ohn B. Judis, a journalist on the left wing of the New Republic, is an often perceptive political analyst, but just as often a mischievous one. His latest article concerns liberal Republican congresswoman Marge Roukema of New Jersey, and the attempt by conservatives to beat her in a primary. This, Judis helpfully advises, would be a disaster for the GOP, which needs to give more power to its “moderates.” A conservative nominee would guarantee that the seat goes to the Democrats.

But Judis stacks the deck to make his case.

1) He picks and chooses data to try to show that Roukema’s district could go Democratic. Judis writes, “Bergen County has become increasingly Democratic — in 1996, it backed Clinton over Bob Dole 48 percent to 42 percent. Even Warren County is becoming more Democratic. In the last election, Democrats won control of its county government.” All very interesting, no doubt, but surely it’s worth mentioning that Dole actually carried Roukema’s district-which is a pretty good definition of a solidly Republican district in the Northeast.

2) The North and the far West, argues Judis, are slowly moving into the Democratic column: “It began in 1992 and then intensified in 1996, when President Clinton made inroads in Republican districts. As these districts went increasingly Democratic, they ousted conservative Republicans — like Rick White and Randy Tate in Washington state and Mike Pappas in New Jersey — as well as recently elected Republican moderates — like Bill Martini in New Jersey and Peter Torkildsen in Massachusetts — who had not yet established strong local followings.”

This is all very misleading. White and Tate, like Martini, came to Congress in the 1994 tsunami; these weren’t longtime Republican districts that suddenly booted them out. And why should the moderate Torkildsen’s defeat be attributed to his having been elected “recently”? He first won election to Congress in 1992; he had been in office twice as long as Tate when they were defeated in 1996. Pappas, too, had only one term in office, winning in 1996 and losing in 1998.

3) Judis cites evidence that Ohio voters like gun control, and then says that conservative ideologues ignore such evidence. Judis quotes an article by Stephen Moore, who’s involved in the effort to defeat Roukema, in which Moore argued that the GOP should write off the Northeast as a “dying region.” Fine. But who considers Ohio part of the Northeast?

4) More Judis: “Indeed, early indications suggest the 2000 presidential election will reinforce this realignment. . . . If Bush wins the presidency, it would certainly delay this trend, just as Jimmy Carter’s victory in 1976 temporarily stalled the move toward a Republican majority. But the forces pushing Republicans to the right and Democrats to the center would eventually reassert themselves, propelling Democrats into office.”

With this hedge, Judis has pretty much made his thesis non-falsifiable. Even if in 2001 the Republicans have both houses of Congress, the White House, a prospective majority of the federal judiciary, and a majority of the nation’s governors. . . they’re still doomed. The Democratic realignment will reassert itself “eventually” — maybe in 2015.