Primary Spending Strategies May Have Shifted General Election Outcomes


In the two weeks since Election Day, Sunlight — along with many others — have examined the impact of outside money. In competitive House seats we found no statistically observable relationship between the outside spending and the likelihood of victory, and found no evidence of spending impacting outcomes for the Senate either. It’s important to note that those who contributed to the $1.4 billion spent by outside groups still matters, though. As Executive Director Ellen Miller notes: “Even if their candidates lost, the influence bought by America’s new class of mega donors will remain.”

Here, we find some indication that outside spending in primary races may have had implications for general election outcomes this cycle. In the competitive races where there was significant primary activity by outside spenders, as compared to a baseline in which parties retaining control of seats they held in the 112th Congress, Democrats over-performed while Republicans significantly underperformed. Furthermore, we found notable involvement by outside Democratic groups in Republican primaries which may have played a roll, while finding little evidence of parallel Republican activity.

We looked at the 90 races in the House that were competitive as of September 6th, according to the Cook Political Report (Likely, Lean or Tossup). Of these competitive seats, in the 19 where there was more than $10,000 in outside Democratic spending, Democrats won 17, a success rate of 89%. This was despite the fact that 12 of those 19 seats were previously held by Republicans. In contrast, of the 25 seats where there was over $10,000 in outside Republican spending, Republicans only won 11, or 44%. 17 of these seats had been held by Republicans prior to the election.

Given the small sample size it is difficult to isolate the effect of significant primary activity, from other factors like incumbency or district partisanship characteristics. Regressions which controlled for the district’s Cook Partisanship Voter Index (reproduced in tables at the end), as well as the incumbent party found that Republicans were 40% less likely to win the general when there was outside spending in their primaries, than if there was no outside spending. Democrats, in contrast, were 2.7 times more likely to win a seat when there was outside spending in their primaries. More than capturing the effect of outside spending per se, this may be a window into the impact of the parties primary nominating process on their general election outcomes. Either way, these findings, while descriptive, lie outside the traditional bounds of statistical significance.

Almost a third (29%) of outside money in Republican primaries was spent by Democratically affiliated groups. Using party alignment of outside groups calculated by previous research here at sunlight, we found that traditionally Democratically aligned groups like House Majority PAC, Women Vote!, and the anti-tea-party Credo super PAC, targeted conservative candidates as early as the primary. Democratic candidates won the general election in 10 of the 17 races where Democratically affiliated outside groups dropped cash in the Republican primary. We find no similar spending trend by Republican affiliated groups during Democratic primaries. Republican aligned groups only spend substantially in two democratic primaries; Now or Never PAC spent against Tammy Duckworth in IL-8 and The New Prosperity Foundation spent against Bill Foster in IL-11.


Democratic spending in Republican primaries signals several possible strategies at play:

  • In some cases it appears that Democratic money helped Democrats run against more conservative candidates, who were more easily painted as out-of-step with their districts. This may have been the case in the Florida 9th, the Democratic super PAC House Majority PAC spent over $93,000 opposing two Republican primary candidates who both lost, John Quinones and William Long. Nominated in their place was Todd Long, a Tea-Party activist who had previously lost 2 house bids, who ultimately lost to Alan Grayson by 25 points. This seat re-districted seat was a Democratic pick-up. This type is the type of strategy we saw play out explicitly in the Missouri Senate race when Democrat Claire McCaskill’s campaign ran ads supporting the extreme Todd Akin.
  • In other cases, early spending by Democratic groups seems to signal that the Democratic machine has targeted an incumbent they believe to be vulnerable, and are simply getting on the ground early to frame the narrative. These cases tend to be relatively moderate Republicans, like Charles Bass in the New Hampshire 2nd or Chip Cravaak in the Minnesota 8th district . Democratically affiliated groups leveled $127,000 of spending at Cravaak during an unopposed primary, before he was defeated in the general election.


It is difficult to know the extent to which Democratic primary spending in outside races shaped the narrative for the coming election and helped label the Republican candidates as out-of-synch with their districts. We may simply be observing a natural correction of the composition the House from the tea-party wave in 2010 back towards a position more in line with the median voter.  In such a case, democratic activity would all have the appearance of succeeding, even though it may not have actually shifted the needle. This sample does, however, offer a window into the strategy employed by Democratic groups, during a cycle in which Democratic candidates were largely successful.


Logistic Regression Results (Republican victory dependent):

Outside Spending in Rep Primary Rep party incumbent No incumbent party District PVI (pos = more consv.)
Odds Ratio 0.588 9.879 5.734 1.429
Std. Error 0.344 7.064 7.122 0.1124


Logistic Regression Results (Democratic victory dependent):

Outside Spending in Dem Primary Dem party incumbent No incumbent party District PVI (pos = more consv.)
Odds Ratio 2.678 10.133 2.492 0.732
Std. Error 1.557 7.663 2.677 0.058

Data used for the regression is available here,  and more detailed descriptive data is available here.