Crazy fans...

[bballref3966]

Who has time to do this research?

Your officials are Karl Hess, Roger Ayers, Michael Roberts | Kentucky Sports Radio

[crosscountry55]

Students on winter break.

[youngump]

Originally Posted by bballref3966

Who has time to do this research?

Your officials are Karl Hess, Roger Ayers, Michael Roberts | Kentucky Sports Radio

Since I haven't made my way over here from the small diamond board for a while to make trouble. Let me make a couple of comments. I suspect there's something of a Rorschach test in how one sees officials impacting a game in the way you look at a story like this. For example, if your initial reaction to this is to believe that Ayers must be out to get Kentucky then that says something about how you see the game. Similarly if you see it as ridiculous that fans would care about an official potentially affecting them negatively, that probably says something too.

Anyway, with that in mind, let me see if I can't throw some fuel on the fans fire. I'm a programmer by day and it's the work of just a few minutes to get my computer to simulate the chances that an official randomly assigned to John Calipari Kentucky games 7 times would only see Kentucky win twice.

In 10 million simulations, it happened about .3% of the time. In about any scientific study, they'd use that to reject the hypothesis that he had no affect on whether the team would win. In fact they'd tend to do that even if it happened 4-5% of the time, so this is pretty conclusive by the way science is done.

So what's going on here? I suspect that mostly it comes down to the fact that officials aren't randomly assigned. Assigners tend to send better officials to games that are bigger or more likely to need better officials. Ayers has seen Kentucky lose to Duke, @Georgia, @Arkansas, and @North Carolina twice. The wins are @Tenessee and Ohio State. 2-5 against that schedule isn't exactly a statistical anomaly of any great perspective.

Which isn't to say anything whatsoever about how Ayers officiates, he could well be doing something that makes it less likely that Kentucky will win but the statistical argument is much weaker than it seems when you know how games are assigned.

Returning to my lurker mode.

[Raymond]

Good for Roger, he take care of business.

[SC Official]

I don't know about you guys, but I'm much more concerned about what stupid comments Jay Bilas will have during the game than whether or not Ayers whacks Calipari.

[Mark T. DeNucci, Sr.]

Originally Posted by SC Official

I don't know about you guys, but I'm much more concerned about what stupid comments Jay Bilas will have during the game than whether or not Ayers whacks Calipari.

Why would you be concerned about what that yahoo thinks or says?

MTD, Sr.

Was the officiating crew announced or something?

[SC Official]

Originally Posted by Mark T. DeNucci, Sr.

Why would you be concerned about what that yahoo thinks or says?

MTD, Sr.

Sarcasm.

[Adam]

Originally Posted by youngump

Since I haven't made my way over here from the small diamond board for a while to make trouble. Let me make a couple of comments. I suspect there's something of a Rorschach test in how one sees officials impacting a game in the way you look at a story like this. For example, if your initial reaction to this is to believe that Ayers must be out to get Kentucky then that says something about how you see the game. Similarly if you see it as ridiculous that fans would care about an official potentially affecting them negatively, that probably says something too.

Anyway, with that in mind, let me see if I can't throw some fuel on the fans fire. I'm a programmer by day and it's the work of just a few minutes to get my computer to simulate the chances that an official randomly assigned to John Calipari Kentucky games 7 times would only see Kentucky win twice.

In 10 million simulations, it happened about .3% of the time. In about any scientific study, they'd use that to reject the hypothesis that he had no affect on whether the team would win. In fact they'd tend to do that even if it happened 4-5% of the time, so this is pretty conclusive by the way science is done.

So what's going on here? I suspect that mostly it comes down to the fact that officials aren't randomly assigned. Assigners tend to send better officials to games that are bigger or more likely to need better officials. Ayers has seen Kentucky lose to Duke, @Georgia, @Arkansas, and @North Carolina twice. The wins are @Tenessee and Ohio State. 2-5 against that schedule isn't exactly a statistical anomaly of any great perspective.

Which isn't to say anything whatsoever about how Ayers officiates, he could well be doing something that makes it less likely that Kentucky will win but the statistical argument is much weaker than it seems when you know how games are assigned.

Returning to my lurker mode.

Fascinating analysis, but I'll have to check the number out of my own nerdish desire. It seems on the surface that, even without taking into account the way games are assigned, the margin for error on a sample size of 7 would be rather large.
3.5 games would be the expected number of wins.
2 isn't that far off.

edited to add: I recognize the expected number of wins is more than 3.5 given UK's standard winning percentage. My math is way off at this point.

Originally Posted by youngump

Since I haven't made my way over here from the small diamond board for a while to make trouble. Let me make a couple of comments. I suspect there's something of a Rorschach test in how one sees officials impacting a game in the way you look at a story like this. For example, if your initial reaction to this is to believe that Ayers must be out to get Kentucky then that says something about how you see the game. Similarly if you see it as ridiculous that fans would care about an official potentially affecting them negatively, that probably says something too.

Anyway, with that in mind, let me see if I can't throw some fuel on the fans fire. I'm a programmer by day and it's the work of just a few minutes to get my computer to simulate the chances that an official randomly assigned to John Calipari Kentucky games 7 times would only see Kentucky win twice.

In 10 million simulations, it happened about .3% of the time. In about any scientific study, they'd use that to reject the hypothesis that he had no affect on whether the team would win. In fact they'd tend to do that even if it happened 4-5% of the time, so this is pretty conclusive by the way science is done.

So what's going on here? I suspect that mostly it comes down to the fact that officials aren't randomly assigned. Assigners tend to send better officials to games that are bigger or more likely to need better officials. Ayers has seen Kentucky lose to Duke, @Georgia, @Arkansas, and @North Carolina twice. The wins are @Tenessee and Ohio State. 2-5 against that schedule isn't exactly a statistical anomaly of any great perspective.

Which isn't to say anything whatsoever about how Ayers officiates, he could well be doing something that makes it less likely that Kentucky will win but the statistical argument is much weaker than it seems when you know how games are assigned.

Returning to my lurker mode.

Originally Posted by Adam

Fascinating analysis, but I'll have to check the number out of my own nerdish desire. It seems on the surface that, even without taking into account the way games are assigned, the margin for error on a sample size of 7 would be rather large.
3.5 games would be the expected number of wins.
2 isn't that far off.

edited to add: I recognize the expected number of wins is more than 3.5 given UK's standard winning percentage. My math is way off at this point.

Excuse me, but wouldn't an occurrence in only .3% of simulations indicate an outside factor influencing the results? I know 7 games is a small sample size, but I demand sample size-adjusted P and T values!

[Adam]

Originally Posted by AremRed

Excuse me, but wouldn't an occurrence in only .3% of simulations indicate an outside factor influencing the results? I know 7 games is a small sample size, but I demand sample size-adjusted P and T values!

Based on z values, there is a 99.9% confidence that Ayers hates Cal (don't we all?) z=-2.9517 p=.001
There is a 99% confidence that he hates UK (again, don't we all?)
z= -2.2935 p=.01

I don't think t values would change this much.