Useless stats (as seen on TV)

I loved watching the NCAA lax telecasts.* But hearing the stats tossed around made me wonder how meaningful they really are. Such as:

* Syracuse is 12-0 this year when scoring first [or whatever the stat was].
* [X team] has won all but one game when leading entering the final period.

Does that really predict the outcome? If you want a useful stat, don't you have to drill down further? I would think the first-goal-scored stat would be statistically meaningful only when the margin is 1 goal. A team like Syracuse wins the majority of its games no matter who scores the first goal. We rolled up a 10-2 lead on Kentucky IIRC that didn't turn out to be meaningful.

Some of these stats seem to be arrived at looking backwards. You look at first-goal-scored, leading-entering-final-period, leading-scorer-has-at-least-three-goals data, and see which seem to have the most significant effect apparent link, then use that. If a team runs off three goals in less than three minutes, is it a predictor of victory because of the quick run, or because the team added three goals? Lacrosse might be ripe for this kind of statistical drilldown that is either overkill or fascinating, and I'm sure if one of us took the tomato side, someone else would then choose to argue the tomahto side.

* I'd add "loved the games especially since they were shown in HD," understanding that one invokes the wrath of ELFers unable to afford the astronomical prices of an HD-capable TV.

billhoward
I loved watching the NCAA lax telecasts.* But hearing the stats tossed around made me wonder how meaningful they really are. Such as:

* Syracuse is 12-0 this year when scoring first [or whatever the stat was].
* [X team] has won all but one game when leading entering the final period.

Does that really predict the outcome? If you want a useful stat, don't you have to drill down further? I would think the first-goal-scored stat would be statistically meaningful only when the margin is 1 goal. A team like Syracuse wins the majority of its games no matter who scores the first goal. We rolled up a 10-2 lead on Kentucky IIRC that didn't turn out to be meaningful.

Some of these stats seem to be arrived at looking backwards. You look at first-goal-scored, leading-entering-final-period, leading-scorer-has-at-least-three-goals data, and see which seem to have the most significant effect apparent link, then use that. If a team runs off three goals in less than three minutes, is it a predictor of victory because of the quick run, or because the team added three goals? Lacrosse might be ripe for this kind of statistical drilldown that is either overkill or fascinating, and I'm sure if one of us took the tomato side, someone else would then choose to argue the tomahto side.

* I'd add "loved the games especially since they were shown in HD," understanding that one invokes the wrath of ELFers unable to afford the astronomical prices of an HD-capable TV.

The key with those kind of stats is that you have to compare the results with condition X to the stats without condition X (or at least to the overall total). If they're 12-0 when scoring the first goal and 0-12 when they don't, well, that would give you some pause. However, if they're 12-0 when scoring first and 12-0 when they don't, it's clearly meaningless. It's also pretty useless with this year's actual data: they're 12-0 when scoring first and 1-1 when not scoring first. The 1-1 just is not statistically significant - there's no way those two data points establish a trend.

For it to be meaningful, there have to be a significant number of trials with X and without X AND there has to be a significant difference between the results. So if Syracuse were 5-2 when scoring first but 2-5 when not, then you could start to wonder.

Good point. In a college sports season, you don't get enough games to have meaningful results if you're looking at a narrow set of conditions: who scores first vs. the outcome of a one-goal game. I just think the announcers dread having dead airtime with nothing but the sounds on the field to entertain fans. I suppose if some statistical myths get debunked, Chico Resch will just find something else to say.

Drift: FWIW, my sense of 1-point NCAA tournament basketball games has been that all those 2-vs.-15 and 3-vs.-14 squeakers wind up in favor of the higher seed. That to me seemed beyond statistical odds. But that's just from parsing all the close games of the last decade or two. Only three NCAA 1-vs.-16 games have been within one point at the end of regulation: two wins for the higher seed plus an OT win (by 4) but that's still just three games. Only once has a 15-seed beaten a 2-seed by 1 point (plus 3 other games by 3, 3 and 4 points) says [en.wikipedia.org], but I don't know how many 1 point wins have gone to the 2-seeds.
Drift: We know teams wearing red win > 50% of the time (based on random assignment of jersey colors in the Olympics).
Drift: The myth of the hot hand in basketball has been debunked. A player who just hit 2 or 3 shots in a row is no more likely to hit the fourth shot than his recent shooting average.

I wonder what are the ten most overused, least valid statistical cliches. Or generalizations, such as every time a first year football player fumbles or throws an interception, you know the words "rookie mistake" are odds-on to follow.

All this is a sign we really should sign off eLynah until just before the Red White game ... and we don't want to.

I don't think "scoring first" is meant to be a meaningful stat. It is meant to be a "hey, how about that" stat. Syracuse's record going into the Army game was 13-1, so the worst correlation that any stat could have would have been "Syracuse is 0-1 when..." - and most of those would be bullshit too.

Most of those statistics are the equivalent of "TEAM is undefeated when scoring first in overtime."

billhoward
I just think the announcers dread having dead airtime with nothing but the sounds on the field to entertain fans.

I could only wish for some dead air. Especially in the lacrosse broadcasts the announcers keep up an astonishing amount of drivel/chatter about peripherally related subjects while the ball is in play. Drives me nuts! What are the lines? Who has the ball? Who's D'ing him up? Zone or man defense? C'mon man! Then the constant repetition of the Yardley Love story during the Virginia rout of Mt St. Mary's without so much as a variation in the words used drove me to another channel where the Club championship between Arizona State and Michigan was a better game anyway.

billhoward
I wonder what are the ... least valid statistical cliches.

Actually the most valid comment by Jack Emmer this weekend, during "Sudden Victory" overtime: "The team that has the ball last will win."

TimV

billhoward
I just think the announcers dread having dead airtime with nothing but the sounds on the field to entertain fans.

I could only wish for some dead air. Especially in the lacrosse broadcasts the announcers keep up an astonishing amount of drivel/chatter about peripherally related subjects while the ball is in play. Drives me nuts! What are the lines? Who has the ball? Who's D'ing him up? Zone or man defense? C'mon man!

Yeah, at one point during the Ivy final I went into the kitchen and didn't realize the game had started again because they were still going on about the UVa thing...

They switch between Lax for Dummies ("when the ball goes out of bounds on a shot, the team closest to the ball ..." and high-level discussions (sliding the wing ... working off an invert, yada yada) that lose the majority of viewers. Mixed between more outpourings of grief for the Virgnia situation best assuaged, they'll tell you, by keepign it tuned right to ESPN.

TimV

billhoward
I just think the announcers dread having dead airtime with nothing but the sounds on the field to entertain fans.

I could only wish for some dead air. Especially in the lacrosse broadcasts the announcers keep up an astonishing amount of drivel/chatter about peripherally related subjects while the ball is in play. Drives me nuts! What are the lines? Who has the ball? Who's D'ing him up? Zone or man defense? C'mon man! Then the constant repetition of the Yardley Love story during the Virginia rout of Mt St. Mary's without so much as a variation in the words used drove me to another channel where the Club championship between Arizona State and Michigan was a better game anyway.

billhoward
I wonder what are the ... least valid statistical cliches.

Actually the most valid comment by Jack Emmer this weekend, during "Sudden Victory" overtime: "The team that has the ball last will win."

Well, he could be wrong. You might have an OWN Goal.

TimV

billhoward
I just think the announcers dread having dead airtime with nothing but the sounds on the field to entertain fans.

I could only wish for some dead air. Especially in the lacrosse broadcasts the announcers keep up an astonishing amount of drivel/chatter about peripherally related subjects while the ball is in play. Drives me nuts! What are the lines? Who has the ball? Who's D'ing him up? Zone or man defense? C'mon man! Then the constant repetition of the Yardley Love story during the Virginia rout of Mt St. Mary's without so much as a variation in the words used drove me to another channel where the Club championship between Arizona State and Michigan was a better game anyway.

billhoward
I wonder what are the ... least valid statistical cliches.

Actually the most valid comment by Jack Emmer this weekend, during "Sudden Victory" overtime: "The team that has the ball last will win."

Unfortunately, Tambroni said the same thing after the OT win over Loyola on Saturday.

Robb
The key with those kind of stats is that you have to compare the results with condition X to the stats without condition X (or at least to the overall total).

Also, the stat means nothing unless you compare it to how other teams do with the same conditions. Classic example of this a fgew years ago was Joe "King of Misunderstanding Stats" Morgan crowing about the Mets playing .900+ ball when leading entering the ninth. The problem being an average MLB team goes about .950 when leading entering the ninth.