Am I an idiot if I don't understand surveys? Even with a sample size of 1000 or 2000, I don't see how you can use that to represent an entire country's population.
eh, you usualy dont get the correct amount, but statisticaly you would get something close to what is correct.
>>29625552
2000 would be woefully small as a representation of an entire country
>>29625828
Nope. You only need 1500 responses for your survey to be respected on a national scale.
>>29625828
Even that the average size of people they use for political polls and put them on cnn and msnbc? I know fox tries to get about 50k people for theirs though.
>>29625552
It has to be random. The greater the sample size, the more likely it is to be representative. There are methods to compute the reliability of it (it depends on the nature of what you are measuring it too; it's not always the same, it's complicated). At some point it becomes so reliable that it is considered representative.
>>29625552
I failed statistics like six times.
It is completely counter-intuitive and broke my brain when I tried to think about it.
Let's say we're looking for favorite color.
If only 10% have a favorite of purple, than 1 in 10 people across the entire country would answer purple. So if you randomly select a single person, there is a 1 in 10 chance of them answering purple.
But we don't actually know what amount of people would say purple, that's what we want to know. But we do know that a real value exists and that every person has some chance of answering purple. So if you randomly selected 1000 people and 100 said purple, because of the fact that this population of people preferers occurs at some real, fixed percentage, and the fact that we drew randomly, if we poll a good amount of people than the percentage we obtain should be representative of the actual amount of people who pick purple in the entire country.
Jesus it's too late for this shit
>>29625838
To add on to that, you have to make sure the people you survey represent the whole population. Say you're surveying people in the US, you can't just survey 1500 niggers or 1500 hillbillies and expect good results.
>>29625838
This is only a guideline based on consensus
It does not represent the hard truth
Stop pretending like law and trends = factuality
>>29625838
Respected by who? I'd certainly take any data drawn from that small a sample size with a grain of salt.
