Monday, January 10, 2005

Tolles Civil Liberties Redux (Part One)


By far my favorite response that you have posted. I think you hit several issues well here. I'll clarify my end a little ordering my responses to your comments. Click read more.

1. Not sure if you mean "mucking" or "making" up facts. I think I'd plead guilty to "mucking" a little. Seven parts and I still didn't explain a lot of stuff very well. Nothing here is "made" up in any sense, I think.

2. Fundamentally: I do think that Republicans/conservatives on the whole are more intolerant and less supportive of an important set of civil liberties. The purpose of the post isn't to defend on party or ideological grounds. There is a defense of country aspect to this. I think to see these findings and feel disappointed in your fellow citizens is a reasonable response. To see them and fear the rise of the fourth Reich is a little too much.

3. I didn't say the survey responses meant nothing; I said "knowing that Republicans are more likely than Democrats to curtail civil liberties tells us nothing." I gave what I thought was a reasonable question: by how much? Let me try it this way: imagine you move to a new town where you know nothing at all about the schools. You ask a real estate agent which of two schools would be a better choice for your kids. She tells you: "Kids at Benjamin Franklin score higher on standardized tests than kids at George Washington". What has she really told you? Anything? What if kids at BF score in the fifth percentile nationally while kids at GW score in the third percentile? Technically, she is right, BF kids score higher. Is that meaningful data or a complete dodge? Both schools totally suck. If you read the Yahoo news report you linked to and don't look at the technical report, it is easy to work yourself into high dudgeon about how bad republicans are. Absent thought about the actual numbers, the news item is simply an invitation to prejudice.

4. Since we have both now seen the report, we can seriously discuss the findings. Let me begin by clearing the air again. There are problems with the weightings and a lot of the questions. On the question you cite with the 40/24 split, I'll say a couple of things. First, I do not think the weighting problems account for this difference. I think it is stunning how high the number are both in aggregate and split out by party. In other words, I would not have guessed that the Republican numbers would be that high, that the Democrats numbers would be that high, or that the overall numbers would be that high. On this question, I think your initial impression is right.

5. Since we are not hiding from the data, it should be pointed out that the split is not 40/24. It is 40/24/17 with independents making up the third group. The 40/24 difference is almost certainly significant. The 24/17 could be, but who knows? It is interesting that the independents score so much more low on this than both the Ds and Rs. Not the case with other questions here.

6. Going backwards a little. Interesting that you complain about my complaining about weighting on the very day that this appears: As late as 7:33 P.M. on Election Day, Mitofsky and Lenski were apparently telling their clients (NBC, CBS, CNN, AP, etc.) that after "weighting" Kerry was beating Bush by 9 points among women and losing by only 4 among men. By 1:24 P.M. the next day (see this file) revised results revealed that, in fact, Kerry won women by only 3 points while Bush won men by 11 points.
So weighting gave Kerry approximately 6.5 million more votes than he actually got if we extrapolate the numbers from sample to population. Weighting is not a trivial issue; it is in fact a very, very hot topic in the survey community. As a very technical response, weighting could, in fact, account for all the difference in the 40/24 split. I don't believe that, but you seem prepared to pick nits. To make it clear: the weighting of the exit polls amounted to 5% of the total of voters (as an aside, wasn't it these exit polls that fueled the fire that Bush had obviously stolen the election in the first place?). Taking 5% off the Rs and giving them to the Ds would make the split 35/29 which is inside the margin of error for the survey.


Post a Comment

<< Home

Listed on Blogwise