I have to say that I have always understood Tip O'Neill's pithy little saying a little different than the generally understanding of what he meant. From Wikipedia:

The former U.S. Speaker of the House Tip O’Neill coined this phrase which encapsulates the principle that a politician’s success is directly tied to his ability to understand and influence the issues of his constituents. Politicians must appeal to the simple, mundane and everyday concerns of those who elect them into office. Those personal issues, rather than big and intangible ideas, are often what voters care most about, according to this principle.

My take has always been one which emphasized the power of the incumbency in determining the outcome of the election.  Some pundit in the lead up to the 2008 election threw out the political truism: people hate congress, but they love their personal congressman.  In other words, no matter what was happening in the country, no matter what was or wasn't happening inside the beltway, what mattered was the incumbent's connection to his or her local constituency.

I hadn't looked at the data recently and I assumed when I did that there would be a noticeable trend over the last three decades.  Not a huge dip, but especially since Newt and his Contract for America, a noticeable trend pointing downward.

But I was wrong when I checked out data at the Center for Responsive Politics's OpenSecrets.  For the House of Representatives, it has remained basically unchanged looking back to 1964.  There was a definite drop from 2008 to 2010, but it was still 85%, the lowest rate over this 46 year span (it tied with 1970).  And some of the years with the highest return rate of 98% were in 1998, 2000, and 2004, with 2002 (96%), 2006 (94%) and 2008 (94%) not far behind. 

The Senate showed a little more volatility, but the worse days for the incumbent Senators were in the 70's and culminating in 1980 with only 55%.  The best year was 2004 with a 96% rate.  After dropping to 79% in 2006, it rose to 82% in 2008 and 83% in 2010. 

This explains why it is so important who wins the House and Senate seats in 2012.  Once the individual gets in, dislodging them becomes a daunting task.  Even with Congressional approval rating hovering around the 20% range.  A huge reason is of course money.  OpenSecrets looks at the money raised by the incumbents vs. the challengers,

This money makes it, as OpenSecrets puts it, so incumbents generally don't have to work as hard to get their name and message out.

This may explain in good part why serious primary challenges are taken so seriously by those inside the beltway.  Once an incumbent makes it back to the general election, their pathway back to the beltway is almost assured.  And the primary/caucus world is more able to be taken over by a grassroots campaign that can suddenly achieve the numbers needed in the generally low-turn out affairs. 

Andrew Gelman over at Nate Silver's Five Thirty Eight has another take on Tip's bit of wisdom in All Politics is Local: The Debate and the Data. Here he looks at the how the campaigns have become more nationalized over the years.  And so not so local in determining the outcome.

It was the nationalizing of the contests which I thought would show a downward trend when I first started to inquire into the data for this blog.  In other words, the reason for the increased partisanship was because politicians were feeling more tied to the perception of their party and its perceived position on the issues and role in the problems.  But even if it is more nationalized, it doesn't seem to have had much impact on the re-election of the incumbent.