MT-AL: Rob Quist (D) vs. Greg Gianforte (R) vs. Mark Wicks (L), May 25

[cinyc]

FYI - I just decided to run a Google Surveys poll for this race.

I have no idea how to weight the results, though.  Weighting to the 2012 GE electorate wouldn't make much sense, since turnout will be lower.  Does anyone have any ideas?  Weight to 2014 mid-terms (if there were any in MT), perhaps?  Last I checked, the 2016 CPS voter survey results weren't available yet, so that's not an option.

[cinyc]

FYI - I just decided to run a Google Surveys poll for this race.

I have no idea how to weight the results, though.  Weighting to the 2012 GE electorate wouldn't make much sense, since turnout will be lower.  Does anyone have any ideas?  Weight to 2014 mid-terms (if there were any in MT), perhaps?  Last I checked, the 2016 CPS voter survey results weren't available yet, so that's not an option.

Weigh it to 2014 House/Senate turnout. That's the best.

The 2014 November CPS seems to have data for Montana.  I guess that's the way I'll go.  As usual, I'll report both the weighted and unweighted numbers.

Google accepted the survey, so we're in the field right now.

[cinyc]

Wow, thanks for doing this! Really looking forward to the results. I wouldn't underestimate Democratic turnout here, though. Generally, the difference between the composition of a midterm and presidential year electorate isn't that significant in Montana, but still...

None of my weighting will be by party.  It will only be by sex and age.  As you may remember, my Google Survey polls are one-question affairs with no screening questions.  The question asked this time is:

"Montanans will go to the polls on May 25 to vote for a new U.S. Congressman. If this special election were held today, for whom would you vote?"

The choices are (in random order, except the last option):
"-Democrat Rob Quist
-Republican Greg Gianforte
-Libertarian Mark Wicks
-I am not likely to vote in this election"

I'm going to have to recreate how I created the 2012 weights to do them for 2014 - that's a bit of a project for the next few days.  Hopefully, I'll figure it out before the poll ends.  I don't remember exactly what I did the last time.  But we're going to need the 2014 weights for the 2018 midterms, anyway.  Now is as good a time to start that as any.

[cinyc]

^Do they allow two questions? I'm interested in a hypothetical Tester vs. Fox Senate matchup, but that should probably wait until after May 25.

Yes, Google Surveys allows 2-10 questions, but the cost is 10x as much as asking 1 question. 

There was a coupon offer for people conducting their first poll ($50 off your first poll), at least as of a few weeks ago.  I'm not sure if the offer is still ongoing.

[cinyc]

^Do they allow two questions? I'm interested in a hypothetical Tester vs. Fox Senate matchup, but that should probably wait until after May 25.

Yes, Google Surveys allows 2-10 questions, but the cost is 10x as much as asking 1 question. 

There was a coupon offer for people conducting their first poll ($50 off your first poll), at least as of a few weeks ago.  I'm not sure if the offer is still ongoing.

How much does it cost to put a 1 question poll out in the field?

For a state poll, $0.15 per respondent, so $75 for my 500-respondent poll.  Past practice shows that a good number of people will choose the "I'm not likely to vote option" (however worded), so I'll probably end up with 300-400 decided voters to weight.

[cinyc]

On the topic of turnout, doesn't the outcome of the mail-in ballot scandal affect that partially. If they do choose to use them that might give Democrats a better turnout edge.

I'm weighting the results to 2014 Montana actual voters and 2014 Montana registered voters, as reported by the November 2014 Census Population Survey.  So far, there isn't much of difference between using either metric - about two points in the most recent data dump.  167 of the expected 500+ respondents have responded so far.  105 of them didn't choose the "I'm not likely to vote" option.  The overall results so far are... interesting.

[cinyc]

My Google Surveys poll is about half done (274/500).  Without spoiling the results, which still can change, there are some trends:

-Right now, "I am not likely to vote in this election" is "winning".  Out of the 274 respondents polled so far, only 176 chose a candidate.  Hopefully, we'll get up to 300 usable respondents by the final tally.
-There is a definite regional divide between Western Montana and Eastern Montana, which is not terribly surprising, though the magnitude of the divide may be.
-Some people are going to be very, very happy with the results, if they hold.  Others, not so much.

Google Surveys hasn't broken down my results by imputed rural/suburban/urban status or imputed income yet, which is disappointing.  They've only given me sex, age and town location data to break down.

[cinyc]

cinyc, how would you grade Google Surveys for the presidential election?  It was fairly off generally, if memory serves.

I'd give the one-question Google Surveys polls that Atlas Users conducted a D+.

A lot were wrong, but most were taken a week or more before the election.  IIRC, the TN poll was right on, and my last-minute NM poll wasn't terrible, when properly weighted.  But we had a lot of stinkers.  Perhaps this is because the one-question methodology just doesn't work.

RRH partially used a one-question Google Survey poll when it polled the VT Governor's race (in order to reach more 18-45s) - and their overall poll ended up okay.  I'm not sure how well Google's in-house, poorly weighted, multi-question, multi-state polls performed.  They were all over the place, but some had very small sample sizes and you'd expect that.

Google Surveys uses different methods to reach users.  They tell you the type of website (News, Reference, etc.) or mobile app used to reach each respondent.  Unfortunately, the different methodologies often lead to different results, and it's not clear that, for example, respondents reading News websites are a better reflection of the electorate than those using Google's mobile app.  I'm not sure that there is a rhyme or reason to which method Google uses more of to reach users except that they can - and I don't know whether any type should be given more weight.  Probably not.

[cinyc]

My Google Survey Poll is complete.  It is either the canary in the coal mine or a huge outlier.

The question:
"Montanans will go to the polls on May 25 to vote for a new U.S. Congressman. If this special election were held today, for whom would you vote?"

The choices were (in random order, except the last option):
"-Democrat Rob Quist
-Republican Greg Gianforte
-Libertarian Mark Wicks
-I am not likely to vote in this election"

502 people were polled.  173 of those 502 chose the "I am not likely to vote in this election" option - 2 fewer than those who chose Democrat Quist.  When weighted for age and sex to 2014 CPS LV weights, the 329 decided likely voters chose Quist by a large margin:

Quist (D) 55%
Gianforte (R) 38%
Wicks (L) 6%
i.e. Quist +17

Unweighted, it's Quist +14 (53/39/8).  Google weighted to the Internet audience is Quist +15 (54/39/8).  The poll was taken from 3/11-14 Mountain Time.

As I hinted, there is a huge regional divide in the results.  Quist has a 33-point 2014 LV weighted lead in Western Montana (which usually comprises about 59% of the vote in a typical election) (64/31/5; N=209).  Gianforte leads by 9 in Eastern Montana (50/41/9; N=116).  Unweighted, it's Quist +29 in Western Montana, and Gianforte +10 in Eastern Montana.

I divided Eastern and Western Montana this way:

It is based on Montana's Area Health Education Regions.

There's not much of a gender gap, at least in the unweighted numbers (women: Quist+16; men Quist +12).  There is a bit of an age gap, with the 18-24 (+18), 25-34 (+34), 55-64 (+34) and 65+ (+15) sets going for Quist, the 35-44 (-9) set going for Gianforte, and the 45-54 set tied, but sample sizes are small.

Google gave me a rural/suburban/urban and imputed income breakdown in their final e-mailed spreadsheet.  I didn't weight them.  Gianforte runs better in rural areas (Quist+4) than suburban areas (Quist+22).  The few urban voters actually broke for Gianforte, but the sample size is really small (N=7).  There's nothing notable about the income numbers except that the overwhelming majority of Montanans polled were $25,000-$49,999 bracket.

[cinyc]

That's insane. Surely this is just small sample noise, right? I mean, is Quist that well-known and liked as a folk singer?

The real question might be whether Gianforte is that well-known and hated statewide.  Quist could be performing as a generic Democrat against a despised Republican.

I want to get excited but I remember that this is the same polling method that gave Hillary large leads in Kansas. Grain of salt, but this is definitely a race to watch.

Definitely take this with a grain of salt.  As I said above, our one-question Google Surveys polls didn't do so well in the 2016 general election.

[cinyc]

That's insane. Surely this is just small sample noise, right? I mean, is Quist that well-known and liked as a folk singer?

Quist's lead in the survey was always double digits once more than 10 respondents were polled.  His smallest lead in the raw data was about 10.5 points.  It bounced around between 10-15 points most of the time.

Methodology issues are more likely to be an issue than sample size, in my view.

[cinyc]

The regional divide is so large as to be unbelievable.  In the 2008 presidential election, the difference between the West and East was about 10 points.  In the 2012 presidential election, it was 12.  A 36 (raw) or 44 (weighted) point difference is way too large.  Small sample sizes leading to large MoEs, terrible Google Surveys methodology or whatever.

[cinyc]

What was the sample size of Western vs. Eastern Montana in your poll?

209/116. 

The West/East divide persisted from the start.  And it always was that stark or very close to it.

[cinyc]

That would probably require a Democratic governor to get elected in 2020, because in the hypothetical scenario of Montana getting a second district, a Republican could just force him out East and draw a new Western district. I'm skeptical of a Democrat keeping the mansion for 20 straight years, and assuming Bullock challenges Daines and Quist is still in office, that'd be asking to send 2 Democrats to Washington and 1 to Helena. That's a tough sell to a state that voted for Trump by 20%.

Congressmen don't technically need to live in the district they represent, just the state.  And he could always move into the new district, anyway. 

I suppose that a Republican could gerrymander two clear Republican-leaning districts, though.  But the power of incumbency is sometimes stronger than district lines, anyway.

[cinyc]

The spreadsheet at the link is to the full spreadsheet I usually use to sift through the raw data for the nuggets like age, sex, time, date, urban/suburban/rural status, income, response type and geographical region.

There were some other trends compared to my poll:

The 55-65s and 65+s in this poll went to Gianforte, unlike in my poll, where the 55-65s were one of Quist's best groups.

The suburban/rural divide persists, with Quist winning suburbanites by 10 and Gianforte leading among rural residents by 15 (both, unweighted).  Again, there were too few urbanites for any meaningful analysis.

The poll again shows that most Montanans are solidly middle class, in the $25K-$50K range.  Higher incomes tended to be more Gianforte, but there are far too few respondents to be sure.

I also break things down by time on the spreadsheet.  As Castro said, Gianforte had a large lead in the beginning, which dwindled at the end.

This poll makes more intuitive sense than mine, even with the lower sample size.

[cinyc]

I decided to run a GCS for this race as well and I got results in between Castro's and Cinyc's. They were,

Rob Quist - 48.4%
Greg Gianforte - 40.5%
Mark Wicks - 11.1%

I left out an "I won't be voting" option to maximize the amount of responses, so take this with an (extra) grain of salt. Here is the link to the actual survey.

Thanks.

Simple weighted to the 2014 November CPS actual voter percentages by age and sex, it's also Quist +8, 49/41/9.  I don't heart decimals.

The regional divide is there, but not as strong as in other polls.  Quist wins Western Montana by 14 (weighted), and loses Eastern Montana by 6 (weighted).  Western Montana was overrepresented in this poll, making up 78% of respondents versus the 59 or 60% they usually are.  The other two polls also overpolled Western Montana, but not by nearly this much.  If you re-weight the weighted regional results, you get Quist+6 statewide.

Agewise, Quist won every group except 35-44s and age unknowns.  Among men, the race is tied.  Quist has a 23 point lead among women.  Note: these results are unweighted.

Weighting spreadsheet here.

[cinyc]

No surprise, but the mail-in ballot bill is dead

It's telling that Republicans are willing to waste taxpayer dollars just to ensure that fewer people go to the polls.  The country needs to be clear-eyed about the Republican Party's increasing authoritarianism.

I don't think this was on as much of a party line vote as you think.  The motion to pass quickly was made by a Republican, and a Democrat representing Native Americans was against the bill because she thought mail-in voting only would discriminate against Native Americans.  And I don't think mail-in voting is totally eliminated without the bill.  It just requires that polling places be opened up, too:

You must be logged in to read this quote.

[cinyc]

I don't think this was on as much of a party line vote as you think.  The motion to pass quickly was made by a Republican, and a Democrat representing Native Americans was against the bill because she thought mail-in voting only would discriminate against Native Americans.

Well it was a local Republican Party official who sent a letter to everyone saying they shouldn't approve the bill because it might hurt their chances.

A Republican motioned to get the bill to the floor quickly, and supported it.  A Democrat representing an Indian Reservation opposed it.  And the bill passed the Republican-dominated Montana Senate by a wide margin, where it was sponsored by a Republican. 

I don't think it is as clear cut as your typical Atlas "evil Republicans" meme espoused by the poster I was responding to, Ronnie, thinks.

And I don't think mail-in voting is totally eliminated without the bill.  It just requires that polling places be opened up, too:

That probably just means the current absentee rules. It's not the same as if everyone got a ballot.

And why should everyone be mailed a ballot?  Believe it or not, not everyone in the country has mail delivery at their house, and not everyone has ready access to a mailbox or post office to drop off outgoing mail.  This is particularly true in rural areas, like Indian Reservations, which is why the Democrat representing one opposed the bill.  Besides, mail gets lost.  In-person votes don't.

Anyone who wants to vote by mail CAN do so in Montana by simply filling out an absentee list request.

I don't like all mail-in balloting (or non-excuse absentee voting, for that matter).  It ruins election day, turning it into election week or election month.  Things can happen during election month that will cause people to regret their initial vote.

[cinyc]

It would be nice if someone actually polled this race, it has received surprisingly little attention outside of Atlas so far.

One reason is that it isn't technically easy to poll Montana by phone.  RRH says that they can't poll Montana due to either a high bond requirement or ban on robo-polling - I don't remember which.  Either way, it's just too costly for them to poll.

I wouldn't be surprised if the Lee Newspapers use Mason-Dixon to poll the race closer to election day, though.  They've done statewide polling of Montana in the past.

[cinyc]

Gravis polled the race, and has it at Gianforte +12:

Gianforte 50%
Quist 38%
Wicks 3%
Breck 2%
Uncertain 7%
http://gravismarketing.com/polling-and-market-research/current-montana-polling-3/
1,222 Voters; MoE +-2.9;  Poll Dates ? - It was released today.


Breck is apparently the Green Party candidate.  Did we miss him getting ballot access, or is Gravis polling someone not on the ballot?

[cinyc]

Apparently, Green Thomas Breck and Independent Steve Kelly are suing to get on the ballot.  They didn't get the requisite number of signatures or pay to get on the ballot, so they're suing in federal court, claiming they didn't have enough time.

One problem is that many counties have already printed or are in the process of printing their ballots.

[cinyc]

Hopefully the DCCC and DNC learn their lesson tonight and start fundraising for Quist.

Hopefully. I'm not prepared to draw conclusions until the end of this year, but so far the party seems to be off to a bad start (imo). It's one thing if the internals are showing the Democrat way down, but with this KS election, it was genuinely close and they had time to react, yet they didn't. Money/etc is not guaranteed to move the needle enough, but in this case tonight it could have at least told the base that they are listening and adapting. Instead, they make no effort and basically tell everyone "we're not changing ****." Perceptions matter.

I know for sure I'd like to know that the party whom I donate much time and money is actually trying new things, taking some risks and moving past failed strategies/ideas. Anyway, we'll see how the next 2 elections go I suppose.

You are assuming that nationalizing races in blood red Trump states makes sense.  The DCCC coming in will allow Gianforte to credibly tie Quist to San Fran Nan and New York Chuck, who I'm sure aren't that popular in Montana.  Who's to say Thompson wouldn't have lost by a larger margin had the KS-04 race been nationalized more?

[cinyc]

You are assuming that nationalizing races in blood red Trump states makes sense.  The DCCC coming in will allow Gianforte to credibly tie Quist to San Fran Nan and New York Chuck, who I'm sure aren't that popular in Montana.  Whose to say Thompson wouldn't have lost by a larger margin had the KS-04 race been nationalized more?

But then what is the alternative? Democrats abandon candidates in deeply Republican territory out of fear of it backfiring? I'm sure there are instances where it will be plenty valid, but the 'nationalization' excuse doesn't seem that strong to me. It seems like a good excuse to keep doing what the party said it would stop doing.

No.  The Democratic establishment should quietly support the candidate in other ways, by sending in surrogates that don't have ties to the Democratic establishment (like Sanders), and finding other ways to quietly get them resources behind the scenes.  Share the GOTV apparatus, microtargeting data and donor lists, for example.  Don't get into a situation where your Kansas or Montana candidate can be tied to unlikable characters from San Francisco and New York City.

[cinyc]

Google Consumer Survey poll, April 6-8, Sample size: 333, MoE: 5.4%

45.4% Rob Quist (D)
43.5% Greg Gianforte (R)
11.1% Mark Wicks (L)

Link.

Woah, that's a really high libertarian vote percentage.

That's pretty much par for the course for a Google Survey when you don't include another opt out.  Those who won't vote or don't care about the question tend to pick minor party candidates.

I'll run the weighting and regional breakdowns when I get a chance.

[cinyc]

Weighting the latest Google Survey poll to 2014 LV by age and sex, I'm getting a two-point Gianforte lead after rounding, 46%-44%-10%.  If you heart decimals, it's a 1.2 point Gianforte lead, 45.6%-44.4%-10.0%.  The reason Gianforte does better in the weighted results is he's doing well among 65+ Males, and worse among the young'uns, who are less likely to vote.

Quist is up by 13 in Western Montana (weighted).  Gianforte leads by 22 in Eastern Montana.