Poll Me Twice, Shame On Me

Trump should end the Iran war, undo tariffs, and elevate other Republicans, but economic anger makes this strategy ineffective now.

Trump should collaborate with Elon Musk to undo the war decision, but is instead grasping at straws with gimmicks.

The midterm convention forces Republicans to attend, highlighting the party's awkward struggle to define a post-Trump agenda.

The convention forces reporters to chase Republicans like Mike Rogers, who decided to attend despite a tough race.

The convention is unhelpful because it keeps Trump front and center instead of letting Republicans run against him, and will pivot to anti-Democrat messaging.

JD Vance is distancing himself from Trump on Iran, reflecting a broader problem Republicans will face defining a post-Trump agenda.

Michigan displays complicated Democratic dynamics with Jewish groups not endorsing El-Sayed, highlighting internal factionalism.

El-Sayed is a below-average candidate who will underperform, but Democrats have a large environmental cushion in Michigan.

The race tests whether a left activist can win in a battleground state, with recriminations likely if he loses.

Data shows candidates like Ilhan Omar and Elizabeth Warren underperform generic Democrats, but talent like AOC's can offset this.

First impressions matter, and El-Sayed's leftward primary positioning makes general election moderation difficult.

Primary polls have a massive margin of error, with a 25-point empirical margin, making them highly uncertain compared to general elections.

The average error in a statewide primary is 13 percentage points, so polls should not be expected to be within two or three points.

The center-left needs a big-tent liberal vision with two major policies, like Obama's appeal to both liberals and moderates.

A successful strategy is going left on one core issue while breaking with the party on others to appear authentic.

A coherent liberal worldview can combine free markets, healthcare decoupling, and controlled legal immigration.

Rob Sand runs above the partisan fray by being direct and pragmatic, avoiding attacks on vested interests like big ag.

Polling importance increases near election day, and fundraising weights are determined empirically via multivariate regression on past data.

The model qualitatively weights dependent data and handles randomization in simulations, but the empirical basis for weighting is unclear.

Unpredictability is predictable, and models must avoid overfitting by incorporating political knowledge about polarization.

Modeling a unique event run only once is philosophically difficult, as it tries to account for unknowable future events.