Being a referee is a thankless job. At best, no one notices you, and at worst, you are the subject of controversy, vitriol and persistent reforms to review and challenge your very decisions.

Like all human beings, referees are imperfect and prone to biases, implicit or otherwise. Some of these biases include having more personal fouls called “against players when they are officiated by an opposite-race officiating crew than when they are officiated by an own-race refereeing crew”.

One particular bias that is widely accepted by the basketball community is the idea of superstar calls. The ideas is that refs give preferential treatment to star players because fans pay to see stars perform and that these stars (especially veterans) have earned the benefit of the doubt. Using data from the NBA’s Last Two Minute Reports, I will test if this bias towards star players exists even in the end of close games.

Background of the Last 2 Minute Report

In March of 2015, the NBA began issuing Last Two Minute Reports (L2M), which review the decisions made by referees in the last 2 minutes of a close game — any game where the lead is 3 points or fewer during the last 2 minutes of the 4th quarter or overtime.

All calls on each report are reviewed by a league operations team. They provide transparency to referee performance and the closest thing we have to objective truth when it comes to the accuracy of a call or no-call.

Using webscraping code provided by atlhawksfanatic, I was able to collect all L2M data from 2015 to 2019 with 38,304 calls over 1,955 games (see below).

To explore the potential bias of superstar calls, additional fields were added to show the number of all-star appearances (to measure popularity/star-power) and years of experience for all players on the report.

Note: All-Star appearances are counted for the entire season, not just from when the selection is announced. For example, D’Angelo Russell is shown as a 1 time All-Star for all plays in the 2018–19 season, not just from February 1st, 2019 onward.

Disadvantaged vs. Advantaged

To understand how a referee can favor a player, we have to understand the ramifications of each incorrect decision. After reviewing the data, I decided to analyze only shooting, personal, loose ball and offensive foul calls. These are the four highest called violations — making up 31,382 or about 82% of all calls — and the interpretation of these calls can be very subjective unlike non-foul violations (double dribble, defensive 3 seconds, etc.). Additionally, they almost always consist of a committing and receiving player, which allows us to understand which player is advantaged or disadvantaged on any given incorrect decision.

• Incorrect Call (IC) is an advantage to the receiving player and a disadvantage to the committing player (i.e. LeBron James is called for a shooting foul on James Harden’s layup attempt, but the call is incorrect)
• Incorrect No-Call (INC) is the opposite: advantage for the committing player, disadvantage for receiving player (i.e. LeBron James makes illegal contact while contesting James Harden’s layup attempt, but no foul is called)

To determine how likely a player will be advantaged or disadvantaged by an incorrect decision, we calculate the rate (%) by dividing the incorrect decisions over the sum of all correct and incorrect decisions:

• Advantaged Rate = (IC’s as the receiving player + INC’s as the committing player) / (IC’s and C’s as the receiving player + INC’s and CNC’s as the committing player)
• Disadvantaged Rate = (IC’s as the committing player + INC’s as the receiving player) / (IC’s and C’s as the committing player + INC’s and CNC’s as the receiving player)

Advantage and Disadvantage Breakdown

After aggregating all calls, we see that James Harden led the league with 708 decisions on L2M reports with 35 advantaged calls and 24 disadvantaged calls. His advantaged call rate means 9.7% of calls made while he was the receiving player and no-calls made while he was the committing player were incorrect, and hence gave him an advantage. Disadvantage rate means 6.9% of calls made while he was the committing player and no-calls made while he was the receiving player were incorrect.

Net Calls is the number of advantaged calls subtracted by disadvantaged calls, and the Net Call Rate is the net calls/total calls. This measures how likely a player is advantaged (positive) or disadvantaged (negative) on any given call in the L2M report. To contextualize this, LeBron James received 17 more disadvantaged calls than advantaged calls on L2M reports from 2015–2019, and received a net disadvantaged call (one that was not negated by an advantaged call) on 3.1% of all referee decisions in the end of close games.

Are Superstars Favored?

While we have an overview of how much a player is advantaged vs. disadvantaged, it doesn’t tackle the central question of whether superstar calls exist in the last 2 minutes. To do this, I broke up all players into 3 groups: superstars (6 or more All-Star selections), stars (1–5 All-Star selections) and non-stars (0 All-Stars).

Note: The number of All-Star selections are counted from 2019. The cut-off of at least 6 All-Stars to be considered a superstar was made to differentiate players like Steph Curry and Anthony Davis (6 All-Stars selections) from the likes of Kyle Lowry, Al Horford and Kevin Love (5 selections). Even then, players like Giannis and Kawhi did not make the super-star group. I recognize All-Star selections is not the perfect metric for star power, but it was the most straightforward one.

The data shows that superstars are the most likely to be advantaged, but also most likely to be disadvantaged by incorrect decisions. However, to better understand the reasons behind these numbers, we need to differentiate a player’s advantage or disadvantage on both sides of the ball.

Defense vs. Offense

In order to do this, we need to 1.) eliminate loose ball foul calls, which typically occur on rebounds and given the way the L2M report is structured, it’s hard to tell if the committing or receiving player is on offense or defense, and 2.) count offensive fouls differently from shooting or personal fouls because the committing player will always be on offense.

• Offensive Advantage: IC’s as the receiving player for shooting and personal fouls + INC’s as the committing player for offensive fouls
• Offensive Disadvantage: INC’s as the receiving player for shooting and personal fouls + IC’s as the committing player for offensive fouls
• Defensive Advantage: INC’s as the committing player for shooting and personal fouls + IC’s as the receiving player for offensive fouls
• Defensive Disadvantage: IC’s as the committing player for shooting and personal fouls + INC’s as the receiving player for offensive fouls

Offensive Breakdown

On offense, superstars again have the highest advantaged and disadvantaged rate out of the 3 groups. However, there is a larger variance in the advantage rate. Compared to stars, superstars are over 20% more likely to be advantaged (i.e. receiving an incorrect foul calls), while only 3.9% more likely to be disadvantaged (i.e. incorrect no calls on an actual foul).

Overall, the referees tend to swallow their whistle (make incorrect no-calls) at a relatively even rate across all player groups, but are more likely to give lenient foul calls to superstars on offense.

Defensive Breakdown

On defense, superstars are the most advantaged with a lower disadvantaged rate than stars. This makes sense because referees are less likely to call fouls on superstars in the last 2 minutes in fear of fouling them out. Interestingly, we see that stars have advantage rates similar to non-stars, but have disadvantaged rates closer to superstars — hence giving them none of the benefits of stardom, but all of the detriments.

In general, the data shows all player groups have positive net advantaged rates on defense, but negative rates on offense. The defense is at an advantage over the offense in the last 2 minutes of close games because referees call fewer fouls, which results in fewer incorrect calls, but more incorrect no-calls. This supports the general notion that fouls are harder to come by in the last 2 minutes when referees tend to “let the players play” rather than call fouls that could determine the outcome of a game.

Conclusion

In summary, superstars receive preferential treatment on both offense and defense compared to stars and non-stars. They are most likely to receive incorrect foul calls on offense and commit incorrect no-calls on defense, which proves the existence of superstars calls even in the last 2 minutes of close games. While disadvantaged call rates tend to have a positive correlation to advantaged call rates, there is a clear indication that superstars benefit more than their peers, albeit at a marginal level.

It’s important to point out this is still a small sample size of all referee decisions that occur over the course of a NBA season. It’s fair to argue that superstars would receive even more preferential treatment in the first 46 minutes of games where games tend to be officiated more loosely.

Lastly in defense of NBA referees, we do see that Net Call Rates for players drop closer to zero as the number of calls increase. The more calls refs make on players, the less likely a player will see a continued pattern of disadvantage or advantage. Of course when we look at the right side of the graphic above, the players with the most calls are predominantly stars or superstars, and there is still persistent referee biases for such players.

Data Sources

• L2M Dataset
• Tableau Public Dashboard