You are viewing an obsolete version of the DU website which is no longer supported by the Administrators. Visit The New DU.
Democratic Underground Latest Greatest Lobby Journals Search Options Help Login
Google

Reply #34: And it really is high time [View All]

Printer-friendly format Printer-friendly format
Printer-friendly format Email this thread to a friend
Printer-friendly format Bookmark this thread
This topic is archived.
Home » Discuss » Topic Forums » Election Reform Donate to DU
Febble Donating Member (1000+ posts) Send PM | Profile | Ignore Tue Nov-22-05 04:57 PM
Response to Reply #28
34. And it really is high time
that TIA found out what the "cluster effect" really is, as he seems to regard it as some kind of peanut butter you spread on an MoE to make it a bit bigger.

It has to be calculated quite precisely, depending on the kind of "clustering" is being compensated for, but it is, in essence, a way of compensating for the underestimate of variance you will get if, instead of giving all voters an equal chance of participating in a poll, you first of all select a number of precincts, and then select voters from within those precincts. Once you have selected your precincts, some voters will have no chance of being selected for the poll, whereas voters on the selected precincts may have a 1 in 10 chance, or higher. Moreover, two voters from the same precinct will tend to have more in common than two voters from different precincts. Thus "clustered" sampling will tend to result in less variance than a true random sample. You therefore need to increase your variance estimate to compensate, by a formula known as the DESR (Design Effect Square Root) which is the square root of the ratio of the variance of the estimate to the variance obtained by assuming the sample is a simple random sample.

More information here:

http://www.spss.com/complex_samples/data_analysis.htm

It's not peanut butter.

And it is also high time that TIA learned that polls have two broad category of error: Sampling error and non-sampling error. All his statistical inferences completely ignore non-sampling error.

Non-sampling error is a huge problem in any survey, precisely because it cannot be easily calculated. For TIA to continue to propagate analyses that blithely state that:

"All things being equal, polling accuracy is directly related to sample size. The larger the sample, the smaller the margin of error" (MoE)."

is, frankly, no longer excusable. All things are NOT equal, by any stretch of the imagination, and TIA knows this (or if he doesn't he needs to learn it). Ignoring all error except for sampling error may work in casinos, but that's about it. It doesn't even work in engineering. It certainly doesn't work in surveys.


Printer Friendly | Permalink |  | Top
 

Home » Discuss » Topic Forums » Election Reform Donate to DU

Powered by DCForum+ Version 1.1 Copyright 1997-2002 DCScripts.com
Software has been extensively modified by the DU administrators


Important Notices: By participating on this discussion board, visitors agree to abide by the rules outlined on our Rules page. Messages posted on the Democratic Underground Discussion Forums are the opinions of the individuals who post them, and do not necessarily represent the opinions of Democratic Underground, LLC.

Home  |  Discussion Forums  |  Journals |  Store  |  Donate

About DU  |  Contact Us  |  Privacy Policy

Got a message for Democratic Underground? Click here to send us a message.

© 2001 - 2011 Democratic Underground, LLC