Men talk as much as Women? Options

[impala454]

http://www.chron.com/disp/story.mpl/front/4945688.html

A University of Texas at Austin psychologist and his colleagues appear to have laid to rest the long-standing myth that women talk more than men. Their study of 400 male and female students in the United States and Mexico found that males and females both say about 16,000 words a day, or 17 words a minute during waking hours.

another great statistic that gets labeled as 'research'.

400 students
recorded 30 seconds out of every 12.5 minutes = 57.6 total minutes recorded/day
each person was recorded for "up to" 10 days.

and they claim they have debunked the myth in our society that women talk more than men, based on recording 4% of 400 student's days for "up to" 10 days.

the things that get labeled as geniune research and rock solid results just baffle me nowadays.

[chook]

Its how you handle the numbers. In a small population you cannot take a small selection, because there is a large chance that it won't represent the whole population.
Lets assume a population where half the people believe that Impala is wrong and half where they think Test is wrong. Population size won't matter, but how many we take will. And assuming that its evenly distributed. If I took this survey in the art department, I'd get skewed results than at the BA. So lets say I asked 3 people, based on the truth here is what they could say.
This is based on combinations of people responding
Who is full of shit?
3 impala 0 cat (.5)^3 = .125
2 impala 1 cat 2!(.5)^3 + (.5)^3 = .375
1 impala 2 cat (.5)^3 + 2!(.5)^3 = .375
0 impala 3 cat (.5)^3 = .125

If the sample size increases, then it is more likely that the reported results match the population. In other words, as more people are asked, it is less likely that the results are reflecting the truth. Statistical methods such as chi^2 and variance from the norm are based on solid math. Biggest problem is how these issues are dealt with.

Impala I am ashamed of you calling statistics just numbers. This is much more defined than computer science.
Cat, the problem is with people doing research methods.