« NYT/CBS News Poll: Kerry, Bush in Dead Heat | Main | Summary of the Post-Debate Polls – Kerry Closes the Gap »

Granite State Gains

Thomas F. Schaller, associate professor of political science at the University of Maryland, Baltimore county, author and executive editor of The Gadflier contributes this important analysis of voter registration trends in New Hampshire

My fellow political scientist Dante Scala, who literally wrote the book on the history and significance of the New Hampshire primary, alerted me earlier today that registration rates in the Granite State don’t look very positive for the Republicans.

New Hampshire’s Board of Elections reports registration rates after its September state-level primary concludes, and Scala reports that the GOP has not made much registration headway since the last registration totals were taken, in advance of the 2004 presidential primary last January.

Digging a bit deeper, I analyzed the final partisan registration totals for each of the past four presidential election cycles, then compared them with this year’s September figures. Now, reader beware: New Hampshire permits same-day registration, so things can change between the September 14 totals and November 2. The figures for 2004, therefore, are only the best set of estimates we have in advance of the election; those registering between now and November could move the numbers.

Year    DEM%    GOP%    IND%    2PD%
2004    28.2    33.6    38.2    45.7
2000    26.2    35.3    38.4    42.6
1996    28.9    38.7    32.4    42.8
1992    33.4    39.1    27.5    46.1
1988    30.4    38.9    30.7    43.9

Having said that, the September figures lend credence to Scala’s finding that the GOP is stagnating this year. Indeed, the Republicans seem to be losing ground. From left to right, the table reports the share of all state voters registered as Democrats, Republicans and Indpendents. The final column, which I calculated, reports the simple two-party share of Democratic registrants (that is, 2PD% = D divided by D+R).

Notice that the Democratic high-water mark in terms of two-party share of registrants was 1992, when Bill Clinton was elected. It dropped back about three points during the following 12 years…until this year, when it has (again, at least through September) returned nearly to its 1992 two-party ratio.

One final note of caution: The share of state registrants self-identifying as Independents has grown steady during the past decade or so. Scala attributes this to a legal change permitting cross-over primary voting for Independents. So,it may be that the GOP is not actually losing registrants to the Democrats, but rather that a significant number of Republicans have re-registered as Independents.

Still, if those independents break proportionally between Kerry and Bush the way they have in previous elections, the registration gains for the Democrats during the past two years may be good news for Kerry.


> Still, if those independents break proportionally
> between Kerry and Bush the way they have in
> previous elections,

I think that you should say "the way they have nationally." You don't know how Independents broke in NH particularly, do you?

Has anyone looked at poll trends or does that not have meaning in polls that have been tied within a margin of error for so long? What I mean by trends is like measuring the change in repeating polls and extrapolating to election day. It's conceivable that someone could be ahead in a poll but dropping and the projection being such that on election day he is behind.

The reason I brought this up is once again that electoral vote poll I mentioned yesterday. Thanks to those who responded. The key point I took away from the responses was that this electoral site
http://www.electoral-vote.com/ can swing a state to Bush when his lead is only within the margin of error and so in that sense can be misleading.

However, the site does have a menu with a "projected final map" option. Presumably this option analyzes trends and extrapolates.

The current electoral vote poll is Bush 285/Kerry 232 but the projected value for election day value is Bush 350/Kerry 178. This is telling me that the poll's movement could be as important as the poll's value.