Computes,
under Gaussian item assumption:
point estimates, analytically; and confidence intervals,
by Monte Carlo simulation.
The data are given from:
a table (such as above), or
given sample sizes (then, with simulated weights).
In the case of "given", the sample weights
(sums) are randomly generated, and
μ and σ are "recovered".
The case of equal sample sizes (simply obtained with
'given sample sizes' all equal) validates the method, since,
by simulation, it leads to corresponding Student's and chi-squared
distributions, as expected.
Other suggested data:
(a) table20.txt,
giving (μ, σ) = (#, #); or
(b) 14 16 27 13 24 10 25 25 46
(copy-paste to the textarea),
giving μ in (24.93, 25.02), σ in (0.14, 0.29); or
(c) 6 6 6 6 6 6 6 6 6 6 6 6 6 6 6,
giving (24.97, 25.02), (0.098, 0.278).
Draws plots of the behavior of locations and
variabilities. |
