The Dome Effect: Is It Real? Has A Journalist Ever Taken A Stats Class? What Is Reality Anyway?

mah depth perception noooo

One of the reasons I'm not too happy to get Kansas in the Sweet 16 is where the games will be played—Jerryworld—combined with Kansas's #1 strength—Jeff Withey going grrr aargh and depositing your shot in the eighth row. Domes have a reputation for being poor environments to shoot in. Meanwhile, the alternative to shooting is challenging this guy.

But just because something is supposed to be true does not mean it is so. Reasons are applied to random chance all the time. Does being in a dome really kill shooters? I wish I had an answer for you without painstakingly combing through every dome game in the NCAA tournament since forever. The data is thin, contradictory and oft-polluted by what can only be termed a journalist's approach to statistics. After googling every which way for any take on the subject, I've come out the other end possibly less informed than I started.

The best I've got: the WSJ published an article in 2011 that appears to be the most comprehensive tackling of the issue. It showed that from 1997-2011, Final Four teams hit 32% of their threes and 42% overall, both four-point drops from—ugh—the four FFs held before the dome came into vogue. Or we could take thousands of games of data instead of 12, WSJ. And maybe account for the fact that the three-point line moved back in 2008-09. Guh.

/shakes fist at journalism school

For what it's worth, Statsheet shows that three-point shooting held steady at around 34.8% from 2003 to 2008 and around 34.2% after the line moved back. I'd imagine tourney teams are on average slightly better than that, so 32% over 15 years represents a small but probably real negative effect that may or may not be caused by domes instead of various other factors that apply to Final Fours like your hand shaking nervously for an hour before the game. "Other factors" didn't impact 12 games almost 20 years ago, FWIW.

KSRCollege put together a chart covering the "open dome"—ie, court on the 50, not in one endzone—era the NCAA instituted in 2009. (It appears Jerryworld is configured with the court on the 50.) It found three point shooting averages dropped from 36% (for the season) to 32%, free throws from 73% to 67%, and eFG from 51% to 44%.

Caveats are rife. For one, the KSR post has a nine-game sample and the WSJ article is rage-inducingly sloppy. For two, some of these effects may be due to the level of competition. For three:

At first glance if you compare the season percentages to the in-game percentages you’d think that all teams are shooting poorer than their season percentages, but this is not the case.  There are three severe outliers taking down the entire sample; 2011 UConn (twice) and 2009 Villanova.  But, when you see their season percentages you’ll see Villanova was an average three point shooting team and UConn was a terrible three point shooting team, so it’s not hard to believe these two teams would have bad shooting nights.  All other Final Four participants in the “Open Dome” era have shot right around their season percentage, so this leads me to believe that distorted sightlines have less to do with the low point totals than I originally thought.

While I'm not sure I agree with the idea that a poor shooting team will be more affected by the depth issues presented, at least this passage underscores the scanty amount of data we're working with.

Other poorly-assembled nine-game samples show no dome effect.

In 2009 and 2010, the NCAA used all three different setups to contest the eight regionals: regular basketball/hockey arenas; traditional domes, with the court set up in the corner of the football field, and the stadium configuration, with the court built on a platform at the center of the football field.

Here is how the shooting stats broke down in those games:

— Arenas (nine games): 42.8 percent, 455-of-1,064.

— Traditional domes (nine games): 43.1 percent, 444-of-1,030

— Stadium setup (six games): 42.4 percent, 290-of-684.

Essentially, there was no distinct statistical variation among the various types of courts.

This study is also tiny and doesn't even bother to separate out threes and free throws, instead hurling everything in one statpile ranging from dunks to prayer heaves. So it's far from definitive itself. Despite that, Mike DeCourcy appears to run it every year without bothering to update it and reference it whenever the topic comes up. No wonder he gets in fights with Kenpom.

Last year, teams didn't seem to have much problem. OSU, Louisville, and Kansas hit exactly 36% of their threes; Kentucky was at 38%. This is probably why there was a flurry of articles about shooting in a dome before that Final Four, but not after.

So. We have a pile of shifty data. Overall I'd suggest it suggests there is a small dome effect that hurts shooting based on the WSJ numbers, which are the closest thing to a real sample we've got. This is advantage Kansas, which takes relatively few threes and forces a lot thanks to Jeff Withey. Probably, anyway. The effect isn't big enough or solid enough to be fate.

Ugh. This post. Just like a young man who thinks Standard Deviation is a Christian goth metal band coming in for a low-sample size study. I am zo unzatisfyed.