Math: Revised Predictions PA Turnout/Popular Vote Ranges

Like the Blowout Principle is the mathematical undermining of the Slate delegate calculator, the Closed/Open Principle is undermining turnout predictions for the remaining states.

A couple weeks ago, I wrote a diary that attempted to predict turnout, but it wasn't a full analysis.  I believed it a sounder model than those predicters who were merely throwing spaghetti at the wall, because I tied the turnout to congressional districts and past performance in the 2008 primary cycle, distinguishing between closed and open/semi-open primaries and comparing like to like.

Still, I began seeing different numbers and methods around the web.  Over at Open Left, fladem tied predictions of turnout straightforwardly to 2004 GE Kerry performance.  PA turnout was pegged at 75% of 2004 Kerry GE turnout, based on an average of 64% of Kerry turnout in primaries and 80% of Kerry turnout in primaries post-Super Tuesday.  I wanted to take a closer look because some 2004 GE states were battlegrounds and others were not.

Michael Barone, favorite slurp target of belligerently non-elite Morning Joe Scrangstownacola Scarborough, tied turnout predictions to things he pulled out of his ass.  Or maybe he asked The Hacktastics (Clinton subsidiaries Sean Wilentz and Craig Crawford) to cook up some numberzes.

The bottom line was I didn't like my first pass because I was uncomfortable without looking closer at blue-ness of the states and wanted a deeper look.  I am suspicious of tying turnout in a contested primary to turnout in the 2004 general election, when many states were not contested (which affects the average).  

Also, the recent PA registration surge made me re-test my own theory and find it incomplete.  [With Friday, April 11 as the NC registration deadline (the numbers will be updated for a few days afterward), PA leads in Dems by about 391,000 over Dems+Indies in NC (4,190,064 to 3,799,076).]

First, a few observations.

Observation #1: Three PA-bordering closed primary states and nearby Connecticut had the following turnout of registered Dems (and there were big drives/big publicity in these states too):

Delaware - 37.3% of registered Dems, 48.2% of Kerry GE vote
New York - 37.5%* of registered Dems, 43.2% of Kerry GE vote
Connecticut - 51.1%* of registered Dems, 41.5% of Kerry GE vote
Maryland - 49.3% of registered Dems, 65.8% of Kerry GE vote
*record

Observation #2: In 2002, Pennsylvania turned out roughly 1,240,000 votes in a highly contested Rendell-Casey gubernatorial race.

Observation #3: Fladem says PA = 75% of Kerry vote and thus just over 2.2M.  That's not inconceivable, but you have to put it in closed primary context and see why it's a high prediction.

Observation #4:
In closed primaries, the average turnout has been 55.39% of that state's 2004 Kerry GE vote.
In semi-open primaries, the average turnout has been 63.46% of that state's 2004 Kerry GE vote.
In open primaries, the average turnout has been 74.99% of that state's Kerry GE vote.

Some may say that the 6 week focus will inflate the number over other averages, and this is partially persuasive to me.  But it's also persuasive to me to say that these primaries are still not generating as much as the Kerry vote, meaning we can't get too carried away with equating a primary to a general.  Many of these numbers in the 40% range are record primary turnouts.  Record Democratic primary turnout was set in 12 states through Super Tuesday:  Alabama, Arizona, Connecticut, Georgia, Illinois, Massachusetts, Missouri, New Hampshire, New Jersey, New York, South Carolina and Utah.  Since these records are still mostly in the 25%-50% range, it's not ok simply to blithely say PA will be higher than 50% of registered Dems without a lot of supporting evidence/argument.

CLOSED PRIMARIES
State        2008 Turnout    #RegDems    %RegDTurnout
Arizona              455,635        904,741            50.4
Connecticut       355,561       695,894            51.1
Delaware              96,374       258,300            37.3
Maryland           878,174    1,782,541            49.3
New York        1,862,445    4,966,942            37.5
Oklahoma          417,207    1,012,594            41.2
                                                      avg. = 44.5%

Pennsylvania   as of apr 6    4,190,064
Kentucky      as of mar 17   1,614,250
Oregon             as of Feb        775,119
South Dakota                        n/a yet

CLOSED PRIMARIES
State          2008 Turnout    #KerryVotes    %ofKerryTurnout
Arizona              455,635            893,524        51.0
Connecticut        355,561           857,488        41.5
Delaware              96,374           200,152        48.2
Maryland            878,174        1,334,493        65.8
New York         1,862,445        4,314,280        43.2
Oklahoma           417,207          503,966         82.8
                                                             avg. = 55.4%

Pennsylvania                          2,938,095
Kentucky                                  712,733
Oregon                                     943,163
South Dakota                            149,244

Note that Oklahoma's turnout has distorted the average in the second table.  There was a lot of room for Dems to improve enthusiasm for Dems in a place like Oklahoma, but but in much more Dem-friendly and Dem-stable mid-Atlantic states the range of Kerry turnout runs from 41-66%.  Of course, one could argue the other closed primary states did not see contested general election races in 2004.  None were battlegrounds (arguably Arizona slightly).  Meanwhile, Pennsylvania was a 2004 battleground, meaning much more effort was put into building Kerry turnout there.  If nothing else, that should cancel out the factor of more time focused on PA than the other states in the primary season.

Looking ahead to NC

SEMI-OPEN PRIMARIES
State           2008Turnout       #RegD         #RegI      #DITot   %TurnRegD     %TurnRegDI
New Hampshire   205,040     258,556      356,023      614,579      79.30        33.36
California         5,066,978   6,749,406   3,043,164   9,792,570      75.07       51.74
Massachusetts 1,254,437   1,472,707   1,987,053   3,459,760      85.18       36.26
New Jersey      1,108,044   1,170,644   2,798,817   3,969,461      94.65       27.91
Utah                  131,403      125,992  *1,040,000   1,165,992    104.29      11.27
*estimated

North Carolina as of Apr10: 2,577,894    1,221,182   3,799,076
West Virginia                        n/a yet

SEMI-OPEN PRIMARIES
State               2008 Turnout    #KerryVotes    %ofKerryTurnout
New Hampshire        205,040        340,511      60.22
California               5,066,978    6,745,485      75.12
Massachusetts       1,254,437    1,803,800      69.54
New Jersey            1,108,044    1,911,430      57.97
Utah                         131,403      241,199      54.48
                                                            avg = 63.46%

North Carolina                           1,525,849
West Virginia                                326,541

The conclusion from the NC portion of this data is unclear.  As a proportion of Registered Dems, NC would look likely to be very high, somewhere in the 1.8M-2M range.  But in the D+I proportion, we'd look to around 1.3M-1.4M turnout.  One of the problems is the wide disparity in the previous semi-open states' proportion between Ds and Is.  New Jersey obviously has many more unaffiliated voters than registered Dems, whereas California is the reverse.  NC has a proportion closer to California with Ds outnumbering Is/unaffiliateds roughly 2-to-1.  

Clearly Massachusetts is as blue as California, yet the percentages of unaffiliateds in Massachusetts is much higher.  I think the distinction is somewhat artificial, based on state by state quirks.  For example, I always voted Dem but only recently had my party registration reflect Dem (was always unstated or independent).

My instinct tells me something like 40% of the eligibles will vote in NC.  That feels about right.  Voter registration is 3.8M and counting with a couple days and official tallies still to tabulate.  I am going to expect roughly 1.5M-1.6M in NC for now, which means I am recanting my prediction in an earlier diary that NC will exceed PA.

Finally, here's a range chart of PA turnout and popular vote margin.  Sorry it's so small, here's a link to a well-sized googlesheets view:

Random Thoughts:
1. Jon Corzine was trying to create a Corzine Club like the Pelosi Club, but the media interpreted his gambit as possibly bailing on Clinton.  Since he was trying to establish popular vote (and he is an absurd person who thinks 0 votes for Obama in Michigan is a legit result), as a justified principle for supers to rally around, I enjoyed the laughable failure of his effort.  In his dismay that such a club was not established, he even wrote a blog at Huffington Post insisting he wasn't vacillating on Clinton.

2. The whole popular vote in the primary compared to the popular vote in the general is dishonest.  New York has more Dems than Texas, yet Texas gets a million votes more say in who the Dem nominee is because it arbitrarily opens its primary to independents and Republicans?    You cannot get this discrepancy in the general.

3. It's also dishonest because had popular vote been the rule, Obama would have spent far more than 1.5 days in California.  Hell, none of the candidates would have campaigned in Iowa, they'd have all been in big states from the beginning of 2007 since only individual voters overall would matter, not delegates.

4. It's also dishonest because it assumes caucus states like Minnesota (which has 72 delegates) would have purposely taken 1/4 of the voice of Missouri (also 72 delegates) in determining the nominee by choosing the caucus method to dilute their voice into near-nothingness.  Missouri had about 823,000 and Minnesota 214,000 turnout.  In the general election, a vote equals a vote equals a vote because every person voting for president is voting for the same main candidates.

5. Sean Wilentz' ridiculous tripe is wonderfully obliterated by Jon Chait.  The main point?  Obama is winning the popular vote, yet Wilentz thinks it would be more "democratic" to change the system so that the person losing the popular vote is winning the delegates.

•  Been a little while

  since I wrote a math diary re: the pretty motherf***er.

  by PocketNines on Thu Apr 10, 2008 at 09:34:13 PM PDT

  •  Your math diaries get an auto-rec from me

    thanks for your research.

    by 1 20 2009 on Thu Apr 10, 2008 at 09:41:36 PM PDT

    •  It took forever to track down some of the numbers

      I wish there was a simple site for voter registration by state and breakdown of Dems/Indies/Republicans, but lots of states don't even seem to keep track of that info, typically the open states (they might keep track of race/age/gender etc. info instead).

      by PocketNines on Thu Apr 10, 2008 at 10:00:34 PM PDT

      •  I've been hoping we'd hear from you!

        You're our go-to guy on these complexities and I've been eager to hear you weigh in on PA.

        by Bugsby on Thu Apr 10, 2008 at 10:15:42 PM PDT

  •  Love that video, thanks

    (Good diary, too)

    by Nancy in LA on Thu Apr 10, 2008 at 09:58:48 PM PDT

  •  thanks for doing

    although actually I think we are all somewhat pulling numbers out of various orifices.

    In Pennsylvania, do we not have to adjust whatever base numbers with which we come up by the additional factor of the increased Democratic registration, both by new registrants and switchers? And dlo we really have any sure basis of determining how those people will split on the 22nd?  Based on the anecdotal evidence, I would argue that they will split no worse than 3-2 for Obama, and that we should expect a 70-80% turnout rate as the floor for their participation.  I am not sure how many of the current polling models include such people as likely voters, since they have never voted in a previous Democratic primary, and some have not voted in the 2006 general.

    Just an additional factor or three to consider.

    by teacherken on Fri Apr 11, 2008 at 04:12:28 AM PDT

    •  I have to believe that every state

      showed a registration surge prior to its primary and that the recent registrants voted in higher proportions than the existing base of voters.  So I think it cancels out.  I think it's close enough to be not worth drilling into and more fits in the "English" you put on the ball when you predict what turnout percentage compared to the average turnout for closed primaries will be.

      by PocketNines on Fri Apr 11, 2008 at 02:46:56 PM PDT

•  Go Pocketnines

  As usual, a brilliant diary which, as usual, gets to the heart of the math.

  Incidentally, I was watching Olbermann last night and he mentioned a study which predicted that had the states which had caucuses held primaries instead, Obama would have 1.3M more votes instead of 7M (not including the caucuses of Maine, Iowa, Washington State, and Nevada of course). What are your thoughts on the matter?

  by NMLib on Thu Apr 10, 2008 at 09:41:04 PM PDT

  •  Oh, and forgive

    my overuse of the words "as usual". I don't usually say as usual, but this unusual and so I will repeat usual against my usual behavior of not saying usual :P

    by NMLib on Thu Apr 10, 2008 at 10:14:38 PM PDT

    •  Here is the link to the

      Wharton School study the KO talked about:

      http://www.dcourage.com/...

      by hyper on Thu Apr 10, 2008 at 11:40:06 PM PDT

  •  My thoughts are that what's important is

    the ideas of different projections of the caucus turnout are out there - to emphasize the critical idea that all votes are not created equal in the primary system.

    I don't know about that particular study, I heard it on KO as well, I am just glad it exists as backup in the event we get into that kind of debate.  If we do get into some dumb debate about popular vote, we have plenty of logical ammo to fight that fight as I noted above.

    by PocketNines on Fri Apr 11, 2008 at 12:13:11 AM PDT

•  Taking Barone at his word

  as follows:

  "Of course my projections could just be plain wrong. Clinton could win Pennsylvania by an unimpressive margin on April 22 and get clocked in Indiana as well as North Carolina on May 6. Then you might see a cascade of superdelegates toward Obama, and the race might effectively be over. But if all those three things don't happen, then I am sure the contest will go on through June 3. And in that case I think my projections are within the realm of possibility."

  PA and NC are headed exactly that way; IN's "clocking" is yet-to-be-seen, but she really needs a solid, double-digit blowout to counteract his certain one in NC that day.

  As for Barone himself, he's actually got a helluva lot of experience at crunching, going back 30 years now. Pardon me while I return to "The Almanac of American Politics 2008" (2006 elections), which I bought last summer as soon as it came out, but haven't quite finished yet.

  by newdem1960 on Thu Apr 10, 2008 at 09:43:09 PM PDT

  •  Well, he sets all the turnout predictions at

    63% of Kerry numbers for closed primaries and 82% for open primaries without explanation, well above their averages to date.

    He arbitrarily gives PR 1 million votes, despite them having an 8am to 3pm primary and having the same number of people as Oregon and Kentucky.  Oregon is a mail-vote system, so the turnout should be enhanced, but the reduced hours for PR should diminish that turnout.  Yet he suggests PR will be nearly twice as big as Oregon in turnout.

    And Clinton winning by 20% in Montana and South Dakota is an instantly discrediting prediction.

    by PocketNines on Thu Apr 10, 2008 at 09:50:25 PM PDT

    •  MT and SD numbers

      are whacked - he oughtta know better!

      PR does have much higher turnout rates than most mainland states of comparable size; if the race is still (technically) contested I think they will have significantly higher numbers than KY.

      by newdem1960 on Thu Apr 10, 2008 at 09:55:53 PM PDT

      •  I think we're close to mythologizing

        Puerto Rico.  I saw him suggest 80% turnout for PR in a primary, despite restricted hours (no evening vote).  The notion that people love democracy everywhere but here is fueling it.  

        This is also the same guy who insisted that all of PR's delegates would go in a bloc and not by proportion, when that is patently false and easily researchable.

        by PocketNines on Thu Apr 10, 2008 at 10:03:07 PM PDT

        •  Barone and PR

          He does have a disclaimer, specifically, that the numbers he inserts there are "pure guesswork". it is sloppy of him to use Cost's chart with a "slightly exaggerated" note I agree.

          by newdem1960 on Thu Apr 10, 2008 at 10:09:41 PM PDT

          •  Given the tendency of the Moroning Joes

            of the world to seize those ideas and run with them right into the media narrative because of Barone's "name," this sloppiness is totally unacceptable and he deserves mockery and derision for it.

            by PocketNines on Thu Apr 10, 2008 at 10:27:24 PM PDT

    •  He didn't really say that?

      HRC winning MT and SD by 20%?  No, wait, HRC winning MT and SD, period?

      Great diary again, great breakdown of the numbers re: turnout, it'll be interesting to see what happens on the 22d.

      And thanks for the links to Wilentz and Chait.  Did you read the comments under the Chait story?  One Clinton supporter said, and I quote: "The fundamental point...is the sense of many Clinton supporters that this campign has strong elements of illegetimacy and unfairness....The sense of illegimacy comes from the role of the caucuses in Obama's totals and his failure to win in large states or among Democrats....
      While one may dispute their feelings and beliefs, the sense of unfairness among Clinton supporters is very strong and quite widespread."

      Yeah, winning more states, delegates and primaries under the rules as put forth by the Democratic party, which all candidates agreed to and are only being challenged by one side after the fact (in the specific areas where they work to her disadvantage) is "unfair and illegitimate".

      At this point it appears the only "fair" thing in the eyes of Hillary's supporters is to let her win, and anything that prevents that is "unfair" because, you know, she SHOULD win, because she's Hilary.

      by synchronicityii on Fri Apr 11, 2008 at 10:06:10 AM PDT

      •  He called it "optimistic" but not "wildly

        unrealistic."

        I call it fucking insane.

        by PocketNines on Fri Apr 11, 2008 at 04:10:17 PM PDT

•  where are the pretty pictures

  such as the ones you put up with your mississippi and wyoming work?    nice work.

  by beeninthewoods on Thu Apr 10, 2008 at 09:43:23 PM PDT

  •  I still have one of the Wyoming pictures-

    As my desktop.

    PA is very photogenic!  We want pix!

    Thanks for your cogent argument.  Heh.  Cogent.  I don't get to say that very often, but it always applies to your work, P9s.

    by Red no more on Thu Apr 10, 2008 at 10:04:04 PM PDT

•  Something I've been wondering about

  How do you account for the fact that Obama got 246,000 more votes in his home state than Clinton got in New York, when Illinois has only about 64% of New York's population?

  by Pangloss on Thu Apr 10, 2008 at 09:48:18 PM PDT

  •  Illinois was open, NY was closed.

    Big factor.  Open and semi-open primaries get about 40% more turnout than closed.

    by PocketNines on Thu Apr 10, 2008 at 09:51:26 PM PDT

    •  Did not know that re: NY

      What do you think the percentages would have been if it were open?

      by 1 20 2009 on Thu Apr 10, 2008 at 10:02:51 PM PDT

      •  Well on average 40% more

        Independents, who would have helped Obama close that gap.

        by PocketNines on Thu Apr 10, 2008 at 10:04:04 PM PDT

        •  How close would Obama have come to winning?

          assuming his normal rate of Indy advantage.

          by 1 20 2009 on Thu Apr 10, 2008 at 10:07:54 PM PDT

          •  I don't know, I haven't done that analysis

            He doesn't win indies by such an overwhelming margin that it would have changed the outcome, it just would have been closer.  Hard to say.  But he'd have gained more delegates and the popular vote total would have been even more in his favor today.  It shows how illegitimate the notion of popular vote in primaries due to inconsistencies is compared with a consistent popular vote count in the general.

            by PocketNines on Thu Apr 10, 2008 at 10:13:45 PM PDT

            •  By the way

              I think your diaries are incredible and I'm awed that you and Chuck Todd are e-mail buddies!

              by newdem1960 on Thu Apr 10, 2008 at 10:16:56 PM PDT

              •  Well, I wouldn't say that we're buddies...

                I just emailed him to ask a couple questions once and he replied, no big deal.

                by PocketNines on Thu Apr 10, 2008 at 10:25:23 PM PDT

            •  Agreed

              but hey how else can you convince low information voters that the nomination is being stolen from you despite losing by every potentially meaningful metric?

              by 1 20 2009 on Thu Apr 10, 2008 at 10:18:31 PM PDT

            •  So let's get this straight

              In Obama's home state, the place where folks are most familiar with him, and a state that favored Kerry by 10% in '04 and Gore by 12% in '00 (IOW, solidly Blue), Obama won an open primary by 31% (64-33).

              Meanwhile, in Clinton's home state,  the place where folks are most familiar with her, and a state that favored Kerry by 18% in '04 and Gore by 25% in '00 (IOW, incredibly Blue), Hillary won a closed primary by 17 points (57-40), and the margin would almost certainly have been less had it been an open primary (my quick back of the envelope calc shows that, using your estimate of 40% greater turnout in an open primary, had Obama taken those indies/Repubs 55-45, the margin would have been ~54-44.  Heck, even if they split those voters 50-50 it would have been 55-43.

              So, the folks who know Obama best love him, the folks who know Hillary best...not so much

              by synchronicityii on Fri Apr 11, 2008 at 01:23:21 PM PDT

•  I know that

  its great for the candidates to have a break and consolidate and get out there campagining

  - but I so can't wait for the next primary to roll on April 22.

  by shiobhan on Thu Apr 10, 2008 at 09:49:25 PM PDT

•  1,627 is the only number that matters

  The popular vote argument is an argument but it is not a winning argument for all the reasons mentioned.  I know it, you know it, Chuck Todd knows it and most important, the superdelegates know it.  Even though few have publically come out to "join" the Pelosi Club, I'm willing to bet that at least 80% of the undeclared superdelegates agree with it.

  We are about 210 short of 1,627 and it will achieved on the night of May 20 when it is likely that the good people of Oregon (maybe Kentucky) push us over 1,627.  On May 21, I believe that David Plouffe will declare victory in some form and the drive will begin in earnest to get the approximately 100 additional superdelegates to push us over 2,024 by about June 3.      

  by mboehm on Thu Apr 10, 2008 at 11:40:54 PM PDT

  •  Yeah, I agree, May 20 is the night that happens

    And I am sure Plouffe will declare victory.

    I also think that May 6 is the more significant date because that is the date when there's just such a tiny percentage of real estate left.

    I agree, let them keep playing out the string, but the reality is that after May 6 all of Superdelegatia will be talking and feeling endgame.  He'll probably be sitting at around 30-36 pledged delegates shy of 1627 and mathematically it's impossible for him not to get those in WV, KY, OR, PR, MT and SD.

    by PocketNines on Fri Apr 11, 2008 at 12:17:13 AM PDT

    •  Well, technically

      It's not "mathematically" impossible, as he could get under 15% of the vote in those states.  :-)  Although the only way that happens is if he starts a rally by abandoning his stump speech to engage in goatf**king onstage.

      FWIW, I think that bigger concern re: Hillary will be the money dramatically drying up after May 6.  Heck, it's already happening now.

      by synchronicityii on Fri Apr 11, 2008 at 10:16:10 AM PDT

      •  True, true. He only has to be viable

        in every CD and statewide total enough to get 1 delegate.  That gets him 36 if exactly 1 in every district and PLEO/at-large breakdown.

        I'm optimistic about Portland, Oregon!  Ha.

        by PocketNines on Fri Apr 11, 2008 at 02:51:06 PM PDT

•  Thank you for pointing out

  the difference between closed, semi-open, and open primaries in turnout.  Using the national popular vote to determine our nominee is already undemocratic enough considering the differences between caucuses and primaries, but even primaries can draw different levels of turnout depending on whether independents and/or Republicans can vote.

  Which is why we use delegates.

  by Skaje on Fri Apr 11, 2008 at 01:32:33 AM PDT

•  Living in PA

  I'm seeing a turnout of 1.8-2.1 mil.  Probably closer to the lower figure.  My guess is the currently registered Dems (3.8 mil) will turnout at about 40-45%, and the new Dems (400 K) will turnout at 75%.

  by IhateBush on Fri Apr 11, 2008 at 05:29:37 AM PDT

  •  That's about 43% turnout

    Which would be about right (I voted for 45% in the poll).  An 8% win for Clinton gives her about a 144K bump.

    by PocketNines on Fri Apr 11, 2008 at 02:53:44 PM PDT