We have to get to the basis of the divergence between base voting for low profile statewide races and the presidential race.  We're never going to bridge the entire gap, but if we can get candidates with the right message we can close most of it.  

More importantly is what's happening in Congressional races.  In many areas Congressional candidates are losing Democrats.

Every few months I am tempted to remove our Cook Political Report ratings and our Partisan Voting Index (PVI) from the public portion of our cookpolitical.com web site, it gets tiresome to respond to people who don't understand what it is,what it is intended to do, how professionals use it and how it fits into the way that our five-person staff analyzes races.  Is it worth it?

By way of explanation, The Cook Political Report and cookpolitical.com is prepared for and used by professionals, government relations, public affairs and political professionals, as it has since founded in 1984, whether they are corporate, labor union, trade or professional association of any other interest group.  Their lobbyists and PAC directors use it, along with a couple of other comparable publications/sites, it aid in their own analyses of races and to help make campaign spending allocation decisions.  These groups range from the left to the right, from environmental, pro-arms control and anti-gun groups on the left to their opposites on the right and everything in between.

When they subscribe to The Cook Political Report, they get rather exhaustive narative analyses of the top Senate, House and gubernatorial races, based on interviews with the candidates in most competitive races, their managers and consultants, sources and journalists in the states/districts, watching the ads and local news coverage and based on our own experience, with over 80 years of political experience between myself and our Senate and House experts.

We invented the Partisan Voting Index in order to have a single-objective measurement of how a district votes in presidential elections compared to the rest of the country.  One single number to look at, based on the average two-party vote in the last two presidential elections.  It is the start of our analytical process, not the middle and certainly not the end.  It's a short-hand way to look at a district at the outset of analyzing it.  We then look at hundreds of other factors, including face-to-face meetings, conversations with camapign consultants, many, many other factors.  If our analysis process could be put on an A-Z scale, looking at the PVI is A, looking at the districts with comparable ratings and seeing the party split among those is B, then we move on to many, many other factors.  Anyone who began and ended their analysis with the PVI would be a fool.

But it is a very useful tool.  If a district has a PVI of R+6 (meaning that the district voted six percentage points more Republican than the country as a whole did), and you look at the other districts that are rated R+6, and for that matter, R+5 and R+7, and see how many (or few)Democrats occupy districts with comparable voting patterns, that is a tipoff that something has to be pretty darn unusual (scandal, tidal wave election) for that district to turn from Republican to Democratic.  But then, you begin examaining and looking for any of those special circumstances that might create that exception.

I thought it interesting and amusing that many, for that matter, many Democrats and liberals, in fact many who periodically post on MyDD and on Daily Kos, were very slow to pick up on the strong pro-Democratic wave in 2006, because for many, their political experiences and expectations were formed in the 1996-2004 period of elections, five election years in which the Tip O'Neill adage of "all politics is local" certainly fit.  But those of us who had witnessed contemporaneously the 1994, 1982, 1980 and 1974 elections, when all politics was anything but local, could see that something very different was happening.

Anyone taking 2006 election results and expecting that to be a good starting point or reference point for 2008, or 2010 would be just as wrong, or more accurately, it would be like someone who expected the Newt Gingrich led tidal wave election to be replicated by similar tidal waves in 1996 and 1998, which obviously didn't happen.

Can Democrats overperform, or underperform their PVI?  Sure.  So can Republicans.  But presidential voting is the only single measurement that can be used across all 50 states and all 435 congressional districts because it is the only kind of race where the same candidates are on the same ballot on the same day.  Use it, but don't expect it to perform miracles.

And don't assume that when you see a Cook Political Report race rating or PVI, you are seeing the totality of what we do and what we are saying about a race and how we reached those conclusions.  99 percent of our content is behind the curtain for our paid subscribers.

There are no magic formulas in this business.  You can start with quantitative data, election results and census data, and later on polling data, but that is where the science ends and the art begins.  It is analysis based on a multitude of other information, and experience having done this for years.  It doesn't mean that we get everything exactly right, no one can do that.  But we come pretty close.

During the first week of August I began writing in my columns that if the election were held then, the House would turn Democratic, or to say it differently, that we were on a trajectory toward Democrats winning the House, and something would have to happen to change that trajectory.  We also said then that the Senate was becoming an increasingly close call, and was nearing the point of being a 50-50 proposition.  That was three months before the election.

Our final projections were a GOP loss of 20-35 in the House (the midpoint was 27.5, the final was 30), four to six in the Senate (final six) and 6-8 in the governorships (final eight).

The PVI didn't make those projections, but we did with a process that started, but didn't end with the PVI.  

Let's get this straight.  Republicans win the Ohio governorship in four consecutive elections (1990, 1994, 1998 and 2002).  Republicans going into the 2006 election with every single statewide office, last time I heard, even every supreme court justice but that might have changed.  But something like all 12 statewide offices.  

Then we have a bottom-falling out, heavily Democratic election in 2006, and this guy builds a model analyzing party strength by county for the state using just 2006 returns, and says that the state preforms almost 55 percent Democratic, ignoring the simple fact that it was an extraordinary election, weird circumstances (governor pleading guilty, massive state scandals) and wants to extrapolate out from just 2006 elections.

Wow.  I am going to watch anything this guy writes from here on out very, very carefully.  I'd fire any employee of mine that used that faulty analsis.

Agreed.

I think that the best thing would be to try to get figures for 2002, 2004, and this year,  That way we can get a trajectory also.  That can be a lot of work depending on the shape the data is in.

My plan was to finish the Graphic Anatomy of Victory series, and then go down to the precinct level in close races.

I jusr want to up the level of analysis for races, so we base assertions in numbers not notions.

That's why our PVI is calculated using the two most recent presidential elections.  Going back any farther than that is not as useful because over a decade or more, the nature of districts change, becoming more urban, more suburban, more or less Democratic, what have you.  

Admittedly this is a crude instrument, we created it and are the first to admit that.  We would never recomend that anyone depend solely on it.  But it is the best there is.

This is very interesting.  But I have a question - Both states you showed were biased against Dems in the PVI index.  Is this the case in ALL states?  Or is there a group of states or a region where the bias goes the other way?

I don't know, but I imagine that in highly Democratic states in presidential races, the opposite would be true.

The jumping off point came from when I worked at the Indiana House during redistricting.  This was during 2001, and the guy drawing up the maps used a slightly modified set of state races.  I think he used 2000 Gov, SOS, and Attorney General.  

My theoretical approach is that in low profile races information is low, so most voters really don't know who the candidate are or what the position does, so they voted according to something they do know, party.  Most voters are rationally ignorant.

Marketers can take advantage of rational ignorance by increasing the complexity of a decision. If the difference in value between a quality product and a poor product is less than the cost to perform the research necessary to differentiate between them, then it is more rational for a consumer to just take his chances on whichever of the two is more convenient and available. Thus, it is in the interest of the producer of a lower value product to proliferate features, options, and package combinations until the average shopper finds it too much trouble to make an informed decision.

This also works for politics. By increasing the number of issues that a person needs to consider in order to make a rational decision between candidates, they can level the playing field by encouraging single-issue voting, party-line voting, and other habits that tend to ignore a candidate's actual qualifications for the job.

Another, more nuanced, political application involves a voter's identification with a political party, much like the adoption of a favorite movie critic. Based on prior experience a reasonably responsible voter will find politicians or a political party that will draw conclusions similar to their own conclusions when the voter had the time to do the analysis. This is just like the way we adopt a movie critic that likes movies that we like. In this way we let the political party, the politician, or the movie critic do the "heavy lifting" for us as we spend our time doing our job, raising our family or just lying on the beach.

The United States is very different from most countries in the number of races we vote for at one time.  In other countries voters cast ballot far fewer races at one time.  Like maybe 10 races maximum at one time.  Here in the US, we often have President, US House, US Senate, Governor's, State House, State Senate, County Council, County Commission, Sherriff, loads of ballot measures, etc.

Eventually most voters who are really interested in the process say to themselves "F**k if I know" and ether cast no vote (a 5% falloff is typical in Indiana and Ohio) or rely on party as a cue to whether the candidate thinks the way that they do.

The PVI doesn't look at party identification, it looks at voting in the last two presidential elections, which tends to correlate very highly with congressional vote in competitive elections.  Obviously when a district is uncontested, or undercontested, with a warm body but not a serious candidate, there is a large differential.  When it is a competitive race, it runs reasonably close.

I emphasize that the PVI is not a predictor.  Only a fool would think it is.  But it is a logical first step of analysis of a district.

Maybe.

That's a long term plan of mine.

It's a time and money issue.

First, I'm going to finish this Graphic Anatomy series which will show them at the county level.

of 233 Democratic Memembers of Congress 9, count them 9 came from districts where George Bush got 60% or more. I am not saying give up, but polarization is clearly  in existence. Indiania  is an interesting exception, which needs to be looked at, also, the Gore-Kerry might be a phase Whereas I think Clinton did better in all three of the districts dems picked up.

Ideally, instead of using a mix of three downballot statewide races, you'd choose one statewide race as an accurate indicator of base partisanship, using the specific characteristics of that race as the reason for choosing it.

So, if studying Nevada, you'd toss out SecState, cause that race obviously had significant candidate effects.  There is almost always one race, however, that has either two typical and evenly matched candidates, or two completely and utterly unknown candidates.  Either will work as an accurate indicator of raw partisanship.

For instance, in California in 2006, the governor race (Schwarzenegger v Angelides) is a terrible indicator.  The Lite Gov race (McClintock v Garamendi) is still no good.  The Insurance Commissioner (someone v Bustamante) is awful.  And Attorney General (Jerry Brown v Poochigian) and Treasurer (Lockyer v somenobody) aren't fair either.  But controller (nobody v nobody) is fair, and is what a lot of people used to measure base partisan turnout.  In Texas, John Sharp v David Dewhurst 2002 is considered the indicator election.  And it might even be the case that in Nevada, governor was the most solidly and evenly contested race and therefore the best indicator.  But usually there is some good election to use, and it's usually not the presidential.   And ideally it's not an average of three random downballot tickets either, but if you can't spare the extra research to figure out the right election to use, it's a pretty decent approximation.

Seriously, how can anyone assert base Democratic partisanship is 54.1% when we haven't owned a major statewide office in many cycles?

Well there are more registered Democrats than republicans in Ohio aren't there?

And as far as not winning a state-wide election for a long time, doesn't that speak to manformiddletown's point that you need run canidates that are appealing to rural as well as urban areas so that the oft-times favorable base-partisanship in rural areas will extend to the up-ballot races as well as the down-ballot races?

Also, the Democrats' share of the presidential vote has been increasing in Nevada now for the last two decades hasn't it? As union strength continues to incease in Las Vegas wouldn't you expect the republicans support to continue to erode away? We're one seat away from taking over the senate now aren't we?

Gary is absolutely correct.  I wouldn't use 2006 (or 1994) as a benchmark for anything other than, I a specific Democratic candidate performed in an election with hurricane force winds at his/her back.  Anyone thinking that this will be replicated in 2008 --- well I would just suggest that they have not been watching politics very closely.  

The last time Democrats enjoyed a comparable political climate to 2006 was in 1974, the last time before that was 1964, the time before that was 1958.  1982 wasn't like this.  Four times in a half century, only once in last 30 years.

For Republicans, the last time was 1994, before that 1980, before that 1966.

Basing any exhaustive analysis of voting patterns solely using 2006 is an inherently flawed approach.  2006 was a change versus status quo election. Democrats controlled nothing, Republicans controlled everything.  That will not be the case in 2008, things will be much more complicated, looking at 2008 through a 2006 prism is insane.

True, there are some difficulties in rural areas, but there are also some advantages. Your point about voter registration in Indiana is probably the most difficult to overcome but the others are not so much.

I'm originally from a rural precinct in central Illinois and my mom and I worked on our precinct's first ever GOTV plan this time around (my mom is Democratic precinct chair).

It's actually not too hard to get your voters out to vote. For one thing, turnover in voters is usually far less in rural areas than in urban areas. So updating the information is not as time-consuming a task.

In addition, you personally know a greater percentage of your partisan (or independent) voters and have more opportunities to persuade them.

And as far as the difficulty of logistics in organizing. I mean, it's not easy, but it's not like rural voters live on the top of mt. everest. You're still likely to find most of your voters living in the main population center of the precinct (i.e. a small town). Which means canvassing and communicating is no different than in a precinct in an urban area. I would say that 75-80% of Democratic voters in our precinct lived in town. The 20-25% that live in rural areas are harder to reach just because you have to drive to everyone's house so it takes longer to talk to fewer people. BUT, since there is often also a better connection between the PCO and the neighbors, a phone-call usually works just as well as a door-to-door visit.

So I think the most important thing for mobilizing rural voters is to just follow Dr./Gov/Chairman Dean's plan. Do the 50-state strategy. Sign-up as the prencinct chair in your local precinct. Get your voter list, get 2-5 volunteers and start canvassing and GOTVing. I don't know for sure yet, but I'm guessing we upped our turnout by at least 10% this time over 2002. And our Democratic turnout was close to 80% (maybe higher).

It can be done.