If Sinema is acting moderate for electoral reasons, she clearly disagrees with the conventional wisdom about how moderate a swing-state senator needs to be. On one hand, maybe she has a point: Donnelly, Heitkamp and McCaskill all lost reelection in 2018, as did Sen. Bill Nelson, whose home state of Florida is about as purple as Arizona but who voted with Trump less often than Sinema did. All four voted with Trump significantly less often than we’d have expected given the partisanship of their state, suggesting that Sinema’s strategy of hewing closer to expectations might have been smarter. (Although this doesn’t justify her approach of voting with Trump more often than expected.) On the other hand, political science research has found that candidates and congressional aides are really bad at assessing where voters stand on the issues. One 2013 study found that politicians overestimated by several percentage points how conservative their constituents were, in direct contradiction of Sinema’s entire theory of the case.

Maybe Sinema just figures she can’t be too careful. But if that’s the case, she’ll just win reelection in a landslide, right? Perhaps — but there are also downsides to moving too far to the center. For every inch of ground she gains with swing voters by demonstrating her independence from the Democratic Party, she risks alienating the Democratic base. And it’s not at all clear whether that’s a winning electoral tradeoff.

In a convenient natural experiment, an OH Predictive Insights poll from early September asked Arizona registered voters their opinions of Sinema and Kelly, who serves as a kind of control variable for Sinema: a Democratic senator from Arizona who isn’t a thorn in his party’s side. Accounting for the poll’s margin of error, the two had virtually identical favorable and unfavorable ratings: 46 percent to 39 percent for Sinema, 47 percent to 43 percent for Kelly.