Simulation Modeling the Midterms Correctly

Wrong election models?  There is no shortage.

For good election models (as with all good statistical models) we require good methodology.

Sound methodology in this field starts with anticipating what the election dynamic will be between dominant variables.  If the right dynamic and dominant variables are chosen, we then sift through the results of the the model for useful information about the election.  From this useful information we then determine what course of action (if any) to take in an election.

And, if you want to compare my c.v. on predictive analytics with others, I refer you to these examples –

From November 07, 2016:

IBD’s number today suggest R-D turnout levels will be even.

Early voting totals suggest IBD is right, but that state pollsters haven’t adjusted their turnout models correctly.

For example, in the CBS/Yougov poll of Florida, which had the state tied 45-45, their weighted sample had whites being only 61.7% of all voters, Hispanics 19.8 and blacks 13.7.

But according to early Florida vote results whites are 66% of the electorate, Hispanics 15% and blacks 13%. If the CBS turnout model is adjusted with these actual figures then Trump is ahead by over 1 point in Florida, and this before election day voting which will break strongly for him.

I’m now confident Trump will take Florida tomorrow.

For another example, most state polls of North Carolina have that state even despite early voting being disastrous for Democrats.

If state polls are generally built around 2012 turnout models and if IBD is right that Republican and Democrat turnout will be even, the state polls, which are very tight, are overestimating Hillary’s actual position.

 

From August 29, 2015 –

Actually, Trump is winning a plurality of Evangelicals and just about every other Republican demographic. Trump is also performing well enough in general election matchups that I’m now comfortable switching my support from Walker to him.

As for Ben Carson, his bump is probably a temporary result of the debate. I suspect Carson will fade and Cruz will pick up most of his social con supporters. But it won’t be enough for Cruz to stop Trump.

 

That last excerpt is from back in 2015 when Scott Walker was still in the GOP presidential primary, and when Ben Carson was in second place, ahead of Ted Cruz.  Cruz, if you remember, did go on to pass Carson and finish second to Trump in delegates.

And in case you ask “why should I care if the Republicans win?” remember that there are a number of very old Liberal Supreme Court justices who may well retire or die during Trump’s first term.  The more Republican Senators the better the odds a Liberal Justice (or more) can be replaced with a Conservative Justice without “moderate” Republicans throwing obstacles in the way.  The Supreme Court alone is enough reason to want a Republican victory.

For the 2018 midterms I choose the electoral dynamics of the Rust Belt because, in order to take the House, Democrats must win multiple districts in this area.  It is also home to many marquee Senate contests.

The dynamic I want to model is how the average Democrat will perform against a Republican in the Rust Belt.  The variables I want to test are related to how the anti-Trump platform of a normal Rust Belt Democrat will fare against the pro-Trump platform of a Rust Belt Republican.

Keep in mind, the dynamic I am focusing on is a national dynamic.  Some races always turn on local dynamics.  Examples of local factors include state-level problems that are not problems in other states, candidate skill levels, personal scandals, and so forth.

To the extent local dynamics are more important than national dynamics, the greater the extent a campaign will decouple from national tends.

Local political dynamics and their ability to decouple from national dynamics are why Republicans such as Mitt Romney can win the Governorship of Liberal states such as Massachusetts but have almost no chance carrying their home states in a Presidential campaign.

With the information we learn from this model, we will then recommend action to maximize the chances of Republicans wining the midterms.

First, let’s analyze where matters stand in the polls.

By historical standards the generic ballot looks promising for Republicans despite being several points behind Democrats.  The reason their position is healthy is because, historically, the generic ballot has shown a seasonal dynamic where Republicans tend to close in on Democrats as election day approaches for a variety of reasons.  Democrats tend to peak before campaign season because the media does nothing but portray Republicans in a negative light while ignoring Democrat faults.  It is only when election season kicks into high gear and Republicans begin their add war campaign against Democrats that Democratic negatives rise and the generic ballot narrows.

Or, to put it another way, pro-Democratic “ads” (in the form of free advertisement through media bias) are already “priced into” polls before Republicans can hit back.

For proof Democrats peak and Republicans close when Republicans run attack ads as November draws closer, see here from October 16, 2016:

Furthermore, there is reason to believe the current lead of Hillary is volatile and can be closed with a good counter-offensive by Trump.

I suspect her lead is fragile for two reasons.

…

When Trump counter attacks, he should not focus on policy but go personal instead. Interestingly, he is preferred by the electorate on a number of issues. But on policy he has maxed out his advantage. The public has already absorbed the policy differences between Trump and Clinton, and there is no further point of reminding the electorate of what they already know.

What they have not thought about are her personal negatives; and it is her personal ratings where the most return on investment for Trump rests.

If Trump attacks Hillary with ads (and he will need to include the scandals in the ad mix, as Republicans usually must do, to get around the media blackout) about her enabling of her husband’s history of sexual assault, Bill Clinton’s connections to Jeffrey Epstein, email and financial scandals, and her role in Benghazi, and any other scandal Trump may have information about, I would expect the polls to once again close between them just as they did during his September offensive which moved him from behind to ahead.

 

And so Trump’s fierce personal attacks and negative ads went on to prove this analysis entirely correct.

Aside from polls, where else can we look for information?

To date, there have been a number of pre-midterm special and off-cycle elections that have drawn attention as potential clues about how the midterms will play out.

Among these pre-midterm elections –

• The 2017 Virginia Gubernatorial race.
• A string of small state legislative races where Democrats won in heavily Republican districts.
• Pennsylvania’s 18th Congressional District won by the Democrats.

 

Surprisingly, and despite media hype, these races provide little information about how the midterms will play in the Rust Belt because they do not reflect what are likely to be the midterms dynamics –

• Virginia’s Governor Election – The Democrats won on the strength of white Federal workers in Northern Virginia’s suburbs (immigrants in these suburbs tend not to vote).  But these suburban whites are atypically Liberal by suburban standards elsewhere in the country because they depend on Federal government work.
• Various State Legislative Races – Turnout in off-cycle state legislative campaigns is very, very small and often not reflective of what will happen in on-cycle campaigns.  Turnout is so limited that it only takes a slight increase in partisan enthusiasm (which Liberal Democrats have had for over a year) to tilt a normally safe district to the other party.  But when normal partisan turnout levels occur in the midterms, these surprise wins are wiped out.  Expect Republicans to recapture most of these state seats when registered Republicans turnout at traditional voter participation rates.
• Pennsylvania’s 18th District – The Democrat won by running as a generally pro-Trump Democrat.  He also had the advantage of having no voting record to point to because he never held office before.  Unfortunately for other Rust Belt Democrats they cannot convert this formula at all to their campaigns because a pro-Trump message is completely out of sync with the anti-Trump message of their party nationally and because they mostly have voted against Trump’s platform in Congress.

 

The only useful information to draw from off-cycle races are that the Democratic base seems enthused.

To counter Democrat turnout, the course of action Republicans should follow is threefold –

1. Depress Democrat turnout as much as possible with negative attack ads.
2. Turn Independents away from Democratic candidates by driving up their negative ratings with attack ads.
3. Maximize Republican turnout to emphasize Trump’s accomplishments by hugging close to his policies which are most popular in the Rust Belt.

 

Issues to attract Independents should include –

• Link the strength of the economy to Trump’s economic policies of tax cuts and deregulation while also highlighting the Democrat’s opposition to the President’s economic policies.
• Constantly accuse Democrats of planning impeachment because impeachment talk is extremely unpopular with Independents.
• Link the Democrats to their radical base on issues such as gun control, illegal immigration, and illegal immigrant crime.
• Embrace Trump’s protectionist measures to win over Rust Belt working class voters.

 

Issues to ramp up Republican turnout should include –

• The need for a Republican Senate to confirm Conservative Judges and Supreme Court Justices.
• Frame Democratic impeachment talk as an attempt to steal the votes of Republicans who voted for Trump.  Framing impeachment as “stealing your vote for President” will help tap Republican enthusiasm for Trump and transfer that enthusiasm to Republicans who will not have Trump at the top of the ticket this year.
• The need to advance Trump’s Conservative agenda in Congress.