Surveys

Why do surveys ask such small numbers of people issues, then apply that to the population at large.

You see things that say "88% of people want an R Rating for videogames" or "76% of people think Rudd is a better leader than Nelson" but when you look at the numbers, it's always between 1000 and 5000 people.

This isn't even 1% of the entire population of Australia!

I've read "scientific" papers where they determine behavior based on the study of little more than 100 people... (assuming, of course, one can accept psychology as a science... which after studying it for 2 years I can't)

I just don't understand it. How can these people claim their research is legitimate if it uses such a small population spread?

Furthermore, why do people keep using such pathetic numbers in their research. Surely, with computing power and networks being the way they are, we can get larger sample rates.

If it's a random sample of the population, and the number of participants is statistically significant then the results are supposed to be reliable and valid. But the thing I never understood was that statistical significance is not related to the percentage of the population that participated (or at least that's what my lecturer's told me). I was told you need at least 49 participants to produce statistically significant results regardless of the population!!!

I read some document where a sample of the population were surveyed, and 90% of them thought that 95% of statistics are completely useless.

Just did the HQ survey

 

Freeeeee drinks blakey August 15th make me a VIP beatch!

97.8% of all statistics are made up, with 76.3% of these made up statistics being made up on the spot. 84.2% of all people know that.

90% of people enjoy gang rape.

just as i have been told with the tv ratings system, i thought there was a demographic snapshot profile that has been assessed/concocted by say statistics australia from the census or something, that was a 'valid representation' of the spectrum of australian society?


please tell me i am a naive idiot and there is no such thing...

cant say ive studies statistics as a subject, but im sure there are some issues with control sizes, variations and extrapolations.


yeah cos the internet is always true