Four questions about one job posting. Each has a right answer and wrong ones, and the arithmetic decides which is which.
Commit to a prediction before you look. These are not opinion questions. Making an implicit belief explicit is what lets you test it, and the useful moment in this exercise is the one where the number disagrees with you. Reconciling an intuition with the evidence is the work; getting one wrong costs nothing and teaches more than getting it right.
What the sequence shows is a well-intentioned fix that does not deliver its stated objective. Nobody in it treats anyone differently. The order is sincere, the deletion is real, and the disparity returns anyway — because an organization is a system of stages, and a change made at one stage does not travel to the others on its own. Disparity-generating mechanisms have to be identified before interventions are designed, and identifying them takes analysis rather than intent. A fix at one stage cannot do the work of a fix at another, and coupling them does. The questions are built so you arrive at that yourself rather than being told.
The numbers here are real. Degree attainment and population counts for adults 25 and over come from the U.S. Census Bureau's Current Population Survey, 2024 — the same public tables an analyst would use to run a four-fifths computation, and cross-checked against the NCES Digest of Education Statistics. What is modeled is the employer: the posting, the reviewer's preference, and how many people apply. The Census publishes who holds a degree, not who applied for which job, and no public source carries applicant flow. That gap is not incidental — it sets a limit on what any hiring analysis can conclude, and the summary at the end returns to it.