The Kate Upton Effect is Real: How Dating Celebrities Affects Athletic Performance

By Austin Tymins and Nick Holmes

Whether it’s partying at exclusive nightclubs before Sunday morning games, or fielding the advances of attractive celebrities, professional athletes typically deal with much larger distractions than the average person. Many observers of this point have blamed poor athletic performance on the supposed distraction of dating a celebrity such as Justin Verlander who has fallen victim to this supposed effect as much as anyone. As Barstool Sports pointed out, his “ERA is like a billion” since dating Kate Upton. In this post, we’re going to try and see if this effect is real, and if so, how sizeable is the effect on athletic performance?

Most casual sports fans are able to name some of these athlete-celebrity relationships. For example, there is Tony Romo and Jessica Simpson, Tony Parker and Eva Longoria, Derek Jeter and half of Hollywood, Delonte West and Lebron’s mom (just kidding, not actually used as data). Some of these relationships seem to have followed the stated negative effect such as when Rory McIlory and Caroline Wozniacki ended their engagement and Rory promptly won the final two majors of 2014. Conversely, there is also anecdotal evidence supporting the opposite notion. Tom Brady and Gisele Bündchen have been together since 2006 and almost all of Brady’s best statistical seasons have come since then.


We considered many quantitative measures to define “high profile” including twitter followers, google analytics etc., but it seemed unreasonable to come up with a stringent rule. Therefore, our definition of celebrity is just that we’ve heard of them and subjectively think they’re famous enough to be included.

We were able to come up with 75 relationship data points across many sports all since the year 2000. Many of these relationships are well known and have reasonably well-defined dates such as Matt Kemp and Rihanna from January 2010 through December 2010. Most of the time however, it is extremely difficult to verify if a relationship actually occurred and even more difficult to find the start and end dates of the relationship. We scoured the People, E! News, Us Weekly and Perez Hilton websites to confirm as much as we possibly could.

It became apparent that it would be difficult to quantify some sports such as skiing (Lindsey Vonn and Tiger Woods) and heavyweight boxing (Wlad Klitschko and Hayden Panettiere) so we’ve restricted the dataset to the four major American sports of Football, Baseball, Basketball and Hockey. While the causal effect is only being observed in these sports, we believe it is reasonable to say these results have external validity in other sports and situations. For these final 57 data points, football production was measured using Approximate Value, baseball using Wins Above Replacement, basketball using Win Shares, and hockey using Point Shares. We gathered numbers for the most recent performance before dating began, while dating, while married/engaged, and after dating/married/engaged.


We first tested the average percent change in yearly performance from the initial level to the dating level. The average in this data was a 15.3% decrease in athletic output, a startling number that supports the hypothesized “Kate Upton Effect”. In the table below, we have the average percent changes from one relationship state to another.

Pre-Dating to Dating

Dating to Married

Dating to After

Pre-Dating to After





The 15.3% decline from pre-dating to dating matches up with the idea that performance declines when dating a celebrity. While the 15.3% decline seems significant, the 25.6% decline from Dating to Married is even more startling. The 23.9% increase in performance after a dating relationship has ended is also a rather large and interesting effect. It is also interesting to note the 6.9% decrease from pre dating performance to after dating. This also appears to be evidence of an aging effect or maybe regression to the mean, two concepts I wish to explore later in depth. Below is a table showing the breakdown of the effect across each of the four sports.

Pre-Dating to Dating

Dating to Married

Dating to After

















Observing percent changes isn’t the only way to measure this effect though. If the initial value is near 0, the percent change could be especially high if the final value differs by much at all. Therefore, I’m going to use Z-Scores to measure an individual athlete’s performance relative to the population mean and standard deviation. This allows us to conduct more statistical tests and come to a more tenable conclusion. The next table shows the tests of statistical significance using p-values derived from a one-tailed t-test on the difference in the two dependent sample means of various relationship state Z-Scores.


Pre-Dating and Dating

Dating and Married

Dating and After

Pre-Dating and After

P Values





We see that the difference between Pre-Dating and Dating is statistically significant below the 5% level and the Z-Scores show that the sign of the relationship is as shown in the percent change in yearly performance above. The difference between Dating and After Dating is also significant below 5% and the difference between Dating and Married is significant at the 10% level. The signs on these relationships are also as shown in the percent change table above.

From all the results above, it seems rather clear that dating a celebrity is correlated with an athletic performance decline. This made us wonder if the effect was dependent on the level of famousness for the significant other. Understandably, we were inspired to look into the possibility of a Kardashian-specific effect and found quantitative support for what we all know to be true. We found a 12.8% performance decline from pre-dating to dating and then a 16.5% increase after breaking up. This seems to be holding true with James Harden who has recently ended his relationship with Khloe Kardashian and has been playing class basketball since then.

About the author


View all posts

1 Comment

Leave a Reply

Your email address will not be published. Required fields are marked *