Exercise 1: Why the 97%-to-36% Drop Is More Alarming Than It First Appears — Possible Solution
=================================================================================================

At first glance, "only 36% replicated" might sound like it simply means
roughly a third of psychology findings are real and two-thirds are
false. But the real implication is worse than that framing suggests.

The 97% figure describes how many of the ORIGINAL 2008 studies reported
a statistically significant result - nearly all of them. If the
underlying effects being studied were mostly real, a well-powered
replication attempt should reproduce a significant result close to that
same very high rate, since a genuinely real effect doesn't disappear
just because a different research team measures it a second time.

Instead, the real replication rate collapsed to about 36% - roughly a
two-thirds drop. Combined with the finding that average effect sizes in
the replications were only about half the size originally reported,
this suggests many of the original 97% of "significant" findings were
never robust real effects in the first place - some were likely
statistical noise, inflated by small samples, flexible analysis
choices, or the publication-bias filter that only lets striking results
through, exactly the real mechanisms Ioannidis's 2005 paper describes.

ANSWER: The drop is alarming because it's not simply "some findings are
false" - it suggests the ORIGINAL near-universal 97% significance rate
itself was inflated well above what real, robust effects would produce,
meaning a large share of what got published and treated as established
knowledge may never have reflected a genuine, reliable effect at all.

WHY THIS WORKS AS AN ANSWER
------------------------------
This explains the real statistical logic behind why the size of the
drop matters, not just its existence, and connects it directly to the
chapter's own Ioannidis material on why inflated significance rates
occur in the first place.