Introduction
I read Nate Silver’s 538 (“538”) article entitled What Would Be the Result if Different Polls Were Used for the Third Democratic Debate. Also, I saw how Tulsi Gabbard felt the DNC, without malice toward her, made poor decisions about selecting pollsters. As a follower of polls, all of this led me to believe the situation was worth, as Steph Meyers would say, A Closer Look”.
Was Tulsi Gabbard a victim of poor decisions by the DNC or just a crybaby? After the “Closer Look” I’m convinced the DNC polling was incredibly negligent in apparently making no inquiry into the polling intentions its anointed pollsters. It gave apparently gave no consideration for a candidate that would emerge from the 0-1% crowd after the second debate and build on that during August to join the 2% and more crowd. It could have been any of the ten in the 0-1% crowd, it just happened that it was Gabbard.
The criteria set by the DNC for inclusion in the debates was essentially “Show us you have (1) numerous donors and (2) a minimal of 2% support in four polls taken from any poll (a “Qualifying Poll”) taken by one of our 16 listed pollsters (a “Qualifying Pollster”) taken during 6/28-8/28 2019 (the ”Qualifying Period”) and which surveyed the nation or any of the first four primary states.” This diary will not focus on the unfairness of the donor requirement but on the choice of the Qualifying Pollsters and the failure to insure there would be a fair group of Qualifying Polls, particularly towards the end of the Qualifying Period.
The difference between polls taken during the early part and the later part of the Qualifying Period is obvious: You should want the debate to reflect the status of the campaign close to the debate more than what it was earlier. This would include the reaction voters had to the last debate and the subsequent campaigning.
The details are below, but consider that the DNC listed 16 Qualifying Pollsters. Sounds formidable, but how many polls were taken by his group during the Qualifying Period? Only 12. How many during the last three weeks in any one of the four early primary states? Zero.
Please note I am not a supporter of Gabbard. I’m just making the argument that better consideration and inquiry into the available pollsters plans would have reflected who were the most popular candidates before the third debate. If all the pollsters would have responded “We never disclose our polling plans”, my arguments are certainly weakened, though not entirely defeated. But I can’t really see why pollsters being considered for inclusion in the DNC debate criteria would not be open about their polling plans and perhaps alter their plans to include polls closer to the third debate.
The Qualifying Pollster Problem
The 16 Qualifying Pollsters where in reality much fewer. Four Qualifying Pollsters provided zero polls during the Qualifying Period and should never have been included, a fifth provided only one poll early in the Qualifying Period. Of the remaining Qualifying Pollsters, three are newspapers which always “pair up” with one of the networks listed as a Qualifying Pollster: The Washington Post pairs with ABC, The New York Times pairs with CBS and The Wall Street Journal pairs with NBC. Consequently, six of the 16 were actually three. There were really only nine active Qualifying Pollsters.
The DNC excluded certain national pollsters that provided many more polls than the network/newspaper groups: Emerson, YouGov, Morning Consult and Harris X. The DNC also excluded two reliable newspapers from New Hampshire (the Boston Globe) and from South Carolina (the Post Courier) which were more active than their choices for pollsters for those states.
Pollsters That Should Never Have Been Included
The DNC should never have included The Associated Press, NPR or Winthrop (a South Carolina University). Not one of these three provided any polls since the beginning of the campaign, much less the Qualifying Period. A fourth pollster, The Des Moines County Register (“DMR”) would seem like an obvious choice based on its Iowa polling reputation, but was also a bad choice since it provided no polling during the Qualifying Period.
The fifth disappointing pollster was the University of New Hampshire (“UNH”) which provided only one poll very early in the Qualifying Period.
Active Pollsters That Should Have Been Included
Emerson College is significant omission. Emerson has a “B+” rating from 538 and provided three national polls during the Qualifying Period, one which gave Gabbard 2% and the last of which, taken in the last days of the Qualifying Period, gave Gabbard 3%. YouGov (which paired with The Economist) is another bad omission. It has a B rating from 538 and has a long history of polling. YouGov had seven national polls during the Qualifying Period, five of which gave Gabbard 2% and one 3%. Note that when YouGov paired with CBS for Iowa and New Hampshire polling its results counted as Qualifying Polls. I guess YouGov is better when it’s chaperoned by CBS.
Note that Survey USA is a glaring omission as well. Survey USA has an “A” rating by 538 and have a long record on one of the most active pollsters. However, Survey USA provided only one national poll during the Qualifying Period and it would not have changed the third debate line up, so this is something of a footnote, but it should be considered for the fourth debate. Morning Consult (which paired with Politico) and HarrisX (which paired with The Hill) were active, but were arguably justifiably excluded because of their poor ratings by 538: a B- for Morning Consult and a C+ for Harris X. In any event their exclusion would not have helped Gabbard and by adding Emerson, YouGov and Survey USA, you don’t really need more.
Also excluded, perhaps justifiably was Gravis which has a C+ rating by 538. The only problem is that Gravis is the only pollster which gave late Qualifying Period results in Nevada and New Hampshire. If their results had been confirmed by one or more of the Qualifying Pollsters, Gabbard could well have far have exceed the four Qualifying Poll requirements.
So let’s take a quick look at the polls nationally and in the early primary states of Iowa, New Hampshire, Nevada and South Carolina as they related to Gabbard.
NATIONALLY
If Emerson and YouGov had been included, Gabbard would have had two more Qualifying Polls to add to Gabbard’s one national Qualifying Poll (taken by CNN). (Note it’s only three because once a candidate gets a Qualifying Poll from a Qualifying Pollster all other Qualifying Polls by that same Qualifying Pollster, in the nation or a given state, do not count).
Note that nationally based on the Real Clear Politics (“RCP”) average of polls taken during the last ten days before the 8/28 deadline, Gabbard had greater support than Klobuchar and Castro, both of whom will be at the debate based on polls taken earlier during the Qualifying Period.
IOWA
Only three polls overall were taken in Iowa during the Qualifying Period and all three were Qualifying Polls. Gabbard got 0 or one percent in these three so you might say she has nothing to complain about. The only problem is the last of these polls was taken during August 1-4, before the period Gabbard saw an increase in her numbers nationally. In short, it would have more fair if there were at least one mid to late August poll to see where Gabbard and others stood. The DNC’s inclusion of DMR was pointless since it did no polling during the Qualifying Period but I wonder if DMR might have responded favorably to provide such a poll.
NEW HAMPSHIRE
Five polls overall where taken in New Hampshire during the Qualifying Period, two of which were Qualifying Polls. One Qualifying Poll (CBS/YouGov), taken in mid-July gave Gabbard 2% and became one of her two Qualifying Polls. The only polls taken in August were from The Boston Globe/Suffolk and Gravis, the former gave Gabbard 3% and the latter gave Gabbard 5%. Obviously Gabbard had an August surge in New Hampshire and if UNH had taken a late poll she would likely have had a second Qualifying poll from the state. Note that, Suffolk had one of its national polls, done with USA Today, count as a Qualifying Poll. Like YouGov, it depends on who sponsors you.
The DNC should have inquired about the late polling plans of UNH and Suffolk and given UNH’s plan to sit on its mid-July poll, it should have encouraged a later poll or included Suffolk. It should also have included The Boston Globe as the state newspaper pollster like DMR. The RCP average of polls showed Gabbard with a 3.3% average for New Hampshire, better than O’Rourke (1.3%), Booker (0.7%), Klobuchar (2.0%), Yang (2.0%) and Castro (1.0%). Even without the Gravis poll she would have exceeded their averages.
NEVADA
Nevada had only two polls overall during the Qualifying Period, neither of which were Qualifying Polls. Of the two non-qualifying polls, one was taken by Gravis in mid-August which gave Gabbard 2%, again showing her August increase in popularity. The DNC should have worked to insure that at least one Qualifying Poll came from Nevada. As it is, Nevada had zero impact on who made the third debate.
SOUTH CAROLINA
South Carolina had four polls overall taken during the Qualifying Period, three of which were Qualifying Polls. The three Qualifying Polls each gave her 0%, but were taken between 7/7 and 7/22 before the second debate. A non-qualifying poll taken in mid-August by the South Carolina’s largest newspaper, The Post Courier, gave her 2%. South Carolina is a state with a many retired military voters and many of the more dovish probably started considering her after the second debate. In any event, the Post Courier is likely comparable to DMR when it comes to statewide elections and should have been included by the DNC just like The Boston Globe. Instead the DNC included Winthrop University (located in South Carolina) as a Qualifying Pollster even though it had done no polling during campaign. With a little bit of inquiry, the DNC would have learned that Winthrop had no intention of polling during the Qualifying Period.
Summary
In short the DNC didn’t do their homework and effectively froze out any candidate that emerged from the 0 to 1% crowd after the last debate and August campaigning. It just so happened that Tulsi Gabbard was that candidate. It was incredibly dumb to include the Associated Press, NPR, Winthrop University and, to a lesser extent DMR, none of which provided a single poll during the Qualifying Period. A little inquiry would have revealed their plans. Certainly the DNC missed the boat by not including The Boston Globe/Suffolk poll and The Post Courier poll. Emerson and YouGov should have been included.
UPDATE
Funny that all the comments have been about how bad Gabbard is as a candidate. The dairy is about how the DNC criteria (1) chose four pollsters who contributed no polls at all, (2) excluded two reliable active national pollsters (3) excluded the best polls from New Hampshire and South Carolina and (4) failed to insure any Nevada polling.