Smallest sample size.

[petrushka.googol]

What is the smallest sample size that we need for a statistical study to be representative of a real population? eg. if there is a population of 100 and my sample size is 10 then even if 8 out of 10 responses are positive what is the possibility that 80 out of 100 people would adhere to the same norms? Are percentages inaccurate in this case?

Please advise.

[studiot]

 

What is the smallest sample size that we need for a statistical study to be representative of a real population? eg. if there is a population of 100 and my sample size is 10 then even if 8 out of 10 responses are positive what is the possibility that 80 out of 100 people would adhere to the same norms? Are percentages inaccurate in this case?

 

You are dealing with statistics so there is no absolute answer as in other branches of maths.

Your answer will be of the form "my sample provides a ..X% chance that the value of interest for the population is...Y"

We call the X% the confidence level.

 

The first thing you need to know is the shape of the probability distribution curve.

You either assume this of have good reason to 'know' its equation.

There are around a dozen standard curves to choose from.

 

The next thing you need to know is the nature of what you are measuring.

Are you looking for an average value for the population (eg an average height)

Or are you looking test if a particular batch of something meets quality criteria. eg is a particular batch of concrete as strong as it should be.

 

The final thing you need is to decide what your confidence level will be.

The bigger your sample, the better your confidence in your measurement.

 

Please provide more detail of where you are coming from for a more detailed answer.

[CharonY]

The thing you want to look into is power analysis.