General framework
A similar approach to that for estimating a population mean is used for estimating population proportions and other parameters.
- Point estimate
- A population parameter is usually estimated with the corresponding sample statistic.
- Error distribution
- The
estimation error has a distribution.
- Standard error and bias
- A good estimator usually has an error distribution whose mean is zero
(unbiased). The standard deviation of the error distribution is called the
standard error.
- From data:
- We can usually find a numerical value for the standard error (or an approximation
to it).
- Confidence interval
- A 95% confidence interval can be found from
estimate - 2 s.e. to estimate + 2 s.e.
- (The interval can sometimes be improved by replacing 2 by another similar
value. )