Saturday, October 04, 2008

Those State Polls ... preface

The Left is getting very happy with itself. Seems that some of the battleground state polls are going Obama's way, and they figure to win big on November 4.

Maybe. But maybe not.

I will admit that I got a little depressed when I saw some of the recent polls, but then I remembered that state polls are, statistically and historically, far less accurate than the national polls, which is one reason the major poll groups generally do not try to do both. An exception is Rasmussen, which is chasing money, and media groups like CNN/Time, which are so biased as to make me question their raw data. More to the point, if you look at all the polls done in a state for a range of time, you see sharp differences in how some of them report things. And as I have always said, an aggregate of polls is actually less statistically valid than any one poll performed under consistent application of NCPP rules. Also, any poll predicting a result which is strongly at odds with the historical results from past elections, is questionable unless a rational explanation is made to explain the sudden change. To be sure, I am not rejecting the polls out of hand, but neither am I so naive, that I forget how many of the 2000 and 2004 state polls were wrong.

So, I have been chewing on the numbers. The thing is, there are 49 different polling groups doing state polls, and so it will take a while to work out the bias in the polls. Once I have something which confirms or dispels the claims being made, I will post it. But for now, keep in mind that most of these polling groups are people you have never heard before, who have a financial interest in the way their polls are reported, and many of whom have hidden their internal data. Treat 'em like used car salesmen until you have better reason to trust them.

On a personal note, though. I would love to hear from someone as to just why McCain would pull out of Michigan, when that state is so winnable and he has gotten a lot without much effort so far. I have made it clear that we cannot lose if we do not give up, but that depends on our candidate staying in the fight himself. Someone needs to remind John what an Obama Administration would do to this country, and why he cannot slack off now or concede anything to that Chicago crook.

Friday, October 03, 2008

The Right Target

All the talk this morning is about the VP debate last night. There’s all kinds of places where you can get details on what was said, how many times Biden lied through his painted teeth, and what the ‘instant poll’ or ‘focus group’ said immediately afterward, so you don’t need me for that. What I am writing about today, is something a bit more intricate, and yet I think important to the race remaining.

My wife wanted to see the debate on CBS, so we got to hear what Ms. Couric thought of it. I don’t usually have much respect for the opinion of Ms. Couric, but she made a good observation, though she did not carry it through to its significant conclusion. Kouric said that Governor Palin was speaking to Senator Biden, the moderator, the audience, and the television viewers, while Senator Biden, in Couric’s words, was “speaking as if Palin was not even there”. That is, Biden was trying to drive home attacks on McCain and pressing his points about the Obama campaign directly to the home viewers. Biden and Palin were going after different target audiences. My hunch is that both nominees accomplished their tactical goals, but only one of them chose the right target.

It’s easy to get caught up in minute details of the election, but it’s critically important to understand how a candidate builds up support. It does not, despite what you may have read or seen or heard, happen in great waves or sudden bursts of enthusiasm. And most people who have made up their mind to support a candidate, do not quickly or casually change their mind. The publicly released polls are not accurate in the image created, that the voters are flighty and chaotic. Quite the opposite, if history is considered.

With firmer facts in hand, we can simply say that the race remains in doubt, because a portion of the voters have not made up their mind. They may well lean towards McCain or to Obama, but there is a portion which has decided and cannot easily be convinced to change their opinion, and there is a small portion which will not make a choice and so will sit out this election, but there is also a significant portion remaining, who are still waiting for a candidate to convince them to give their vote to that candidate. And so it is happening, small pieces at a time, and along the way the candidates may well not be certain themselves about just how much support they have in certain key places. This is true in many states, where the race is close enough that a well-focused effort could make all the difference.

Without overdoing the numbers, it is safe to say that most Democrats are supporting or will support Obama, and that most Republicans are supporting or will support McCain. There are, in the end, just three groups in play:

Republicans unsure about McCain-Palin, and Independents leaning towards McCain, but still not locked in, a group of about 2.9% of voters;

Democrats still uncomfortable with Obama-Biden (including the PUMAs), and Independents leaning towards Obama, but still not committed, a group of about 3.6% of voters; and

The pure Independants, who will not vote or who will go for a minor candidate unless McCain or Obama convinces them they deserve their vote, a group of about 8.0% of voters.

Palin spoke to the undecideds, while Biden spoke to his base. Guess which mission was more important to this election?

UPDATE – How the Math Was Done

Some folks are curious as to how I got my numbers for undecideds. I got them from the internal party affiliation results from the three major polls in the past month which have published them.

Gallup, CBS, and Fox News all publish the internal party-identified support in their internal data. When you take out the support for Obama, McCain, and ‘Other’, what’s left is your undecideds. When you consider that the poll starts with registered or ‘likely’ voters, that’s pretty significant. When you plug the undecideds back in for each group, multiplying by their proportion of the population, the result is that undecided sub-group’s portion of the total voting population. So, taking the numbers from each of the three polls, and averaging them together, I get the numbers I used for today’s article.

Now, I have to admit that the actual number could be a little higher or lower than I claimed, for three reasons. First, since the three polls differed in their numbers (though not greatly), averaging them may lessen the accuracy of the analysis. Of course, I have no way to say which of the three is the ‘most reliable’, so averaging them is the best course to acknowledging all of them to an equal degree. Second, folks who do not make up their mind by the election may just not vote at all. I do not think this is very likely, however, because at a minimum these are voters who have registered, and who show some interest in the election as evidenced by their participation in the poll. So I believe these people are valid respondents, representing an actual demographic whose importance to the election is growing as we move ahead. And third, while I do try to be objective in analysis, as a human being I impose some subjectivity into my opinions, even when I try to avoid it, and certain assumptions were made in developing the template. On the other hand, while we all have biases, awareness of such bias, open acknowledgement of the bias and attempts to test for and correct against material flaws in the work, and the evidence from past tests using similar assumptions lead me to believe that I have produced a valid model.

Thursday, October 02, 2008

The Fourth Quarter

I generally don’t like sports analogies when discussing politics, because 1 – it’s more important than any game, and 2 – the analogy usually misses by a bit. But in this case, the analogy is close enough to be useful, and the point is important.

Back when I was younger, I was an official in several sports, including high school football here in Texas. Some games were tough, hard-fought matters, and when the last quarter started, both teams were tired and hurt. It came down in those contests, to who was willing to stick it out for twelve more minutes on the clock, maybe a couple or three dozen plays of bruising tackles, hard hits, and fighting your own body to go a little farther, carry the play through to the end, keeping up your block or running your route all the way through. Because you knew that the game was still anyone’s to have, even if you or they were ahead. Sometimes you could beat a big lead, if you were tough enough and refused to give up.

The media is completely in the tank for Obama, we know that. So it should not surprise anyone that the polls they publish all show him in the lead. Numbers can be fudged, and the polls have done that. The funny thing is, if we give up and don’t vote, if we sit on our butts and whine about the world and do not get our neighbors, friends and citizens motivated to support John McCain, then those weasels will have succeeded in getting us to quit and they will get the lopsided party turnout they are predicting. Make no mistake, though, if Obama wins it will be because too many good men did nothing.

I explained yesterday that when the polls are reweighted to historically valid norms, McCain leads Obama. It’s tight but very real, and if we just fight through to the end the way we have through August, then we will certainly win.

Want more? There’s a lot of folks still undecided out there. Taking the levels of support by party from four major polls which published their internals, here are the portions of undecided voters as of October 1:

GALLUP: 15.52% of voters are still undecided

CBS NEWS: 9.70% of voters are still undecided

FOX NEWS: 15.56% of voters are still undecided

Franklin & Marshall: 17.19% of voters are still undecided

When you add that information to the margin of error and correct for the historically invalid weighting the polls are using, it becomes absolutely clear that this race has a long way to go. Trusting the MSM is like believing what the other team’s cheerleaders tell you.

Gut it up, this one’s no where near over, we’re just starting the fourth quarter. Now is where we find out who’s got what it takes.

Wednesday, October 01, 2008

The Secret Poll, October 1 Edition

I have been working through the poll numbers for quite a while now, sorting out valid patterns from the fakes. I held off posting the true state of things for a long time, for a number of reasons, but I notice that some on the Right have begun to lose hope and make sounds of giving up. So I will tell you plainly, that

We Are Winning

and can only fall in this election if you give up. It's been a long road and the enemy has been his usual foul self, with lies and smears and everything we have learned to expect from people who put power above any moral or honorable precepts. It's close, but here's where we have been, and where we are:

August 31: McCain 41.77%, Obama 41.06%

September 7: McCain 42.45%, Obama 42.04%

September 14: McCain 45.71%, Obama 39.62%

September 21: McCain 44.48%, Obama 42.06%

September 28: McCain 42.73%, Obama 41.62%

And based on the demographic responses, once the undecideds shake out if we work as hard as we can and continue to keep faith, the final popular vote will be

McCain 51.59%

Obama 48.41%

Keys to remember:

This is not a football game or a baseball game, it's politics. Support is built up gradually and won bits at a time. Also, some of the best gains are not obvious at first, because some significant actions take time to develop. McCain and Obama both fell back a bit the last week of September, McCain because Republican support fell off a bit, Obama lost independents' support. This is a salient factor in where the candidates' opportunities and weaknesses lay.

State Polls and the Shadow States

Many readers have observed that the election of our next president will be decided not by a national popular vote, but by the states and the District of Columbia in the Electoral tally. The problem with the states as they relate to opinion polling, is that while the national polls have some points of concern, the state polls are relatively even less certain. The folks at Real Clear Politics show forty-nine different polling groups which have published results for states since the end of August, and few of these have publicly-released verification of their methodology, let alone access to their internal data. Caveat Emptor indeed , in these waters!

Even so, such information as has been presented can serve a purpose, to address curiosity if nothing else. I have reviewed and collated the information presented by RCP, and once again it is all here at the link if you want to do the same yourself. I would warn the reader that an average of polls is not statistically valid, but it does satisfy a certain curiosity. More, while there is a range of opinion among the pollsters, it is interesting to see how many polls were done for a given state. So, in the interest of public service, I am posting here a reference for the current polling average, along with the historical average since 1988, and the statistical trend for the state over the last 20 years. Please also note the columns for undecided, “Clear” and “Polls”, which will be explained below the report:

State (historical adv, trend adv) poll avg. (undecided) Clear? # polls
Alabama (h R+15, t R+32) 59.7-36.0 McCain (4.3 und) Clear, 3 polls
Alaska (h R+22, t R + 20) 57.7-35.0 McCain (7.3 und) Clear, 3 polls
Delaware (h D + 7, t D + 4) 43.3-56.0 Obama (0.7 und) Clear, 3 polls
Illinois (h D + 10, t D + 9) 37.7-55.0 Obama (7.3 und) Clear, 3 polls
Kentucky (h R + 8, t R + 21) 55.0-38.7 McCain (6.3 und) Clear, 3 polls

Michigan (h D + 4, t D + 2) 43.9-47.9 Obama (8.3 und) Unclear, 14 polls
Ohio (h R + 2, t R + 1) 46.4-45.4 Mccain (8.1 und) Unclear, 14 polls
Florida (h R + 5, t R + 9) 46.8-45.6 McCain (7.5 und) Unclear, 13 polls
Pennsylvania (h D + 6, t D + 2) 43.2-48.6 Obama (8.2 und) Unclear, 11 polls
Virginia (h R + 9, t R + 9) 47.0-46.7 McCain (6.3 und) Unclear, 10 polls
New Hampshire (h R + 3, t D + 2) 45.4-47.3 Obama (7.4 und) Unclear, 8 polls
New Jersey (h D + 6, t 0.0) 42.0-49.6 Obama (8.4 und) Unclear, 8 polls
Colorado (h R + 4, t R + 2) 45.1-48.9 Obama (6.0 und) Unclear, 7 poll
Indiana (h R + 14, t R + 21) 47.3-45.0 McCain (7.7 und) Unclear, 6 polls
Iowa (h D + 5, t R + 1) 41.8-51.0 Obama (7.2 und) Unclear, 6 polls
Minnesota (h D + 8, t D + 5) 44.5-48.8 Obama (6.7 und) Unclear, 6 polls
North Carolina (h R + 9, t R + 11) 49.3-45.2 McCain (5.5 und) Unclear, 6 polls
Oregon (h D + 5, t D + 7) 40.5-48.2 Obama (11.3 und) Unclear, 6 polls
Wisconsin (h D + 4, t D + 3) 44.2-47.8 Obama (8.0 und) Unclear, 6 polls
California (h D + 9, t D + 9) 39.4-51.8 Obama (8.8 und) Unclear, 5 polls
Georgia (h R + 10, t R + 17) 52.3-42.3 McCain (5.5 und) Unclear, 4 polls
Missouri (h R+ 2, t R + 9) 49.0-45.8 McCain (5.3 und) Unclear, 4 polls
Nevada (h R + 5, t R + 1) 46.3-46.3 tie (7.5 und) Unclear, 4 polls
New Mexico (h D + 2, t R + 1) 43.8-50.3 Obama (6.0 und) Unclear, 4 polls
Washington (h D + 8, t D + 9) 42.8-48.8 Obama (8.5 und) Unclear, 4 polls
Maine (h D + 8, t D + 18) 42.7-50.3 Obama (7.0 und) Unclear, 3 polls
New York (h D + 18, t D + 13) 40.3-52.7 Obama (7.0 und) Unclear, 3 polls
North Dakota (h R + 18, t R + 25) 49.3-41.3 McCain (9.3 und) Unclear, 3 polls
West Virginia (h D + 3, t R + 14) 48.0-42.3 McCain (9.7 und) Unclear, 3 polls
Connecticut (h D + 9, t D + 4) 39.5-53.5 Obama (7.0 und) Unclear, 2 polls
Idaho (h R + 27, t R + 35) 60.0-29.0 McCain (11.0 und) Unclear, 2 polls
Kansas (h R + 16, t R + 26) 55.5-39.5 McCain (5.0 und) Unclear, 2 polls
Montana (h R + 12, t R + 14) 53.0-41.0 McCain (6.0 und) Unclear, 2 polls
Oklahoma (h R + 15, t R + 35) 64.0-32.0 McCain (4.0 und) Unclear, 2 polls
Rhode Island (h D + 19, t D + 13) 34.5-54.5 Obama (11.0 und) Unclear, 2 polls
South Carolina (h R + 12, t R + 17) 54.5-42.0 McCain (3.5 und) Unclear, 2 polls
Utah (h R + 32, t R + 46) 63.0-28.0 McCain (9.0 und) Unclear, 2 polls
Vermont (h D + 13, t D + 26) 36.0-57.5 Obama (6.5 und) Unclear, 2 polls
Wyoming ( h R + 25, t R + 36) 57.5-37.5 McCain (5.0 und) Unclear, 2 polls
Arizona (h R + 8, t R + 14) 59.0-38.0 McCain (3.0 und) Unclear, 1 poll
Hawaii (h D + 15, t D + 3) 27.0-68.0 Obama (5.0 und), Unclear, 1 poll
Louisiana (h R + 3, t R + 17) 55.0-40.0 McCain (5.0 und) Unclear, 1 poll
Maryland (h D + 11, t D + 10) 37.0-60.0 Obama (3.0 und) Unclear, 1 poll
Massachusetts (h D + 23, t D + 23) 39.0-55.0 Obama (6.0 und) Unclear, 1 poll
Mississippi (h R + 14, t R + 19) 55.0-37.0 McCain (8.0 und) Unclear, 1 poll
South Dakota (h R + 13, t R + 20) 54.0-37.0 McCain (9.0 und) Unclear, 1 poll
Tennessee (h R + 5, t R+ 19) 55.0-39.0 McCain (6.0 und) Unclear, 1 poll
Arkansas (h D + 1, t R + 11) no polls, Unclear
Nebraska (h R + 24, t R + 34) no polls, Unclear
Texas (h R + 13, t R + 22) no polls, Unclear
District of Columbia (h D + 75, t D + 82) no polls, Unclear

Only five states are statistically clear at this time. The reason is what I call ‘the shadow’. That is, given the number of people interviewed for each of these state polls, the statistical margin of error is about +/- 4.0 percent. That means that McCain or Obama could each be as much as 4.0 points higher or 4.0 points lower in support than the results of the poll indicate, assuming the poll is valid as defined by the NCPP. This creates an 8.0 point range of uncertainty. The 8 points for the margin of error, when added to the undecided portion of the vote, gives the range of shadow. If the leader does not have a lead greater than the range of shadow, the poll is unclear. Also, if there are fewer than three polls performed on a state, the stated results are inconclusive, which is why some states with large leads for one candidate are still statistically “unclear”.

This is what I mean by the state polls not being very helpful. And to make matters worse, this aggregation does not remove outliers, and of course when there are only two or three polls for a state, it’s impossible to know which numbers would be outliers for a state.

Monday, September 29, 2008

Is There a “Good” Poll?

Looking at Real Clear Politics this morning, I noticed that there is still a range of opinion about the presidential race. Obama supporters will be glad to hear that Rasmussen Tracking, Hotline/FD Tracking, and Gallup Tracking all say Obama is leading McCain by 5 points or more. McCain supporters can be consoled by the fact that Battleground Tracking says McCain is still leading by 2 points. Neither would be happy to hear that trusting the headlines from a “tracking” poll would be about as good an idea as eating a sandwich made in Galveston 17 days ago. The short version of tracking polls is that they are trendy ways for polling groups to get a headline and maybe some attention, but they are simply unproven as empirical analysis. Tracking polls are volatile, depend on rough methodology which tends to oversample some demographics in the interest of speed, and have never been peer-reviewed the way more traditional polling is done. These guys even get cute, and try to put these hash polls together to imitate a real effort to get a sense of the national opinion. And, as I have warned many times before, when a poll does not release its internal data to the public, and conceals any significant part of its methodology and weighting, you should treat the pollster like a Bear Stearns banker. Having said this so long, I hear a lot of people say that they will ignore all polls, while others want to know which poll I think is “best”. That, however, is not so easy to answer.

First off, almost all polls have some validity. With very few exceptions, I do believe that polling groups with more than four years of experience in the business will try to follow a standardized methodology, and I also believe that most polls make a determined effort to avoid bias as much as possible. When I say that a poll has bias, therefore, what I mean is that the poll may be accepted for its findings, provided the reader is aware of the poll’s history and tendencies. If the poll is consistent, you can compensate for bias and get a generally objective idea of the situation. What’s more, if a poll is consistent in its methodology and weightings, then movement by a candidate over time within the same poll can be considered valid to show growing or diminishing strength of support.

So, you don’t care about all the details, you just want to know who’s winning? Good luck with that, the national polls may not be much real help. That’s due to a number of things. First, you should know that the polls taken before the last week of the campaign have no statistical value in predicting the election winner. Also, as I said there is always bias present, so the fact that one poll or many says something does not make it necessarily so (just one reason that both Obama and Mccain have hired private firms to poll for them). And then there is a crucial fact to consider about the election; it’s not one race, actually, it’s fifty-one races to decide the matter. As Mister Gore realized a bit late in the 2000 election, it is the Electoral vote which determines the presidency.

So, as several people have asked, how do the state polls look? Well, that’s a tougher question. First off, you know how I am about transparency in poll internal data, but very few polls make that information public (and some, like Rasmussen, will provide it but only if you pay a fee, which goes against my principles – if you want the publicity of announcing your poll results to the media, you have the moral duty to provide all the supporting data). A significant exception is Survey USA, which is the best of the state polls in following clear NCPP rules. Also, I should warn the reader that in 2004, a lot of state polls were well off the mark. During the last month of the 2004 election, for example, I remember some polls which gave President Bush leads in Michigan and Oregon, and Kerry leads in West Virginia and Florida, none of which turned out to be correct in the actual election. State polls, it should be understood, are smaller budget than the national polls, and often have a much smaller base, which creates a much greater margin of error.