Team Hitting Percentage by Game (Set)

Even a casual reader of box scores would probably realize that the same team on the same night can record vastly different hitting percentages in different games (sets). I was curious if there were systematic differences, so I decided to conduct an analysis. Why might there be such differences?

As a match progresses, coaches may devise adjustments to take away a source of offensive success the opponent had been enjoying. Or coaches may come up with ways to overcome the opponent's defensive approach to increase their own team's spiking success. Some coaches may be able to implement adjustments between Games 1 and 2, but if not then, perhaps between Games 2 and 3, when there is a full-fledged intermission.

I looked at the Pac 12 conference, as it appeared to reach its halfway point of league play faster than other major conferences. The Pac 12 plays a full home-and-away round-robin, meaning that each member team plays 22 conference matches. I started compiling the statistics about a week ago, when every Pac 12 team had completed 11 or 10 conference matches (the latter is the result of the Arizona-ASU, Oregon-OSU, and Washington-WSU rivalries playing back-to-back matches at the end of the season, instead of once during the first half and once during the second half of the conference schedule).

Pac 12 matches before my cut-off date predominantly were three-game sweeps (43 out of 63 total matches; 68.3.%); 16 four-game (25.4%) and 4 five-game (6.3%) matches occurred. Note that, in any given match, both teams produce hitting statistics, so there were 126 three-game data sequences from the 63 matches. I used data from all matches, so if a contest went four or five games, I took hitting statistics from only the first three games.

Averaging over all teams in all matches, hitting percentages did not differ statistically between the first (.210), second (.228), and third (.198) games. (For those with some statistical training, I used repeated-measures Analysis of Variance.) To probe further, I plotted the results separately by team, as shown in the following graphs (the three panels were created for ease of viewing; you may click on the graphic to enlarge it).

As can be seen, there is no consistent pattern. Some teams -- particularly UCLA, Cal, and Arizona -- started fast and then declined in their hitting percentages. Others -- such as Stanford, Washington, and Arizona State -- started off relatively low and then increased their hitting prowess. Other teams appeared to peak in Game 2.

Because of the small sample sizes, it is hard to know if these are random fluctuations or substantive trends. If a given team experiences the same type of trend in the second half of the Pac 12 season as it did in the first half, then there really may be something going on. With UCLA, for example, the intermission after Game 2 conceivably could take the Bruins out of their offensive flow (more so than other teams, who also, of course, have the intermission) or perhaps allow opposing coaches to make defensive adjustments to the Bruins' attack.

On the other hand, the Stanford and Washington coaches look like they may be taking advantage of the intermission to find ways to take their respective teams' offensive attacks to higher levels. (ASU's improvement seemed to occur between Games 1 and 2.)

The larger the number of data points, the more reliable the statistical analysis. However, if a coach wanted to show his or her team how it was hitting by game in order to motivate improvements, the coach probably wouldn't want to wait too long to start doing so. Thus, in practice, most of the data compilations would end up being based on relatively few matches. The 11 or 10 matches on which the above averages for each team were based may therefore be a reasonable number.