Replies

Most of the difference is probably how people are counted. If you count everyone you survey, you'll get a mix of people who will vote, people who won't, and people who aren't even registered. So you restrict it to registered voters, but even that will get substantial numbers of people who won't vote.

So you try and make statistical adjustments for who actually votes and who doesn't. If my memory serves, the LA Times poll was "likely voters". This simply means they ask people:

Are you going to vote?

and tag "Yes" and "I think so" as "likely". The bulk of Democratic voters are less likely to vote than the bulk of Republicans, so this will bias you slightly towards Democrats.

The ABC poll is "certain" voters, which tend to go Republican. Of course in the actual election, we simply don't know who will go out to vote until it happens. This election may have high turnout thanks to the high interest in it; it may have more Latinos voting because they are keen on having "one of their own" as governor. Or it may have more fans of Arnold's movies vote, in which case Arnold will get a lot of support the polls won't necessarily show. Because this is an unprecedented election, we really have no clue how it's going to play out, and as a result polling companies have a wide latitude to indulge in their own theorizing.

So the ABC poll has Arnold winning, and the LA Times poll has Bustamante winning. These different results most likely satisfy different agendas and interest groups within news organizations, tempered by the professional judgement of those working for the companies.

It may be more sensible to look at trends. For example, the last SurveyUSA poll (same company as ran the ABC one) had Arnold at 51%. Now he's at 45%. So his support has dropped, and Bustamante has risen from 17% to 29% (!).

What it looks like to me is that there is a middle ground vote shifting between Arnold and Bustamante, and people are coalescing around the three major candidates. Since Bustamante has become more visible by announcing his program, it seems natural that his support has increased. Most likely, this reflects people who are his natural supporters who had not yet heard much from him before. However, I think he has hit a limit because his program is bound to be unpopular with anyone with a car worth over $20,000. There are plenty of non-rich people who have cars over $20k. I don't think they'll appreciate having to bear the Bustamante burden.

So we will see.

D