Barely winning an election as a natural experiment

Question

You want to estimate the incumbency advantage: does holding a seat now make a party more likely to win that seat again next time? Comparing past winners to past losers is unfair, because winning parties were stronger to begin with.

Explain how close elections give a regression discontinuity design, why barely winning is nearly random, the key assumption, and the limitation.

Accepted Answer

Why the naive comparison fails. Parties that won last time were, on average, more popular, better funded, and better organized, all of which also help them win next time. So winners re-winning could just reflect that they were stronger all along, not any advantage from holding the seat. Past victory is confounded with underlying strength. The regression discontinuity idea. The running variable is the vote margin, and the cutoff is 50 percent (win versus lose). Look only at very close races, decided by a fraction of a percentage point. A party that squeaks in with 50.1 percent and one that jus…

Barely winning an election as a natural experiment

Your answer

Discussion

Learn the concepts

Related questions