Holding the Line: Social Norms and Party-Line Voting

Earlier this year, Senate Minority Leader Mitch McConnell voted to acquit former President Donald Trump on a single article of impeachment, charging him for “incitement of insurrection” over the January 6 riot at the Capitol. Although he voted to acquit Trump, McConnell harshly criticized him shortly after the impeachment trial, saying that rioters had been “fed wild falsehoods by the most powerful man on Earth.” McConnell also added, “Trump is practically and morally responsible for provoking the events of the day.”1

Like McConnell, many Republican senators criticized Trump for his actions but voted to acquit him anyway.1a Ultimately, Trump was found not guilty in a 57-43 vote, 10 votes short of the supermajority needed to convict him. 

It is not surprising that a majority of Democratic senators voted to convict, while a majority of Republican senators voted to acquit. The degree of party-line voting in the United States has increased substantially over the last few decades.2 The percentage of party unity votes for both the Senate and the House of Representatives rose from 32% in 1970 to 70% in 2020.3,4

There are many reasons why parting-line voting has increased substantially over the last few decades. From a behavioral science perspective, social norms are one major explanation for the increasingly distinct voting records of Democrats and Republicans in Congress. 

Social norms

Voting along party lines may have become a social norm for many Democrats and Republicans in Congress. A social norm is a rule of behavior to which individuals tend to conform, on condition that they believe that:

(a) most people in their reference network conform to it (empirical expectation); and

(b) that most people in their reference network believe they ought to conform to it (normative expectation).5

People’s reference networks include people they care about when making particular decisions. In McConnell’s case, his reference network may have included his fellow Republican senators and his constituents. 

If voting along party lines has become a social norm for McConnell, then he preferred to acquit Trump because:

(a) he saw other Republican senators acquit Trump or believed that other Republican senators would  acquit Trump; and

(b) he believed that most of his fellow Republican senators and constituents expected him to acquit Trump. 

Note that preference does not equal “liking better.” For instance, I love pineapple pizza, but I am allergic to pineapples. So, all things considered, I would prefer plain pizza when given the choice. Similarly, McConnell probably did not want to acquit Trump given his harsh criticism of Trump after the impeachment trial, but all things considered, he preferred to acquit Trump because normative expectations play a critical role in compliance.5 If others believe one ought to conform to a norm, nonconformity could result in ostracization or even punishment from those who conform to or support the norm.5 

Consequences of norm violations 

Although some studies suggest that violating a norm can make the transgressor feel good, most studies agree that transgressors experience negative emotional consequences, such as feelings of guilt, shame, and disappointment in the self.6,7,8,9 Given that feelings of guilt and related emotions foster compliance, these feelings prevent the transgressor from violating the norm in the future.10,11

Interpersonally, norm violations often trigger negative affective reactions in observers. Some of the most common reactions are anger and blame.12,13,14 If McConnell had voted to convict, most of the anger and blame would come from his fellow Republicans (i.e., his in-group members). 

Researcher Saaid Mendoza and colleagues investigated whether individuals would punish violations of fairness norms more strictly when they were committed by an in-group member than by an out-group member, even if it is costly to the individual.14a To do this, the researchers recruited 115 undergraduates and told them that the study was a collaborative project between their college (Amherst College) and a nearby rival college. Thus, group membership was defined by college affiliation. Participants were seated in private computer cubicles and assigned an avatar based on their college affiliation. 

Prior to the experiment, the researchers assessed how strongly the participants identified with their college. For the experiment, participants played an ultimatum game in which one player (the proposer) divided $20 between themselves and another player, who had the choice of accepting or rejecting the offer (the responder). If the responder accepted the offer, then both players received the proposed amount—but if they rejected it, both players received nothing. Unfair offers (e.g., 70:30 splits) were often rejected, since fairness is strongly valued by most people.14b

After a practice trial, the participants considered offers (e.g., $4, $6, $7, $8, or $10) from avatars similar to their own (i.e., in-group members) and avatars different from their own (i.e., out-group members). After the experiment, participants then reported the average amount of money they expected to receive from in-group versus out-group members.

Results showed that participants did not differ in their tendency to accept fair offers (i.e., $10) from in-group members and out-group members. However, participants were more likely to reject an $8 offer from in-group members than from out-group members. ($8 is a meaningful threshold for fairness, since on average, participants expected to receive $8.61 from in-group members and $7.52 from out-group members.)

In addition, if the participants identified strongly with their college (i.e., the in-group), then they were more likely to enforce fairness norms by rejecting unfair offers from other in-group members.


The seven Republican senators who voted to convict Trump faced backlash almost immediately from their in-group members after the impeachment trial. For example, Republican Senator Pat Toomey was censured by a handful of county-level Republican parties for his vote to convict Trump.15 Washington County Republican Chairman Dave Ball said, “We did not send him there to vote his conscience or to do the right thing, we sent him there to represent us.”15 This statement clearly demonstrates the normative expectation that Toomey’s reference network had for him: to acquit Trump. Perhaps if Toomey were seeking reelection, he would have succumbed to the social norm of voting along party lines.  

If most politicians in Congress subscribe to the social norm of voting along party lines, then there is another reason to question the integrity of the U.S. government. Politicians being influenced by voting norms likely means that a large percentage of laws are passed based on which party sponsors them, rather than the benefits or protections they could provide to citizens. The impact of voting norms also potentially means that when making decisions, politicians do not consider what’s right or wrong, but rather, what will make their reference network happy. 

Read Next


Beyond Irrational Politics

What can behavioral science tell us about politics? A lot, it turns out. Political polarization has intensified to the extent that we give our trust based on who says something, not what they say.


Is a Biased Vote Better Than No Vote?

The US Presidential Election is fast approaching, and pressure is mounting for every eligible person to cast their ballot. But these calls to vote can play into our cognitive biases.