The New Progressive Agenda Project

Many campaigns and most incumbents do not perform regular issue polling at the state or congressional district level, meaning that they typically have to guess whether a given issue will be popular among their specific electorates based on national support. Simply put, there isn't district-by-district evidence regarding the electoral viability of the progressive agenda. Until now.

For our New Progressive Agenda Project, Data for Progress (@DataProgress) enlisted Civis Analytics (@CivisAnalytics), a leading data science firm, to poll a dozen progressive policies. Our goal is to give policymakers the most reliable numbers to date on support for a new and vibrant progressive agenda. For each policy, we asked a leading politician or expert to explain why it’s necessary. But in addition, we tested it with the public. These policies have durable support and can stand up to predictable right-wing counterarguments. We have good reason to believe that even when they become politicized, they’ll remain popular with persuadable voters and the base in many geographies.

About the Project

The New Progressive Agenda Project gives policymakers and advocates reliable congressional district and state-level polling data that would normally be out of reach for even the best-funded campaign. Using the state-of-the-art modeling techniques employed by leading campaign professionals, we are confident that these policies will remain popular in the electorate at large while also engaging the progressive base. They have been carefully vetted by veteran campaigners from Civis Analytics, which was formed by the data scientists who oversaw Barack Obama’s 2012 re-election campaign. Civis’s political data science arm is one of the most reputable in the business. Civis correctly forecasted the winner in 383 out of 394 contested 2018 races (97%), and their estimate of the national popular vote in 2018 was accurate to within tenths of a percent.

Civis Analytics fielded support for more than a dozen progressive policies to determine public support for the policies. Full question wording is available below, but here it is important to note that questions included a revenue pay-for where needed, as well as both partisan cues and counterframes throughout. In other words, respondents were told that these policies were being proposed by Democrats, and were given reasons why Republicans they oppose these policies. The average sample for analysis was more than 6,000 respondents and these surveys were fielded on the Civis omnibus. Using using modern machine learning techniques, Civis generated estimates for Clinton voters, Trump voters and Independent voters, nonvoters and likely voters overall. Because our goal is to provide information that can be immediately relevant to politicians, these numbers with the exception of non-voters reflect a likely voter electorate, not national adults. Policies are presented in terms of their two-way support (that is, excluding respondents who did not register an opinion one way or the other).