Off the Kuff Rotating Header Image

Neil Dikeman

What about Neal?

Ross Ramsey reminds us there is a third person in the Texas Senate race.

Neal Dikeman

Libertarians and other third-party candidates have never won state elections in Texas and rarely make a meaningful difference in election results, with one big exception: As spoilers.

If recent indications of a close U.S. Senate race between U.S. Sen. Ted Cruz, a Republican, and U.S. Rep. Beto O’Rourke, a Democrat, prove valid, a third candidate’s voters could spell the difference on Election Day.

“It will be the Libertarian voters who win this race,” says a hopeful Neal Dikeman, the Texas Libertarian candidate for the U.S. Senate.

[…]

Most polls have Cruz ahead of O’Rourke — but only by single digits. The most recent survey, from Quinnipiac University, had Cruz ahead by 9 percentage points among likely voters. Dikeman wasn’t included in that one, and none of the respondents said they would vote for someone other than the two major-party candidates.

For what it’s worth, covering a nine-point spread would be a big reach for a Libertarian candidate. Most of the time, in races with both Democrats and Republicans, third parties do well to get half that amount. Their mileage varies: Mark Miller and Martina Salinas, a Libertarian and a Green, combined for 8.6 percent in the 2016 race for Texas Railroad Commission; that same year, the presidential candidates from those parties, Gary Johnson and Jill Stein, combined for 3.9 percent.

But several late summer polls in the Texas race for U.S. Senate are closer than Quinnipiac’s latest, raising at least the possibility that support for Dikeman could amount to more than the final difference between Ted and Beto.

This was written before that Ipsos poll came out, but that doesn’t change the main point. The two points of interest is that there is a Libertarian candidate, which will affect the win number in this race, and that the polling we have seen so far has not really taken this into effect.

On point one, by “win number” I mean the actual minimum amount needed to finish first in the race. It’s not fifty percent because there are more than two candidates. I did a broad look at this before the 2014 elections, so let’s revisit that here. Each Senate race in recent years has had at least three candidates in it. What percentage of the vote was actually needed to win those races? Here’s a look:


Year    Lib  Green   Else  Total     Win
========================================
2014  2.88%  1.18%  0.02%  4.18%  47.91%
2012  2.06%  0.86%         2.92%  48.54%
2008  2.34%                2.34%  48.83%
2006  2.26%                2.26%  48.87%
2002  0.79%  0.55%  0.03%  1.37%  49.32%

“Win” is the minimum amount that would have won that year. There were write-in candidates in 2014 and 2002. The third party vote hasn’t amounted to much in these races, but it’s not nothing. As you can see, in each year after 2002, 49% was enough to win, and in 2014 48% was enough.

What about this year? Obviously, that depends on how much support Neal Dikeman ultimately attracts. History suggests that will be in the two to three percent range, but it’s at least possible it could be more. Given that nobody likes Ted Cruz, it may be that the number of Republicans who refuse to vote for him but won’t vote for a Democrat is higher than usual. If that’s the case, then Dikeman will be the beneficiary of that. It wouldn’t shock me if he got more like three or four percent.

We might get some feel for that if pollsters specifically included Dikeman in their candidate choices, especially now if everyone is switching to a likely voter model. Not because polling for third party candidates is particularly accurate – they almost always overstate third party support – but because it might give a clearer picture of the gap between Cruz and O’Rourke. I have to imagine that the Quinnipiac poll would have Cruz at something lower than 54 had Dikeman been named as a choice. Yes, the polls have included “don’t know” as a choice, but it’s not the same as an actual person. It’s my hope we’ll see polls like that going forward. After all, that 47% support Beto got in the Ipsos poll may be closer to a win than you might think.

Harris County poll: Hidalgo 53, Emmett 47

From the inbox last week:

Lina Hidalgo

The Lina Hidalgo campaign for Harris County Judge today released the results of its first county-wide poll, showing the Democratic challenger leading the Republican incumbent by a stunning six percentage points; among Harris County voters who plan to vote in the County Judge race, 53% plan to vote for Lina Hidalgo and 47% say they will vote for Ed Emmett.

The poll, conducted by Texas Democratic Party-authorized polling firm, Change Research, surveyed more than 1700 registered voters in Harris County on May 11, 12, 13, 19, and 20, and has a margin of error of +/- 3%.

“This poll supports what I am hearing as I travel to every corner of Harris County – that people are ready for new, authentic leadership for the future,” said Hidalgo. “In spite of the poll’s heartening results, I plan to campaign every day as if we are six points down, not six points up. I will work my heart out to make sure that every voter in Harris County feels heard and included.”

Other poll findings of note include:

94% of Harris County voters report feeling more interested (56%) in or equally as interested (38%) in the 2018 election as they have felt about prior elections.

President Trump is viewed unfavorably by 60% of Harris County voters

Voters report that the three issues that will drive their voting behavior most in November are:

1. Government transparency
2. Education
3. Jobs

Like me, you probably had a lot of questions when you saw this. I went ahead and emailed the Hidalgo campaign to get more information about the poll, and they graciously provided me this executive summary and this spreadsheet with the questions and answers broken down by race/age/gender/etc. I think the best way to present the fuller data set and discuss the points I want to raise are to go through the questions and responses in the spreadsheet. So with that said, here we go.

Question: Which of the following best desribes you? “I live in Harris County, am registered to vote, and identify as a”:


               All  Trump  Clinton  No vote
===========================================
Democrat     41.6%   1.2%    74.9%    23.2%
Republican   33.5%  78.9%     2.0%    14.2%
Independent  24.9%  19.9%    23.1%    52.6%

Question: Do you plan to vote in the November 6, 2018 elections?


               All  Trump  Clinton  No vote
===========================================
Yes          81.4%  89.9%    87.9%    56.8%
Maybe        16.5%   8.8%    11.4%    30.0%
No            2.2%   1.2%     0.7%    13.2%

Question: How interested are you in the election in 2018 compared to previous elections?


               All  Trump  Clinton  No vote
===========================================
More         56.3%   46.5%   69.1%    39.8%
Same         38.0%   50.4%   26.2%    37.4%
Less          1.9%    2.2%    0.8%     9.5%
Unsure        3.7%    0.9%    3.8%    13.3%

First things first, all responses are given as percentages rather than number of respondents. You can reverse engineer that, of course, but I think it’s more illustrative to provide both. That will especially be the case with some later questions. I sent a separate email to the contact for the polling firm about that; I’ll update if I get a response.

In the questions above, “Trump” and “Clinton” refer to the subset of people who said they voted for Trump or Clinton in 2016, while “No vote” are the people who said they didn’t vote in 2016. There isn’t a question asking why someone did not vote in 2016, so it could be the case that they were not eligible – too young, or not yet a citizen – or not registered. Basically, this says there are more people who identify as Democrats in Harris County – I don’t think that is a surprise to anyone – and a larger share of self-identified Republicans voted for Trump than Dems voted for Clinton. As for questions 2 and 3, it sure seems like everyone is excited to vote this fall, with Democrats perhaps more so. Needless to say, that remains to be seen. How true these sentiments are will be the million dollar question for candidates, pollsters, and loud-mouthed pundits.

Question: In the 2016 election, did you vote for:


Trump      36.8%
Clinton    48.7%
Johnson     2.8%
Stein       2.4%
No vote     9.4%

As a reminder, 53.95% of voters in Harris County actually voted for Hillary Clinton, while 41.61% voted for Trump. Gary Johnson took 3.03%, while Jill Stein had 0.90%, which means this poll oversamples Jill Stein voters. Make note of the date, you may never see that again. Another 0.43% wrote in Evan McMullin, and a further 0.09% wrote in someone else. If you go back to question 1, that’s why the Trump/Clinton/No vote subsets didn’t add up to 100%.

(Yes, I’m jumping around a little. This is how I want to present the data.)

Question: On a scale of 1-10, how do you feel about President Donald Trump today? 1 = strongly oppose, 10 = strongly support


               All  Trump  Clinton  No vote
===========================================
1            39.7%   0.3%    71.8%    35.5%
2            10.0%   0.0%    18.3%     3.5%
3-8          20.3%  15.2%     9.5%    47.9%
9             5.6%  14.2%     0.0%     4.3%
10           24.4%  64.1%     0.4%     8.8%

Allow me to point to this tweet by Amy Walter of the Cook Political Report to explain what this means.

90.1% of Clinton voters have the strongest negative feelings about Trump, while 78.3% of Trump voters have the strongest positive feelings about him. ‘Nuff said. Oh, and the non-voters mostly don’t like him, too.

Question: For whom do you plan to vote in the 2018 election for US Senate?


                 All  Trump  Clinton  No vote
=============================================
Ted Cruz       42.0%  93.4%     3.6%    31.2%
Beto O'Rourke  49.3%   2.1%    90.5%    52.2%
Neal Dikeman    1.9%   1.1%     0.7%     4.1%
Bob McNeil      6.9%   3.4%     5.2%    12.5%

Neal Dikeman is the Libertarian candidate. Bob McNeil is an independent who could be fairly classified as farther to the right than Cruz. He’s also not yet officially on the ballot yet, as he has to turn in some 47K petition signatures to the Secretary of State by June 21. Good luck with that. His presence in the question is basically noise, so don’t be too distracted by it. There won’t be a Green Party candidate. The 3.6% of Clinton supporters for Cruz is a reminder that there were a non-trivial number of Republicans who crossed over to vote for Clinton in 2016. Note here that all the numbers add up to 100, which is something that never happens in polls. You will see a possible mechanism for this in the next section.

Oh, and as for that Quinnipiac poll, don’t try to reconcile these two results. I think it is unlikely that O’Rourke could win Harris County by seven points while losing the state by double digits, but that doesn’t imply in any way that one poll is more “valid” or “correct” than the other. They are their own separate data points.

Question: For whom do you plan to vote in the 2018 election for Harris County Judge?


                 All  Trump  Clinton  No vote
=============================================
Ed Emmett      34.3%  74.9%    13.9%    14.0%
Lina Hidalgo   33.5%   2.8%    63.5%    30.4%
Won't vote     32.2%  22.4%    22.7%    55.6%

Question for undecided voters: If you had to choose for whom to vote for Harris County Judge in the 2018 election, who would you select?


                 All  Trump  Clinton  No vote
=============================================
Ed Emmett      24.7%  67.9%     9.8%    14.6%
Lina Hidalgo   44.7%  14.8%    74.7%    45.1%
Won't vote     30.7%  17.3%    15.5%    40.4%

Totals excluding undecided voters:


                 All  Trump  Clinton  No vote
=============================================
Ed Emmett      47.2%  93.7%    16.7%    28.5%
Lina Hidalgo   52.8%   6.3%    83.3%    71.5%

And here is how we get to the headline number. I don’t care for this construction. Having “won’t vote” as a choice rather than the more standard “don’t know” is a weird decision, one that casts some doubt on the “enthusiasm for voting” question. Regardless, any way you look at it, one may reasonably conclude that these voters as a group may be less likely than those who picked a name. As such, you can’t add them together. It’s my presumption that the pollster went through a similar exercise in the US Senate question (this might help explain the bizarrely high percentage for the candidate who probably won’t be on the ballot, who I’d bet none of the respondents had ever heard of – basically, he’s the “none of the above” choice), though they didn’t show the individual steps for how they got there.

I mean look, Ed Emmett has to be the best-known politician in the county, while Lina Hidalgo – who was unopposed in March and didn’t have much money as of January – surely has low name recognition. The fact that she was within a point of him in the first question, assuming the sample is reasonable, is pretty encouraging on its own. It’s a reflection of the partisan split in Harris County – remember, Emmett gets a significant number of crossovers – and demonstrates that Hidalgo has a lot of room to grow, as surely a decent number of those “won’t vote” respondents are actually likely Dems who just don’t know who she is yet. I don’t understand the need to push it further than that. And in thinking about it, I’m a little concerned that the O’Rourke/Cruz first-question numbers were a few points closer, with the “but if you had to choose” question being the reason for the larger gap.

So what do I make of this? As I say, it’s a data point. Maybe it will be in line with others – I’m sure we’ll see other polls – and maybe it won’t. I expect we’ll see plenty of conflicting results – again, so much of this depends on who shows up in November, and right now no one knows how that will look. We’re guessing. Some will guess better than others, and will base their guesses on better data. I think this particular result is optimistic, but reasonably so. Plausibly so. I’ll feel better if and when I see more results like it, or results from other races that correlate with it. But it’s one result, and the Quinnipiac experience reminds us again to not put too much stock in any one result.