Playtime in a Pandemic
The COVID-19 pandemic has caused some of the largest disruptions to society’s routines in recent memory. Major League Baseball (MLB) was not immune from these disruptions and had its first non-strike-shortened season since World War II. The 2020 MLB season was also remarkable because no attendance was permitted during the regular season games. The energy and sound of a crowd can motivate or distract players on the field, so the absence of an audience may have also had an effect on players’ performance. The effects of a crowd on a game may also be different based on the stadium: stadiums can be indoors or outdoors, smaller or larger, older or newer. What teams saw the greatest difference in performance during the 2020 season?
The 2020 season being both shorter and having no attendance thus provides an interesting natural experiment to compare against other seasons. The 2021 season started occurred under similar pandemic pressures as 2020 but allowed attendance at games, which makes 2021 a good “control” to compare against the 2020 season. I’m going to focus on the duration of games rather than any specific defensive or offensive statistics because players, coaches, and umpires all can influence the speed of the game. My hypothesis is that games in the 2020 season were shorter because it was demoralizing for players, coaches, and umpires to be on the field with no one in attendance. I’m going to test this by using game-level data from the Retrosheet gamelogs for 2020 and 2021.
First, I visualized the distribution of game lengths by home team for the 2020 and 2021 seasons. The x-axis are home teams’ stadiums, the y-axis is the duration of games in that stadium, and the blue bars are for games in 2020 and the yellow bars are for games in 2021.
I calculated the difference between teams’ average game length in 2020 and 2021 and ordered the x-axis from negative (2020 games were shorter than 2021 games) to positive (2020 games were longer than 2021 games). There’s no consistent pattern than 2020 games were shorter than 2021 games across these teams. For example, the Milwaukee Brewers played over 25 minutes longer in 2021 than in 2020 while the Toronto Blue Gays played 9 minutes shorter in 2021 than in 2020. Here’s a table of the teams’ differences between 2020 and 2021.
The causes of these differences may not be attributable only to the pandemic. We could compare the differences by indoor vs. outdoor stadium, large vs. small stadiums, time of year, weather conditions, and teams with stronger vs. weaker statistics. To begin to explore whether these differences are because of the time of year (cold springs and falls or hot summers may speed up games), I explored the 2020 vs. 2021 differences by month of the game. Because the 2020 regular season did not start until late July compared to a normal April start, we only have three months to compare. There is a consistent trend that 2021 games are longer on average than 2020 games. A simple t-test reveals these differences are statistically significant (p < 0.05) for August and September.
There is mixed but robust evidence that there were significant changes in the length of MLB games between 2020 when there attendance was not permitted because of the COVID-19 pandemic and in 2021. There is a large amount of variation in the duration of games across stadiums in general and some teams played faster in 2021 than in 2020. But looking at all teams together, they played significantly faster in 2020 months of August and September than in 2021. This supports my hypothesis that the lack of attendance in 2020 significantly impacted how games were played, particularly in the later months of the season.