Tuesday, August 28, 2012

Another Cooked Poll: 3 Swing State Poll with Stacked Sample

Well friends, it's becoming almost comical how frequently the Drive-By Media keeps giving us cooked polls. This one's about a week old, but with so much going on, I haven't had time to delve into it. But today campers, we're going to be doing just that.

The perpetrator this time is a CBS/New York Times/Quinnipiac Poll of three critical swing states: Florida, Ohio and Wisconsin. One problem that presents right at the start is that it's essentially three polls in one and it only has about 1200 "likely voters" in the poll. If you figure that it's only 400-500 individuals per state, that makes the margin of error ridiculous. The poll claims the following:

                           Obama                            Romney                                    Result

Florida                  49%                                46%                                     Obama +3%

Ohio                     50%                                44%                                     Obama +6%

Wisconsin             49%                                 47%                                     Obama +2%

So all three are pretty close. However, all three polls have oversampled Democrats significantly, especially for a poll of Likely Voters, not Registered Voters. As we've discussed time and time again, Republican voters are more reliable to the tune of about a 2-4% swing in the GOP's favor whenever the poll goes from Registered to Likely voters.

We're going to be a little imprecise due to lack of data on what the samples OUGHT to be, because the best I can find in those three states are statistics of their party registration. We'll then use a conservative estimate from that registration sample of +2% to the Republicans to account for likely voters.

                   Democrat             Republican           Advantage             Adjusted Advantage with Likely Voters

Florida           41%                     36%                   D +5%                                   D +3%

Ohio               36%                    37%                   R +1%                                    R +3%

Wisconsin       38%                     34%                  D +4%                                    D +2%

So friends that's our baseline. Florida with a Democrat +3% Advantage, Ohio with a Republican +3% advantage and Wisconsin with a +2% advantage. We can expect likely a bigger swing in favor of the GOP considering, at least nationally, given that when Independent leaners are added into party registration you end up with an even split.

So let's look at what the Poll used as THEIR party sample:

                   Democrat             Republican           Advantage              OVERSAMPLE vs. Baseline                                                                                                                      (Without Baseline Adjustment)

Florida           34%                     28%                   D + 6%                            D +3%   (D +1%)

Ohio               34%                    26%                   R +1%                              D +11% (D +9%)

Wisconsin       32%                    28%                   D +4%                              D +4%  (D +2%)

The most ridiculous is Ohio, that oversampled Democrats by 11% against our Likely Voter baseline! Even if we don't use our baselines, there's still a 9% oversample of Democrats! That's just plain ridiculous! The rest of the oversamples happen to be JUST ENOUGH to turn the polls in Obama's favor.

So let's go right ahead and adjust that poll, considering that if we added the proper number of Republicans back into the poll, and presuming (and this is a conservative estimate) that 3/4 of Republicans will vote for Governor Romney, shall we:

                          Obama                           Romney                                  Result (vs. Cooked Poll Result)

Florida                  47%                               48%                                     Romney +1% (Obama +3%)

Ohio                     43%                                51%                                    Romney +8% (Obama +6%)

Wisconsin             46%                                50%                                    Romney +4% (Obama +2%)

Now isn't THAT interesting? When you remove the party biases and oversampling of Democrats, now all of a sudden, shock of shocks, Instead of winning these three crucial swing states, now Obama is LOSING these three states! Amazing!

Friends, this exercise we've just performed is by no means intended to be scientific. However, I believe it does do a fair job showing the difference in the Drive-By Media's outcomes with completely cooked samples vs. honest samples based on voter registration and voter probability.

Rest assured on this final note, my friends: The Drive-By Media wants you to think Obama is winning. As we have continued to show you in poll after poll, the facts simply do not show it to be true when looking at a real electorate and not a fantasy electorate that come from the mind of a liberal news manager's imagination.

I leave you with this final thought, from the classic baseball movie "A League of Their Own":


  1. Correct me if I'm wrong, but you are assuming that because (supposedly) party leanings are 50-50 nationally, that each state is 50-50 when it comes to part leanings? That means, by your logic, any Texas poll that has more Republicans in its sample than Democrats is wrong because it's not skewed to the 50-50 national average??????????

  2. Anonny,

    Actually, he's ony assuming that for swing states. Though the assumption is probably not perfectly correct, it is more likely to correct that polling "registered voters". Texas on the other hand isn't a swing state, so naturally that expectation wouldn't be there.

  3. BTW, Chris. Dick Morris recently did a poll for Hannity which gave Romney a significant lead. Mind you, it was for Hannity, but there you go.

  4. Who said each swing state will be the same as yhe national average? They do swing one way or the other, you know. That is the point; where it is a swing state or not, comparing a state to the national average is completely flawed and useless. The electorate is almost never continuous across the states.

  5. You two really ought to click some of the source links included in the post. I used as my baseline for the polling samples the Party Registration statistics for each state (listed within) then adjusted the percentages as +2% for Republicans because when you go from registered voters to likely voters you see a 2-4% advantage for the GOP (because Republican voters are more reliable). So my stats are based on party registration.

    Another thing I did not note was the fact that Wisconsin is an Open Primary state where Independents can vote in either primary (which would mean the state almost certainly has inflated Independent registrants that personally lean one direction or the other).


All posts will be reviewed subject to the Rules for Commenting. Any post that does not abide by these rules will not be posted, entirely at the discretion of the blog editor.

Commenters who repeatedly violate these rules will be permanently banned from commenting, and thus none of their comments, regardless of content, will be posted.