Right, sorry! You got burned by the reframe. Read more
Reuben Fischer-Baum
reubenfb
There are more detailed predictions if you click on the games in Bloomberg's model, and it actually agrees with you (Portugal is heavily favored over the U.S., and the U.S. is slightly favored over Ghana). However, the single most likely outcome for both games is, by their model, a tie:
Ok maybe that one. Read more
Because that group is also tough! It's been between these two groups the whole time. For what's it worth, the models above are weird about Portugal: SBNation and The Economist—probably the least-seasoned models in the set—have them in the top four, but they're not nearly as close anywhere else. Read more
This is more one of those weird problems you run into with these sorts of models. For any given game, it's more likely that a team will score 0-2 goals than 3+ goals. Obviously, across all 64 World Cup games, it's extremely unlikely that a team will never score 3 goals, but which games would you change? Read more
Ringo Starr is funny too, trapped in his "clearly the worst Beatle" triangle. Read more
Exactly, and I think Delaware has a similar "come and buy tax-free booze" deal to New Hampshire. North Dakota is sort of in gold rush mode right now, so I think that alcohol is actually being drunk by (recent) residents. Read more
Another behind-the-scenes tidbit: Jason Whitlock once stole a rib right off my plate at a BBQ restaurant. (This is a real thing that actually happened.) Read more
Counterpoint: Those sheep are pretty herded. Read more
This is a good critique, and I certainly didn't mean to take a shot at your initial response. (The vast majority of our statistical nitpicking comes from people who don't seem to have much of a statistics background.) I've circulated to some other statisticians and their take seems pretty aligned with yours, so I… Read more
This is fair. Like I said, using ANOVA doesn't remove the multiple comparisons problem, but it is preferable to running multiple t-tests off the bat. (Unless I'm way off, my understanding is that running an ANOVA on the data presented in the comic probably wouldn't yield a F statistic with a significant p-value.) Read more
I didn't say that they didn't know, I said that they had to figure them out for themselves because they weren't listed by the NBA. Read more
Not that this completely removes the multiple comparisons problem, but the researchers ran an ANOVA test on all the potential response variables together. That means that actually could reject the null hypothesis that none of the variables varied with ref height (with 95 percent confidence) before diving into the… Read more
Agreed. Read more
I had the pleasure of being able to work on it before 10:30 last night. Read more
Chill dude, it's 9:24. You'll have your recap in a bit. Read more
That's a weird catch-22 though. You can't really have it both ways with "LeBron James isn't one of the greats until he wins some championships" and "If you like LeBron James because he's won championships then you're a bandwagon fan." Read more