Political Ads: Not as Powerful as You (or Politicians) Think

Over at the Washington Post’s Wonkblog, I have a new post on the effects of political advertising in the 2012 presidential race.  I first use the map below to illustrate where it was that each candidate had advertising advantages when considering all national advertisements by the campaigns and their allies starting in April 2012.  I then estimate the effect of advertising in non-swing states, making use of the uneven mapping between television markets and swing state boundaries.

The takeaway?  The 2012 electorate was hard to persuade through television, with advertising effects much smaller than in 2008 but comparable to those in 2004.  In all likelihood, even major shifts in advertising would have produced only minor shifts toward the candidate benefiting from those shifts.

5 Responses to Political Ads: Not as Powerful as You (or Politicians) Think

  1. Jonathan January 20, 2013 at 10:38 pm #

    Dan, you might be interested in a talk Romney’s chief data scientist Alex Lundry (adjuncting in your Department at Georgetown no less) gave at the DC Data Science meetup a month or so ago.


    His slides can be found on Twitter as images too from here going forward


    Romney’s data people were tasked w/ creating an optimal advertising strategy…so your point would seem rather contentious.

  2. Comment January 21, 2013 at 7:31 pm #

    Why are the West Coast and Northeast lopped off?

  3. Dan H. January 21, 2013 at 9:35 pm #

    @Jonathan, thanks for the pointer to Alex Lundry’s presentation–that’s definitely of interest. That said, I don’t see anything in the description or slides that necessarily contradicts the evidence presented in the blog post. I don’t see a contradiction between the claim that 1) the 2012 electorate was very hard to move with TV advertising, and that 2) the campaigns expended enormous financial and personnel resources to optimize their advertising.

    @Comment, there are a few reasons. One is just to let people see a bit more detail on the states of interest, which are for the most part the swing states. This wasn’t 2000: no one threw ad money at California. A second is because New England requires additional data work, given that my mapping provides county boundaries and some New England states report election results by town.

  4. Andrew Gelman January 21, 2013 at 9:39 pm #


    The black boundaries between the counties are distracting. I suggest gray. I also recommend a red/blue intensity scale (i.e., bright red down to gray, then gray to bright blue in the other direction). I’ve found that to be much clearer than the red/purple/blue color scheme.

  5. payday loans May 29, 2013 at 2:57 pm #

    I simply could not go away your web site before suggesting that
    I actually loved the usual information a person provide in your guests?
    Is gonna be back frequently in order to investigate cross-check new posts