Joe. Posted 3 September 2007 Posted 3 September 2007 :laugh: Somebody actually carries out these tests?
AyewJoking Posted 3 September 2007 Posted 3 September 2007 So 1788 people speak for the rest of Britain?
BigGibbo Posted 3 September 2007 Posted 3 September 2007 So 1788 people speak for the rest of Britain? jees thats good for statistics some beauty product thats being advertised on t.v at the min was saying works in 94% of cases & i noticed in small print they only tested it on 90 odd women & somehow its legal to pass that percentage off as a fact
Koke Posted 3 September 2007 Author Posted 3 September 2007 So 1788 people speak for the rest of Britain? Obviously its a load of tosh but it was jut a cheap shot at the inbreds down the road.
AyewJoking Posted 3 September 2007 Posted 3 September 2007 Obviously its a load of tosh but it was jut a cheap shot at the inbreds down the road. Wasnt having a go at you NJ, just thought the survey was stupid.
Geo V Posted 3 September 2007 Posted 3 September 2007 Wasnt having a go at you NJ, just thought the survey was stupid. I cant believe you had a go mate, thats just disgraceful. Say sorry
lookwhaticando Posted 4 September 2007 Posted 4 September 2007 In much the same way government approval ratings are said to be representative of public opinion when only 1,001 people (sometimes fewer) are asked for an opinion. A sample of 1788 is actually quite high for such a survey... often surveys are conducted with many many fewer respondents - particularly in government related surveys, oddly enough. That's the way statistics works. You decide how much error you're willing to tolerate (1%, 2%, 5% or something in-between?) and then work out how large your sample needs to be - making a few assumptions about the distribution of the population - to achieve such an error level. They don't just pick 1788 people and hope for the best... there is a process to ensure the numbers are meaningful. There is a level of error in the data that was considered from the start... although that wasn't mentioned in the article. Error of up to 5% is typical (particularly in those government/opposition opinion polls)... 2 or 3% is also quite common (and where I suspect this survey was aimed).
Rincewind Posted 4 September 2007 Posted 4 September 2007 It's like picking one represensitive from the UN for each country. They vote to go to war so it is statistically assumed that the whole of GB's public agree. In Iraqs case the error margin was a little out I believe. This is how you receive all the junk mail in the post. By looking at the surveys the advertisers estimate which group gets which item of junk mail.
Recommended Posts
Archived
This topic is now archived and is closed to further replies.