Following the leader: Congressional Communication on Social Media
Over the last decade, social media has become a key platform for political leaders to communicate with the public; however, political elites also use these platforms to strategically communicate with one another. In a recent paper, Hao Yan, Mike Alvarez, Sanmay Das, Betsy Sinclair, and I find that patterns of communication on social media follow predictable patterns from a game theoretic model of message coordination among the party leadership and their rank-and-file members. These findings hint at the possibility of using social media to understand the strategic interactions of members of Congress.
Our results suggest that leadership’s ability to influence the terms of discussions depends on balancing two key forces: coordinating around a policy stance and successfully aggregating information where policy outcomes are uncertain. Party leaders and rank-and-file must balance coordinating around any position (the coordination problem), but also selecting the correct position (the information problem).
Depending on which problem dominates, party rank-and-file will lead or follow with respect to their strategic communication strategy. When the party has a low need for policy direction, the coordination problem dominates. So, the rank-and-file turns toward the leadership whose messaging acts as a rallying point around which to coordinate. Thus, we expect that topics where the need for policy direction is low will be the topics where House leaders initiate discussion for the rank-and-file. By high need for direction, we mean that the electoral, political, or policy costs of coordinating on the “wrong’’ policy stance are large. In the case of coordinating on the wrong stance, when need for direction is high, it might be more advantageous to avoid coordination. For example, coordinating on the wrong stance could exacerbate a government shutdown crisis, leading to electoral defeat for the party. By low need for direction, we mean that the effects (good or bad) for coordinating on the “right’’ or “wrong’’ policy stance are not large — perhaps only a positive or negative cycle of news coverage.
To find topics where need for direction is low, we follow the subsequent procedure: we identify topics topics that drive separation in between the two parties (for example, immigration), we expect the information problem is least acute. For the most polarizing topics, this is precisely because these are policy stances which define belonging to a particular party. Thus, in theory, it is on these types of partisan topics that leaders have the most influence over the rank-and-file, since the outsized costs or benefits of coordinating on the wrong messaging are low. Here, the coordination problem should predominate, and we expect party leaders to take the initiative in defining the messaging.
We show results from each party above. Specifically, Figure 1 shows that Democratic House leaders exerted the most influence over the propensity to discuss Trump immigration policies (approximately a 1 percentage point increase for each standard deviation shock) and protecting health care (a ~0.5 percentage point increase). House leaders exerted a ~0.5 percentage point increase for gun policy, presidential power, combating climate change, child separations, and women’s reproductive health. On these same topics, Figure 1 shows that House rank-and-file exerted a 4 percentage point increase on the leader’s propensity to discuss President Trump’s immigration policies, a 3 percentage point increase for protecting health care, and 2 percentage point increases for gun policy, presidential power, combating climate change, child separations, and women’s reproductive health. Notably, these effect sizes are an order of magnitude higher than the leadership’s influence on rank-and-file members.
The Republican party exhibits behavior consistent with the Democratic party. Figure 2 shows that impulses of ~10 percent to the GOP House leaders’ daily propensity to discuss a particular issue results in a less than 1 percent increase in the rank-and-file members’ daily propensity to discuss that issue. In particular, Republican leaders induced a 1 percentage point increase in their rank-and-file members’ propensity to discuss impeachment. Leaders induced a 0.5 percentage point increase for trade deals and USMCA. For protecting health care, equality for women, jobs and the economy, presidential power, and agricultural policy, GOP House leadership exerted a 0.2 percentage point increase on the rank-and-file members’ propensities to discuss these sentiment-topics. Figure 2 also shows that members induced a 3.5 percentage point increase in their leadership’s propensity to discuss impeachment, and a 1.5 percentage point increase for trade deals and USMCA. For protecting health care, equality for women, jobs and the economy, presidential power, and agricultural policy, GOP leadership exerted a 0.5 percentage point increase on the rank-and-file members’ propensities to discuss these sentiment-topics. Again, members’ influence is an order of magnitude larger than the leadership’s influence. Notable, the magnitudes derived for Republicans leadership and rank-and-file members are smaller than for Democratic leaders and members. This suggests that Democratic party leaders and members are more responsive to each other with respect to their messaging around their propensity to discuss sentiment-topics.
This evidence from the topical analysis suggests that leaders exert influence over their rank-and-file members on topics that come to dominate social media discussion, and vice versa. Importantly, although it is clear House leadership and rank-and-file messaging on Twitter influence each other, we find consistent with game theoretic frameworks of Congressional strategic communication, leaders tend to initiate discussion and messaging on topics where the need for policy direction is low and the coordination problem predominates. Interestingly, rank-and-file members are able to exert nearly double the influence over their leaders as in the reverse. This suggests the power of leaders to drive messaging is not necessarily absolute in the US House, even on a highly scripted platform such as Twitter.
