What We're Doing with This Campaign Finance Stuff

In partnership with Dartmouth Rootstrikers we've been keeping track of monthly polling averages for all of the races we've been watching, including the 34 Class 2 Senate elections plus the two special Senate elections, as well as all 36 gubernatorial elections this year. Rootstrikers, if you didn't know, is a group (largely students) devoted to reducing the influence of money in politics. In part this is an extension of our study over the winter, in which we found a negative correlation between the amount of money spent in a Senate race and the margin of victory in that race: the more money was spent, the more competitive the race was.

There were some limitations to this study. First, while the correlation was fairly strong, it didn't allow us to infer any causal relationship--that is, we couldn't tell whether it was the competitiveness of a race that attracted more donations, or large amounts of donations that made races more competitive. Second, since the spending data we have is aggregate--all campaign funds and independent expenditures are pooled together to form a single dollar value, which, combined with a margin of victory, produces a single data point. As a result, even if there were a causal relationship, we wouldn't be able to tell which candidate was making it more competitive as a result of the spending.

This project will improve upon both of those drawbacks. What we'll essentially be doing is keeping the averages like we've been doing, but also noting down what the average for each candidate was at the end of each month. For example, on March 31 we found that in Arkansas, Rep. Tom Cotton led incumbent Sen. Mark Pryor about 45.0-42.1 in our average. A month later, on April 30, we had the senator leading 42.2-41.5. We'll do this again for the end of May, and the end of June, and the end of July, etc., until the end of October.

The key here is that after the elections, the FEC releases its candidate and independent expenditure disclosures after the elections are held. As a result we can track all of the campaign expenditures within a given month and compare those numbers to the changes in the polling averages. This will give us a rough sense of how much each dollar of campaign spending changes the polling situation. For the Arkansas Senate example above, we'll basically see how much was spent during the month of April to change the poll standing from a 45.0-42.1 Cotton lead to a 42.2-41.5 Pryor lead.

Note that I say a "rough sense": first, things may change faster than our month-to-month track can account for. Second, there's variability between states in the money-to-percentage-point ratio, and there are also state-by-state events that can outweigh any amount of spending. And in the most uncompetitive states--which still include over half of the Senate and gubernatorial elections this year--we expect polls to be less accurate and more sparse and campaign donations to be lower, with a wider gap between the winner and the loser. In short, this will give us the best results when campaign spending is high and the race is close.

By its nature, this will be on the back burner for most of the year. A lot of it is passive: we'll be waiting as polls come in to do our averages as we usually do, and the FEC is pretty slow at releasing its records--the most up-to-date records for candidates at the present time are from September 2013. We hope that they'll be quicker on the draw once the next cycle has begun, and we'll be sure to let you know if anything weird happens.