasonable. But 2, 3 & 5 were way off on the "yes" vote as well.

Posted by: DavidNYC at November 30, 2005 10:19 AM | Permalink | Edit Comment | Delete Comment

Unfortunately, being an Ohio person with personal knowledge, I would have to agree with the Dispatch people that the last push ad campaign was heavily NO. With that being said, once again sadly, I would also have to agree with the Dispatch analysis of why RON was defeated so handily (confusion, huge last-minute NO campaign, etc) but I don't think that explains the bizarre polling. Honestly, most of those elements were in play prior to and during the polling.

Posted by: OH-09Dem at November 30, 2005 12:38 PM | Permalink | Edit Comment | Delete Comment

The last minute blitz was definitely PRO the issues. The anti forces had been blitzing for at least two months non-stop and virtually had no way to up the level of blitz in the last few weeks. Most of the discouragement feedback I had from friends supporting the issues came at least a month prior to the election when they'd call me and report that they were seeing the anti commercials nonstop. So no, I'm not satisfied with the "last minute blitz" explanation - unless they came up with a brand-new, incredibly effective take, and they didn't.

I think confusion was definitely a factor in the defeat of #4 and #5, but #2 and #3 were not really confusing anyone, and most people seemed to be in favour of #2 totally, although I'm starting to get feedback, mostly from the progessive side, that the spin about increasing fraud was taking hold, and many on the left, suspcious of all Ohio elections now, voted against #2 for that reason.

I will grant an underestimation of the oppostion to these issues but I cannot grant these polls being this radically outrageously off for the first time in the Dispatch's history, especially given the accuracy on #1, which for my money was the most opaque and confusing of all the issues.

I still haven't seen an explanation that makes sense and I would love to hear one.

Posted by: Ansatasia P at November 30, 2005 02:22 PM | Permalink | Edit Comment | Delete Comment