I was playing with my daughter yesterday with the TV on in the background. I wasn’t really paying attention to the screen. Hungry for college football content, I had TiVoed a Texas Longhorns Season Preview on Fox Sports Southwest. They had interviews with coaches mixed with footage of the 2008 season, so I was kind of paying attention when I heard head coach Mack Brown say “stats are for losers.”
Yeah, LOSERS with their stupid stats. Losers are always whining about numbers as Sean Connery says in the edited-for-television version of The Rock: “Winners go home and date the prom queen.”
My first though was: “stats are for losers” in what context? So I skipped backward to hear more. During the Longhorns’ national championship season in 2005, Brown and his coaching staff told Vince Young–a major running threat–to not pass the line of scrimmage at certain times. They wanted him to use his arm, much like they want Colt McCoy to do now. But your quarterback is probably worried about his rushing stats, right?
“We do not care about stats. Stats are for losers. We care about winning the game.”
Ah, okay, so it’s not that he doesn’t care for stats, it’s that winning comes first. That’s fine. I can see where he is coming from.
I love looking at stats, and I have a good memory when it comes to numbers, so I spend a lot of time thinking quantitatively about life. For instance, I remember that I was a disappointing 12 for 23 in slow-pitch softball this spring, but had 8 hits in 10 at bats in a summer tournament. I played catch with my daughter yesterday and she used her left hand to throw about two-thirds of the time (she is stronger with the left but has more control with the right). I have spent 5 of 11 weekends this summer outside of Texas. Stats help give me a more objective view at what has happened in my life. I have made a baby with 100% of my spouses.
Stats comprise knowledge. They guide our reasoning, and we use them to make decisions, and I believe that to be true for anyone from the highest levels of professional sports to the most successful businesspeople to me trying to improve my free throw percentage in a church league. To take the phrase “stats are for losers” at its face means that only losers use that knowledge.
I don’t think Coach Brown meant it like that. I don’t make 3.8 million dollars a year (with an additional 1.2 million dollar incentive for still being coach on January 15, 2010), but I do respect Coach Brown greatly so I am going to assume he meant one of two things. Either stats are a way to comfort losers after a loss, or stats are for losers given certain conditions. I will expand on the latter explanation.
Stats are for losers if their statistical goals get in the way of winning. The classic example of the conflict this can cause comes from incentivized salaries in team sports. Suppose Derek Jeter gets a million dollar bonus for ending the season with a .300 batting average or better (3 hits for every 10 at bats). Going into the last game of the season, he has a .300 batting average, having 180 hits in 600 at bats. The opposing pitcher for the game is Scott Kazmir, who Jeter is 4 for 33 (.121) against lifetime. Should Jeter play? If his statistical goal (and the bonus) is most important to him, then most people would say he shouldn’t play because he is likely to dip below .300. Of course, Jeter would never do this for a multitude of reasons: he would incur the fans’ and media’s wrath for one, and Jeter seems like the kind of guy who is confident enough to believe he can get 1 hit in 3 at bats on the last day of the season. But might he be more inclined to look for a walk or sacrifice bunt, two outcomes that do not affect batting average?
Stats are for losers if context and variables are not taken into account. Let’s say you move to a new city for a job. Your first few days at work, it has taken you an average of 45 to 60 minutes to get to your desk. So how long will it take you today? Before I even finish asking that question, you have probably processed three or four variables: day of the week, time of day, route, and parking (and, if you work at UT, how willing you are to risk a parking ticket). So let’s look at those variables. It’s Saturday, it’s early afternoon, you’ll take the same route, and parking should be easier because less people work on the weekends. But your job is in San Francisco and the Giants are playing an afternoon game. Those variables matter, because now it will take you 75 to 90 minutes. If you just looked at previous stats without thinking about the variables of this event, you would be banging your head against your steering wheel around the time you hit King Street.
This is my biggest frustration with using the points-per-game statistic, both on the individual and team level. Let’s say a team leads their league in scoring per game. That’s great, but you have to ask two contextual questions. First, who have they played? The opposing team is obviously a huge variable. UC Davis football almost led the country in scoring in 2000 but they were playing Division II teams almost exclusively; did they have a better offense than every Division I team? Second, what is their defense like? The Golden State Warriors were 2nd in the NBA in scoring this past season but it’s partially because of a subpar defense that doesn’t slow the other team down, therefore giving their own offense more chances and time to score. Points-per-game is a stat without context, which is why I prefer points-per-possession to evaluate teams.
Stats are for losers if they eliminate experience from the decision-making process.* The interesting thing about people who claim to hate stats is that they use them all the time, except they call it “experience” or “gut instinct.” When a football coach decides on a running play on 3rd and 5 instead of a pass play, he is doing it for one of a multitude of reasons, but the reason will be based on the X times out of Y that he has seen it work in the past. Coach Brown didn’t want Vince Young to run for first downs during particular games because…well, I don’t know the reason, but it was probably something like: “Most of the time when we tell Vince it’s okay to run, he doesn’t look downfield” or “When Vince runs for first downs, he usually tires out too early” or “We are trying to teach Vince to be a better passer” because Coach Brown and his staff know that championship teams almost always have good passers. “Most of the time”, “usually”, and “almost always” are statistics. They are not precise, but they are statistics in a general form.
And this is why experience matters: we just don’t know the numbers. I don’t know what Heather wants for dinner tonight, but if I were to have kept stats during the course of our marriage, I would find that she almost always wants to eat out on Saturdays, and she wants Mexican, Thai, or Vietnamese–stats that I technically could have kept but didn’t. But who knows how she is feeling today? Maybe when she is tired she tends towards one cuisine or the other. There are an infinite number of variables, some of which we just don’t have the tools to measure yet, so we use our own past to fill the gaps. Experience is a substitute for unknown–or unknowable–data.
***
I’m already interested to see how the Longhorns season plays out: I’m a big fan, this should be a great season, and I have students on the team. But now I have another reason: Erin Andrews’ halftime interviews with Coach Brown. “Coach Brown, your #1-ranked offense had two turnovers in the first half. How does that stat affect you?” “Erin, you know that as a coach I hate turnovers but I also think stats are for losers. So I think what you mean is: ‘Our real good guys had a couple’a whoopsies in the early part goin’ on.'”
See? He’s still using stats, and I still love Coach Brown.
*Let me say that I don’t think experience is not more important than statistical data. The biggest problem with using our own past experiences to understand future events is that we tend to rely only on our own experience. I’ll give a personal example. I hurt my ankle a few weeks ago in California and wasn’t planning on going to the doctor. See, I hurt my ankle once before doing the exact same thing I was doing this time, and two days later it felt fine, so why go to the doctor now? The problem was that I was relying on my own experience, which is a tiny sample size (n=1), and my own orientation towards getting professional medical help, which I would describe as “reluctant until bleeding”.
I neglected two things. First, if I looked at the larger sample size of everyone (and not just experience with my own body), I would have realized that most people who cannot walk on their ankle because of the pain need to seek medical attention. I put my own experience ahead of worldwide medical statistics. Second, I used an experience from when I was younger, neglecting this annoying trend of aging one year for every year that I live. I was 23 when I hurt myself the first time, but now I’m 26. Maybe that’s not a big difference, but I’m at the age now where my friends are starting to wear knee braces when we play basketball. The trend of aging probably played into the damage to my ankle this time. This is a very simple example that anyone can follow without a spreadsheet, but it’s hard to pull out more complex trends without stats.