In statistical analysis the usefulness of confidence
intervals is undeniable. Taking into account the uncertainty
associated with samples, the estimation of the population
parameters can be performed by confidence intervals. In other
words, as long as there is randomness and finite sample size
(which define the reality) estimation of statistical indicators will
be done by confidence intervals. Inspired by the methods used
for proportions (frequency of an event), we propose and study a
new confidence interval for the mean of a continuous random
variable. The main idea comes from the Wilson interval and we
shape a similar formula for computing the mean bounds.
Numerical simulation results show advantages but also
limitations of the newly proposed method.