95% confidence interval of the mean of a populationsample

Hi,

Have not been able to locate an answer on this forum to this (seemingly, but for me unsolvable) simple question: how to calculate a 95% confidence interval of the mean of a populationsample, i.e. what is the absolute range of values that fall within 95% of all values.

Jan

Answer 1

[Oleg Komarov]

If the sample follows a particular distribution and you have the stats TB, then you can estimate the parameters with the confidence intervals you want.

Give a look at normfit

Oleg

Answer 2

[Germán]

Another good option, if you have a reasonably "large" data set and dont know the underlying distribution, is to use bootstrap. The function is bootci. Personally, I favor the bca confidence interval because it works on non-normal cases; unimodality is still a requisite though.

Bootstrap is a resampling statistic, it takes your data as an accurate depiction of the population and samples it (with replacement) to estimate the distribution of a particular statistic you are interested in, e.g. the mean.

[Germán]

Note: since it uses resampling it will vary slightly from call to call (depending on the number of samples; more samples, less variation), so you want to aim for a large number of samples. n >= 1000.

[Jan]

Useful advice, but my dataset is not as large as that. Thx.