Statistical ball-spotting tendencies

[Morningrise]

Here's an interesting article that uses NFL statistics to show that balls are spotted on five-yard lines significantly more often than chance would predict. The author's hunch is that, when in question, officials gravitate toward those easily identifiable marks on the field. This effect diminishes in the red zone, when inches and feet matter more.

https://gutterstats.wordpress.com/2015/11/03/are-nfl-officials-biased-with-their-ball-placement/

[mishatx]

So, statistics show officials are following what is a commonly accepted philosophy?

[ncwingman]

This reminds me of a tale I heard from former graduate students commiserating about their graduate work. One student had decided to study the psychology of how songs get played on the radio and spent hour after hour documenting which songs got played, how often, how close to other songs by the same/similar artists, fast song vs. slow song... huge statistical analysis.

After compiling all the data, he discovered that 99% of music radio stations use a piece of software to select/play songs. This software is designed, coded and intentionally implements each one of the trends he was noticing in the playlists.

He could have saved a lot of time and effort if he had simply asked how things were done.

[bkdow]

oh my....that was a wasted amount of statistical analysis time.

[Legacy Zebra]

Which commonly accepted philosophy says to put it on a 5 yard line? I'm well aware of and am a huge proponent of putting it on a "tick mark", but who is teaching to put it on a "big line"? Not saying that there's some giant ball spotting conspiracy obviously, but it is an interesting point to think about.

[NorCalMike]

None of this is conscious. If you read the article this is a well know statistical abnormally. So what happens when the ball stops between the 29 and 31, the official is more likely to place the ball on the 30 than the other yard lines. It is all subconscious. You don't even know you do it.

[ncwingman]

Despite the fact that I can't see his raw data (not wanting to pay $49 for it), I'm actually thinking that his raw data suffers from statistical heaping as well. Chances are, the ball is reported as being on a discrete yard line in all cases. If you look at the play-by-play data that gets posted to nfl.com or espn, plays are always recorded as starting and end on integer yard lines, even if the ball is halfway in between. You never see a play such as "1st and 10 at the 20 -- E. Smith up the middle for 2.5 yards. 2nd and 7.5 at the 22.5 -- E. Smith up the middle for 3.25 yards. 3rd and 4.25 at the 25.75"

A lot of drives start at the 20 due to touchbacks. Players see the sticks and get the ball just beyond for a first down. Statistically, unless they made it all the way to the 31, it's a 10 yard gain and the ball is at the 30 -- even if the ball is actually spotted at the 30.5.

Maybe there is an unconscious bias to the whole thing, but that bias could be amplified by the manner in which the stats were collected in the first place.

[Morningrise]

Maybe there is an unconscious bias to the whole thing, but that bias could be amplified by the manner in which the stats were collected in the first place.

Very good point - the rounding off could be due to the official scorer putting the ball on a five-yard line, instead of the H and L.

I would also be interested in seeing whether this effect occurs on all types of plays. On a long gain, when all the officials are rather far away at the end of the play, and neither team cares whether the play gains 23.5 yards versus 24.5, I would not be terribly surprised to learn that officials just spot the ball at the 45 if it's a choice between that and the 46. I would be much more surprised if this effect persisted on short gains as well - when the line to gain matters, and when both wings plus the umpire have a good view right near the runner when he gets stopped.