.-
help for ^qchi^                                                 (STB-29: sg47)
.-

Chi-square Q-Q plot
-------------------

    ^qchi^ var [^if^ exp] [^in^ range] ^,^ ^d^f^(^#^)^ ^tra^nsform [ graph_options ]


Description
-----------

^qchi^ produces a chi-square quantile-quantile (Q-Q) plot of var.


Options
-------

^df()^ is not optional, and specifies the degrees of freedom for the chi-square
    variable

^transform^ transforms the values on both axes to their cube roots, which
    transforms a chi-square variable to approximate Normality.  Since the
    straggly upper tail is suppressed, the graph may be easier to interpret.

graph_options are any of the options allowed with ^graph, twoway^; see help
    @graph@


Remarks
-------

If var has an approximate chi-square distribution with ^df()^ degrees of freedom,
the resulting plot will be roughly linear, with a greater concentration of
points in the lower left quadrant of the diagram than in the upper right. The
values in the upper right quadrant will be very variable and may not appear to
lie on the `ideal' line even if the var really does follow a chi-square dist-
ribution.

The slope in the lower left quadrant indicates whether the ^df()^ are about
right or not.  If ^df()^ is too low, the slope will be >1 and the points will
form a curve above the line.  If ^df()^ is too high, the slope will be <1 and
the points will mostly fall below the line.


Examples
--------

 . ^set obs 200^
 . ^gen chi = invnorm(uniform())^^2^
 . ^qchi chi, df(1) xlabel ylabel^


Author
------

        Patrick Royston
        Royal Postgraduate Medical School, London
        FAX:  (011)-44-181-740-3119


Also see
--------

    STB:  STB-29  sg47
On-line:  help for @diagplots@