You cannot select more than 25 topics
Topics must start with a letter or number, can include dashes ('-') and can be up to 35 characters long.
163 lines
5.6 KiB
Plaintext
163 lines
5.6 KiB
Plaintext
9 months ago
|
.-
|
||
|
help for ^praccum^ - 1.6.4 - 2Nov2005
|
||
|
.-
|
||
|
|
||
|
Accumulate results from ^prvalue^
|
||
|
-------------------------------
|
||
|
|
||
|
^praccum^, [^xis(^value^)^ ^u^sing^(^matrixnm^)^ ^s^aving^(^matrixnm^)^ ^gen^erate^(^rootname^)^]
|
||
|
|
||
|
where either saving() or using() are required.
|
||
|
|
||
|
Description
|
||
|
-----------
|
||
|
|
||
|
^praccum^ accumulates predictions from a series of calls to ^prvalue^ and
|
||
|
optionally saves these accumluated values to variables. These variables can
|
||
|
then be plotted. This command allows you to plot predicted values in
|
||
|
situations that cannot be handled by ^prgen^ (e.g., nonlinearities).
|
||
|
|
||
|
The command works with cloglog, cnreg, intreg, logit, mlogit, mprobit, nbreg,
|
||
|
ologit, oprobit, poisson, probit, regress, slogit, tobit, zinb, zip, ztnb,
|
||
|
and ztp.
|
||
|
|
||
|
|
||
|
Options
|
||
|
-------
|
||
|
|
||
|
^xis(^value^)^ specifies the value of the x-variable associated with the predicted
|
||
|
values that are being accumulated. If ^xis^ is not specified, new values
|
||
|
are not accumulated.
|
||
|
|
||
|
^using(^matrixnm^)^ specifies the name of matrix to which accumulated results
|
||
|
should be added. ^matrixnm^ will be created if it does not exist.
|
||
|
|
||
|
^saving(^matrixnm^)^: is only used to save the initial results and differs from
|
||
|
differs from ^using()^ in that it will overwrite ^matrixnm^ if it exists.
|
||
|
|
||
|
^generate(^rootname^)^: root name of variables to be created from the matrix
|
||
|
specified by ^using^. This is only used when you are done accumulating
|
||
|
results and are ready to generate the variables.
|
||
|
|
||
|
Examples of included squared terms
|
||
|
----------------------------------
|
||
|
|
||
|
Consider the logit:
|
||
|
|
||
|
^. use binlfp,clear^
|
||
|
^. gen age2 = age*age^
|
||
|
^. logit lfp k5 k618 age age2 wc hc lwg inc^
|
||
|
|
||
|
If you want to plot the predictions against age, you cannot use ^prgen^ since
|
||
|
when age changes, age2 must also change. The command:
|
||
|
|
||
|
^. prvalue , x(age=20 age2=400) rest(mean)^
|
||
|
|
||
|
computes predicted values for age=20 and age2=20*20=400. The command:
|
||
|
|
||
|
^. praccum , saving(mage) xis(20)^
|
||
|
|
||
|
creates a matrix named mage that contains three columns. The first column will
|
||
|
have a 20 for the value of age; the second the probability of a 0 given the
|
||
|
values of the independent variables used in ^prvalue^, and the third column
|
||
|
will have the probability of a 1. We now change the value of age and add this
|
||
|
to the matrix mage:
|
||
|
|
||
|
^. prvalue , x(age=25 age2=625) rest(mean)^
|
||
|
^. praccum , using(mage) xis(25)^
|
||
|
|
||
|
Here we are just adding a row to mage. This process repeats for other values:
|
||
|
|
||
|
^. prvalue , x(age=30 age2=900) rest(mean)^
|
||
|
^. praccum , using(mage) xis(30)^
|
||
|
^. prvalue , x(age=35 age2=1225) rest(mean)^
|
||
|
^. praccum , using(mage) xis(35)^
|
||
|
^. prvalue , x(age=40 age2=1600) rest(mean)^
|
||
|
^. praccum , using(mage) xis(40)^
|
||
|
^. prvalue , x(age=45 age2=2025) rest(mean)^
|
||
|
^. praccum , using(mage) xis(45)^
|
||
|
^. prvalue , x(age=50 age2=2500) rest(mean)^
|
||
|
^. praccum , using(mage) xis(50)^
|
||
|
^. prvalue , x(age=55 age2=3025) rest(mean)^
|
||
|
^. praccum , using(mage) xis(55)^
|
||
|
^. prvalue , x(age=60 age2=3600) rest(mean)^
|
||
|
^. praccum , using(mage) xis(60) gen(agsq)^
|
||
|
|
||
|
Produces the output:
|
||
|
|
||
|
^New variables created by praccum:^
|
||
|
|
||
|
^Variable | Obs Mean Std. Dev. Min Max^
|
||
|
^---------+-----------------------------------------------------^
|
||
|
^ agsqx | 9 40 13.69306 20 60^
|
||
|
^ agsqp0 | 9 .4282142 .1752595 .2676314 .7479599^
|
||
|
^ agsqp1 | 9 .5717858 .1752595 .2520402 .7323686 ^
|
||
|
|
||
|
Which can be plotted:
|
||
|
|
||
|
^. graph agsqp1 agsqx,c(s)^
|
||
|
|
||
|
Example using ^forvalues^
|
||
|
-------------------------
|
||
|
|
||
|
The ^forvalues^ command makes using ^praccum^ much simpler. The
|
||
|
following yields the same output as the example above:
|
||
|
|
||
|
^. capture matrix drop mage^
|
||
|
^. forvalues count = 20(5)60 {^
|
||
|
^. local countsq = `count'*`count'^
|
||
|
^. prvalue, x(age `count' age2 `countsq') rest(mean) brief^
|
||
|
^. praccum, using(mage) xis(`count')^
|
||
|
^. }^
|
||
|
^. praccum, using(mage) gen(agsq)^
|
||
|
|
||
|
Example using global macros
|
||
|
---------------------------
|
||
|
|
||
|
^forvalues^ is not available for Stata 6. Here, the task can still be
|
||
|
simplified by using global macros. The advantage of this approach is
|
||
|
that you can let Stata do the multiplying:
|
||
|
|
||
|
^. global age = 20^
|
||
|
^. global age2 = $age*$age^
|
||
|
^. prvalue , x(age=$age age2=$age2) rest(mean)^
|
||
|
^. praccum , saving(mage) xis($age)^
|
||
|
^. global age = 25^
|
||
|
^. global age2 = $age*$age^
|
||
|
^. prvalue , x(age=$age age2=$age2) rest(mean)^
|
||
|
^. praccum , using(mage) xis($age)^
|
||
|
^. global age = 30^
|
||
|
^. global age2 = $age*$age^
|
||
|
^. prvalue , x(age=$age age2=$age2) rest(mean)^
|
||
|
^. praccum , using(mage) xis($age)^
|
||
|
^. global age = 35^
|
||
|
^. global age2 = $age*$age^
|
||
|
^. prvalue , x(age=$age age2=$age2) rest(mean)^
|
||
|
^. praccum , using(mage) xis($age)^
|
||
|
^. global age = 40^
|
||
|
^. global age2 = $age*$age^
|
||
|
^. prvalue , x(age=$age age2=$age2) rest(mean)^
|
||
|
^. praccum , using(mage) xis($age)^
|
||
|
^. global age = 45^
|
||
|
^. global age2 = $age*$age^
|
||
|
^. prvalue , x(age=$age age2=$age2) rest(mean)^
|
||
|
^. praccum , using(mage) xis($age)^
|
||
|
^. global age = 50^
|
||
|
^. global age2 = $age*$age^
|
||
|
^. prvalue , x(age=$age age2=$age2) rest(mean)^
|
||
|
^. praccum , using(mage) xis($age)^
|
||
|
^. global age = 55^
|
||
|
^. global age2 = $age*$age^
|
||
|
^. prvalue , x(age=$age age2=$age2) rest(mean)^
|
||
|
^. praccum , using(mage) xis($age)^
|
||
|
^. global age = 60^
|
||
|
^. global age2 = $age*$age^
|
||
|
^. prvalue , x(age=$age age2=$age2) rest(mean)^
|
||
|
^. praccum , using(mage) xis($age) gen(agsq)^
|
||
|
^. graph agsqp1 agsqx,c(s)^
|
||
|
|
||
|
.-
|
||
|
Authors: J. Scott Long - jslong@@indiana.edu
|
||
|
Jeremy Freese - jfreese@@ssc.wisc.edu
|
||
|
www.indiana.edu/~jslsoc/
|