by Paul Goldman
Being the only one who has predicted the first Democratic sweep since 1989, the new poll from the highly respected Quinnipiac University folks showing McAuliffe up a statistically significant 6% – the second biggest margin for a winning Democratic candidate at this stage of a GUV race in the modern era (Warner led by more, Robb was about the same, Wilder/Kaine/Baliles were less) should be signs that my crystal ball is a lot better than Professor Sabato’s. But is it?
Historically speaking, the Q-Poll is making a bold statement: namely, the Virginia is a now a Democratic state, “blue” as they say, no longer “purple” having left “red” in the rear view mirror. Why? If you dig into the poll just a little – indeed it is clear from the first set of numbers to anyone who knows VA politics – the pollsters are predicting that contrary to all previous history, the 2013 in an off-year GUV race will match the presidential year turnout model for 2008 and 2012. THIS HAS NEVER HAPPENED BEFORE, NOT EVEN CLOSE.
For example: In 2009, the exit polls showed that the voting electorate contained 4 percentage points MORE Republicans than Democrats, fueled by the turnout of Evangelical Christian voters according to the way the pollsters use demographic stats (like it or not, in the polling business, everyone is part of a least one subgroup, and for understandable reasons). What are the understandable reasons?
Like it or not, the overall choices of voters is NOT THAT HARD TO PREDICT if you know what you are doing. This is why polling works so well: it can slice and dice all of us into groups, and then predict how we are going to vote based on how those in that group are trending.
Thus, since most of the electorate is easily put into a group that STRONGLY TRENDS ONE WAY OR THE OTHER in a two-way race, polls can take a small sample of a huge state – and the Q-Poll actually uses close to double the sample of most polls – and magically predict the outcome within a rather small margin of error. It is quite magical at least to me, fascinating since I first read about it and became enamored with statistics of all kind (I also got into national trouble by writing a column predicting how many freshmen women were engaged in a certain type of sexual practice based on my mathematical model). In that case, for reasons of tabloid journalism, I confess to have screwed with the numbers a little due to the sample size. But I was trying to make a larger point, which had to do with censorship and other things not relevant. But it taught me the power of statistical equations.
It also taught me their key limitation when it comes to the 2013 VA GUV race: The pollsters’ projected turnout model. Or in English: Who the heck will actually show up and vote this November for GUV, LG and AG. According to Quinnipiac’s poll today, Democrats will make up 7 percentage points MORE of the electorate than Republicans. Independents will make up nearly 40% of the electorate.
But you say: It was 4 percentage points more Republican in 2009, why should it be 7 percentage points more Democratic in 2013? Moreover, this 7 percentage-point Democratic advantage tracks the 2012 presidential turnout, yet this is the off-year election cycle when historically the presidential year model has NEVER APPLIED. How cam this be?
That’s a fair question. But let’s remember that in 2012, GOP voting statistic gurus predicted the 2008 presidential turnout in Virginia – which was 6 percentage points more Dem than GOP – couldn’t be duplicated.
What happened? The 2012 exit polls, as indicated, said the VA presidential turnout was 7 percentage points more DEMOCRATIC, which means on a math basis, they were statistically the same in terms of party ID at that macro data level.
THUS THE QUESTION: If the Q-Poll is correct, and the PRESIDENTIAL turnout model is now the GUBERNATORIAL turnout model, this is not only fascinating to those of us who like stats.
It is, to borrow the title from the book, A Virginia GUBERNATORIAL GAME CHANGER.
Why? SIMPLE. If you look at the Q-poll, basically ALL self-identified Dems and GOPers are voting for their respective party’s GUV nominees, although Cuccinelli is approaching a critical mass of party defectors at this stage of a race. But as of right now, on a math basis, there isn’t sufficient data to say with statistical certainty that it will definitely hurt.
Meaning: If the Q-poll has 7 percentage points more Dems than Republicans, and the top line of the poll says McAuliffe leads Cuccinelli by 6%, you don’t need to be Dr. Sabato or Not Larry Sabato to do the math: all other things being equal, Terry should have a 6-7 percentage point lead in the poll depending on how you round off the data.
Or put another way: The $20 million or so that has been spent TO DATE by the candidates attacking each other HAS NOT MOVED A SINGLE VOTER on a collective basis.
The Partisans – Dems and GOPers – are doing what they always do, back their respective candidates. At this stage of the race, those who consider themselves INDEPENDENT, are basically split as is often the case when you have a race that is seen as overwhelmingly negative, since these voters are the most repelled by the partisan he said/she said part of the game.
INDEED, if you presume that the 2009 turnout model will prevail again, then the Q-Poll would have Cuccinelli up by 4 percentage points with the same numbers.
INDEED # 2, IF YOU STUDY THE Q-Poll, the normal markers of a VA GUV race make it a DEAD EVEN RACE. Why?
The normative turnout on election day would not be either +7 Democrat or + 4 Republican. It would be closer to +2 give or take depending on what happens in the last month of the campaign.
If you look at the Q-poll, I defy anyone to predict how the large block of independents still undecided will break: unlike the partisans, the independents don’t seem to be getting, on a collective basis, any of those Beach Boy “good vibrations” from the political system here in Virginia these days.
MEANING: If independents split roughly 50-50, then this election, despite all the gazillions spent on negative ads, etc, will be based in large measure on THE TURNOUT OF PARTISANS. That is to say, will be electorate look more like 2009 or more like 2012 on a DEM v GOP basis?
So, the Q-Poll raises this question: Has President Barack Obama changed the politics of Virginia in a fundamental way, creating a new electorate of likely voters in a GUV year election? The Q-Poll says: YES, at least based on their math. But does it do so based on the politics of 2013?
There are at least 3 reasons – 2013 specific – to suggest that while the Q-Poll seems to predict a brand new off-year turnout normative equation for Virginia, this might not be the case, it might actually only be a reflection of the unusual circumstances of this year’s election.
(1) MCDONNELL’S SCANDAL WILL KEEP REPUBLICANS HOME THIS NOVEMBER.
There is plenty of statistical history correlating, or at least seemingly so, a “depressed” turnout for a scandal ridden chief executive’s party at the polls. The Q-Poll has the number of Republicans way, way below any number I have seen in all my years of looking at Virginia polls. We are talking numbers that are below the bedrock. The Q-Poll had a large sample. It was weighted, meaning the pollsters had some statistical model they felt comfortable with using. AND IT CAME OUT A PLUS 7 DEMOCRAT, unheard for a GUV year. Based on other data, you can not attribute this amazing stat to any quality of either major party candidate for Governor [in that regard, Q-Poll didn’t include the Libertarian party dude. It would be interesting to see a three way for analysis purposes, in the end the guy will get a small percentage if that, and so not likely to be a factor unless you have the type of recount margins for say John Warner in 1978 or Doug Wilder in 1989 or of course McDonnell v Deeds in 2005].
SO: Is the McDonnell Scandal factor real, is it going to lead to a silent “no show” GOP protest vote? Or is it spurring a “show up” protest DEM vote? As I say, history suggests that the Republican Governor’s problems – imagine if the guy gets indicted or resigns – are likely to keep a good number of Republicans home UNLESS the Cuccinelli campaign changes its course assuming Q-Poll is correct.
2) THE CUCCINELLI FACTOR IS REALLY ENERGIZING DEMOCRATS
The Q-Poll poll basically shows an amazing stat: near universal support of self-identified Dems for McAuliffe. If it were to hold precisely as in the poll, it would surpass anything ever seen in VA politics on a purely statistical basis. This can only happen with a huge negative energy: there is no positive vibe that can do it. The negative Cuccinelli energy among Dems is confirmed by what seems to be an unprecedented negative energy among a cadre in the GOP who while agreeing with Cuccinelli on most issues for most of their careers, seem to have turned on him with a vengeance never before seen in a VA GUV race at that level. Thus, it is a 2013 specific event, not something one would build into a normative model.
The money gap between TMac and KennyC is going to be more favorable to the Democratic candidate than anytime in the modern era for a statistically competitive race. Warner had a big lead over Earley from jump street and 9/11 froze everything at the very moment an underdog would try to make his move. Earley never was “in” that race on a statistical basis. Truth is, he would have lost if the money gap was reversed.
THUS: In 2013, Cuccinelli faces the 100-year flood type of situation on the money. He has enough to be “competitive” in terms of the dictionary definition of the term. But in reality, the type of campaign he is running can not be competitive in reality over the long haul to election day. He is like the British in WW2, under attack from the German air force. He can’t win unless the (a) the Germans fail to figure out how to use their advantage in material things; and (2) the RAF figures out some brilliant strategies to offset the material advantages not to mention the courage of their pilots.
The Q-Poll could be an “outlier” in terms of the turnout. There is no way, now or perhaps even after the vote in November, to know for sure. A 7-percentage-point Dem over GOP advantage on election days in a GUV year would defy the statistical history of Virginia in the modern, two-party era except during the transitional phase when former Democrats where moving over to the GOP.
But that era is over: this is clear from all the polls including the 2009 McDonnell landslide. He got a small slice of the Democratic vote, enough to win an otherwise 50-50 race, but at the same time, you don’t get that slice in a 50-50 race, the dynamics are far different.
SO are the Q-Poll boys right, have we entered a new era where VA has to be considered a “blue state” due to demographic changes?
Good question, but I don’t have the answer, at least statistically. However, the 7 percentage-point Democratic advantage in the Q-Poll suggests to me the following: right now, this is a scandal-driven election, which means it is an election to be decided by the CHARACTER ISSUE, NOT by the SUBSTANTIVE ISSUES.
“Would we have won a character fight?” famously asks Michael Douglas, playing a President seeking re-election, to his chief political guy, played by actor Martin Sheen (who would go on to be elected President himself, Hollywood wise, in the TV Series The West Wing).
A “character election” for GUV is unusual for Virginia if not unprecedented. A case can be made that 1981 was a character election for Governor, in terms of the dominant theme/issue. As for the others, one or more issues, often hot button, proved decisive in terms of how it played out.
So far, there is no substantive issue that has dominated this election because in large measure, they are playing out by the campaign strategies as character issues by and large.
The Q-Poll gives statistical support for this analysis: and for this being the likely arc of the campaign.
In this regard, a “character issue” election might be logical this year, since it is overdue on a statistical basis and would more likely occur in a scandal-driven situation, none more high profile than now in the history of the state.
The Q-Poll then says the following: net, net, all things being equal, a scandal-driven election favors McAuliffe over Cuccinelli, and should likewise play down the ballot.
This suggests the following to me in closing. Cuccinelli has made a HUGE BLUNDER in NOT RETURNING THE WILLIAMS GIFTS or, in the alternative, NOT AGGRESSIVELY TELLING VOTERS ABOUT being cleared by a Democratic prosecutor as a legal matter (only the voters can clear you as a political matter). It is a blunder worthy of losing an election.
Likewise, Democrats have made a mistake in allowing the GreenTech Auto stuff to get the traction it has IN WHAT SHOULD HAVE BEEN TANGENTIAL AREAS. But by sitting on their lead, they have opened themselves up to stuff THAT THE KAINE ADMINISTRATION WARNED THEM ABOUT IN 2009. This is a totally unforced error in my book.
So while I think the Q-Poll turnout model is not yet proven, the Q-Poll tells me to stay out on the limb for longer, indeed buy up all the seats. Right now, the Democrats are headed for a sweep, although I am surprised that the Q-Poll has independents tied in the horse race. I suspect this might indicate that the McDonnell Scandal has made a certain percentage of Republicans embarrassed to so identify, thus they are calling themselves “independents” but in reality will vote a straight GOP ticket.
But right now, the Democrats are more eager and motivated to vote. And I believe true independents are more disgusted with McDonnell – and thus his party – than the Governor and the GOP high command seems to want to admit.
Indeed, the Q-Poll may reveal a Freudian slip among the pollsters. Based on what I saw, they DIDN’T EVEN BOTHER TO ASK A SUBSTANTIVE QUESTION, it never struck them as important. Personally, I don’t believe this is correct, I do believe issues are important this year. But it may be that the issues that will move voters are, as indicated, those that make character statement, as opposed to something else.
So the Sweep Lives, you heard it here first.