Re: Well, admittedly the name Amos Tversky didn't ring a bell, so I ...
- --- In APBR_analysis@yahoogroups.com, igor eduardo küpfer
>3ms 2ms 1ms all 1md 2md 3md
> ...found one statistically significant instance of a cold streaky
> players: Darryl Dawkins's 80-81 season....> from Gilovich et al.:
> Probability of Making a Shot Conditioned on the Outcome of
.88 .73 .71 .62 .57 .58 .51 -.142**
Ed, I've inserted "column headers" over Dawkins' numbers. I hope I
got them right.
This situation seemed mighty curious to me. Why did Darryl shoot
almost 90% after 3 missed shots?
My source shows that he averaged just over 9 FGA per game, in '81.
9 FGA is a rather finite number; it's also hard to hit 62% of 9
Let's suppose Darryl's average game is 8 FGA, with 5 made (.625
shooting). That leaves 3 misses per game.
After 1 made shot, he's 4-7 (.571) for the remainder of the sample,
for this game.
After 1 missed shot, he's 5-7 (.714).
These numbers are startlingly close to the above "revelations" about
his inverted hot hand.
Consider a game in which the guy is 2-3. After his miss, there is
0% chance of a miss. After a hit, the chances are 50%.
Dawkins was a high-%, low-volume shooter. As such, he serves the
purpose of refuting the "hot hand"; and then he serves to spoil the
If you evaluate the 3-point-shootout charts (subtracting the miss/
make from the sample totals), you might find a similar self-
contradiction. The samples are bigger, but as expected, the
conclusions are less dramatic.
> The column headers represent each player's FG% in the specifiedsituations;
> after 3 missed shots, after two misses, etc. Notice that none ofthe players
> have a positive r -- showing that these players tended to shoot_worse_
> after making shots, and better after missing shots, the oppositeof the
> hot-hand theory. Of course, none of those correlations werestatistically
> significant, except for Dawkins.