The Washington Post poll out this morning makes it clear that public perceptions of the GOP took a significant hit during the budget shutdown. Polling even before the shutdown suggested that the GOP was more likely to be blamed than the Democrats, so the question becomes, why would a party pursue a course of action likely to damage its standing with voters?

One answer is that the GOP is a group of people, and so faces collective action problems. As we have seen so publicly in the past few weeks, politicians’ incentives as individuals don’t always align with their incentives as party members. But there’s another explanation I want to focus on here: politicians’ tendency to exaggerate their capacity to reshape public opinion through messaging.

Consider the mid-shutdown conversation between Kentucky’s two senators, when Mitch McConnell and Rand Paul were caught on a live microphone discussing strategy. Here’s Paul: “I think if we keep saying ‘We wanted to defund it. We fought for that but now we’re willing to compromise on this,’ I think they can’t … well, I know we don’t want to be here, but we’re gonna win this, I think.”

The quotation captures a widely held belief among elected officials that messaging matters — and that the messages parties choose play an important role in the public’s response. It’s not just GOP leaders who think that. Here’s President Obama in 2012: “The mistake of my first term … was thinking that this job was just about getting the policy right. And that’s important. But the nature of this office is also to tell a story to the American people that gives them a sense of unity and purpose and optimism, especially during tough times.” That’s a view of the president as narrator-in-chief.

If you believe that messaging can influence public opinion in powerful ways even in the short-term, that belief can heighten the incentive to pursue policies that are out of step with public opinion. In their 2000 book recounting the Clinton-era budget and health-care debates, Lawrence Jacobs and Robert Shapiro find evidence of precisely that. Rather than taking positions that are broadly in line with centrist public opinion, politicians on both sides of the aisle used poll-tested phrases to “simulate responsiveness” — that is, to dress up their party’s positions in broadly popular language.

Still, the evidence about the limits of short-term messaging is strong, as journalists and political scientists have pointed out repeatedly. If anyone can effectively shape public opinion through messaging, it should be the president, a high-visibility actor who doesn’t face the same collective action problems as do members of Congress. And yet, all of our recent presidents have managed to push public opinion away from them during their time in office, including one dubbed “The Great Communicator” and another who rose to prominence thanks to a keynote address at a national convention.

Or take a concrete example: the messaging battles over the federal health-care law in 2009 and 2010. The two parties exerted tremendous effort to explain their positions on health-care reform in catchy soundbites, just as they had done during the debates in 1993 and 1994. But the evidence I’ve amassed is at odds with a strong role for messaging.  Public opinion on health-care reform shifted only gradually, and the words that American citizens used to describe the law’s advantages and disadvantages remained almost constant as the debate unfolded. What’s more, the sharpest changes in public opinion don’t match well with the parties’ changing rhetorical strategies. Despite pundits’ fixation with the “death panels” frame, that’s not a term that virtually any voters use to argue against the law.

In part, the myth of messaging relies on the idea that there are lots of voters who are at once engaged with politics and without strong party loyalties. But as John Sides has pointed out, such voters are few and far between, since it is the strong partisans whose rooting interest keeps them tuned into C-SPAN. Just as you don’t find a lot of people at football games who will root for whichever team plays the better game, the core audience for contemporary politics doesn’t have many attentive, neutral voters who are simply listening for the best argument. Instead, the voters who follow the ins and outs of politics most closely are those with a strong commitment to a party, making them very unlikely to abandon that party at the turn of a phrase.

Does that mean that messaging is wholly irrelevant? No. As Frank Baumgartner, Suzanna De Boef and Amber Boydstun demonstrate, changing understandings on the death penalty played a central role in reshaping opinions and policies. But importantly, they didn’t do so overnight — and it wasn’t politicians who led the shift. Instead, as activists and journalists gradually adopted a new perspective on the death penalty emphasizing the possibility of wrongful conviction, voters and public officials followed suit. Messaging can matter, but it can’t invert public opinion in a matter of weeks.

Even if messaging doesn’t shape public opinion directly in the short-term, it might influence the outcome of policy debates precisely because politicians and their aides believe it to matter. Arguably, that’s precisely what we saw during the shutdown, as the belief among some GOP leaders in the efficacy of messaging increased their willingness to let the battle play out.