Mostly True
Poll results from the past 35 years show that State of the Union speeches "do not move" presidential approval numbers.

Matthew Dowd on Sunday, January 23rd, 2011 in an interview on ABC's "This Week with Christiane Amanpour"

Matthew Dowd says there's no evidence of a State of the Union 'bounce'

CORRECTION: The initial version of this item erroneously attributed the comment to New York Times columnist Maureen Dowd.

With President Barack Obama about to give the State of the Union address, pundits of all stripes have been pre-analyzing and prognosticating what he may say before a joint session of Congress on Jan. 25, 2011.

On the Jan. 23, 2011, edition of ABC’s This Week with Christiane Amanpour, former Republican strategist Matthew Dowd offered some historical context to how State of the Union speeches actually change public opinion -- or rather, how they do not.

"To me, the State of the Union (addresses) … they don't affect the American public," Dowd said during the program’s roundtable segment. "If you look at … approval numbers going into State of the Unions over the last 35 years and coming out, they do not move the numbers. Even Ronald Reagan, who was lauded as one of the best communicators in the history of this country, never moved the American public."

We thought we’d see whether Dowd was correct.

While a number of media-affiliated polling organizations track presidential approval ratings, we turned to Gallup because polling experts frequently used the company’s polls for long-term studies. Gallup, which began polling before World War II, has the longest and most consistent track record of publishing such polls.

Mark Blumenthal, a polling analyst who founded the website, published a thorough analysis of the topic in 2006. It was augmented in a 2010 study by Gallup’s own Jeffrey M. Jones.

The Gallup data goes back to 1978 -- to President Jimmy Carter’s first State of the Union address -- and includes a total of 28 addresses. (Most presidents have given an address to a joint session of Congress shortly after their first inauguration, but technically these are not State of the Union addresses. Because of that, neither Blumenthal nor Jones included them in their surveys, and we won’t either.)

Here’s the bottom line as Jones puts it: "A review of Gallup historical data suggests these speeches rarely affect a president's public standing in a meaningful way, despite the amount of attention they receive."

By our count, presidents are actually somewhat more likely to see their polling numbers decline than see them rise after they give a State of the Union address. In all, the numbers rose 10 times, fell 15 times and remained unchanged three times. (You can see the year-by-year figures here.)

Even more strikingly, the shifts were usually quite small. If you were to define a presidential "bounce" as a 5-point rise or more -- a bit outside the typical margin of error of 3 points -- then there have only been three "bounces," or about once in every 10 speeches.

In fact, during the period studied, a president’s standing rose by more than 6 percentage points only once -- the 10-point rise for Bill Clinton in 1998. And that speech was somewhat exceptional, since the Monica Lewinsky scandal had just broken and Clinton used the address to refocus Americans (with some success) on his policy agenda.

Clinton was the only one of the five presidents prior to Obama whose numbers rose when the poll numbers after each president's addresses were averaged together. Clinton saw his approval ratings rise by 3 points on average, while Carter and President Ronald Reagan saw their ratings drop by 1 point, President George H.W. Bush saw his ratings drop by 4 points and President George W. Bush saw his drop by 1 point.

We should note that there is another type of poll taken during State of the Union season, usually with strikingly different results -- so-called "instant polls." When Gallup does these polls, they ask people who responded to a pre-State of the Union poll if they minded getting a quick call after the speech to gauge their immediate reaction. Typically, these polls show favorable results toward whoever the president is.

However, Blumenthal discounts the value of these polls, citing data showing that, in most cases, viewership for the State of the Union tilts heavily toward the president’s party. "Simply put, people who don't like a particular president are considerably less apt to tune him in," he quotes ABC polling director Gary Langer as saying.

Since Blumenthal’s published analysis is now five years old, we checked with him to see whether he still felt the same way. "I definitely think the argument still holds," he told us.

Dowd was very close to correct that State of the Union addresses "do not move the numbers." Over the past 33 years at least, that has been the case more often than not. And the one president Dowd specifically cited, Reagan, never got a bump greater than 3 points. But there are a few exceptions, including Clinton's 10-point bounce in 1998. So we rate his statement Mostly True.