By Adam Gilfix and Henry Johnson
Twitter and the NBA go together like Jordan and Pippen. Not only has the NBA made Twitter an open highlight repository, the site also serves as a public forum for major announcements, conspiracy theories, and Kevin Durant’s early musings.
The undisputed champ of NBA Twitter is LeBron James. Akron’s prodigal son has over 33 million followers, more than the likes of Bill Gates and Beyoncé. This massive following exists for good reason: in addition to being an icon of culture and style, LeBron James is very good at basketball.
As it turns out, the relationship between NBA skill and Twitter followers is quite strong. Kevin Durant, a league MVP and four-time scoring champion, is next most followed after LeBron. Rounding out the top five are Shaquille O’Neal, Kobe Bryant, and Carmelo Anthony, all of whom are past or present superstars.
To quantify the link between basketball ability and social media popularity, we built a model* that predicts an NBA player’s Twitter followers count based on Value Over Replacement Player (VORP), a popular measure of skill. The model also uses the player’s time spent in the NBA, the year of his rookie season, his time since creating a Twitter account, and the number of tweets sent from his account as predictors. The NBA Twitter accounts, as well as VORP and season data – all prior to the start of the 2016 season – come from Basketball Reference. We limited the dataset to accounts with a blue “verified” checkmark and at least 5,000 followers.
By looking at the residuals of the model, we can see which NBA players’ Twitter accounts are the most over- and under-followed. Rather than simply using the residuals themselves (actual value minus predicted value), we opted to use the standardized residuals (which are on the log-scale due to the nature of the model). These standardized residuals are a better measure of over- and under-followed-ness because they pick out the accounts that the model did a poor job of forecasting based on the order of magnitude of the error rather than the absolute difference. The following table shows the top 20 accounts for which the expected number of followers differs from the actual number according to this standardized residual value, with under-followed accounts in red and over-followed accounts in green:
At the top of the list is Nick Young, who has amassed more than 368,000 Twitter followers, several orders of magnitude above his expected count of 1,576. Young may have a sub-par VORP of -5.7, but he’s built for internet stardom. He has a catchy nickname, he’s the face of an inescapable meme, and his ex-fiancé is the chart-topping Australian rapper Iggy Azalea. Kris Humphries, like Young, is likely over-followed in part because he is the former romantic partner of a celebrity, Kim Kardashian West.
Another category of over-followed accounts includes NBA analysts like Kenny Smith and Jay Williams, whose broadcasting skills have probably earned them more followers than their time on the hardwood. Kobe Bryant and Derrick Rose, meanwhile, have had wonderful seasons in the NBA, but they have amassed brands in excess of their abilities according to our model. Interestingly, King James’ Twitter account has over 33 million followers, well above the 4 million mark predicted by the model. This discrepancy is quite large on an absolute difference scale, but it’s fairly reasonable at the standardized residual level, as it is only off by one order of magnitude.
Royce White, the 16th most over-followed account, is a humanitarian and prominent face of mental health. Sebastian Telfair is a former high school star and documentary subject whose NBA career failed to meet unfair expectations. Joel Embiid, whose injuries have kept his career VORP at 0, is simply hilarious.
The most under-followed accounts, meanwhile, belong to solid players whose careers occurred before Twitter exploded. Dan Majerle played mostly in the 1990s, while Theo Ratliff’s best years were in the early 2000s. Chauncey Billups peaked a bit later, but he didn’t join Twitter until August 2016.
With the NBA season in full swing, it will be interesting to see which accounts gain and lose followers relative to expectations. In the meantime, if you’re curious whether your favorite player is over- or under-followed, please leave a comment or find us on Twitter at the eternally under-followed @harvard_sports.
*Here is the model (terms are logged due to right-skewness or squared due to left-skewness; some terms have constants added to them to account for these transformations):