By Bob Benenson, CQ Staff

Over the past couple of weeks, CQ Politics published individual stories in which we changed our ratings on the key Senate contests in Minnesota, Colorado and New Hampshire, then followed that with a roundup of 14 other races in which ratings were adjusted. This series of switches was capped on Tuesday when the indictment of longtime Alaska Republican Sen. Ted Stevens prompted CQ Politics to flip its rating from Leans Republican — a reflection of the strong challenge the long-dominant incumbent had drawn from Democratic Anchorage Mayor Mark Begich — to Leans Democratic, as the long-simmering ethics allegations hanging over the senator exploded into front-page news.

The handicapping of all of the nation’s Senate, House and gubernatorial contests has been a staple of Congressional Quarterly’s election coverage for more than a half century, and it naturally has reached many thousands more readers since 2006, when the launch of the free CQ Politics site expanded our reach nationwide and well beyond our traditional audience within the Washington political community.

With that more widespread attention have come more questions about how we arrive at our ratings, and whether our ratings amount to hard and fast predictions of who is going to win. So I am taking this Jigsaw Politics column to explain and hopefully make things a little more transparent.

CQ invented the ratings system — which many other analysts and news organizations came to adopt and adapt to their own styles — in the early 1950s. Originally a product of the Congressional Quarterly Weekly Report, now CQ Weekly, the ratings evolved into seven categories:

— No Clear Favorite, for races that are deemed to close to call.

— Leans Democratic and Leans Republican, for highly competitive races in which the named party is perceived to have at least a slight edge, but in which an upset is a highly plausible possibility.

— Democrat Favored and Republican Favored, for races in which the named party has a clearer advantage and is likely to win, though an upset cannot be completely ruled out.

— Safe Democratic and Safe Republican, for races in which the named party is viewed as certain to win.

One point that needs to be clearly understood is that the contests rated “Leans,” as well as those rated “No Clear Favorite,” are very competitive and could go either way — and so it behooves those with partisan rooting interests to pay close attention and make no presumptions about those contests’ outcomes.

The slightly favored candidate in a Leans race has something going for him or her that we believe justifies the rating: years of popular incumbency, their party’s strong track record in the state’s or district’s election, a big edge in campaign fundraising or other advantages. But the challenger also has qualities that makes her or him a highly viable contender. If that weren’t the case, the race would be rated in the “Favored” category for less competitive races, or the “Safe” category for non-competitive races.

We have maintained the tradition of limiting the number of categories to keep our race ratings from getting overly complicated for readers. We are aware, though, that this results in a lot of races falling into gray areas between categories.

For example, there are 23 House races that we currently have rated as Leans Democratic (19 for seats currently held by Democrats, four for Republican-held seats subject to serious Democratic takeover bids) and the 23 contests rated Leans Republican (all for GOP seats). Among these are indisputable battleground races between opposing top-tier contenders that border on tossups and could ultimately go the other way. There are others in that category in which the favored candidate is battling a strong opponent, but is somewhat more likely to win. In fact, one of our wonky in-house jokes is that these races would be more properly rated “Lavored,” for those on the cusp between Leans and Favored, and “Savored,” from races on the borderline between Favored and Safe.

We have for years tried to apply a mental framework in designating these ratings. No Clear Favorite races are those we believe will end up the closest, with the winner prevailing by a razor-thin margin. Leans races are those we think, in greatest likelihood, will end up with a winner taking between 51 percent and 54 percent. Favored races are those we think could end up with a winner taking roughly 55 percent to 59 percent. And most Safe races should end up with the victor taking 60 percent of the vote or more, a landslide in our books.