Do demographic trends help predict the outcomes of US elections? We checked in our new @nberpubs.bsky.social paper "Pitfalls of Demographic Forecasts of US Elections" with Richard Calvo and Jesse Shapiro. 👇👇 www.nber.org/papers/w33016 A thread on what we do and find. (1/n) - ThreadSky

vinpons.bsky.social • 138 days ago

Do demographic trends help predict the outcomes of US elections? We checked in our new @nberpubs.bsky.social
paper "Pitfalls of Demographic Forecasts of US Elections" with Richard Calvo and Jesse Shapiro. 👇👇
https://www.nber.org/papers/w33016
A thread on what we do and find. (1/n)

Comments

vinpons.bsky.social•138 days ago

Scholars and pundits have, in the past, forecast large partisan shifts in the US electorate based on demographic trends, such as suburbanization, or the growth in the Hispanic share of the population. (2/n)

vinpons.bsky.social•138 days ago

For instance, in the 1969 book The Emerging Republican Majority, Phillips argued that demography would doom the Democrats. In the 2002 book The Emerging Democratic Majority, Judis and Teixeira argued precisely the opposite. Subsequent events have often defied such forecasts. (3/n)

stephenfhwebb.bsky.social•137 days ago

By defintion in a 2 party system you'd expect positions to change to maintain a basically 50/50 split. The demographic changes might still have influenced the underlying policy programmes

vinpons.bsky.social•138 days ago

We test this approach systematically using data from 1952-2020. In each election, we use survey data to fit a model relating a person’s vote to their age, gender, race, income, education, and the type of area in which they live. (4/n)

vinpons.bsky.social•138 days ago

We use the fitted model, in tandem with a projection of demographic trends, to predict future elections. (5/n)

vinpons.bsky.social•138 days ago

Even a forecaster with perfect knowledge of future demographic trends would have performed poorly over this period—worse even than one who simply guesses that each election will have a 50-50 partisan split (see graph below). (6/n)

vinpons.bsky.social•138 days ago

Enriching the set of demographics available or allowing for a rich set of interactions among demographic characteristics do not change this conclusion, even though they help the model fit the survey data better. (7/n)

Comments

Posting Rules

Reply