0

I have problem understanding following example in parentheses. Maybe you can help me.

Ideally, researchers know the full extent of the population they want to study, and they can select a sample from this population at random. Statisticians can calculate the probability that such random samples represent the population; this is usually expressed in terms of sampling error (for example, there might be a 95 percent probability that the distribution of responses in a sample will be within 3 percent of the distribution in the population).

Michael R. Chernick
  • 39,640
  • 28
  • 74
  • 143
Amir
  • 1
  • 2
    This is explained in various contexts, such as the thread on sampling errors for polls at http://stats.stackexchange.com/questions/16413/. You may also find that searching our site for [sampling error](http://stats.stackexchange.com/search?q=%22sampling+error%22) uncovers helpful explanations. – whuber Aug 13 '12 at 21:37

1 Answers1

7

The question in parentheses is incorrectly stated. It is not the distribution of response that can be said to be within x% of the population distribution with some probability, it is the estimate from the random sample that can be claimed to cover with an interval x% wide the true parameter value with a specified probability. The statement in bold in the parentheses is not understandable because it does not make sense as stated.

Michael R. Chernick
  • 39,640
  • 28
  • 74
  • 143
  • Thanks michael. I have studied something about statistics at university, but i;m not a statistics professional. however, this phrase is stated in a book around social science – Amir Aug 13 '12 at 23:16
  • 3
    +1, throw away this book, it uses misleading language. Technically speaking, the design probabilities ARE the true probabilities, so one does not have to hide behind a shaky "repeated sampling" frequentist argument to justify the confidence intervals. – StasK Aug 14 '12 at 20:46