“…people cross party lines a lot, yet you say they don’t.”

I have stated a number of times that party loyalty is between 85% and 95%. Thus, 15% to 5% cross lines. If you characterize what I’m saying as people don’t cross party lines and to counter that you say they do a lot, then you’re obviously saying it is much more than 15%.

only about 50% of voters are registered with a party

TD – http://www.elections.state.ny.us/NYSBOE/enrollment/county/county_nov07.pdf

says 120,479 of 677,006 aren’t registered with a party. That’s 17.8%.

R – well beyond my claim of 15%ish?

TD – Where did I say they had to go well beyond 15%? I objected to the “pigeonholing”, not any specific percentages.

Half of Erie county are registered Dem, if all of Keane’s votes were Dem then that means 14% voted for Collins to make up his 64%, if the people who showed up to vote were representative of the whole, and if every non-Dem voted for Collins. Heavily Democratic Erie County has elected 6 County Executives, 5 of them Republican.

I need to go to bed now, I have to get up early to go to my job which DOESN’T involve haranguing innocent people with life-stealing telephone polls.

Now, onto the actual study. Let’s look at Collins last year because that’s one where a good number of voters, namely Dems, had to cross lines. Let’s be very generous and say that 1/3 of all Dems crossed party lines and voted for Collins. (If that’s actually true, the Dem party in this area would be in great trouble.) That’s an exageration for our example, his numbers weren’t quite that good, but we’ll run with it. If Dems have a 2-to-1 edge in our county (it’s not quite that large, but it’s close) and only about 50% of voters are registered with a party that means that only about 8% of the voters crossed lines in this race. The number of Republicans crossing lines to vote for Keane had to be miniscule if at all because if Keane couldn’t keep party registrants that would be more amicable to his message he’d have a hard time convincing registrants of the other party to cross over to him and the election results play that out. So, in this election alone, about 8% of registered voters crossed lines. I was saying it was somewhere between 5% and 15%. You objected, but this is still in line with that. And this is the best case scenario in recent Erie County elections.

Now, consider that there were other elections last year, seats for town boards and county legislature were up. Most of those, if not all, were pretty much decided along party lines. Look at BP’s run. His vote percentage was almost exactly the percentage of Dems in his district. So, if you factor in 4 more races all decided along party lines with the results of Collins’ race, that 8% of voters only crossed 20% of the time. That means we’re down to less than 2% party crossing in the last election in our county.

Sure, there may have been a little more party crossing going on in those other races, but even if we had 1/3 of all party registered voters crossing lines in each election, an unheard of amount by any standard in a developed democracy, you’re still only talking about 15% of the electorate crossing lines. Again, within my 5% – 15% range which comes from the technique you despise but don’t understand.

This is not a discussion about politics. It has become an academic discussion about research methods in studying politics, but I guess you are so unqualified you can’t even recognize that.

Party identification and party registration are two entirely different things. Party registration is not universal but party identity is. If you were concerned with party registration, you couldn’t conduct your poll in at least 15 states in the US let alone most countries where only people active in politics are considered members of a party. Party identification works because it can be used anywhere and is not constrained by local laws. Party registration rules are different in every state in the US and in every country. Many states in the US do not even have party registration or enrollments.

It is imperative that your research method fits the case you are studying. There are questions that only matter in a two-party system, some that only matter in a multi-party system, and some only work in a three-party system. You wouldn’t ask the same questions in a parliamentary system as you would in a presidential system. Many of your questions would be pointless.

Besides, Timothy, your opinion is completely pointless in this regard. Those who actually do this work day in and day out have developed this method, use this method and respect this method. A rank amateur with no credentials or background knowledge of this is not going to change this practice, no matter how many Wikipedia articles you might find, even if those articles do pull single lines here and there from much much larger bodies of work and areas of study. Wikipedia does not make you an expert and does not make up for years of advanced study, application and experience.