Several pollsters, such as CNN/Time (PDF), Marist, and Gallup, have been releasing their poll results showing totals for both all registered voters and “likely voters.” The difference between how Democrats are performing this year with all registered voters and how they are doing with the group these pollsters believe is most likely to vote is striking. Democrats are often polling 10 points worse–or more–with likely voters.

Design of the likely voter model has a huge impact on polling results. Normally, in a midterm election, only about 40 percent of those eligible to vote actually do, so determining who is going to be in that 40 percent is critical. In theory, even in the worst political environment, with a “perfect” turnout operation, one party could still score a massive victory if they could get every single potential supporter in that non-voting 60 percent to turnout. Of course, that is an impossible task, one neither party has ever accomplished, but it shows how important turnout is to American elections.

Recently, Gallup and Mark Blumenthal at Huffington Post have tried to examine the issue. For people interested in this subject, I highly recommend Mark Blumenthal’s whole article on the many ways pollsters determine who is likely to vote. The short answer is that there is no one correct method; it is part art and part science.

Some pollsters use a few of these methods, some use nearly all, so there are probably as many likely voter “models” out there as there are pollsters.

The tools that pollsters apply also reflects a philosophical difference I have written about before: Most national media pollsters, especially those that apply the classic Gallup index-and-cutoff model, are wary of making a priori judgments about the demographics or attitudes of the likely electorate. They prefer to set up a theoretically-objective mechanism to determine what a likely voter is and trust it to determine the demographics and partisan balance of their sample.

On the other hand, most (but not all) campaign pollsters — those who (like me in a former life) conduct surveys on behalf of political campaigns — have grown comfortable about drawing on data from other sources and applying their own judgment about the demographics and even the party identification of the likely electorate.

All of these factors help explain why different polls that all theoretically measure the same “likely” electorate can produce very different results.

Gallup recently explained how their method for determining likely voters has resulted in more favorable numbers for Republicans. It is mainly because they are projecting self-described conservatives will make up a historically large percentage of the overall electorate. At this point, Gallup estimates 54 percent of voters will be conservatives. In comparison, in 2002 and 2006, only 42 percent of voters were conservatives. From Gallup:

Likely voters skew more conservative this year partly because the underlying population has become slightly more conservative. According to Gallup’s Sept. 23-Oct. 3 poll, 40% of national adults are conservative, up from 37% in Gallup’s final 2006 pre-election survey, and 34% in 1994. However, conservatives also appear more activated to vote this year relative to moderates and liberals, thus sharply expanding their segment of the likely voter pie.

Of course, in the end, a vote cast by a very engaged individual counts same as a vote from a relatively disinterested. How much the enthusiasm gap will translate into a turnout gap and how successfully the Democrats’ field operation can drag disengaged supporters to the polls are the two, big, unanswered questions on which the accuracy of the poll results will depend.