Monday, October 30, 2006

More Nataf Follies

To follow up on yesterday's post about Dan Nataf's poll, there are other problems that need to be pointed out:

- EVERY question identified the candidate's party affiliation. This has also been a problem with some of the Sun's polls in the recent past.

- Question 12 also presents a bigger problem:
12. With which political party, if any, are you registered?
Unweighted Weighted
(1) Democratic 42 41
(2) Republican 38 47
(3) Independent 11 12
(5) Other (e.g., Green Party) 1 1
(0) No Answer 5 0
This goes back to the point yersterday of using random-digit dialing as a polling sample. The ENTIRE partisan breakdown is self-selected, not selected using any scientific polling methods. How can truly know if this number is accurate? We cannot, which defeats the entire purpose of doing the poll in the first place if you know that the information you get is statistical garbage.

What's sad is that part of the CLSI mission is to teach students statistical research methods, which really gives me pause as to what these students are actually being taught.

As my wife noted, "Nataf screwed this poll up nine ways to Sunday."


Post a Comment

<< Home

Site Feed