Repurposing Public Data

The Challenge

In March of 2011, before the Wisconsin recall process formally began and prior to the selection of any Democratic challengers, Lincoln Park Strategies examined the openness of voters to recalling their State Senators in eight districts across Wisconsin utilizing publicly available data from IVR polling (Interactive Voice Response or “automated surveys”) conducted by Public Policy Polling for the Daily Kos. While the questions asked in the poll were not initially designed with this exercise in mind, Lincoln Park Strategies was able to use the existing data to rank the eight districts on their likelihood of electing a Democrat in the recall election.

The Strategy

Lincoln Park Strategies rank ordered the eight districts based on six indicators: approval/disapproval of Scott Walker, approval/disapproval of the district’s State Senator, support/oppose the recall effort, support/oppose re-electing the incumbent State Senator, support of Scott Walker or Democrats, and household union membership. A district with a rank of one was deemed the most open to Democratic candidates in a recall election, while a district with a rank of eight was conversely deemed least open to Democrats. Beyond the simple rank order, the difference between the support for the Democratic side and the Republican side on each of the six indicators was then averaged across questions, creating a mean spread measurement. A higher mean spread meant the district was more fertile ground for a Democratic victory, and a negative spread indicated it would be difficult for a Democrat to win.

The Results

Overall, the predictive study held up very well. To judge the effectiveness, we compared the mean spread to the difference in Republican performance and the Democratic performance in these districts. The two districts where Democrats were successful in their recall efforts (the 32nd and 18th) were rated as the two most open to a Democratic State Senator in our analysis and also were the two districts where the difference in performance yielded a net negative for the Republican candidate. The other district identified as a Low Probability Democratic Pick Up district (the14th) was the district where Democrats came closest to winning a coveted third seat needed to take control of the majority in the State Senate. In other words, the districts we calculated as the first, second, and third strategic targets for Democratic victory ended up providing our two victories and our most narrow defeat. However, since it is impossible to accurately predict the exact path of a campaign, including media and outreach strategies, studies such as this offer a unique means of assessing the situation well in advance of Election Day.

The techniques used in our Wisconsin study could never be relied upon as an exact indicator of outcomes, but neither could any polling five months out from the election.  What this study shows is that IVR polling can be used in a reliable and cost effective manner to determine the territory a candidate will be dealing with and to help organizations decide which districts are the most strategic targets. The type of candidates running, fundraising, and many other variables will always change the equation, however knowledge of the likelihood of success is a piece of information that is critical to any campaign.