In a regression discontinuity design, assignment to treatment is determined by a cut-off on an assignment variable. This design is appropriate when assignment to treatment is determined by what?

• A. A random assignment
• B. A sharp threshold in assignment
• C. A longitudinal follow-up
• D. A cross-sectional snapshot

Answer: (B)

Regression discontinuity hinges on a fixed rule that assigns treatment based on crossing a known cutoff on a running variable. This creates a sharp jump in the probability of receiving treatment exactly at the threshold, so units just above and just below the cutoff are comparable in all other respects. Any observed jump in the outcome at that point can be attributed to the treatment, assuming other factors change smoothly at the cutoff. If assignment were random, you’d be in a randomized design instead; if it relied on following people over time, that would be a longitudinal study; a single cross-sectional snapshot wouldn’t exploit the discontinuity needed. So this design is appropriate when there is a sharp threshold in assignment that produces a clear change in treatment probability at the cutoff.