Separately measuring home‐field advantage for offenses and defenses: A panel‐data study of constituent channels within collegiate American football

Abstract

We improve constituent‐channel estimates of home‐field and neutral‐site advantage for collegiate American football's top division by utilizing a richer, 12‐season data set and by exploiting the COVID‐19 pandemic as a random shock. Novel to the literature, we separately examine points scored by each team, allowing us to identify impacts on each team's offense and defense individually. The information set provided by our model is a strict superset of that provided by the previous standard in the literature, making ours a strictly dominant modeling choice. We demonstrate this improvement theoretically and empirically. Physiologically, away‐team travel distance does not impact their own score, but it increases home‐team scores, consistent with the notion that defenses tire faster than offenses. There is also similar but limited evidence of this effect for neutral‐site teams. Time zones may play a minor role, too. Psychologically, crowd size and density hurt away‐team scores but do not impact home or neutral‐site teams. The away‐team effect disappears in 2020, however, indicating that the pre‐2020 effect is caused by the crowd's noise, not their mere presence. We also find that increasing stadium capacity while holding crowd size constant hurts home‐team scores, highlighting the importance of considering ticket demand when considering stadium expansion. Tactically, stadium familiarity helps offenses, not defenses, while team‐opponent familiarity has the opposite effect. Weather also plays a role. At median values for key variables, we find an overall home‐field advantage of 4.1 points.