| John Navas 2006-05-23, 5:48 pm |
| [POSTED TO alt.cellular.cingular - REPLY ON USENET PLEASE]
In < 446db089$0$96931$742
ec2ed@news.sonic.net> on Fri, 19 May 2006 04:48:36
-0700, SMS <scharf.steven@geemail.com> wrote:
>Scott wrote:
>
Only USA statisticians are credible? How droll!
[color=darkred]
>Look at the first line, "In surveys with low response rates." That's the
>key point that Navas is ignoring. Once you have a high response rate,
>non-response bias isn't an issue, unless you're doing some sort of
>survey where there is a reason that one group would be much less likely
>to respond. In terms of wireless, the surveys by Consumer Reports, JD
>Power, and others, aren't asking people "which carrier do you think is
>the best?" they are surveying individuals regarding their experience
>with their own carrier. Thus, it's not a valid complaint to claim that
>maybe the reason that Cingular always does so poorly is because only
>people that are unhappy with their wireless service choose to respond
>because as long as the response rate for each carrier, in each region,
>is sufficient, the results are valid. Occasionally CR will leave out
>ratings for a carrier in a specific region if they did not receive
>enough responses. No one takes Navas's criticism of the CR surveys
>seriously, it's simply sour grapes.
The issue of non-response bias pertains to ACSI, the subject of this thread.
"Stay on target, Luke!"
CU/CR suffers more from a different issue: non-random, self-seleted sample.
Finally, any survey that lumps different technologies together (D-AMPS and GSM
in the case of Cingular, CDMA and iDEN in the case of Sprint-Nextel) isn't a
meaningful measure of either, since it's roughly comparable to saying the
average person has one testicle.
That you don't acknowledge these real statistical issues makes your bias all
the more apparent.
--
Best regards, SEE THE FAQ FOR CINGULAR WIRELESS AT
John Navas <http://en.wikibooks.org/wiki/ Cingu...less_FA
Q>
|