Exercise 1: What Should Have Prompted Scrutiny in 1998 — Possible Solution
============================================================================

Three real, methodology-level red flags were visible in 1998, entirely
separate from whether the study's conclusion turned out to be true or
false:

1. SAMPLE SIZE. Twelve patients is far too small a group to support a
   population-wide causal claim about a vaccine given to millions of
   children every year. A finding at that scale needs replication in a
   much larger, independently run study before it can support any real
   policy or medical advice - a single small case series can only ever
   suggest a hypothesis worth testing further, not confirm one.

2. SELECTION OF PATIENTS. The twelve children weren't a random or
   representative sample - they were referred to Wakefield's own
   hospital, several through a legal aid scheme tied to litigation
   against vaccine manufacturers. A sample selected this way is
   genuinely prone to bias before a single measurement is even taken,
   regardless of how carefully the measurements themselves are done.

3. UNDISCLOSED FUNDING. Wakefield's own financial relationship with a
   lawyer preparing a lawsuit against MMR manufacturers was a direct
   conflict of interest that was never disclosed to the journal or its
   readers. Readers in 1998 had no way to know this specific fact - but
   the broader habit of asking "who funded this, and do they have a
   stake in the result?" is a real, applicable check for any study,
   regardless of whether a specific conflict is ever confirmed.

ANSWER: The sample size (12 patients), the non-random way those
patients were selected (several through vaccine-litigation-linked legal
aid), and the simple habit of asking who funded the research and
whether they had a stake in the outcome were all real, methodology-level
questions available in 1998 - independent of any fraud investigation,
and independent of whether the conclusion later turned out right or
wrong.

WHY THIS WORKS AS AN ANSWER
------------------------------
This identifies methodology problems that were genuinely visible at
publication time, rather than hindsight-only facts that only the later
investigation uncovered - directly answering what the exercise actually
asked.