One of the most prominent challenges in calculating the APR for 2003-04 is the issue of small sample sizes for particular teams and their impact relative to the application of penalties.

The Committee on Academic Performance agreed to address that issue by using a statistically derived margin of error for each sport, which will help ensure that penalties given to teams with small sample sizes are as statistically valid as those for teams with large sample sizes.

"Because we are in the very early application of the APR, the Committee on Academic Performance has made allowances for squad sizes to ensure fair treatment among sports," said Kevin Lennon, NCAA vice-president for membership services.

Lennon said a consistent margin of error set at a confidence boundary of 84 percent will be applied to determine each team's subjectivity to contemporaneous penalties. That means the "upper confidence boundary" of a team's APR score will have to be below 925 for that team to be subject to contemporaneous penalties. Some smaller squads that may be identified as under-performing in the 2003-04 APR reports will not be subject to penalty once the confidence boundary is applied.

However, it may be unwise for institutions to become comfortable if their teams avoid contemporaneous penalties only because the upper confidence boundary puts them above the 925 APR cut-off.

"The confidence boundary provides only a temporary reprieve, particularly for some smaller, under-performing, squads until more years of APR data are available," Lennon said. "These confidence boundaries will be reduced and possibly eliminated as additional years of data are added to the APR."

Teams that rely on the confidence boundary to escape penalty rather than improve their APRs right away also might find the "pay me later" syndrome hard to accept. As more years of APR data become available, teams will find scores harder to change. For example, an APR score of 890 with four years of data and no confidence boundary is a hard score to recover from when only one year of data changes each year in the rolling four-year average. Also, by the time four years of APR data are available, the historically based penalties will be in effect, which hit teams harder than contemporaneous penalties do.

"All indications are that the easiest -- and wisest -- time for teams to improve their APR is now when fewer years of data are involved," Lennon said. "Institutions should take appropriate action for all teams just above, at or under 925, regardless of the confidence boundary. Teams that wait because they think the confidence boundary is a safe haven may regret it in a year or two when they face much more serious penalties and the confidence boundary has gone away."

-- Gary T. Brown