In 1993, journalist Richard Ben Cramer published a landmark work of political journalism, What It Takes: The Way to the White House. The 1,072-page book, seven years in the making, tracked six major candidates as they made their way through the 1988 presidential campaign, which ultimately helped shape the coverage of every campaign since.

I thought I'd use this model to look at incumbent governors -- but turn it on its head. Conventional wisdom would say that incumbents have been pretty hard to unseat in recent years. I looked at how many incumbent governors overall have sought a new term since 1998 and found that there were 97 such candidacies. This means that the reelection rate for incumbent governors during this period was 82 percent, making incumbent governors almost five times as likely to win as to lose.

For governors to have such high a reelection rate is noteworthy, since politicians holding executive positions would appear to be at greater risk of suffering from voter frustration than a single lawmaker, who is only one out of many in a large body and most likely has deeper ties to constituents in their home district than the governor does statewide.

So it begs the question: What does it take for an incumbent to lose?

I analyzed every election cycle from 1998 through 2010 in search of governors who were elected to office but failed to secure a second term. I didn't count governors who voluntarily decided against seeking a new term, and I didn't count those who assumed the governorship without an election and subsequently lost a bid for a full term.

However, I did count governors who tried to secure their party's re-nomination for a new term but lost to a primary rival. And I counted California Democratic Gov. Gray Davis, who was elected twice but then was turned out of office in a recall election. His tenure provides a few good anecdotes about incumbents, which I will explain later on.

All told, I found 17 incumbent governors who lost over the 12-year period -- a group almost evenly split between Democrats and Republicans. These governors' losses were widely spread out, occurring in nine different years, including off-year elections in states such as Kentucky, Mississippi and New Jersey. The only year that provided as many as three separate examples was 2002, a campaign cycle of unusual upheaval in the gubernatorial ranks.

The eight incumbent Republicans that lost, in chronological order, were Fob James (Alabama, 1998), David Beasley (South Carolina, 1998), Cecil Underwood (West Virginia, 2000), Craig Benson (New Hampshire, 2004), Frank Murkowski (Alaska, 2006 -- primary), Bob Ehrlich (Maryland, 2006), Ernie Fletcher (Kentucky, 2007) and Jim Gibbons (Nevada, 2010 -- primary).

As for the Democrats, nine incumbents weren't reelected: Don Siegelman (Alabama, 2002), Roy Barnes (Georgia, 2002), Jim Hodges (South Carolina, 2002), Ronnie Musgrove (Mississippi, 2003), Gray Davis (California, 2003 -- recall), Bob Holden (Missouri, 2004 -- primary, Jon Corzine (New Jersey, 2009), Chet Culver (Iowa, 2010), and Ted Strickland (Ohio, 2010).

To get a better sense of what caused these 17 incumbents to lose, I drew up a list of factors I thought might have played a role in one or more of the races. I then ran these by political experts in the states that ousted their governors, hoping to glean which of the factors played major roles, minor roles or no role in the incumbent's loss.

Here are the factors, in declining order of impact.

Troubles for the state combined with ineffectual leadership. This was a major factor in six losses -- those of Republicans James and Benson, and of Democrats Holden, Corzine, Culver and Strickland. It was a minor factor in two contests, Alaska's Murkowski and Nevada's Gibbons.

Perhaps the classic case is Ohio's Strickland, who had the misfortune of presiding over one of the nation's hardest hit states during the Great Recession. While Strickland was hardly the least effective politician on our list of one-termers, the economic troubles he presided over were too much for him to overcome in the 2010 election. The race was close, but he ended up losing by less than three percentage points to current governor, Republican John Kasich.

Long-term partisan trends in the state shifting away from the governor's party. This was a major factor in five races -- those of Republican Ehrlich in Maryland and Southern Democrats Hodges, Barnes, Siegelman and Musgrove.

This has been a much bigger factor for the Democrats than the Republicans. All four Democratic incumbents on this list were from the South, where Democrats at any level have become increasingly endangered. The last of these four to lose -- Mississippi's Musgrove -- went down eight years ago, and given the long-term trends, there's a good chance we won't see any Democratic governors in these states for some time now.

Poor campaign skills. My sources and I found this to be a major factor in two races -- Benson's and Holden's -- and a minor factor in six (James, Murkowski, Fletcher and Gibbons on the Republican side and Hodges and Corzine for the Democrats). Such numbers suggest that being a poor campaigner can help sink an incumbent governor's candidacy, but not usually by itself. Defeat typically requires other problems beyond weak campaign skills.

Personal scandal. Scandals helped torpedo the reelection bids of Republicans Murkowski (a frivolous jet purchase, among other things), Fletcher (alleged corruption) and Gibbons (an affair), and Democrat Siegelman (alleged corruption). This suggests that scandals weren't common, but when they happened, they were deadly to an incumbent's reelection prospects.

A bad national electoral environment for the governor's party in that election cycle. I found this factor to have a surprisingly weak influence. It was a minor factor in Ehrlich's loss in 2006 (when Democrats surged nationally) and a bigger factor for two Democrats in 2010 -- Culver and Strickland.

In addition, I found two other factors whose influence was too scattered to draw major conclusions about them. West Virginia's Underwood lost in part because he had been elected with the presence of a significant third-party candidate on the ballot, and then lost in a year when there wasn't one. Meanwhile, South Carolina's Beasley lost after taking a principled but unpopular stance on a hot-button issue -- the Confederate battle flag.

Some broad conclusions can be drawn from this data:

1. The most common problem leading to an incumbent governor's loss is a practical one -- economic trouble in their state, combined with the governor's inability to get in front of the problem. This essentially boils down to voters exercising their prerogative to throw the bum out.

2. Scandal-tarred incumbents often bow out rather than run for reelection, but when they do decide to seek another term, the outcomes are uniformly bad. Bottom line: Leave gracefully.

3. Long-term partisan trends in a state, or the good fortune of running in a favorable election cycle, can make a big difference when a weak incumbent is on the ballot.

For the sake of comparison, I also looked at eight incumbents who were only marginally popular when they ran for another term, but who managed to win anyway.

Two were Republicans: California's Arnold Schwarzenegger in 2006 and current Texas Gov. Rick Perry in 2010. Six were Democrats: Gray Davis (California, 2002), John Baldacci (Maine, 2006), Jennifer Granholm (Michigan, 2006), Ted Kulongoski (Oregon, 2006), Jim Doyle (Wisconsin, 2006) and Christine Gregoire (Washington, 2008).

Here too, I selected a few potential factors -- largely the reverse of the ones we looked at above -- and asked sources in the states about which ones played a major or minor role in the incumbent's victory. Here's the rundown:

A good national electoral environment for the governor's party in that election cycle. For each of the five Democrats on my list who won in 2006 and 2008, this was a hugely important factor. (It was also a minor factor for Perry, a Republican, in 2010.)

In 2006, the Democrats surged nationally, taking over both chambers of Congress. Two years later, the Democrats gained ground again, thanks to big turnout driven by Barack Obama's presidential bid.

The favorable environments for Democrats in 2006 and 2008 unquestionably helped Baldacci, Granholm, Kulongoski, Doyle and Gregoire. Each had a somewhat bland, technocratic style but represented a state with a purple-to-blue tinge in a very good year for Democrats. In each case, it was enough to put them over the top. It's worth noting that no Democratic incumbent governor anywhere lost in either campaign cycle.

The opponent had worse campaign skills than the incumbent. This one also mattered a lot. It helped Republican Schwarzenegger and Democrats Davis, Baldacci, Granholm, Kulongoski and Doyle.

The two California governors offer the purest case studies.

Davis, whose approval ratings were mired in the high 30s to low 40s, went so far as to fund attack ads in the GOP primary that were aimed squarely at former Los Angeles Mayor Richard Riordan, the strongest of Davis' possible Republican rivals. The ads helped secure the nomination for William Simon -- who was not only a more conservative candidate, making him less appealing in a general election, but also one with far less experience on the campaign trail.

"One Republican consultant after the race likened the Davis-Simon choice to having to decide whether you liked Lyle or Eric Menendez better," recalled former Davis aide Garry South, referring to the two men convicted of shooting their parents to death in the early 1990s. "I publicly said it was a race between damaged goods and a defective product."

Meanwhile, Schwarzenegger -- despite some significant early stumbles after his 2003 recall victory over Davis -- won a full term in 2006 in part because he faced Democrat Phil Angelides, whose personality and campaign style wore so poorly on California's ordinarily Democratic-leaning voters that he couldn't even reach 39 percent on Election Day.

Long-term partisan trends in the state favoring the governor's party. This, too, was a notable factor in helping preserve weak incumbents. It helped Democrats Davis, Kulongoski and Gregoire, but it was probably most notable for helping Perry win his third reelection bid in 2010.

Perry went into the election with middling popularity, after having won a second full term in 2006 with just 39 percent of the vote in a four-way contest. And in 2010, the Democrats got as good a candidate as they could have found -- former Houston Mayor Bill White, who had moderate credentials and a largely successful record in office. By then, Texas had become so solidly Republican that Perry managed to win over theoretically stiff competition by a 13-point margin.

The incumbent's lack of scandal. This helped in a minor way in a number of cases, but it was never a deciding factor.

A neutral-to-positive economic picture in the state. I was surprised to find how infrequently a decent economy helped reelect an endangered incumbent. It aided Perry in 2010, but rarely helped the others, with the possible exception of Doyle.

Finally, two incumbents -- Maine's Baldacci and Oregon's Kulongoski -- benefited from having third-party candidates on the ballot for their reelection bids, but in neither case made this the deciding factor.