Do campaign ads have any effect on elections?

The Freakonomics blog had an article some time ago that argued that there is little evidence that money buys elections.  Here is the conclusion of the article:

I have examined several other natural experiments and found similar results. For example, large shocks to campaign spending from changes in campaign finance regulations do not produce concomitant impacts on electoral success, nor do candidates with vast personal wealth to spend on their campaigns fare better than other candidates.  

These findings may be surprising at first blush, but the intuition isn’t that hard to grasp.  After all, how many people do you know who ever change their minds on something important like their political beliefs (well, other than liberal Republicans who find themselves running for national office)?  People just aren’t that malleable; and for that reason, campaign spending is far less important in determining election outcomes than many people believe (or fear).

There is an article today by John Sides that argues, yet again, that campaign ads have little to do with whether candidates win elections.  John Sides however explains his rationale by citing four common reasons for this effect:

1. Campaign activity can be overwhelmed by other events in the campaign.
2. It can be hard to win the messaging war (essentially your counterparties are saying much the same things)
3. The effects of campaign activity are short-lived.
4. In most presidential elections it is hard to get a big advantage in campaign activity.

Note, the two conclusions while superficially similar, may not in fact be similar at all.  

Freakonomics argues that once opinions are set in, ads have little effect.  Here's Steven Levitt's paper on the subject.  Steven Levitt concludes that campaign spending had negligible impact on elections and spending by your opponent often had more of an impact.  In Freakonomics' view a vast spending difference between the two candidates would not necessarily handicap one or the other.  In John Sides' view, it may.  John Sides' explanation is that the campaigns were fighting themselves to a draw.

Here's my issue with John Sides' argument that seems illogical.  If as they say "effects of campaign activity are short-lived" why even aim for a draw? They cite their and others’ analyses of 2012 advertising effects as well as previous research to demonstrate their point.  

What caught my attention were two charts they reproduced in their article:

Here's what they say about the charts:

This graph shows that Romney’s time at Bain Capital was a major part of advertising and news coverage only in July  In part, this was because the Obama campaign emphasized Bain less as the general election campaign heated up. In part, this was because the news coverage of Bain Capital — stimulated in July largely by a Boston Globe story – waned as the media moved on to other topics.

However, I don't buy their conclusion.  Firstly, the articles they cite in support actually studied the relationship between vote shares and advertising intensity.  Their conclusion that long term advertising intensity did not have an effect appears to be drawn for lower correlation for advertising intensity over longer terms.  Hmmm ... but since this isn't a controlled experiment, wouldn't it be more logical to assume that the ad teams reallocated media dollars to regions where the ads appeared to be working.  So, I am not sure how they draw that conclusion.  This does not exactly undermine their conclusion, but it does raise some eyebrows.

Next let's take those charts.  Actually, that's what got me thinking.  My recollection was that what Obama was trying to do was to define Romney before Romney could define himself.  So, what I thought might be interesting to look at would be Google Search trends for "Romney Bain".  Here is the chart:

What I noticed almost immediately was that post Obama's massive ad campaign, he was able to raise the level of interest in "Romney Bain" so that it settled at a new level.  This isn't a "short term effect".  Is it possible that how short term the effect is depends on how well formed your opinions are.

I have two takeaways.  In essence, I actually agree with both Freakonomics and John Sides.  I also generally agree with John Sides' reasons.  However, my other takeaway is that the problem with these intensity and spend studies is that they ignore message and "earned" media - i.e. what people are talking about.  As the Google charts show, done right, paid media may move the conversation.  It is likely though that spending cannot really compensate for the lack of appeal of the underlying product (candidate) or message.