Something to Run For (It’s Not Trump)

By Meredith Conroy (@sidney_b) and Jon Green (@_Jon_Green)

Whichever #wave you’re anticipating on November 6th--Blue, Pink, or Rainbow--it will have been made possible by the undisputed surge in individuals ready and willing to run for office since January 2017. The sheer number of qualified and representative candidates running this year has caught our collective attention because for so many decades political candidates largely resembled only a small subset of American citizens, and they emerged from traditional candidate pools -- legal and business fields, for example. In many cases where districts were not competitive, no candidates emerged at all. This midterm year is different.

But what is driving this new wave of candidates? In explaining the response from Democrats and progressives, there has been no shortage of stories that credit individuals’ interest in running as a reaction to Donald Trump’s election. And while we don’t discount that Trump’s election and his administration’s agenda roused progressive candidates in particular to action, we suspect there is more to this story.

To better understand what is driving a new era of progressive individuals to run for office, we partnered with Run for Something, a non-profit with the goal of recruiting young and diverse progressive candidates to run for political office. Of the thousands of people who visited to express an interest in getting involved, we identified who was most likely to translate that interest into an actual candidacy, and how prospective and actual candidates articulated their interest in running for office.

Run for Something provided us with about 10,000 de-identified intake forms from their website collected between April of 2017 and July of 2018. The intake form was accessible from, where individuals could visit and submit a form notifying the organization of their interest in potentially becoming a candidate. Forms asked respondents their birth year, race, gender, ZIP code, and the reasons why they were interested in running for office. Run for Something also provided us with tags associated with each form, indicating whether or not each respondent actually wound up running for office.

We present a selection of some of our findings below. You can find additional results in our full report here. First, the charts display results of a model estimating the likelihood of a given respondent running for office based on demographic characteristics: race, gender, age, and geographic region. We find main effects for race (conditional on being in the dataset and controlling for other factors, people of color were more likely to run than white respondents) and an interactive effect between age and gender (the likelihood of running increases with age, but this relationship is stronger for women than it is for men). Rural respondents are also somewhat more likely to translate interest into action overall, possibly reflecting a relatively smaller pool of progressive candidates in rural areas..

Screen Shot 2018-11-01 at 2.30.25 PM.png

We were also interested in how people explain their interest in running for office. Why do they want to get involved? To begin, we pick out a handful of keyword categories -- words related to health, education, local issues, political dynamics, and Trump -- to get a preliminary look at how members of different social groups may have articulated their interest in running for office, differently. Here, a few things stand out. First, relatively few respondents -- regardless of whether they eventually ran for office -- mentioned terms related to President Trump in their statements of interest. Of the keyword categories we examine here, Trump is the least commonly-mentioned (less than ten percent of the pool mentioned Trump or Trump related terms). Second, respondents who ended up running for office were much more likely to talk about education, their local communities, and political dynamics than respondents who did not wind up launching a campaign. This suggests to us that the people who are running for office are motivated to solve specific problems facing their communities. The below chart also shows evidence of gender differences in keyword prevalence. Men are more likely to mention political dynamics when describing why they want to run for office, while women were more likely to mention education and health.


We then conducted a more systematic text analysis by specifying a structural topic model to identify groups of words that tended to appear together across the statements of interest, and whether those groups of words were more likely to be used by members of different social groups. While further discussion and visualization of these results can be found in our full report, a few takeaways are worth noting here. First, in line with the preliminary text analysis shown above, prospective candidates who talked about their local community, the political dynamics associated with the district in which they were planning on running, and specific issues such as education and health care were more likely to translate their interest into action than respondents who simply mentioned a general interest in getting more involved in the political process or a general frustration with Donald Trump’s administration.

Most notably, when it comes to predicting who runs for office, talk isn’t cheap. Our analysis shows that the information from the prospective candidates’ statements of interest is much more informative than their demographic information. About ten percent of the people who filled out Run for Something’s intake form wound up running for office -- a relatively rare event. This being the case, a model predicting whether each respondent becomes a candidate based on demographic characteristics alone is right about 90% of the time by simply predicting that no one runs for office. However, when we predict candidate emergence using the output from our text-based model, not only do we get respondents who are predicted to run for office, but these predictions are fairly accurate, correctly classifying 74% of the respondents who it predicted ran for office.

Screen Shot 2018-11-01 at 2.29.53 PM.png

How can we account for the new wave of candidates running for office at all levels of government in 2018? Although reports have attributed the surge to Trump himself, our analysis of Run for Something’s candidate pool indicates that when it comes to explaining interest in running for office, relatively few prospective candidates are articulating their interest in reference to the administration -- and those who do are significantly less likely to convert their interest into an actual candidacy. Instead, respondents who emerge as candidates mentioned issues they care about (often health care and education), concerns regarding representation gaps, and matters that directly relate to their local communities. In short, regardless as to whether Trump’s election spurred interest in running for office, the progressives who are acting on that interest have something to run for.

Jason Ganz