Remember in school when you had to do science projects? The teacher, stifled by the so-called scientific method, forced you to shoe-horn your findings into a rigid set of hypotheses, data collections, graphs and conclusions.
While those were hardly my favorite moments in school, there is something to be said for the method behind the madness: you start any study with a hypothesis, then gather the data to prove or disprove it.
And that's where I found myself following last season's offensive debacle. Obviously, a lot of the blame went toward quarterback Garrett Gilbert. But was it too much? The eye test said that Gilbert struggled in part due to a mediocre (at best) running game built around a trio of running backs who rotated in and out of the infirmary all season, an offensive line that seemed to be less physical than a Jane Fonda workout tape and a set of receivers that were covered more often than a Beatles song.
But eyes are often deceiving. And so, I set out to form a hypothesis, and through data collection, either prove or disprove it.
HYPOTHESIS: If Garrett Gilbert were put in situations to be successful (read: non-obvious passing downs), then he would put up better numbers.
Sounds obvious, right? But I don't want there to be just a little bit of improvement. I want Gilbert to show me that he's capable of managing a game when he's supported, say somewhere around 60 percent completion percentage, a 2-1 touchdown-to-interception ratio and a conversion rate of better than 45 percent on third downs. And that's our base line. To compare, Gilbert completed 59 percent of his throws (not far off there) with a 10-17 touchdown-to-interception ratio (way, way off), and converted just 37.5 percent of his passing third downs.
In order to do this, I tracked every Gilbert passing attempt from last season — all 441 of them — and classified them by down and distance. I thought that would be the best course of action, rather than something a bit more artificial than yards per carry. Because if Texas gets a holding penalty, then Fozzy Whittaker runs for five yards on each of the next two downs, Gilbert is still facing third-and-10 right in the face.
I've broken each down into three distance categories: long, medium and short. Long refers to any situation where a the team had to travel seven yards or more for a conversion (either a first down, or a touchdown if inside the 10). Medium was any down where the team had between four and six yards to a conversion, while short was any situation with three or fewer yards needed for a conversion.
I've also listed the number of conversions in each situation. If the play was a first-and-10, and Gilbert passed for 11 yards and a first down, it was a conversion. If the play was a second and goal from the four and he passed for a touchdown, conversion. This will help with our conversion rate.
And here are the final season totals:
First-and-long: 98-174 for 1,202 yards and three touchdowns to five interceptions (38 conversions)
First-and-medium: 0-1 (no conversions)
Second-and-long: 81-118 for 712 yards and three touchdowns to five interceptions (25 conversions)
Second-and-medium: 9-18 for 85 yards (five conversions)
Second-and-short: 3-4 for 30 yards (three conversions)
Third-and-long: 29-60 for 346 yards and five interceptions (14 conversions)
Third-and-medium: 28-46 for 308 yards and one touchdown to two interceptions (24 conversions)
Third-and-short: 7-14 for 53 yards and one touchdown (seven conversions)
Fourth-and-long: 0-1 (no conversions)
Fourth-and-medium: 2-2 for 12 yards and one touchdown (one conversion)
Fourth-and-short: 3-3 for 6 yards and one touchdown (two conversions)
Total: 260-441 for 2,754 yards and 10 touchdowns to 17 interceptions (119 conversions)
So, What Did We Learn?
* The study confirmed the original hypothesis. In non-obvious passing downs, or second-through-fourth downs of medium and shorter distances (we'll get to first down later), Gilbert completed 59.8 percent of his throws for 494 yards and four touchdowns to two interceptions.
He converted 51.7 percent of his third downs and 60 percent of his fourth down attempts at those distances.
* He was significantly worse in long situations. On obvious passing downs, or second-through-fourth downs, facing distances of more than seven yards, Gilbert actually completed a slightly higher percentage of his throws (61.5 percent). His yards per completion were also up slightly, from 9.5 in medium and short to 9.6 on long situations.
But his touchdown-to-interception ratio was an abysmal 3-to-10, and he converted just 23.3 percent of his third downs and failed on the lone fourth-and-long attempt.
* When adding first downs, long distances were responsible for a whopping 15 of Gilbert's 17 interceptions. The other two interceptions he threw were in third-and-medium situations. In long situations, Gilbert averaged an interception every 23.5 attempts. In medium-and-short situations, he threw an average of one interception every 44 attempts.
He was also much less likely to throw a touchdown in long situations. Six of his 10 touchdown passes came on first- and second-and-long, but he averaged one touchdown pass every 58.8 passes. Four came on his 88 attempts in medium-and-short situations, or one every 22 throws.
* Out of situations where Gilbert had more than 10 attempts, he was most accurate, by far, on second-and-long. On those plays, he completed 68.6 percent of his throws. Of course, there's a potential explanation: typically offenses in second-and-long go for shorter throws to try and create manageable third downs. In fact, out of those situations, second-and-long had the second-shortest yards per completion, ahead of only third-and-short.
Of course, Gilbert converted every third-and-short pass he hit on, while he converted just 25 of his 81 completions on second-and-long. It's also important to note that Gilbert's increased accuracy on the down didn't stop him from throwing five second-and-long interceptions.
First and long saw Gilbert's largest yards-per-completion (12.3) as he tried downfield more. He was also less accurate (56 percent) and also threw five interceptions at that down and distance.
* Passer rating can be an imperfect science. On the non-obvious passing downs, Gilbert completed nearly 60 percent of his passes with a 2-to-1 touchdown-to-interception ratio and put up a passer rating of … 118.4. That would put him 10th in the Big 12 last year.
But here's where the passer rating doesn't work: it doesn't factor in conversions. Take the following hypothetical situation with two separate outcomes.
Gilbert faces a pair of third-and-twos. In the first situation, Gilbert completes a pair of three-yard passes. In each situation, the drive continues, and with it, a chance at points. His passer rating there is 125.2.
In the second situation, Gilbert completes a 20-yard pass on the first third-and-two, then fires incomplete on the second. Sure, the first play is a larger gain, but the second results in the end of a drive. His passer rating for that situation: 134.
So even though he converted on both throws in the first, he's knocked because passer rating favors longer completions.
It is worth noting however, that his passer rating in medium and short situations is significantly higher than that in long situations. On second-through-fourth-and-long, Gilbert's passer rating was 105.5.
I used to have a professor who told me that numbers numb. And there's certainly an argument to say that I've used far too many here. But if you take nothing from this study, take this: those worried about whether Garrett Gilbert can perform at quarterback for Texas should also take a long look at his supporting cast. Because if the Longhorns can run the ball and put him into favorable situations this season, the statistics would indicate that he's up for the job.