It seems likely that Sen. Kyrsten Sinema (D-Ariz.) has a plan. Most people do, on most things, even if those plans are externally inscrutable or internally shaky. Whatever this plan might be — and it is certainly the case that any plan she has for this particular political moment is hard for an outsider to ascertain — it is also probably the case that it’s rooted in questionable assumptions about the electorate.

Why do I say that? Because research has consistently shown that politicians and staff for elected officials make incorrect assumptions.

We should start by noting that there’s nothing about Sinema that necessarily stands out politically to a significant effect. We can measure the Senate on two metrics: how ideological they are (as indicated by a measurement called DW-NOMINATE calculated by Voteview) and how red or blue their state is, as revealed in its 2020 presidential vote margin. When we do so, we find Sinema near the middle on both metrics: Arizona narrowly preferred President Biden and she is among the most ideologically moderate members of the Senate, much less the Democratic caucus.

But, then, so is her colleague, Sen. Mark Kelly (D-Ariz.). Sinema is the second most-moderate member of the Democratic caucus, behind only Sen. Joe Manchin III (D-W.Va.). But Kelly is ranked fifth, trailing Sens. Angus King (I-Maine) and Thomas R. Carper (D-Del.). In other words, the difference between Sinema and Kelly on these two metrics is vanishingly small — but the difference in their responses to the twin policy proposals the caucus is trying to pass is miles apart.

Manchin and Sinema have been at the forefront of discussions about the legislation. Manchin, of course, represents one of the reddest states in the country, giving him strong incentive to push back on the caucus. But that’s not a requirement, of course. Sen. Jon Tester (D-Mont.) has been engaged in the discussions but hasn’t been a constant part of the media circus. Sinema has.

Again, it may be the case that Sinema has a plan, perhaps one that views the proposals — which would, among other things, increase spending on infrastructure, tackling climate change and health-care coverage — through the lens of what Arizonans seek from government policy. If it is the case that her objections to the bill in its original form are based on perceived hostility from her voters, it is worth considering what we know about such perceptions.

In 2013, David E. Broockman and Christopher Skovron published a paper called “What Politicians Believe About Their Constituents: Asymmetric Misperceptions and Prospects for Constituency Control.” A very academic descriptor but one with several useful findings.

For example, they asked a number of state legislators to estimate support for universal health care and same-sex marriage among their constituents and then compared those responses to the actual level of support in the districts. In general, Democratic legislators underestimated support for both issues and Republican legislators underestimated that support even more dramatically. You can see that on the graphs below; the black line indicates a perfect alignment of prediction with support. For both the blue (Democratic) and red (Republican) dots, the trend is well below the actual levels of support.

“These systematic distortions in elite perception suggest an extremely biased view of the American electorate among those who vie to exercise political authority,” Broockman and Skovron wrote.

There’s another part of that paper that’s relevant to Sinema. The senator has come under fire for being inaccessible to her constituents, leading to confrontations in a public restroom and on a plane. Broockman and Skovron’s research finds that having Sinema interact more with voters probably wouldn’t matter much anyway in informing her of their views.

“Did candidates who had spent more time interacting with potential supporters gain a better appreciation for their constituents’ views? Our results suggest that they did not,” they wrote, later adding that “[p]oliticians may sincerely believe that they are learning about district opinion writ large from these activities, but our results suggest that they are learning little that is systematically diagnostic of constituency opinion.”

Well, Sinema has staff, right? Perhaps they can fill her in on what Arizonans think.

Perhaps not. A few years ago, researchers Alexander Hertel-Fernandez, Matto Mildenberger and Leah C. Stokes posed a similar experiment to congressional staffers. They summarized their paper in an essay for the New York Times with the pointed title, “Congress Has No Clue What Americans Want.”

They found that Republican Hill staffers underestimated support for universal background checks by almost 50 points on average, for example. Interestingly, they also found that the estimates for support were fairly uniform regardless of district, perhaps suggesting some Hill-centric consensus. You can see the responses below, in graphs from the paper. The axes are reversed from the graphs above; the vertical axis measures actual support and the horizontal the estimates of staffers. Note the limited vertical range in which the responses sit for each question.

In describing the experiment, the researchers are again direct.

“[I]n none of the five areas are staffers estimating their constituents’ preferences with any degree of relative or absolute accuracy,” they wrote. “Staffer perceptions are far more extreme than the public’s actual policy preferences.”

Another recent bit of research adds additional complexity to the question. In a paper titled, “Incongruent Voting or Symbolic Representation? Asymmetrical Representation in Congress, 2008—2014,” researchers Adam Cayton and Ryan Dawkins considered, among other things, what sort of legislative agreement voters were actually looking for. That is, did voters want a legislator that was ideologically aligned with them, even if they deviated on matters of policy? Or were they more focused on policy support and less on ideology?

What they found is that it depends. Among Democrats, legislator approval was more positively influenced by agreement on policy positions. Among Republicans, it was more positively influenced by ideological alignment. (The metric used on the graph below is the difference between complete agreement and complete disagreement on either policy or ideology; higher values indicate more positive benefit in approval.)

If this holds in Arizona, Sinema might see some benefit from independents by holding to a moderate ideological position. But she’s likely to frustrate hard-left Democrats on both ideology and policy — which may be why there have been rumblings of a primary challenge already, despite her reelection bid still being three years away.

“While Democrats have a stronger incentive to follow their constituents’ policy preferences,” Cayton and Dawkins wrote, “Republican representatives are incentivized to follow voters’ symbolic preferences, even if it means betraying their district’s policy preferences to do so.” Sinema is taking a position that would seem to be more positively received by the political right than the left.

Again, maybe there’s a plan. Perhaps there isn’t; politics is not usually driven by perfectly logical analyses. But if there is a plan, and it depends on Sinema’s assessments of where Arizona wants her to be, her plan may not unfold as anticipated.