Part of our promise here at Politics Nation is that you, our readers, will be able to find the latest polling data on any House, Senate or Gubernatorial race you happen to find interesting. There will be no shortage of them -- either races you're interested in or surveys from those hot spots -- in the coming months, but it is important to remember that not all polls are created equal.

It will be our goal to bring you the latest numbers from pollsters whose methodology is commonly accepted as scientifically sound. That means we will include numbers from respected partisan polling firms, of which there are many throughout the country.

Partisan polls, though, ought to be taken with a grain of salt. When polls from partisan sources become publicly available, they are released to make a point. Still, if they are published on Politics Nation, we are confident that they are reported accurately. That means a horse-race matchup between two or more candidates will be asked near the beginning of the survey, before positive or negative messages about one or more of the candidates are read.

Some pollsters offer "informed ballots," meaning respondents are read brief biographies of candidates and then asked for whom they will vote. It will be our policy not to publish informed ballot tests unless the full biographies for both candidates are made available. In those cases, we will run the biographies next to the informed ballot tests.

When it comes to non-partisan pollsters, there are three primary means of surveying respondents: Live interviews, interactive voice response (IVR, for short) and web-based surveys. Live interviews are widely considered the most accurate in terms of horse race numbers in the months and weeks leading up to an election. For one thing, monitoring survey quality is easiest with live interviews, and a live person talking to a respondent can verify that the person on the other end at least claims to be the voter pollsters are trying to reach.

In an examination this reporter made, for The Hotline, of pollsters who conducted surveys in key contests in 2006, live-calling pollsters came closest to the margin of victory in fourteen of fifteen races, with one tie between live and IVR pollsters.

IVR, on the other hand, offers less stable numbers. The process -- a message asking voters to press a number corresponding with a certain view or candidate -- is cheaper and more efficient, in terms of the amount of time it takes to complete a survey, than live interviews, but they don't always offer the same consistency: In several 2006 races, IVR pollsters showed dramatic swings in just a matter of days. Barring a disastrous performance, the utterance of some unfortunate slur or other collapse, voters don't turn on a candidate that quickly.

That being said, some IVR pollsters have shown remarkable accuracy this year, and it has been our experience that the last poll some firms do before Election Day can turn out to be the most accurate. In California this year, while some pollsters showed Barack Obama winning by wide margins, SurveyUSA, one IVR practitioner, predicted a ten-point win for Hillary Clinton. Clinton won by 9.6%. SurveyUSA also came closest to predicting outcomes in last year's Connecticut Senate race, while Rasmussen, another IVR pollster, predicted the Democratic win in Missouri's Senate contest within one point, tying live-calling firm Mason-Dixon.

The third methodology, like IVR, looks like an exciting new way to conduct polls. Interactive and internet polling shows promise for the future. But like IVR, interactive and internet polls are too uncontrolled; in essence, it is impossible to know who is a part of the sample. Too, they are subject to manipulation, as evidenced by Ron Paul's incredible showing in a number of post-debate polls on the Drudge Report and other sites. Politics Nation will not run any interactive polls this year.

A final note: Some pollsters will use methodology that may seem a little out of the ordinary. A recent poll from the University of Washington, for example, attracted several comments about the survey's accuracy, and rightfully so: The poll was conducted over an eleven-day time period, far longer than the three or four days it takes to complete most surveys, and respondents were a part of a panel -- that is, they are a part of the same pool of respondents who answered pollsters' questions in the UW's earlier poll. In cases wherein a question ought to be raised about a poll's methodology -- or at least something ought to be brought to readers' attention -- Politics Nation will make our best effort to do so, as we did with that particular survey.

This year will likely be the most polled political season in the history of modern polling. Each competitive race will get enough attention to be surveyed many times. But not all polls are created equal, and not all polls will get the same treatment from Politics Nation. We hope this is an effective guide explaining our own methodology.