Hall of Fame quarterback Brett Favre recently made headlines for suggesting that Taylor Swift’s relationship with Travis Kelce is a “distraction” that fans would blame if the Chiefs lose in the playoffs. “If they don’t win it or get to the Super Bowl,” Favre told TMZ, “people are going to say, ‘That’s why.'”
Of course, it wasn’t too long ago when all of the buzz in the sports and entertainment world was about how Swift’s attendance at Chiefs games was propelling her boyfriend to perform better on the gridiron. The narrative became particularly pronounced in late October after a CBS graphic went viral showing that Kelce was averaging over 50 yards more in the games that Swift attends.
But, as I noted back then, this so-called Swift effect on Kelce’s yards per game (YPG) failed to reach conventional levels of statistical significance. Or more simply put, Kelce’s higher production with Swift in the audience could have merely been the result of random variation in his performance across games.
Kelce’s performance in the second half of the 2023 season is clearly consistent with that more mundane explanation, too. The Swift effect for how much better Kelce performs with her in attendance regressed from 58 YPG over his first seven games of the 2023 season down to just 4 YPG over his final nine games of the year (including last week’s playoff victory against the Dolphins).
Taken as a whole, Kelce averaged 75.5 yards per game this season in the 10 games Swift attended, compared to 50 yards per game in the six that she missed. That 25 yard per game difference again fails to reach conventional levels of statistical significance (p = .26). A simple difference in means test, for example, suggests that Kelce performs anywhere between 20.5 yards worse to 71.5 yards better per game with Swift in attendance than he does without her in the stadium.
That large 95% confidence interval for the size of the Swift effect grows even larger [-39 to 71.2 YPG] after accounting for the fact that 7 out of the 10 games she has attended were in Kansas City and Kelce averaged more receiving yards this season at home than on the road (78.8 vs 53.1 YPG respectively). This uncertainty serves as yet another important reminder of how difficult it is to make meaningful statistical inferences from small samples.
It also means that you should inevitably ignore narratives claiming that there’s a Swift effect one way or another. After all, there just isn’t evidence to reject the most plausible expectation headed into today’s playoff game: Taylor Swift will have no impact whatsoever on how her boyfriend and his team perform against the Buffalo Bills.