Cite as: 596 U. S. ____ (2022) 13 KAGAN, J., dissenting Doc. 65–1, p. 31. In other words, officeholders did more in exchange for donations repaying their personal loans than for other donations. The analysis next looked at Section 304’s effect. Here, the data showed that politicians with debt exceeding the law’s $250,000 threshold became “significantly less responsive” to contributions than before: They began to “behave remarkably similar to their debt free counterparts.” Id., at 28; see Ovtchinnikov, Self-Funding 3 (similarly stating that those politicians became more “independent of contributions from special interest[s]”). In other words, Section 304 did just what Congress thought it would. By preventing post-election contributions from personally enriching politicians, the provision diminished donor-responsive voting. The majority tries to undermine those findings by quoting the kind of careful caveats always accompanying good social science. See ante, at 17; Ovtchinnikov, Self-Funding 21 (noting that the study is a “first step in understanding” and that more work is needed to “fully pin down” all aspects of causation). But the authors are confident—and rightly so—in the findings just described: that Section 304 markedly decreased the frequency with which officeholders voted as donors would like. And although the authors could not responsibly claim that all the shifted votes they tallied were part of quid pro quo deals— they are, after all, professors, not the FBI—they deduce from the data that politicians carrying campaign debt were “less likely to [be] sell[ing] access” than to be “sell[ing] votes.” Id., at 18. Finally, the record evidence addresses the “almost equal[ly]” important matter of the appearance of corruption. Shrink Missouri, 528 U. S., at 390; see supra, at 6–7. A Government-commissioned survey of public opinion found that 81% of respondents believed it “very likely” or “likely” that a person who “donate[s] money to a candidate’s campaign after the election expect[s] a political favor in re-