Natural Cubic  Spline                                           [snp7: STB-22]
---------------------

   ^spline^ yvar xvar [[^if^ exp] [^in^ range] [^, n^knots^(^#^)^ ^kn^ots^(^#,...,#^)^
	^logit^ ^reg^ress^(^string^)^ ^gen(^newvar^)^ ^nog^raph graph_options ]


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

^spline^ fits a natural cubic spline in xvar to yvar and, by default, plots
the original data together with the fitted smooth curve.  The regression
can be carried out with any estimation command, the default being ^regress^
(ordinary least squares).  ^spline^ optionally saves the smoothed variable.  

Options
-------

^knots^ specifies the exact location of the interior knots.

^nknots^ specifies the number of interior knots. (Ignored if the locations
	are specified).

^regress(^estimate^)^ fits the cubic spline using the command "estimate".
	The default is ^regress^. Use ^sp_adj^ instead of ^spline^ for
	^blogit^, ^bprobit^, ^clogit^, ^glogit^, or ^gprobit^  or ^cox^ with 
        ^dead^.

^logit^ transforms the fitted values and plots the y-axis on a logit scale.
	The observations are automatically jittered in the vertical scale and
	are plotted just outside the range of the smoothed curve.

^nograph^ suppresses the graph. 

^gen(^newvar^)^ creates newvar containing the smoothed values of yvar.

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


Remarks
-------

The natural cubic spline has knots at the ends of xvar as well at the interior
sites.  It is constrained to be linear beyond the the extreme knots and to be
everywhere twice continuously differentiable.  By default the interior knots 
are placed at equally spaced quantiles.


Author
------

        Peter Sasieni
        Imperial Cancer Research Fund, London
        FAX 011-44-71-269-3429


Also see
--------

 Manual:  [4] estimate
    STB:  snp7 (STB-22)
On-line:  ^help^ for ^graph^, ^spbase^, and ^sp_adj^