Lynn Vavreck for NYT’s The Upshot:

The evidence suggests that campaign ads have small effects that decay rapidly — very rapidly — but just enough of the impact accumulates to make running more advertising than your opponent seem a necessity.

It sets off an arms race of ads as candidates try to neutralize or displace their opponents. But will the 2016 general election be different? Mr. Trump has used unconventional campaign tactics and has relied on free media to get his messages out. All of this may render advertising less relevant.

A study estimated that most of the impact of an ad in a presidential election is gone within a day or two of its airing (I am one of the authors of this paper). In governor, congressional and Senate elections, the effects last a bit longer: three or four days. Fleeting effects on campaigns have been shown by various authors in the lab; in Canada; in the 2000 and 2004 general elections; in the 2006 midterm elections; in the 2012 general election; and in field experiments in a Texas governor’s primary in 2006 and a general election in 2014.

The takeaway from these studies is simple: Even though the effects from an ad imbalance are small and go away fast, candidates cannot allow them to pile up. Election Day may be far away, but candidates may still want to match their opponents’ daily advertising in the months before the vote because they care about publicly released news polls that convey information to voters — and donors — about their viability and the closeness of the race.