Exercise 2: Identifying Type I and Type II Errors — Possible Solution
====================================================================

THE GIVEN SCENARIO: TYPE I ERROR
------------------------------
The team rejected H0 (concluded the feature had a real effect) when,
in reality, H0 was actually true (no real effect existed - the
observed difference was just sampling noise). Per this chapter's own
2x2 table, rejecting H0 when H0 is actually true is precisely the
definition of a TYPE I ERROR - a false positive. The team's
significance threshold (alpha) was exactly the mechanism that was
supposed to control how often this happens, but a Type I error is
still possible any time a test is run, purely by chance, at whatever
rate alpha allows.

A SECOND SCENARIO SHOWING THE OPPOSITE ERROR: TYPE II ERROR
------------------------------
Suppose a different team tests a new checkout flow that genuinely does
improve conversion rate in reality (H0 is actually false - there IS a
real effect). However, their test sample size was small, or the true
effect size was modest, so the resulting p-value came out above alpha
(say, p = 0.11), and the team failed to reject H0, concluding "no
significant effect found" and discarding a genuinely beneficial
feature. Per this chapter's own 2x2 table, failing to reject H0 when
H0 is actually false is precisely a TYPE II ERROR - a false negative,
a real effect that went undetected.

WHY THIS WORKS AS AN ANSWER
------------------------------
The given scenario is matched directly to this chapter's own 2x2
error table by identifying which cell it falls into (reject H0 / H0
actually true), and the second scenario is deliberately constructed to
land in the opposite cell (fail to reject H0 / H0 actually false),
demonstrating a clear understanding of both error types rather than
just naming one and describing the other only vaguely.