I don’t see what’s wrong with a low sample size. That seems pretty standard and it’s enough to rule out noise in this case. Almost all of the participants improved and by a statistically significant amount.
They are saying that at the start their intervention group was 15 IQ points below the control group! And post-training the intervention group mostly closed the gap with the control group (but still did not quite get there).
They actually selected the test group for having the lowest score on the synesthesia test. So this fits with my theory of synesthesia being correlated with IQ, but it’s also interesting that synesthesia training improves IQ.
The usual things—the results are at best brittle and worst just a figment of someone’s imagination.
Almost all of the participants improved and by a statistically significant amount.
Yeah, well, that’s a problem :-/
I eyeballed the IQ improvement graph for the intervention group and converted it into numbers. By the way, there are only 13 lines there, so either someone’s results exactly matched some other person on both tests or they just forgot one.
The starting values are (91
96
99
102
105
109
109
113
122
133
139
139
145)
and the ending values are (122
113
109
118
133
99
118
123
151
133
145
151
151)
The deltas (change in IQ) are (31
17
10
16
28
−10
9
10
29
0
6
12
6)
So what do we see? One person got dumber by 10 points, one stayed exactly the same, and 11 got their scores up. Notably three people increased their scores by more than one standard deviation—by 28, 29, and 31 points.
Y’know, I am not going to believe that a bit of association training between letters and colors will produce a greater than 1 sd increase in IQ for about a quarter (23%) of people.
I don’t see what’s wrong with a low sample size. That seems pretty standard and it’s enough to rule out noise in this case.
The replication project in psychology just found that only a third of the findings they investigated replicated.
In general studies with low sample size often don’t replicate.
I don’t see what’s wrong with a low sample size. That seems pretty standard and it’s enough to rule out noise in this case. Almost all of the participants improved and by a statistically significant amount.
They actually selected the test group for having the lowest score on the synesthesia test. So this fits with my theory of synesthesia being correlated with IQ, but it’s also interesting that synesthesia training improves IQ.
The usual things—the results are at best brittle and worst just a figment of someone’s imagination.
Yeah, well, that’s a problem :-/
I eyeballed the IQ improvement graph for the intervention group and converted it into numbers. By the way, there are only 13 lines there, so either someone’s results exactly matched some other person on both tests or they just forgot one.
The starting values are (91 96 99 102 105 109 109 113 122 133 139 139 145)
and the ending values are (122 113 109 118 133 99 118 123 151 133 145 151 151)
The deltas (change in IQ) are (31 17 10 16 28 −10 9 10 29 0 6 12 6)
So what do we see? One person got dumber by 10 points, one stayed exactly the same, and 11 got their scores up. Notably three people increased their scores by more than one standard deviation—by 28, 29, and 31 points.
Y’know, I am not going to believe that a bit of association training between letters and colors will produce a greater than 1 sd increase in IQ for about a quarter (23%) of people.
The replication project in psychology just found that only a third of the findings they investigated replicated. In general studies with low sample size often don’t replicate.