Twenty-five years ago, I worked on polling for candidates for governor in Puerto Rico. Puerto Rican politics revolves around the issue of status – the current commonwealth versus statehood – and voters were and are closely divided between the two parties that back each alternative.

Questions routinely used in polls in the mainland didn’t work in Puerto Rico. Ask voters there for favorable impressions of the statehood party candidate, and you got a glorious rush of positive comments from statehood voters – while from commonwealth voters, not a single word of praise. Ask for unfavorable impressions, and the statehood voters were mute, while the commonwealth voters gush forth with an abundance of complaints.

And so it went: Voters thought everything about their own party and its leaders was good and everything about the other party was bad.

Things seem an awful lot like that on the mainland United States these days. Voters are closely divided, and Democrats have almost nothing good to say about George W. Bush, while Republicans have almost nothing but contempt for Democrats.

One consequence is that elections have become less about persuasion and more about turning out the base. In 2004, both parties did so: John Kerry’s popular vote was 16 percent higher than Al Gore’s, while George W. Bush’s popular vote was 23 percent higher than four years before.

In last week’s elections, neither party did so well. Turnout in New Jersey and Virginia was not up much from four years before, while in California turnout was down 26 percent (on preliminary figures) from the recall election two years before.

There’s another consequence, I thought after examining the 2005 returns: Partisan preference is not closely connected to job ratings of party leaders. Voters stick with their parties no matter what. The two parties’ percentages for governor in Virginia were almost identical in 2001, when Bush’s job rating was hovering around 80 percent, and in 2005, when his job rating was hovering around 40 percent.

In New Jersey, the Republican actually ran 6 percent better in 2005 than in 2001 (just as Bush ran 6 percent better in the state in 2004 than in 2000). In California, three of the four ballot propositions backed by Gov. Arnold Schwarzenegger got yes votes significantly higher than Schwarzenegger’s current job approval.

This has actually been the case for some time, since the budget showdown between Bill Clinton and Newt Gingrich in 1995-96. In 1996, Democrats won percentages well under Clinton’s robust job rating in both the presidential and House races (49 and 48.5 percent). Republicans that year and again in 1998 won percentages in House races (49 percent both times) well above the poll numbers of Gingrich, the party’s most visible leader.

In 2000, Al Gore, with 48 percent, ran way behind Clinton’s high job approval and way above Clinton’s low personal favorability numbers. In 2002, 2004 and now 2005 Republican percentages were in the same narrow range, even though Bush’s job ratings were significantly different in those three years.

You can see a similar phenomenon at other times in our history when voters have been sharply polarized. During World War II, Franklin Roosevelt’s job ratings stayed around 75 percent – people recognized that he was a great war leader. Yet he was re-elected in 1944 by only a narrow partisan margin of 53 percent to 46 percent. (He would have gotten more, but Republicans and Dixiecrats then, like Democrats in Florida in 2000, worked to keep soldiers from voting.)

In other periods, we don’t see this polarization. Between 1988 and 1994, support for the parties (and for Ross Perot) oscillated wildly, in the polls and in election results. Large numbers of voters weren’t firmly anchored to either party and, discontented, were prepared to seek alternatives.

Are we entering such a period again? Polls show great discontent, and Democrats look with hope at polls showing them well ahead of Republicans in the House elections generally. But Democrats have been ahead in this so-called generic vote for the House in most polls since 1994, even as Republicans won more seats and more popular votes in six straight House elections. I haven’t seen poll results from specific districts showing such movement, but I’m keeping an eye out.

In the meantime, the 2005 elections – state contests, to be sure, and so not necessarily indicative of trends in congressional elections – suggest that we have the same kind of Puerto Rican style politics we’ve had since 1995-96. If so, turnout will decide the outcome in 2006.