Who’s in charge here?

Three polls show a tightening race, with the Democrats leading by 4-7 points. Three others, taken at more or less the same time, show Democratic leads increasing to between 13 points and 20 points. WTF?

Right then. This is now officially weird.

Three polls (USAT/Gallup, Pew, and ABC/WaPo) show the race tightening dramatically, with the Democratic lead on the generic ballot question falling into the mid-single digits. Three others (Newsweek,CNN, Fox/OD) show the Democrats’ lead either increasing a little bit or jumping, and have the current lead between 13 and 20.

The explanation doesn’t seem to be timing:

The three polls showing narrowing were in the field W-Th-F-Sa or Th-F-Sa-Su.

The three polls showing widening were in the field Th-F, Sa-Su-M, and Su-M. [Time polled W-Th-F, and had the Democratic lead stable at 15.]

Maybe Saturday, the 4th, give an unusually Republican-leading sample?

Chris Bowers at MyDD says the Pew poll, the one showing the most dramatic tightening, has an unrealistically conservative and Republican-identified “likely voter” panel, and shows that a reweighting produces dramatically different result.

Has there ever been an election where the polling results coming to the finish were so mutually contradictory? You might think that all of this will be irrelevant by midnight tomorrow, but that isn’t quite right. Now that we know how straightforward it is to cheat with touchscreen machines if you have the cards in your possession, and now that we know that the Republican Party is running a national effort to suppress votes with harrassing “false flag” robo-calls, it doesn’t require an especially suspicious mind to think that there may be widespread efforts to cheat in the count (on top of the usual widespread efforts to prevent Democrats from voting). The more unclear the polling picture is going into the election, the less obvious the cheating will be if it happens.

Update Democracy Corps says it’s the volatile “generic” vote collapsing toward the stable “named” vote; in their polling, the Democratic generic advantage has slipped, but there’s been no slippage in the “named” vote. Named Democrats remain, on average, 5 points up on the named Republicans in 50 Republic-held districts thought of us pickup possibilities.

(My speculation: If the Dem is an average of 5 points up in 50 Republican-held districts, that sounds like a pickup of way more than 25, especially if there are a few districts that weren’t thought to be competitive where a Democrat wins anyway.

Author: Mark Kleiman

Professor of Public Policy at the NYU Marron Institute for Urban Management and editor of the Journal of Drug Policy Analysis. Teaches about the methods of policy analysis about drug abuse control and crime control policy, working out the implications of two principles: that swift and certain sanctions don't have to be severe to be effective, and that well-designed threats usually don't have to be carried out. Books: Drugs and Drug Policy: What Everyone Needs to Know (with Jonathan Caulkins and Angela Hawken) When Brute Force Fails: How to Have Less Crime and Less Punishment (Princeton, 2009; named one of the "books of the year" by The Economist Against Excess: Drug Policy for Results (Basic, 1993) Marijuana: Costs of Abuse, Costs of Control (Greenwood, 1989) UCLA Homepage Curriculum Vitae Contact: Markarkleiman-at-gmail.com

7 thoughts on “Who’s in charge here?”

  1. Somebody put it that many Republicans and Republicans-calling-themselves-independents were coming back home, now that the decision moment is here. They might have been disgusted with the GOP, but letting Democrats in would be too much.

  2. CNN lists the Dem margin for registered voters at 15 points, but still nothing on that from Fox. I suspect that this time the margin among actual voters will match the margin among registered voters (whatever it actually is) closely. I'm also going to use my semi-trusty calculator to apply that Greenberg poll of the 50 contestted districts to how they went in 2004 and try to project how many seats the Dems may pick up with the total margin that Greenberg reports.

  3. Projection onto the 2004 returns complete; they indicate a Dem pickup of 34 seats among the 50 seats polled by Greenberg. This sounds about right.

  4. "Now that we know how straightforward it is to cheat with touchscreen machines if you have the cards in your possession, and now that we know that the Republican Party is running a national effort to suppress votes with harrassing "false flag" robo-calls, it doesn't require an especially suspicious mind to think that there may be widespread efforts to cheat in the count (on top of the usual widespread efforts to prevent Democrats from voting). "
    This is not to dismiss your concerns. Not at all, I think they are legitimate concerns. But in terms of the standard of proof you set for fraud due to lack of voter-ID, you are going to have some trouble. Like fraud about voter identity, vote fixing via machine is essentially unprovable unless you catch people in the act. Is it better to leave obvious holes open and hope to catch people in the act or is it better to close obvious holes?

  5. Chris Bowers is not a reliable poll analyst. In 2004, he predicted a Kerry victory because the polls were allegedly oversampling Republicans. Now he says the polls are oversampling conservatives and I'm supposed to take this seriously?
    Note that Bowers' "analysis" is that the Pew Poll has 5 percentage points too many conservatives and 5 points too few liberals. Bowers claims that reweighting the poll to his preferred ratio of conservatives and liberals changes the results from +4 Dem to +20 Dem. But this is mathematically impossible. 5 points of biased sampling can change the results by at most 10 points.
    Finally, I don't think this divergence in polls is too surprising. First, there's sampling error (note that some of these polls include only 600 LVs). Second, all the polling firms have difference LV models. Third, they all differ in how hard they push undecideds to make a choice. So, I don't know which poll is right, but I'd suggest averaging them.

  6. Some back-of-the-envelope calculations:
    If Democracy Corps is right, and the +5% for the named Democratic candidate is a 52.5/48.5 split, my binomial calculator says the following:
    Speaker Pelosi is a lock (99.92%);
    A 25 seat pickup is likely (~61%);
    A 30 seat pickup is not probable (~10%).
    As far as trying to catch the thieves in action, I don't think we should be subject to black-box voting. Period.
    All code used to count votes should be required to be in the public domain. Let 50,000 geeks at the code and the holes and trapdoors will be found.
    Myself, I think we ought to go back to punch cards and maintain the goddamned equipment properly. There's not much question about card sorters working properly, and you don't need doctorates in computer science and statistics to verify the count.
    BC

Comments are closed.