Blackpool Liberal Democrats

TV debates: Do they matter?

With the party leaders having agreed to televised debates, I thought it would be useful to look at the evidence about the influence of the debates on the outcome of US elections.

The best study on this subject is that of Professor James Stimson is his excellent book Tides of Consent.

Stimson notes all the famous stories about turning points in US elections but says they don't really stand up when you look at the newspaper and TV records.

"the decisive stories that come to be told don't seem to have been noticed by those writing fresh from just having witnessed the real thing".

The reports generally record debates as having been boring, devoid of real action and incapable of deciding the results.

Polls about who won often start by showing no clear winner, changing only later following the commentators declaring who won. Eventually this becomes legend and is folded into the narrative of the election.

So how do the stories of turning points stand up to the evidence?

Not too well.

Thomas Holbrook in his book Do Campaigns Matter? makes a very simple point. The debates take place very late in the campaign, the last one maybe little more than a week before the election date.

How then can they be influential? By this point, voters are rooting for their side rather than weighing up options.

Stimson investigates this idea.

He looks at the way that the winning candidate's margin changes over five elections (1976, 1980, 1988, 1992 and 2000). He takes the five campaigns together as an average.

And he finds? The debates took place after the winner's margin was already determined and the opinion line was flat.

As he puts it:

Important things do happen [in the course of the months before the election], every eventual winner coming from behind. But these important movements have already happened before the first debate is ever aired.

So what might explain this?

Here's one theory. Most of the big movements in the campaign are not from people moving from one candidate to another, but from those who have no involvement or opinion (even if they may have expressed one in a survey), forming one. By the time of the debates this process is over.

Another theory?

On average the audience for a Presidential candidate debate is about half the number of people who will go to the polls. Most of those watching are partisans rather than swing voters and enjoy rooting for a side.

But this looks only at the average of the contests. This might smooth out all the interesting stuff, disguising the fact that individual constests were settled by the debates.

So what about the famous 1960 debates between Kennedy and Nixon, where JFK's clean looks trumped Nixon's 5 o'clock shadow? Did the debates decide that? No. It turns out that Kennedy was very, very slightly further ahead on the eve of the famous first debate than on election day.

And what about 1976, with Ford's famous gaffe about Soviet influence in Eastern Europe? No again. Carter actually lost some of his lead after the debates.

The Reagan-Carter 1980 debates and the Bush-Dukakis 1988 debates may, just possibly, have put a point or two onto the anyway clear winning margin, but nothing more than that.

Perhaps only the 2000 Gore-Bush debates, where the election was absurdly close support the idea that debates are really influential. Gore's performances may have confirmed a downward trend, although this is little more than a hypothesis. The outcome is consistent with the debates having an influence, but doesn't confirm that they did so.

Here are a few conclusions based on this evidence.

First, the timing of the debates is critical. Gordon Brown would surely want them sooner rather than later if they are going to influence the tide of opinion.

By the time of the campaign proper they are probably too late. We should be having these debates now if we want them to be influential. Otherwise they will just add a little drama and good TV pictures to what is going to happen anyway.

Second, the influence of the debates even at their most potent seems to be fairly small. It may not seem like it from the media narrative but in the US they haven't really been game changers.

Third, in the US, of course, the debates are preceded by conventions which sometimes do shift opinion. Perhaps we have already had these (the conferences) but more likely, the reason why conventions work in the US is that they introduce the candidate to the electorate. The UK system is different with party leaders being well established by election time.

But perhaps the debates will prove the equivalent of a convention for Nick Clegg.

Posted by Daniel Finkelstein on December 22, 2009

http://timesonline.typepad.com/comment/2009/12/with-the-party-leaders-having-agreed-totelevised-debates-i-thought-it-would-be-useful-to-look-at-the-evidence-about-the-infl.html