You are on page 1of 3

Strange

U.S. Senate

TO: Interested Parties


FROM: Travis Smith
RE: Alabama Senate Race Update
DATE: 8/29/17

Current State of the Alabama Republican Primary Race For US Senate

Strange Has the Momentum

As we know, our internal polling and analytics shows the race to be very competitive, just as we
anticipated. Two recent public polls reflect this as well.

1. Strange 45%-Moore 47% (8/29/17): Just today Harper Polling released a poll that has
the race a virtual tie. More importantly, it is clear that Strange has the momentum.

o Strange leads undecideds: Among the 8% of undecided voters, 30% say they
lean Strange, compared to just 6% who lean Moore. 63% are truly undecided.

o Strange is the Trump candidate: A majority of likely voters say Strange is the
candidate who is most supportive of President Trump, not Moore (Strange 50%-
Moore 40%).

o Voters know Moore cannot deliver: Moore has a net-unfavorable rating with
voters who want a candidate who can get things done for Alabama (39%/45%).

2. Strange 41%-Moore 45% (8/24/17): A recent SLF poll reported by the Washington
Examiner showed the race within the margin of error.

Fake Polls

The SLF and Harper polls reflect the true state of the race. The methodology in other polls
(JMC; DDHQ) published in the first days following the primary in Alabama make their numbers
extremely suspect.

The difficulty in polling the low turnout universe in the GOP primary for U.S. Senate in
Alabama was evident as in most cases, these polls overestimated Roy Moores winning margin
on primary day and underestimated Luther Stranges actual vote share. The two incorrect
Trafalgar polls had Moore up 14% and 12%, when his margin over Strange in the primary was
only six points, 39% to 33%. Several published polls right before the primary showed Stranges

Info@LutherStrange.com | PO Box 3670, Montgomery, AL 36109, United States | www.LutherStrange.com


vote share in the low 20s, but he obviously received a healthy 33% of the vote vs. eight other
GOP primary opponents. This pro-Moore bias is clearly holding in these published polls.

Before we get into the specific methodology of these polls, there are several things about these
results that just do not pass the smell test and makes sense with the reality that happened on
primary day.

1. These surveys have Strange receiving less vote than the more than respectable 33% he
garnered in the actual primary against eight other candidates on Election Day. This
means that instead of getting any of the 28% that went to the other 7 candidates, besides
Moore and Strange, Strange not only did not get any of the other candidates vote, he
actually lost some votes. That just does not jive with reality.

2. Why are their such high undecideds in these surveys? Both these candidates are known
quantities to the primary voters, and it just defies believability that 17% plus of the
electorate would be undecided at this point and this is not close to what internal polling
analytics has shown. This leads us to believe due to lack of a rigorous screening process
and probably faulty voter lists, there are a lot of non-primary voters in these surveys
which also accounts for negative image of Strange in these surveys.

3. The geographic modeling in these surveys is off. For whatever reason, they have
significantly under-sampled voters in the Birmingham area, coincidentally Stranges base
of geographic support. On Primary day, Strange actually beat Moore by 15% in Jefferson
County and 9% in Shelby County, the two largest counties in the Birmingham media
market, yet these polls actually have Strange losing in those areas in these surveys. They
also have Birmingham accounting for about one third of the statewide vote. In the end,
Birmingham will in all likelihood account for over 40% of the primary vote, like it has
historically in GOP primaries in Alabama and like it did during the first primary.

Remember, these are the same folks who said Mo Brooks was closing on Strange and could beat
him, when in reality, Strange beat Mo Brooks by more than 13% statewide.

Here are some of the specific problems with these two surveys:

JMC (8/19/17) Flaws:

1. Robo Poll: This is a robo poll that did not institute any supplementary methods to capture
voters who do not have landline telephones.

2. Weak Screen: The screener is incredibly weak. They ask people simply if they are going
to vote (yes/no) instead of asking likelihood to vote. Considering that, this poll has more
of a likelihood to capture people who may not actually vote. In all of our primary internal
polling, Moore did better with lower propensity voters and those who were unaware of
the actual election date. This bias, then, in the JMC poll likely favors Moore.

Info@LutherStrange.com | PO Box 3670, Montgomery, AL 36109, United States | www.LutherStrange.com


3. Wrong Sample: It is very surprising to us that the Trump endorsement is a one to one.
Trumps numbers are 85%+ favorability/approval in our surveys. This leads me to
believe that their survey is capturing too many non-Republicans or liberals/moderates
instead of conservatives.

4. Geographic Model: Their poll has Strange losing in his home market of Birmingham
36/49. This does not make sense with our internal polling and analytics, nor the actual
results from the election. We ran even in the entire market and won Jefferson County
43%/28% over Moore.

5. Poor Track Record: Their primary poll was wrong, so this is wrong too. These were the
same folks that said Brooks was closing in on Strange in the first primary and could beat
him. They had us and Brooks in a statistical tie ten days before the election. We beat him
by more than 13-points and our own internals ten days out had us up double digits over
Brooks.

DDHQ (8/22/17) Flaws:

1. Weak Screen: The list they used is way too wide, with registered voters. When you do not
select a universe based on past vote history you will get a lot of unlikely voters in the
survey. Not to mention they have a very weak screen on voter likelihood, like JMC.

2. Sample Error: They do not ask a question about political party affiliation. It begs to the
notion that they have non-Republican primary voters in here. Even though Democrats can
vote in the runoff, the vast majority of the non-Republicans will be Independent. This is
even more questionable when you see 6% of the survey being African American. This is
a special Republican primary runoff election in Alabama.

3. Geographic Model: Their geographic distributions appear extremely incorrect. They are
saying 25% of the vote is coming out of the Montgomery media market. When looking at
past vote history, the share of the vote there should be only 12% - they have doubled that.
While they overestimate Montgomery, they undercount Birmingham, which is Stranges
home market and where he does best. They have Birmingham at 36% and it should be at
least 40%. Depressing the Birmingham vote depresses Stranges vote.

The outcome will be decided by which candidate turns out their voters, not fake polls. Strange
has the momentum and he will win this race because no one is going to outwork the campaign,
the volunteers, or the candidate.

Info@LutherStrange.com | PO Box 3670, Montgomery, AL 36109, United States | www.LutherStrange.com

You might also like