Mathematical Model Offers Insight Into What Drives Partisan Politics 

"The division started to grow up from there."  

The campaigning around the midterm elections provided a concrete reminder of the current state of US politics: They’re increasingly divided. As a way to address the division, a new mathematical model out of Dartmouth College offers some hope as to what social factors might be widening this gaping gulf.

Previous data analyses have shown us that we are a divided nation. It’s no surprise. A new model published in Royal Society Open Science, built and pioneered by Feng Fu, PhD., an applied mathematician at Dartmouth and undergrad student Tucker Evans, tries to boil this pattern down to a few distinct variables. In doing so, it reveals how politicians today are much less likely to work across the aisle than they were in the mid-20th century. Not only that, but both Democrat and Republican members of Congress are just as unlikely to work together no matter which party is in power. And the researchers say it goes deeper than simple political disagreements.

Fu tells Inverse that the model takes into account the extremity of a representative’s political opinion, the benefit that they might gain from sticking by their party (a factor they label homogeneity), and the benefit of forging connections with others who are not necessarily affiliated with their group. When they applied this framework to voting data from the US House of Representatives from 1949 to 2009, one change in these variables predicted how polarized the Congress became.

politics Washington DC

How Does It Work?

“The fundamental that drives the division is how people value the benefit of homogeneity versus the benefit of connections,” Fu explains. Their model — and the real-world data — suggest that these senators value homogeneity more than connections.

To explain this, the data puts forth a model for how groups form. Initially, extreme opinions tend to drive people apart on both sides. In short, there will always be extremists on both sides, but whether those extremists tend to lead to overall polarization, came down to the relationship between two other variables: the benefit to the individual of maintaining a wide array of social connections or the benefits of doubling down and agreeing with the group.

When the researchers ran test after test on their model, they found that if wider social connections were more highly valued that in-group homogeneity, the network would converge around a more central point. But it homogeneity was more highly valued, the group would fracture into two camps.

While there are benefits to doubling down, sticking with one’s party, and pushing for a cause that the other party strongly opposes, there are also times when reaching across the aisle may make sense to pass crucial legislation. Feng explains that when they applied their model to voting records from the US House of Representatives, there was one period of time where this seemed to be the case. Though he’s hesitant to guess at why this might have happened based purely on a mathematical formula.

“There was a period in Congress when people valued connections more than homogeneity in the Sixties and Seventies, the maximal cohesion in history,” Fu says. “The division started to grow up from there. I don’t know what kind of social or political factors led to that.”

political polarization mathematical model
In 1969 and 1979, this model suggests that there was less polarization because the value of social connections was higher than the value of sticking by party lines.

What Does This Actually Mean?

Teasing apart the political context this model describes in the sixties and seventies is a job for a political scientist or historian, not a mathematical model. Ideally here, we’d hope to find a solution in the data, though the paper doesn’t seem to offer one.

Other research has at least illuminated this dynamic a bit further. For example, a study in PLOS One from 2015, also on House of Representatives voter data (from 1949 to 2012), highlighted that in general, cooperative pairs (representatives from opposing parties who vote together) are hard to find.

But as the number of different politicians who vote together dwindles, a different trend has taken its place. Since 1990, there is a handful of representatives who tended to vote against party lines more often — which the authors of the PLOS One paper call “super cooperators.” For example, during the 110th Congress — between 2007 and 2009 — 98.3 percent of the cooperative pairs were within a network of seven individual congress members. The researchers suggest that for each of these individuals, it was in the best interest of their constituents — not their political careers — to work across the aisle:

The few super-cooperators, who hand pick legislation and cooperate with members from each party, despite threat of alienation from his or her party, may be today’s hallmark example of carefully representing a constituency.

Whether that’s what was at work on a larger scale during the time period identified by Fu’s model is outside of the math wonk’s purview, but perhaps in the hands of historians, the mathematical model of this shift will be illuminating. Data might hold clues as to how we got where we are, even if it can’t yet show us a solution.

Media via Flickr/, Royal Society Open Science, Flickr