You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
In function descr(), the statistics can be divided into 5 types according to units of measurement and scales:
Mean, Std.Dev, Min, Q1, Median, Q3, Max, MAD, IQR are measured on the same scale as an input variable (rounding could be provided for each variable separately);
CV has relative units of measurement and its scale depends on the values of mean and SD;
Skewness, SE.Skewness, and Kurtosis are coefficients, centered around 0. So 2 decimal places are usually sufficient.
N.Valid is always an integer. No decimal places should be displayed.
Pct.Valid is measured on the scale of percentages so 0 to 2 decimal places are usually sufficient.
Could these groups of measures have different default rounding options?
The text was updated successfully, but these errors were encountered:
For the ascii tables displayed in the console, this currently is not possible, as the number of digits shown depends on a pander argument that applies to the table as a whole. For html results, the N.Valid value is already shown without decimals (unless sampling weights are used). Specifying variable-specific rounding is feasible, although would require some work. I have a pretty busy schedule so I can't promise anything... But I'll leave the issue open and if you or others want to work on this feature, by all means do so!
In function
descr()
, the statistics can be divided into 5 types according to units of measurement and scales:Could these groups of measures have different default rounding options?
The text was updated successfully, but these errors were encountered: