For a sport with rabid fans, historic rivalries and trivia buffs, college football does not have a single, official repository of historic rosters. Each school prints its own rosters and is responsible for its archives.
To analyze changes in weight over time, The Associated Press obtained official rosters from 2001 to 2012 from all 120 Football Bowl Subdivision schools, the big-time teams that make up what used to be known as Division I-A. School rosters have been recognized by peer-reviewed scientific journals as a legitimate source for studying athletes.
For some schools, the AP already had media guides. Other schools published their media guides and official rosters online. In many cases, reporters asked schools for their historic rosters. In some instances, when a school did not have a roster readily available, the AP used cached versions of the school's official football website, capturing the roster as it was published by the school at the time.
From there, the AP studied more than 61,000 individual athletes whose name appeared on rosters for the same teams for multiple years. The AP did not attempt to track transfer students.
The AP tracked each player's change in weight over each year and over his career. The AP also calculated each player's yearly body mass index, which calculates the ratio of height to weight. A gain of 20 pounds is more significant to a 5-foot-8-inch athlete than to a 6-foot-6-inch athlete. In such cases, comparing BMI is useful.
For decades, scientific studies have shown that anabolic steroid use leads to an increase in bodyweight. The weight gains observed in the studies varied depending on the type of drug, the sport and the duration of use. In 2004, two Dutch doctors analyzed all the research and concluded that short-term steroid use typically helped athletes gain 4-11 pounds. Some athletes reported gains of 20-33 pounds, but that was outside of well-designed clinical studies, the authors wrote in the journal Sports Medicine.
Kathy Turpin of the National Center for Drug Free Sport, which conducts testing for the NCAA and about 300 schools, said that rapid, significant weight gain is something her organization considers a potentially suspicious indicator.
Changes in weight and body mass do not prove steroid use, and the purpose of the AP's analysis was not to prove that individual players were doping. The analysis was one part of a larger effort to test the question: Does the NCAA's incredibly low rate of positive steroid tests - it was .64 percent in 2009 and as low as .26 percent in 2006 - accurately reflect a near absence of steroid use? Former drug testers, players, dealers and trainers said otherwise.
In its study, the AP conducted several tests.
First, the AP compared all players' body mass gains against everyone else in big-time college football and found outliers. That process, known as linear regression, factored in other variables that could have accounted for weight gain, including position, the school's athletic conference, how much money was spent on football, the team's win-loss record and even whether the school's drug policy gave it the authority to test for steroids.
Second, the AP looked at players who gained more than 20 pounds in a single season - the high end reported in the 2004 study of bodyweight gain by athletes. More than 4,700 players fell into that category, although it's unclear based on the data how much of that amount was muscle.
Finally, reporters examined players who, in any one year, also increased their body mass by more than 21 percent. That's the extreme change that former NFL star and admitted steroid user Lyle Alzado saw when he first started doping in college. During the last decade, more than 130 players had done so.
As with any statistical analysis, the tests are as good as the available information. Schools don't routinely make available their team body composition, strength training or speed data. One explanation for the unusual gains is that NCAA players said they were just getting really fat.
``I could easily increase your weight just by pouring a quarter cup of olive oil on everything you eat. Believe me when I tell you, your weight is going to go up,'' said Dan Benardot, director of the Laboratory for Elite Athlete Performance at Georgia State University.
While former players who have admitted using steroids said they quickly put on lean muscle, it's impossible to extrapolate those anecdotal accounts to the entire population without knowing about each player's body composition.
``There's a big mass increase that occurs. Is it muscle? Is it fat? Is it a combination of the two? You don't know,'' Benardot said. ``If it's mainly muscle then there are suspicions that there are anabolic hormones being used to aid that accrual. If it's predominantly fat, then they're just eating a lot more to increase.''
The AP consulted on its methodology with George Shambaugh, a statistician and professor at Georgetown University.
``The outliers suggest there are some underlying factors that make these players different,'' Shambaugh said. ``Sure, it could be steroids, or it could something else. But steroids are certainly a possible culprit, so it's worth opening up the box and taking a look inside.''