If you want a representative research for a 1000 customers, how many of them should you interview?

Question

That is if you want a pretty high per cent accuracy,like 90%

Hm, are you positive? 10 % seems too little and 30% seems too much. And what per cent of accuracy/ liabilaty does it provide?

First of all, thank you for taking time to answer my question.
By "90% accuracy" I mean that this should be the chance of the sample accurately representing the general population.

Answer 1

say 10% i.e 100 customers

Answer 2

What I think you are asking is how do you sample 1,000 customers in your market base to be reasonably assured your sample represents the population (the 1,000).

First you have to select your sample. That is typically a simple random selection if other factors (like groupings) do not indicate otherwise. Number your customers 1 through 1,000; then run a random number generator (Excel has one) for a uniform distribution; so that all 1,000 have an equal opportunity to be selected.

Then you have to designate a sample size. The larger the sample size, the more likely your results will accurately indicate the population you are sampling. As some others have said, 30 is probably an OK sample size because sampling errors get really small at that threshold. But if you have the time and money, a larger sampling would improve your probability that the sample truly represents the population from which that sample was extracted.

The next step is rather complex...it's called hypothesis testing. This is just a fancy way of saying you are going to validate your sampling results. If you feel that is necessary I suggest you read over an intermediate statistics book.

Answer 3

First you have to ask yourself about your customers.

Are all customers the same? Or do you have 'big', 'medium' and 'little' customers. In which case you would want to do a stratified sample.

How about you define what you mean by 90% accuracy before we answer further.

Answer 4

It is usually admitted that 30 is a representative sample of any larger population, provided that the sample is truly random.