/cdn.vox-cdn.com/uploads/chorus_image/image/31977873/20131231_rvr_sz2_029.0.jpg)
The original idea was to test whether there are any strategic tradeoffs in running a particular variety of offense. For instance, are offensive linemen physically specialized in different types of offenses? Last week I found that, on average, spread-to-run offensive lines were 6.5 pounds lighter than their pro-style counterparts (though this missed statistical significance at the 95% confidence level). Are there any differences when you account for BMI instead of just weight?
BMI is obviously an imperfect measure of both fitness and physical specialization. It notoriously doesn’t take muscle into account, so people with a high percentage of muscle – such as football players – rate as overweight or obese when they might not otherwise using more detailed body-fat measurements. However, BMI is a step up from simply looking at body weight to capture physical specialization for offensive linemen, so I ran the numbers and present the results below.
I first ran a simple ANOVA test to see if there were any differences between the offense-type group means. This was just to see if there was variation in the average BMIs between any dyad of offense types – pro versus spread-to-run, pro versus option, pro versus spread-to-pass, etc – but it won’t tell us which ones are different.
Analysis of Variance | |||||
Source | SS | Df | MS | F | Prob>f |
Between groups | 279.56 | 3 | 93.19 | 13.24 | 0.000 |
Within groups | 2680.67 | 381 | 7.04 | ||
Total | 2960.23 | 384 | 7.71 |
The ANOVA test results are pretty clear: there is sufficient reason to believe that the BMI means are not the same across groups. But we kind of expected that, since we knew previously that option teams were significantly lighter than all other offensive lines, and this could account for the high F test statistic.
Next, I ran a few follow up tests to show post-hoc pairwise comparisons – in this case, a more detailed comparison of BMI by offense type. I ran Bonferroni, Scheffe, and Sidak methods, but I’ll only show the Sidak ouput since all three produced very similar results.
Comparison of BMI by Offense Type (Sidak) P values are in parentheses |
|||
Style | Pro-style | Spread-to-run | Option |
Spread-to-run | -0.63 (0.37) | ||
Option | -2.11 (0.00) | -1.49 (0.00) | |
Spread-to-pass | 0.47 (0.77) | 1.10 (0.02) | 2.58 (0.00) |
The average differences in mean BMIs between spread-to-pass and pro-style offensive lines, and spread-to-run and pro-style BMIs aren’t statistically significant. The latter is the comparison we were most interested in originally – whether, for instance, Saban’s offensive linemen had greater BMIs than Urban Meyer’s group of linemen. At least according to this test, there’s no reason to suspect that the mean BMIs of pro and spread-to-run offensive lines differ by any meaningful amount.
As you might have guessed, option offensive lines’ average BMIs are significantly lower (ranging from 1.49 to 2.58 points) than other offensive schemes’ lines. Again, military restrictions play a key role here.
Finally, maybe the most surprising result is that spread-to-pass offensive line BMIs are an average of 1.1 points higher than their spread to run counterparts. This gives us additional evidence that "spread" teams are a highly diverse group of offenses, with statistically significant physical variation between spread-to-pass and spread-to-run lines.
What will be interesting to watch is how the rise of packaged plays – in which the quarterback makes a run/pass post-snap read while the offensive lines run block for up to three yards from the line of scrimmage – affects the physical differences between offensive lines. While spread-to-pass teams might currently be larger than their running counterparts, packaged plays allow teams to pass as much as they want but still run block. It’s possible that spread-to-pass teams might then prefer leaner linemen for run blocking, even if the ultimate goal is to put the ball in the air. A core group of teams have run packaged plays for years, including West Virginia, Oklahoma State, and Texas A&M, though that number is likely to grow significantly this upcoming season.
Loading comments...