Via Goddard, Eric Ostermeier makes that case over at smart politics:
[H]istory has not been kind to candidates who test the presidential waters a second time - with just five winning the presidency on their second attempt over the last 200+ years. The far more common outcome is that which has been experienced by the dozens of candidates who reprised their role as failed White House hopefuls the second time around.
That's true as far as it goes, I think...but it's not exactly meaningful. After all, the most important point is that history isn't kind to candidates who test the presidential waters, period.
So: five candidates, by their accounting, have won on their second try. Ronald Reagan won on his third try. And there's a long list of those who have failed on their second try.
But there's also a long list who have failed on their first try! After all, the overwhelming majority of those who run for president are defeated -- somewhere in the neighborhood of 90%, I'd guess, over time.
Or, to put it another way: we've only had 44 presidents. But Ostermeier is excluding LBJ, who won in his second try -- but only after becoming an incumbent. That's the right way to do it. However, if we exclude all those who inherited the presidency (and Cleveland's second term) then we're down to 34 who became president by being elected. And I'd be reluctant to count either Washington (since no one had previously run) or John Adams (since only Washington had previously run). So that gets us down to 32, of whom 26 won on their first try, 5 on the second, and Ronald Reagan on the third try. Now, I have no idea how many candidates in a typical cycle are first timers or repeaters, but a five-to-one ratio doesn't sound particularly unlikely to me. If so, then the chances of a repeater would actually be just the same as a first-timer.
To put it yet another way...the best way to win a presidential election is to run as the incumbent. Barack Obama has a far better chance of winning next year than any particular GOP candidate!
It's also worth noting that since reform, George H.W. Bush and Reagan were repeaters, while Carter, Clinton, George W. Bush, and Obama were first-timers. Adding the postwar presidents puts two more (Ike and Kennedy) in the first-time column, and Nixon in the repeaters.
Or, more to the point, there's nothing in any of this which would make me think that Romney has less of a chance of winning the presidency than Pawlenty or Barbour.
(I have no reason not to trust Ostermeier's data, but I will note that there are lots of disagreements over who counts as a candidate for president, especially in pre-reform contests when one could be a viable candidate without entering primaries -- if there were any -- and there were no FEC regulations to comply with. I'm not aware of any presidents who he miscategorizes as a result, however).