Quote:
|
988 households surveyed (according to the article Chewbacca linked to) is actually not that many in a country of however many million people.
|
True enough Aerich, but I'll add something I learned from a social science stats class (very amateur I know lol). It's only 12 away from 1000, which is an acceptable number of individuals assuming a reasonably random sampling OR if they're proportioned by the percentages reflected in the population (city vs village, Sunni, Shi'i, and Kurd, man and woman, young and old) AND randomized within those groups. You aren't likely to fall victim to simple mathmatical anomalies assuming that your sample is as random as you hope. That being said, it probably falls into a 5% margin of error, as so much statistics do. By the way, I've often seen samples three times the size used for the whole of the United States. Practically, it can take researchers a lot of time to compile all of those answers and run the tests depending on how many variables (questions and personal info usually) there are.
However, one study is never enough to conclude anything by. There's a lot of room for error in statistics, so it's important to do more studies.
[ 10-30-2004, 02:54 PM: Message edited by: Lucern ]