| Historical Stats & Info |
| "It is the most important victory of my career." -- Head Coach Dick Voris, after the Hoos’ 15-12 victory over Duke on September 27, 1958. Voris finished his UVA career with a record of 1-29. |
| "We've stopped recruiting young men who want to come here to be students first and athletes second." -- Former Virginia head coach Sonny Randle, describing his strategy for turning around UVA's football program |
| "As the score mounted, to 20-0 and finally 26-0, his movements slowed. With two minutes to go and South Carolina threatening once more, Voris stood behind several rows of substitutes, staring at his shoes." -- Sports Illustrated, describing Coach Voris’ stellar coaching performance during the Hoos’ 26-0 loss to South Carolina in 1960 |
| "Really, Texas wasn't as good as I thought they'd be." -- Ted Manly, Virginia's freshman quarterback, after Texas had spanked the Hoos 68-0 |
| Curing Saturday Night Fever, Clearing the Stench of POO, and Other Half-Truths Gleaned from Uniform Color Data |
 |
| |
 |
| |
 |
| |
 |
| |
 |
| |
 |
| |
 |
| |
 |
| |
 |
| |
 |
| www.hoosfootball.com is not affiliated with or endorsed by the University of Virginia. |
| All photographs on this site are the property of the respective copyright owners. They are presented here solely for educational and/or editorial purposes and may not be reproduced for any other purpose. |
| Copyright © 2008 www.hoosfootball.com All Rights Reserved |
The Hoos reportedly will be sporting a range of new looks this fall, with the latest reports
indicating that UVA will have at least three jerseys, with orange returning to the regular
rotation. As the phrase "Oregon of the East" gets thrown around -- either in excitement
or horror, depending on your point of view -- we at Hoosfootball.com thought it would be a
good time to look back and see if we can draw any conclusions about which color
choices have served UVA well in the last few decades.
We're going to look at the colors of the jerseys and pants UVA wore in its games from
1982 to 2009, essentially the entire George Welsh and Al Groh eras, and see how they
compare to game results. Obviously, the statistical revolution has taken over baseball,
so why not college football as well?
To do a serious, statistically rigorous study of how UVA football performance has varied
with uniform color choices, we'd need to design a study that incorporated any
independent variables that we think could have an impact on the outcome of games.
These would include measures to account for the strength of the opponent and whether
the game was home or away, among other things. Some basic statistical tools would
give us some idea of any relationships between game results and those independent
variables. Then we'd need to address some obvious multicollinearity problems. For
example, any apparent effect from the colors would most likely be the result of the
traditional use of white pants at home and dark on the road. Only after addressing those
issues would we be able to make a fair determination of whether uniform color choices
had any relationship to game results.
There are two problems with this rigorous approach:
First, it would almost certainly tell us that once you eliminate the home/away issue or
other factors, there's no true relationship between uniform color and results.
Second, it would be a gigantic pain in the ass.
So we're just going to compare results for the different colors and combinations and leave
out all of those other factors that "football experts" think are important, like the strength of
the opponent or whether the game is home or away. In effect, we're going to assume that
it's possible that blue pants make you faster than white pants or that orange jerseys
make you hit harder.
There are seven different color combinations in the time period we studied. The win-loss
records for the combinations included in our data set are below.
indicating that UVA will have at least three jerseys, with orange returning to the regular
rotation. As the phrase "Oregon of the East" gets thrown around -- either in excitement
or horror, depending on your point of view -- we at Hoosfootball.com thought it would be a
good time to look back and see if we can draw any conclusions about which color
choices have served UVA well in the last few decades.
We're going to look at the colors of the jerseys and pants UVA wore in its games from
1982 to 2009, essentially the entire George Welsh and Al Groh eras, and see how they
compare to game results. Obviously, the statistical revolution has taken over baseball,
so why not college football as well?
To do a serious, statistically rigorous study of how UVA football performance has varied
with uniform color choices, we'd need to design a study that incorporated any
independent variables that we think could have an impact on the outcome of games.
These would include measures to account for the strength of the opponent and whether
the game was home or away, among other things. Some basic statistical tools would
give us some idea of any relationships between game results and those independent
variables. Then we'd need to address some obvious multicollinearity problems. For
example, any apparent effect from the colors would most likely be the result of the
traditional use of white pants at home and dark on the road. Only after addressing those
issues would we be able to make a fair determination of whether uniform color choices
had any relationship to game results.
There are two problems with this rigorous approach:
First, it would almost certainly tell us that once you eliminate the home/away issue or
other factors, there's no true relationship between uniform color and results.
Second, it would be a gigantic pain in the ass.
So we're just going to compare results for the different colors and combinations and leave
out all of those other factors that "football experts" think are important, like the strength of
the opponent or whether the game is home or away. In effect, we're going to assume that
it's possible that blue pants make you faster than white pants or that orange jerseys
make you hit harder.
There are seven different color combinations in the time period we studied. The win-loss
records for the combinations included in our data set are below.
|
| To answer your first question, yes, we did once wear all orange in the George Welsh era, although it's a safe bet he would like to forget that it ever happened. There was some discussion of this on the radio pregame show before the 2009 Virginia Tech game. As I recall from the discussion, Coach Welsh begrudgingly agreed to the all-orange look after some pleading from players. But after the 42-10 beat down by Maryland, he apparently said it would never happen again. If you see someone throwing their hat to the ground in disgust the first time we bust out Nike's latest all-orange combo this year, I recommend that you pick up the hat, hand it back to Coach Welsh, and thank him for his years of fine service to UVA. |
| So despite the small sample size, we still think it's safe to conclude that we should rethink the whole "Power of Orange" (POO) concept. Moving on, we see that the best winning percentages come when wearing orange jerseys with white pants (.679) and blue jerseys with white pants (.670). But when wearing white jerseys with those white pants, things don't go so well (.388). We'd recommend that the all-white look be reserved for post-game Saturday night activities. There's only so much you can do with win-loss records, so to have a little more fun with this, we |
decided to pull in some more data. For all of the games in our data set, we gathered the
points scored and points allowed and ran something called analysis of variance (ANOVA)
tests to see how those totals varied depending on the uniform color.
The idea behind ANOVA is that it takes average values for different groups and then tries
to tell you if there are statistically significant differences among the means for each
group.
Starting with UVA's points scored, the basic numbers from the ANOVA test give us the
mean points scored for each uniform combination, along with some additional information,
like the standard deviation, a range for estimating the mean points scored for each color
combo, and the minimum and maximum for each. Here's what we find when we look at
the points scored for each of the color combinations UVA has worn since George Welsh
took over in 1982:
points scored and points allowed and ran something called analysis of variance (ANOVA)
tests to see how those totals varied depending on the uniform color.
The idea behind ANOVA is that it takes average values for different groups and then tries
to tell you if there are statistically significant differences among the means for each
group.
Starting with UVA's points scored, the basic numbers from the ANOVA test give us the
mean points scored for each uniform combination, along with some additional information,
like the standard deviation, a range for estimating the mean points scored for each color
combo, and the minimum and maximum for each. Here's what we find when we look at
the points scored for each of the color combinations UVA has worn since George Welsh
took over in 1982:
| UVA Points Scored |
| In the "mean" column, we see the average number of points scored by UVA for each of the uniform combinations. Setting aside the Orange Crushed game, we see that the lowest-scoring uniform option is the white jersey/blue pants combo at 22.55 points; the highest-scoring combination is white pants/blue jersey combo at 27.04 points. If you're thinking that doesn't seem to be that big of a spread, you're right. In addition to breaking down the means for each uniform combo, ANOVA also tells us if the differences we see in the means for each uniform combination are actually statistically significant and show a real relationship, or if they are within the range you'd expect with random variation. In this case, ANOVA says they're not statistically significant, but before we move on, there are a couple fun nuggets here:
wouldn't get that excited about it. But there it is.
there's less than one point of difference between the points scored for orange jerseys and white pants and vice versa. As mentioned above, the spread between the blue jersey/white pants combo and the white jersey/blue pants combo is more than 4 points. This makes some sense, when you think about it. With the exception of one "throwback" game, the games with orange jerseys or pants are all in the Welsh Era, from 1982 through 1993, after which we switched to the initial blue and white combination as the sabre logo was born. The numbers for the blue and white combinations encompass seven Welsh seasons and nine Groh seasons, so they certainly can't be contributed purely to Al Groh. But one oft-noted aspect of the Groh era was difficulty on the road, relative to at home, so seeing a bigger spread in the blue uniform days isn't a complete surprise. Let's jump over to the defensive side of the ball to see if we can find anything interesting. |
| UVA Points Allowed |
| Setting aside the all-orange game again, the values range from a mean of 15.88 points allowed while wearing all blue to 26.69 points allowed when wearing all white. Clearly, there's a wider spread here, and ANOVA tells us that in this case, the differences actually are statistically significant. A couple thoughts on those numbers:
ass nickname for the all-blue defense (Blue Thunder? The Blue Man Group? ) We only used all blue for eight games, the second smallest sample size in the data set, so we shouldn't take the result too seriously, but those defensive numbers do look pretty good.
about nutty Oregon-style uniforms, maybe you'll feel a little better knowing that the results can't be much worse than with the old-school Penn State look. |
| As we prepare to enter a new, multi-colored era, we must remember that the numbers don't lie. Hopefully the powers that be will still have the opportunity to take this analysis into consideration and make the right moves, like avoiding all-white and pairing those new colored jerseys with some stylin' new white pants. And finally, we should take a lesson from the best Major League Baseball teams, which combine advanced statistical analysis with old-school scouting. While one all-orange game doesn't give us enough data to make strong statistical conclusions, sometimes we have to trust our eyes. |
_______________________
but that took me 9 months to get around to. (Note: Since first inserting this footnote, I've had to update it from 7 months to 8 months, then from 8 months to 9. I have a bit of a procrastination problem.)
Mary, 9/9/95 at N.C. State, 10/2/93 v. Ohio, 10/30/93 at N.C. State, 10/3/92 at Wake, 9/16/89 at Ga. Tech, 10/15/88 at Louisville, 11/7/87 at Ga. Tech, 9/18/82 v. JMU, 11/6/82 at Ga. Tech, and 11/13/82 at UNC. If you have information on the uniforms worn for any of these games, please let us know.
becoming the multi-uniformed "Oregon of the East," future charts will be a hell of a lot more complicated. Plus, it’ll take 100 years to have a big enough sample size for each uniform combination to enable us to know if there are statistically significant differences between "Orange Jerseys with Those Metallic Mud-flap Traction Grips on the Shoulders with Blue Pinstriped Pants" and "White Togas with Orange Roman Numerals."
analysis of variance in the "real world." I bet you feel like an ass now.
design department.
remind you of this.
|
| 1 |
| 2 |
| 3 |
| 4 |
| 5 |
| 6 |
| 7 |
| 1 |
| 5 |
| 6 |
| 7 |
| 4 |
| 3 |
| 2 |
| So I somehow managed to convince one of my old college buddies, Brian, to take time out of his busy schedule to write the article below for my website. I enjoyed it immensely, depsite the fact that I'm way too dumb to understand all the math. |
| Brian is a 1992 graduate of UVa and a long-time Hoos fan. He lives in Richmond with his wife and two sons. Brian's likes include puppies, rainbows, and unicorns, and his dislikes include "mean people." |
|
|
|