Re: st: RE: updated package -binscatter- in SSC

[Michael Stepner <stepner@mit.edu>]

George, thank you for catching this.  I had forgotten to update the
name of one local.  The fit line was actually being computed correctly
(it is computed off of the microdata, not the binned scatterpoints),
but the line wasn't being plotted over the full x-range.

Version 7.01 will be in the SSC shortly with this fixed.  But if
anyone wants a quick fix in the meantime, you can change one word in
the ado-file manually.  You can find the location of the ado file by
typing -which binscatter-.  On line 617, replace the following line:

local fitline_ub_rindex=`nbins'

with

local fitline_ub_rindex=`nquantiles'


Best,
Michael

On 14 October 2013 12:13, Hoffman, George <ghoffman@mcw.edu> wrote:
> Michael -
> Thank you for this great binscatter.ado. In the updated version when run on Stata/SE 13, I have noticed that the regression lines are only calculated from the first 20 bins, if option nquantile(n>20) is used. Is this the desired behavior?
> Thank you in advance.
> George Hoffman
>
>
> -----Original Message-----
> From: owner-statalist@hsphsun2.harvard.edu [mailto:owner-statalist@hsphsun2.harvard.edu] On Behalf Of Michael Stepner
> Sent: Sunday, October 13, 2013 12:46 PM
> To: statalist@hsphsun2.harvard.edu
> Subject: st: updated package -binscatter- in SSC
>
> -binscatter- has been updated in the SSC, with thanks to Kit Baum.
>
> binscatter is a Stata routine which draws binned scatterplots, providing a non-parametric method for visualizing the relationship between two variables.  It groups an x-axis variable into equal-sized bins, computes the mean of the x-axis and y-axis variables within each bin, then creates a scatterplot of these data points.  The result is a non-parametric visualization of the conditional expectation function.
>
> This update (version 7.00) corrects issues reported to me by e-mail, with thanks to Art Burke, Laszlo Sandor and Raj Chetty for reporting them.  It also runs faster than previous versions, and has added documentation on how binscatter handles by-variables.
>
> When given a by(variable), -binscatter- plots separate series for each by-value, but there are 2 important steps in which it performs computations that do not treat the by-values separately.  These are now described in detail in the help file, along with suggested procedures for users who wish to obtain an alternative behaviour.
>
> Changelog from version 6.02:
> - version requirement increased to Stata 12.1
> - refusal to accept weights fixed. -binscatter- supports fweights and aweights.
> - crash with string by-variable fixed.
> - correctly plots conditional x-means when combined with by().
> previous versions incorrectly plotted the unconditional x-mean against the conditional y-mean.
> - bundled -fastxtile- updated to 1.21 (corrects loss of numerical precision)
> - added documentation on by-variable procedures
> - runs faster than earlier versions
> - can be run quietly
>
> If you've already installed binscatter, you can update it by running -adoupdate-.  To install it for the first time, run -ssc install binscatter-.
>
> Best,
> Michael
> *
> *   For searches and help try:
> *   http://www.stata.com/help.cgi?search
> *   http://www.stata.com/support/faqs/resources/statalist-faq/
> *   http://www.ats.ucla.edu/stat/stata/
>
> *
> *   For searches and help try:
> *   http://www.stata.com/help.cgi?search
> *   http://www.stata.com/support/faqs/resources/statalist-faq/
> *   http://www.ats.ucla.edu/stat/stata/

*
*   For searches and help try:
*   http://www.stata.com/help.cgi?search
*   http://www.stata.com/support/faqs/resources/statalist-faq/
*   http://www.ats.ucla.edu/stat/stata/