Skip to content

Kentucky Sports Radio

University of Kentucky Basketball, Football, and Recruiting news brought to you in the most ridiculous manner possible.

Are We Being Fed Bad Stats?

When fans of opposing teams get into arguments regarding which team is superior, often the fight moves to stats to prove without a reasonable doubt that Team X is clearly better than Team Y (That is if the argument ever moves away from name calling or probation smack).  When fantasy sports junkies look to find that up-and-coming Running Back that no one else is considering to boost their team, they turn to stats.  When Owners and GM’s of poor professional teams look to find a competitive advantage against their rich counterparts, they look for stats that others don’t know the true value of.  Statistics are all around us as sports fans and are readily available for consumption, but are the numbers we receive from the countless sources reliable or even meaningful?  Are numbers like points per game, turnovers per game, or yards per game truly worthwhile in determining how good a team is, or do they merely tell us how a team plays?  To get this discussion started I went back and investigated the most important factors in winning college football games from my most recent articles and compared those to stats like passing yards per game to see which ones were more meaningful in determining winners.

 

The below chart contains data from the 2007 season onward and shows the correlations to winning percentage of various efficiency stats like Passing Yards per Attempt compared to the traditionally shown stats like Passing Yards per Game.  Remember here that the closer the number is to 1 or -1 the higher the correlation.

The first striking thing is how different the correlations are from one another, for instance an efficient passing game is one of the most relevant factors in advanced stats but its basic sibling, Pass Yards per Game, is largely irrelevant when it comes to winning games.  Another place where the relative importance is largely different is pass defense, advanced stats say stopping opponents from passing is critically important while its simple cousin says stopping the pass is highly irrelevant.  One category stayed similar to its advanced counterpart, Rush Yards per Game, so should this lead us to believe that efficient passing on both sides of the ball is irrelevant or critical to success?

 

To understand this predicament further I’ll let my good buddy Mark Twain have a word on the subject matter.  “There are three kinds of lies: lies, damned lies and statistics.”  Thanks Mark!  My interpretation of his underlying theme (perhaps oversimplified) in that statement is be careful of the numbers you believe as they can be easily manipulated.  So, should we believe the advanced way of thinking or the long held belief of traditional stats like Yards per Game?  We should accept the advanced way as it takes numerous factors into account and puts them in one simple stat.  Take the difference in passing stats for instance, Pass Yards per Attempt takes into consideration multiple things like incompletions, yards gained, and pass attempts whereas raw pass yards per game just takes into account what its name implies.  There’s such a low correlation with winning in that stat because college football, as I proved yesterday, is a predominant running game, so naturally pass yards per game wont consider that.  If a team gives up a low amount of total yards through the air it doesn’t necessarily mean they have a good pass defense, it could mean that their opponents rarely pass.  The high correlation to Rush Yards per Game is significant to me for the previously stated reason that college football, as a whole, is a run first league.  Since teams run at a high rate, naturally teams who stop their opponent should be more successful.  So when interpreting a stat like Pass Yards per Game, just know that you’re not getting very good information because it takes little else into consideration.

 

So, if stats like Yards per Attempt are readily available why is it that we don’t often see them on a TV broadcast?  I think the simple answers are that the audience doesn’t demand it and it’s convenient for the respective network to put in a simple number.  A wide portion of the TV audience has little interest in Yards per Attempt because they want something quick and easy to satisfy their knowledge.  Yard per Game stats give convenience but they don’t give good information.  Yard per Attempt stats and other similar advanced numbers tell a much more compelling story and are simple to understand when an explanation is given.  While not always the most convenient, it would be beneficial for the sports public to use more meaningful numbers in their evaluations so a more concrete truth can be reached.

Article written by Jonathan Schuette

16 Comments for Are We Being Fed Bad Stats?



  1. Homer Simpson again
    8:08 pm July 31, 2012 Permalink

    uuhhhhh… DONUTS…



  2. Smyrna_Cat
    8:16 pm July 31, 2012 Permalink

    maybe not everyone is so obsessed with stats.



  3. Owen
    8:17 pm July 31, 2012 Permalink

    It’s not about any individual stat, it’s about them in conjunction with one another in order to compare players and teams. Obviously you want the guy who averages say 10ypa compared to the guy who averages 5ypa if they have identical completion ratings.

    Obviously defense ratings are harder to quantify and qualify but I’m sure there are some metrics that give more valuable information than simply “yards/points allowed.” Heck, a “yards per attempt allowed” would be a pretty meaningful statistic.



  4. TampaCat
    8:22 pm July 31, 2012 Permalink

    just win baby! just win!



  5. UK Homer
    8:26 pm July 31, 2012 Permalink

    If you’re leading, then you run the ball. If you’re losing, then you pass the ball. Makes sense why running beats passing when trying to correlate to winning. Schuette, I’d love to see 1st half/2nd half breakdown. I’m guessing 2nd half stats drive this, but wonder if the style a team chooses (i.e.: what they run in the 1st half when the game is closer) is as distinctive.



  6. Bill
    8:27 pm July 31, 2012 Permalink

    Good job, good effort



  7. burger
    8:28 pm July 31, 2012 Permalink

    I think the score does a good enough job.

    Oh… and the cheerleaders. You can see all of theses stats and more in how bouncy the girls are.



  8. ShelbyCatFan
    8:40 pm July 31, 2012 Permalink

    How many time did we run a HB Draw on third and 5 yds+ last season? That’s a stat I would wanna see! How bout a quick slant at least for Christ sakes. Take the leash off ur offense Joker!



  9. WB
    8:43 pm July 31, 2012 Permalink

    I didn’t understand a word I just read. @7 I think that that’s why Oregon has improved so much lately.



  10. Grover Johnson
    8:51 pm July 31, 2012 Permalink

    “Are numbers like points per game, turnovers per game, or yards per game truly worthwhile in determining how good a team is, or do they merely tell us how a team plays?”

    Serious? A team that plays great is a great team. A team that plays poorly is a poor team. Great teams don’t play poorly and poor teams don’t play great.



  11. Jonathan Schuette
    8:52 pm July 31, 2012 Permalink

    #5 I’ve been wondering that myself while researching, so that’s something that I’ll be monitoring this season. I’d love to look at last season’s data but I just don’t have the time to look at game by game/play by play/ situational data of all teams at this time with other projects I have going. But, I will research that this year when watching the games.



  12. Jonathan Schuette
    9:02 pm July 31, 2012 Permalink

    #10 You’ve missed the point of the entire article.

    Here’s a question, which team is less interception prone? Team A who threw 16 interceptions on 554 attempts or a Team B who threw 8 interceptions on 167 passes? Answer: Team A. They threw an interception on 2.8% of passes whereas Team B threw a pick on 4.8% of passes. Raw data would say Team B was better in the regard, but that’s obviously not the case. The main point of this is to use better data in evaluations.



  13. LMAO
    9:36 pm July 31, 2012 Permalink

    Like Billy Bob UK fan understands any of this.



  14. WB
    9:50 pm July 31, 2012 Permalink

    #13 The only thing you understand is what setting to put your trimmer on when you want that perfect line beard.



  15. Go Cats
    8:53 am August 1, 2012 Permalink

    As a professional statistician, I can say with confidence that my field is abused and misused then any other out, except perhaps sports talk radio which anybody can do. Couple of notes:
    – Don’t look at correlations as they only show the linear relationship between two variables. So, the correlation between weather and month is zero because there is a sinusoidal, not linear, relatinship.
    – Stats can only show a mathematical relationship between two variables. It does not, as you imply, show a casual relationship. For example, ice cream consumption and murder rates are very related. There are often lurking variables (in the ice cream case, the lurking variable is weather).
    – Stats are always based on series of assumptions that often never gets properly tested (independence, constant variance, normality, and others). Unless you know what a durbin-watson test, Kolmogorov-Smirnov, Anderson-Darling, and Shapiro-Wilk tests are, then step away from the stats software.



  16. yikes
    12:30 pm August 1, 2012 Permalink

    #15 Ugh.. you beat me to it ! that is EXACTLY what i was going to say