People were shown pairs of candidates with randomly assigned priorities, and picked their favorite. Each bar shows how much a candidate gained or lost support when the issue was on their agenda, everything else held equal. Light bars show when it was one of the candidate's priorities, darker bars show when a candidate made it a top priority.
Source: national survey of US registered voters; a randomized candidate-choice experiment (six head-to-head picks per person). Effects are average marginal effects on the probability of being chosen, weighted to Census benchmarks; whiskers are 95% confidence ranges. Each view is scaled to its own audience.