If Members of Congress voted on a timetable for withdrawing from Iraq based on whether more people in their districts favored or opposed such a timetable, it would pass the House by a margin of at least 418-17, and the Senate by a margin of 98-2. If only those Members voted yes who represent districts where at least 50% of the public supports such a timetable, it would still pass with a margin of at least 329-106 in the House and 78-22 in the Senate, a three-fourths majority in both chambers. In either case this would be far more than the margin needed to override a Presidential veto.
This conclusion is based on estimating opinion on withdrawal in 435 House districts and the 50 states by matching national poll data on a timetable for withdrawal, where the party affiliation of respondents is known (the Pew poll in March) with data on the partisan breakdown of individual districts (the 2004 Bush/Kerry result.) In other words, it assumes that Republicans in different parts of the country are equally likely to support a timetable for withdrawal, and the same for Democrats.
While this is surely not true, it is certainly true "on average" (by definition) and the results of this analysis are so lopsided that regional variation could not possibly affect the overall result that Congress would have a veto-proof majority for a timetable for withdrawal if Members voted their districts. The calculations are given in a spreadsheet here, where you can check the result for your own Congressional District.
Moreover, the simple technique used here surely overestimates the number of Republicans and therefore underestimates the support for a timetable for withdrawal. By using the 2004 Bush/Kerry result as a proxy for the partisan breakdown of the district, it assumes that everyone who voted for Bush was a Republican, which we know was not true; we also know that Bush voters were more likely to be "Democrats" than Kerry voters were to be "Republican." The Bush/Kerry result was 51% Bush, 48% Kerry (and 1% Nader.) Thus, the technique here assumes that 51% of Americans are Republicans. CNN's exit polling characterized people as 37% Republican (of whom 6% voted for Kerry) and 37% Democratic (of whom 11% voted for Bush) with the rest Independents, who split slightly for Kerry. Pew (counting Republican leaners as Republicans and Democratic leaners as Democrats) gives the partisan breakdown in 2004 as 47% Democratic, 41% Republican.
In addition, the country has trended Democratic since 2004. Pew now finds 50% Democrats and 35% Republicans. So the technique clearly overcounts Republicans and therefore undercounts withdrawal supporters. (Since we are tallying with the Pew poll, what is relevant here is not who is truly a "Republican" but who would have been counted by Pew as a Republican.)
Moreover, a CBS/New York Times poll taken in April found even greater support for a timetable for withdrawal (64%) than Pew found in March (59%.)
The results are driven by the following. According to the Pew poll, 77% of Democrats, 34% of Republicans, and 61% of Independents supported a timetable for withdrawal. If we assume that a district is made up of Democrats and Republicans (i.e. ignoring Independents, who favor withdrawal) the breakeven point for a plurality for withdrawal is roughly 70%/30% Republican - any district less Republican than that will have a plurality for withdrawal. Only 4% of the House districts in the country are that Republican. Many states have no such district. If we use the higher, 50% for withdrawal standard, the breakeven point is roughly 62%/38% Republican, and just under a quarter of the districts are Republican enough. Still there are large states, like Illinois, Michigan, and New York, that have no such House district.
From this analysis we can conclude that the majority of Republican Members of the House and Senate are simply not representing their districts on the question of a timetable for withdrawal from Iraq.
Opinion surveys which attempt to gauge Republican support for a withdrawal timetable tend to rely on self-identification. That is, the respondent is asked some variation of, "would you describe yourself as a Republican." The group of people who answer yes to this question are a smaller and "more Republican" group of people than people who are actually likely to vote Republican in elections. (For example, the pollster could ask, "how did you vote for President in 2004?") But presumably, it's the larger group of "Republicans" who Republican Members of Congress should be concerned about, because few of them could get elected to Congress based on the votes of the smaller group alone. Thus, pollsters, by relying only on self-identification, are under-reporting "Republican" support for a timetable for withdrawal.
It may well be the case that with Election 2008 more than a year away, many Republican Members of Congress are more concerned with the smaller group of self-identified Republicans, who they are more likely to hear from on a regular basis, and who would form the constituency for a right-wing primary challenge, than the larger group of Republican voters whose support they will need on general election day. But as general election 2008 looms closer, this will change, and news media should also turn their attention to this larger group.