By default gretl prints regression coefficients and standard errors
to 6 significant figures, and also by default the "summary" command
prints a large array of descriptive statistics -- too large to fit
all the stats for a given series on a single line -- using 5
significant figures.
Nothing wrong with that, but for presentation purposes one sometimes
wants to cut down on the number of figures (and/or statistics)
shown. And it's a bit of a pain to have to round numerical output
manually. There are now some mechanisms on place to make this
easier.
* For scripting use there's a new "set" variable, "gretl_digits".
This controls the number of significant figures in regression output
(including LaTeX and RTF), with a minimum of 3 and maximum of 6.
It also governs the figures shown by "summary", though in this case
we impose a maximum of 5.
* I've revised the output shown by "summary" with the --simple
option to include the median. With both mean and median shown one
can infer skewness (roughly), making the simple printout more
useable. In this case we now have a hard-wired precision of 4
significant digits (and each series fits on a single line).
* In the GUI, the right-click popup menu in the model window now has
a "Digits..." entry which lets you adjust the precision on the fly.
This also governs the precision used by Copy (as plain text, RTF
or LaTeX).
* Also in the GUI, the "Summary statistics" menu item (when multiple
series are selected) now offers the choice of "Main statistics"
(corresponding to the --simple option) or "Full statistics".
Allin