C[omp]ute

From: "andrew cooke" <andrew@...>

Date: Sun, 31 Dec 2006 14:55:49 -0300 (CLST)

Ipsos did a telephone poll for AP and AOL last week.  1000 people were
asked questions, which gives an error of about 3% (that's just the
stochastic error - sqrt(1000)/1000 ~ 30/1000).

The AP released 3 stories:

A neutral survey "Poll: How Different Groups Feel"
http://hosted.ap.org/dynamic/stories/O/OPTIMISM_POLL_GLANCE

A negative summary "Poll: Americans See Gloom, Doom in 2007"
http://hosted.ap.org/dynamic/stories/2/2007_PREDICTIONS_AP_POLL

And a positive summary "AP Poll: Americans Optimistic for 2007"
http://hosted.ap.org/dynamic/stories/O/OPTIMISM_AP_POLL

Presumably the idea is to get the best value for money possible - each
media outlet can purchase the article that best reflects the message they
are trying to send to their readers.  So it is interesting to see which
stories are used.

Searching on news.google.com for the following (quoted) text, I found:

"Democrats were much more likely than Republicans to have a negative view
of the past year." (neutral) - 1/25

"Another terrorist attack, a warmer planet, death and destruction from a
natural disaster." (negative) - 1/103

"The news from Iraq and other national headlines may be grim, but in
Greenville, N.C., John Given has a new baby and his first home, and life
is good." (positive) - 2/100

When two numbers are given, the second is without "similar results excluded".

Unfortunately, there may systematic bias in the above - the stories were
released at different times:
 neutral (1/25): in last 24 hours
 negative (1/103): in last 10 hours
 positive (2/100): in last 24 hours

Looking at individual newspapers (searching for "poll" in last few days):
  IHT: -ve (slightly edited)
  Wash Post: all three
  NY Times: all three
  NY Post: +ve (heavily edited)

Oh, and the Jesus thing?  1/4 of surveyed americans predict his return in
2007...

Andrew