Did anyone watch that programme on channel 4 last Sunday about
astrology (I think it was Sunday - Mars was in Leo and Virgo was
rising)? Basically, they got a conman to provide two participants
with their "readings" - completely fabricated and almost identical
for both subjects. By picking up on expressions and feedback, he
managed to convince both of them that his "reading" was at least 95%
correct.
They also got an astrologer and a city analyst to predict the
FTSE-100 index for a certain date. The analyst got it right as far
as points are concerned, but got the trend wrong (it was falling when
he said it would be rising). The astrologer was a lot more vague -
saying the trend would have changed as compared to the last 2 weeks.
Judging by this, I'd have to go with Sean's side. All of that stuff
I wrote on selective hindsight applies in foresight too -
self-fulfilling prophecies and confirmatory biases, as Sean says.
This ties in with this week's debate, sort of. Hey - something's got
to - wher did this digression into astrology come from?
The Peters & Ceci paper demonstrates just how selective we can be,
and this has obvious implications for a meta-analysis such as
Prioleau et al. (Incidentally, how did Peters & Ceci get published in
a peer refereed journal...?!)
Null effects virtually never get published, the file-drawer effect at
its most extreme illustrating the selective nature of research, even
in our Science.
Not quite sure where this is going, just wanted to get the message
across about confirmatory bias, and the implications of Peters & Ceci
on a meta-analysis. Anyone want to finish my sentence for me?
This archive was generated by hypermail 2b30 : Tue Feb 13 2001 - 16:23:15 GMT