Popularity Contest

Over the past five days, Clinton supporters have been floating the notion that Hillary Clinton is actually winning or is extremely close on the popular vote. It appears that they are coming to terms with the fact that Senator Clinton can't catch Barack Obama in the pledged delegate count -- even with Florida and Michigan in the mix. So they must build the case for the superdelegates to reverse the cumulative result of the primaries

So out comes this notion that Hillary Clinton is winning the popular vote. On Sunday, it was reported that a senior Clinton official had this to say:

The momentum is shifting to us right now. If we are the leader in the popular vote and we have closed the gap in pledged delegates, that's a very persuasive argument.

While last night's 24-point win by Obama might diminish the value of the Clinton's popular vote talking point, the meme is out there and we can expect to see it trotted out once or twice between now and the Pennsylvania primary on April 22.

It's a convenient argument for the Clinton camp to make because it is appealingly simple. And while the premise seems rational (just add up the votes) it's also flat wrong. As has been widely noted, Obama has outperformed Hillary in the caucuses. And vote totals from caucuses are extremely low when compared to vote totals from primary elections.

For instance, consider New Hampshire and Maine. These two states not only border one another. They also have the same population (more or less). Maine has 1,322,000 residents and New Hampshire has 1,310,000.

Maine has caucuses, New Hampshire has primaries. Maine's preliminary caucus results show about 3,500 votes, while New Hampshire's results show about 217,500. A little math shows that New Hampshire's contribution to any popular vote total would thus be about 62 times that of Maine. So, if you simply add up the votes, you punish Maine severely for its choice of format.

It seems reasonable to expect those making the inherently populist argument that the overall vote should weigh on the minds of superdelegates would find a way to fairly deal with caucuses in their calculations. On the other hand, if they are just interested in adding up a bunch of numbers in such a way that it makes their candidate look stronger than she actually is, then ignoring the impact of caucuses on vote totals might be the way to go.

As you will see, there really is no fair way to assess the will of the voters outside of the pledged delegate count. Which leads me to the personal conclusion that the Clinton campaign is engaged in little more than pure spin on this matter.

But let's assume for a minute that there is some practical validity to using the popular vote analysis as an argument for upsetting a lead in the pledged delegate totals. And let's further assume that we are actually trying to achieve a fair evaluation. How would we figure it out? How would we deal with the extremely low values coming out of caucuses? With a mix of caucuses and primary elections, there's no direct way to use votes cast as an overall measurement.

However, one indirect way of conducting such an analysis would be to use the popular votes of John Kerry in 2004 to determine the relative number of Democratic voters in each state. For example, a state like Connecticut might have fewer people than Kentucky but more Democratic voters. Then using primary results, we can establish an adjusted popular vote that proportionately weights each state based on the number of Democratic voters.

So this is the approach that I took in the analysis below. I started with the primary results for each state that's voted so far (except Michigan - see below). For the sake of simplifying things, I ignored all votes going to a candidate other than Clinton or Obama and I calculated the percentage that each candidate got in head-to-head voting (all other votes excluded). I then multiplied each of those percentages (expressed as a decimal) by the number of general election votes received by John Kerry in 2004 in the state in question. This gave me an adjusted popular vote for each candidate in each state.

As a hypothetical example, assume there were a million votes cast in the the primary for a certain state. Clinton gets 55% and Obama gets 45%. Now, if Kerry received 2,000,000 votes in the general election in 2004, Clinton would have an adjusted popular vote of 1,100,000 (55% of two million) and Obama would receive an adjusted popular vote of 900,000 (45% of two million).

Of course, after doing this there is an inflation of votes across the board since more people vote in the general election than the primary. But what we are really after is an understanding of the percentage of Democratic voters who support each candidate. And since this inflation is the only way to understand the relative weights we should establish for each state, we are sort of stuck with it.

Upon doing these calculations for every state, I found that Barack Obama appears to have the support of just over 52% of Democratic voters and Hillary Clinton that of just under 48% (this assumes a two-person race).

Now, there is one hitch to all this. The number given leaves out Michigan. I'm not sure what the best way is to include Michigan. Barack Obama removed his name from the ballot (as did other candidates). Hillary Clinton left her name on the ballot and received 55% of the vote. We can't really give the remaining 45% to Senator Obama but it also would be silly to assume that he would have received zero had his name been on the ballot.

So for the sake of argument, let's give Senator Clinton a generous win in Michigan of 60% vs 40% for Obama. This is significantly better than she did in Ohio But even with under this assumption, Barack Obama would still lead Clinton nationally by 51.47% to 48.53% for a total lead of almost 3 percent (2.94).

Observant readers will have noted that the methodology used fails to account for varying voter turn-outs from state to state. This brings me to the problem of talking about the popular vote at all. There's no good way to do it. If your try to put all the states on the same playing field, you more or less have to assume equal turnout between states (or add in even more assumption about how varying turnouts could be applied). If you try to account for turnout, you again have a bias towards states which hold primaries.

In actuality, we could manipulate these numbers in a variety of ways and none would be completely accurate. Whether a state allows cross-over voters, conducts caucuses or allows early voting will all impact how many votes actually go into the tin.

There is, of course, a certain populist appeal to simply rolling up the votes state-by-state. But that would ignore the differences in how states conduct their primaries. That's why the best measurement of Democratic voters' preference for a nominee is the pledged delegate count. There's no non-partisan need to introduce another standard.

Unless Clinton supporters can point out specifically why the pledged delegate count is not representative of the people's will, talk of Hillary leading the popular vote should be viewed as nothing more than spin.

You can view a spreadsheet detailing this work here.

For more information on how pledged delegate totals have tracked against popular vote, go here. (updated sporadically)

(Note that the Texas caucuses were not used in this article's calculations, only the primary.)

Comments

[S1m0n]

You are dignifying Clinton's desperate snatch at a clutchable straw waaaaay too much with this post.

[NoGoodNamesLeft]

Whoa.... Talk about circular logic. This just amazes me. Throughout this campaign whenever an argument has been made that caucuses aren't as democratic as primaries because fewer people are ABLE to participate in them, it has been met with comments such as anyone who wanted to vote in a caucus could have. Of course this ignores the fact that people who have to work during a caucus time period couldn't participate even if they wanted to. People who are housebound cannot participate. People who have children and cannot afford childcare cannot participate.

Now on the flip side, those who have always defended the idea of caucuses, want to find a way to adjust the total vote from a caucus to approximate what it MIGHT have been if it were a primary. What happened to the theory that caucuses don't disenfranchise voters and they could have shown up if they wanted? And, why would you make the automatic assumption that those who didn't participate in the caucus would necessaily vote in the same percentages for each candidate as those that did? What we do know is that the demographics of the voters who are least likely to be ABLE to attend a caucus are the same as the demographics of voters who support Hillary Clinton.

In fact all your entire argument does is prove how a caucus is not a valid barometer of the actual voice of the people of the states that use caucuses.

When you say there is no vaklid way to count the popular vote, are you saying they don't actually have a count of the number of people who participated in a caucus? If that is true, how are we to trust the results from these caucuses anyway?

Finally, if you want proof that pledged deletgates do not reflect the will of the people, look no further than Texas and Nevada where the people voted for Clinton and the pledged delegate system gave more delegates to Obama. Look no further than the results in any delegate district in any state where there was to be an even number of delegates awarded and those delegates were split evenly unless one candidate won over 65% of the vote. But, an odd number of delgates in a district were split favorably to any candidate winning by 1 vote.

Here's another example of how caucuses don't reflect the will of the voters. You get to keep changing your mind in a caucus. What happened to Mitt Romney in the Iowa caucus? He led the vote in round 1. Had it been a primary, he would have won the state. But, the McCain and Huckabee camps colluded in round two of this process to steal the win from Romney by joining forces to give Huckabee the win.

Here's a thought, get rid of caucuses, they are stupid.

How can any number be any more clear than counting up the number of people who actually voted?

[Markometer10000]

A few comments. I will agree that it is more difficult for people to participate in caucuses. But for whatever reason, the state parties have chosen them as the means by which the people's will is expressed. But overall, turnout is inherently lower in caucuses. That doesn't necessarily mean that the results aren't representative of the will of the people, just that fewer people participate.

You can't simply say the caucuses aren't representative simply because you don't like the results. Are there polls showing that Obama out performed public opinion in the polls? I would be interested in seeing them.

I don't feel that I am making an assumption either way. As I point out at the end of the piece, it is problematic no matter how you look at it. And really the point of the piece is to show that these are apples and oranges and there is no way to accurately roll them up into a national popular vote.

Caucuses are the chosen format of a number of states. Never before has it been suggested that the influence of a state on the election should be diminished because it chooses the caucus format. But that is the tacit argument of those who want to look at the national popular vote.

Also, the delegate count is skewed from the vote in every state. and overall that skew has hurt Obama (even though he's done better in the caucuses). See the last link in my the article for more detail.

Your last statement simply confuses simplicity with accuracy -- which is exactly what the Clinton camp is banking on.

[NoGoodNamesLeft]

i would ask you to back up your position that the results of caucuses actually mirror the will of the people based on public opinion. Are the public opinion polls you are relying based on registered voters or likely caucuse goers?

[Countess]

None of the Obama votes should count. Only the votes for Hillary should count and then the Clintons would stop their relentless smear campaign.

[Thatcher]

That's an interesting response ... if Obama was out of the race, then the "Only thing they have to Smear is Smear themselves..."

[Thatcher]

Actually, Mark - it's even simpler than that ... I understand what you are saying about the "votes" in caucuses - but it's only 4 states where "votes" don't really mean "votes" ... those are the 4 that haven't released hard-count numbers: Iowa, Nevada, Washington, Maine. All the other caucuses had hard counts released (totals are on Real Clear Politics website)

However, they have released "approximate" turnout numbers and "percentages".

Iowa had about 239,000 turn out.

Obama 37.1%, Clinton 29.9%. Do the math:

Obama 88,669 - Clinton 71,461

Nevada had 117,599 turn out.

Delegate votes: 10,742 - Obama 4,844 (45.1%), Clinton 5,459 (50.8%). Do the math:

Obama 53,030 - Clinton 59,763

Washington had about 250,000 turn out.

Delegate votes: 32,220 - Obama 21,768 (67.6%), Clinton 10,038 (31.2%). Do the math:

Obama 168,901 - Clinton 77,886

Maine had 44,667 turn out.

Delegate votes: 3,496 - Obama 2,079 (59.5%), Clinton 1,396 (39.9%). Do the math:

Obama 26,563 - Clinton 17,836

That gives good estimates of:

Obama 337,162

Clinton 226,946

Obama leads by 110,216 in those 4 caucuses. Add that to the 703,742 lead he has now (not counting FL and MI) and Obama has a lead in the popular vote of 813,688 votes.

I've gone one step further doing calculations that show based upon recent polls and the similar increased turnout from 2004-2008 among states left in play (not counting FL and MI) - including giving Clinton a 290,000 vote win over Obama in PA ... that, at most, she adds 450,000 votes to her total (that means that in KY, PA, WV she wins big the ones she is supposed to win and in the ones Obama is supposed to win, I have them as a tie - IN, MT, OR, SD,GU, NC).

She's still short 360,000 votes.

So, under only the best-case (one might say miraculous) circumstances could she hope to get the popular vote and that would have to include FL and MI being seated based upon the January results - which is not going to happen.

[Markometer10000]

Excellent comment.

I find that there's a bit of duplicity from the Clinton camp in even arguing the point because on the one hand they have been making noise about how the caucuses aren't democratic and they depress turnout etc. But then on the other hand, they want to roll all these 'votes' into a national popular vote and then make the case for a superdelegate overrride.

In the system we have there is no better measure of people's will than pledged delegates. And there might be other ways to look at the numbers (as both you and I have done), none can reasobably be asserted to be a better gauge than pledged delegates.

[Thatcher]

Mark - agreed. The rules state pledged delegates - and super delegates should remember those rules when they make their decision on whom to support. The fact is, each time something changes against the favor of the Clintons, they have tried to find some new wrinkle they think will "redefine" or give "definition" to the rule. When the rule states delegates.

[Thatcher]

Sorry - I want to make sure I nip this in the bud - I did forget to type in PR for Puerto Rico in Clinton's "450,000" total - it is included - I just forgot to type in this post that it included PR.

[PioneerKing]

Yo Dude:

What are you talking about? You've gone through all of that convoluted blather for nothing - just count the number of people who've voted regardless of whether it was a primary or caucus and Obama will still be ahead.

Now the reason that the Clinton's want the popular vot to be of greater weight is that there is chicanery afoot. Lots of non-exsisting people, dead people, and folks who should have been purged from voting rolls be voting in the upcoming elections.

Hillary will get her popular vote numbers even if they have to count the cows in PA and the crows in Florida.

[sarah99]

awesome! This is the first in-depth analysis of this issue I've seen. I hope it spreads.