|Does Getting Those Ground Balls Really Matter?|
Statistics and Their Relationship to Win-Loss Records in Men’s and Women’s Division I Lacrosse
Failed clears, turnovers, and ground balls lost to opponents. They may cause coaches to scream or otherwise “overheat,” but how much do they ultimately matter? Is there a relationship between commonly recorded statistics and teams’ win-loss records?
Selected end-of-season statistics were gathered for all men’s and women’s Division I teams. Not all statistics were available for all teams, but there is relatively little missing data for most results presented here. Some statistics (e.g., turnovers for the men) were not available frequently enough to warrant inclusion. In any case, no attempt was made to be exhaustive.
Statistics such as goals and face-off wins are, with few exceptions, unambiguous. On the other hand, most of the statistics analyzed here are subjective, and the tendencies and habits of the scorekeeper can have a big influence. To the extent that scorekeepers record, say, ground balls differently, it may obscure the relationship between GBs and win-loss percentage.
However, one hopes that, if a statistic is recorded too liberally or too conservatively for the home team, the visiting team gets the same treatment. Therefore, the team-opponent difference (on a percentage or per-game basis, as appropriate) may be more predictive. And that is in fact what we generally found, so you’ll see a number of “margin” statistics below.
Results for men’s and women’s lacrosse were very similar, so what follows is organized by statistical area rather than gender. For each statistic analyzed, teams were placed in five groups, with the cut points representing quintiles (20% bands), but only roughly, as more natural cut points were used whenever possible. The average win-loss percentage (WL) and computer power rating (PR) were then computed for the teams in each of the five groups.
There are admittedly better ways to do the statistical analysis than to use a grouping arrangement, but this method makes the results more digestible. Suffice it to say that all of the relationships described here are statistically significant, although they are not always neatly linear in nature.
We begin our look at the Division I statistics with shooting percentage.