Reproducible Research Using Stata

Managing Statistical Output

Reproducible Research Using Stata

reStructuredText

Reproducible Research Using Stata L. Philip Schumm

Ronald A. Thisted

Department of Health Studies University of Chicago

July 11, 2005

Examples



reStructuredText

Common practice

data analysis

do-file

writing

cut & paste re-enter by hand

paper/ report

Examples



reStructuredText

Common practice

data analysis

do-file

I

writing


Inefficient and time-consuming

paper/ report

Examples



reStructuredText

Common practice

data analysis

do-file

writing


I

Inefficient and time-consuming

I

Can lead to non-reproducible results

paper/ report

Examples



reStructuredText

A big improvement: Intermediary files

data analysis

writing

individual results do-file

graphs & tables

paper/ report

Examples



reStructuredText


data analysis

writing


graphs & tables

I

Not all results automatically transferred

paper/ report

Examples



reStructuredText


data analysis

writing


graphs & tables

I


I

Can be difficult to manage

paper/ report

Examples



reStructuredText


data analysis

writing


paper/ report

graphs & tables

I


I

Can be difficult to manage

I

Data analysis and writing still asynchronous

Examples



reStructuredText

Reproducible research

What is reproducible research?

I

Emerging literature (e.g., Buckheit and Donoho, 1995; Gentleman and Lang, 2003)

Examples



reStructuredText

Examples



I


I

Dynamic document composed of code chunks and text chunks



reStructuredText

Examples



I


I

Dynamic document composed of code chunks and text chunks Literate programming (Knuth, 1992)

I

I I

tangling weaving



reStructuredText

Examples



I


I

Dynamic document composed of code chunks and text chunks Literate programming (Knuth, 1992)

I

I I

I

tangling weaving

R package called Sweave (Leisch, 2002)



Dynamic do-files

A “dynamic” do-file

data analysis

do-file

writing

stata2doc.py

reStructuredText

Examples



reStructuredText

Dynamic do-files

Comments, commands, and docstrings // Here is an example dynamic do-file. * here is the docstring for the first command sysuse auto * weightsq equals weight squared gen weightsq=weight^2 reg mpg weight weightsq foreign /* As you can see, commands don’t have to have docstrings. */

Examples



reStructuredText

-stata2doc- and -s2d-

Two stata commands: stata2doc and s2d data analysis

writing

do-file

stata2doc.py

stata2doc.ado log file graphs scalars

Examples



reStructuredText


Syntax

stata2doc using do-file, [dirname(dirname) linesize(#) as(type) replace override options] s2d [exp list, nodisplay table noisily warn name(name)] :

Examples



reStructuredText

Examples


Examples of -s2d- usage . s2d w2coef=_b[weightsq] rsq=e(r2): reg mpg weight weightsq foreign . scalar li s2d_rsq = s2d_w2coef =

.69129599 1.591e-06

. s2d two = (1 + 1), noi s2d_two =

2



reStructuredText

Final output

Putting it all together data analysis

writing

do-file

reST document

stata2rst.py

stata2doc.ado log file graphs scalars

Examples



reStructuredText

What is reStructuredText?

I

A plaintext markup syntax and parser system

Examples



reStructuredText


I


I

Intuitive, readable, and easy-to-use

Examples



reStructuredText


I


I


I

Powerful and extensible

Examples



reStructuredText

Examples


I


I


I

Powerful and extensible

I

via Docutils may be translated into a variety of formats (e.g., HTML, LATEX, PDF, Open Office) (see http://docutils.sourceforge.net for more information)



reStructuredText

Simple command

Simple command: do-file /* ---------------A Simple Example ---------------This is a *very* simple example in which I shall demonstrate the following: 1) 2) 3) 4)

a simple command graphs substitution tables

The Venerable Auto Data ----------------------Let’s start by reading them in: */ sysuse auto

Examples



reStructuredText

Simple command

Simple command: reStructuredText ---------------A Simple Example ---------------This is a *very* simple example in which I shall demonstrate the following: 1) 2) 3) 4)

a simple command graphs substitution tables

The Venerable Auto Data ----------------------Let’s start by reading them in: :: . sysuse auto (1978 Automobile Data)

Examples



reStructuredText

Simple command

Simple command: PDF via LATEX A Simple Example

This is a very simple example in which I shall demonstrate the following: 1) a simple command 2) graphs 3) substitution 4) tables

The Venerable Auto Data Let’s start by reading them in: . sysuse auto (1978 Automobile Data)

Examples



reStructuredText

Graphs

Graph: do-file * Now lets look at a boxplot comparing mpg between * domestic and foreign. * Boxplot comparing domestic and foreign. gr box mpg, over(foreign) name(fig1)

Examples



reStructuredText

Graphs

Graph: reStructuredText Now lets look at a boxplot comparing mpg between domestic and foreign. .. gr box mpg, over(foreign) name(fig1) .. figure:: fig1.pdf :scale: 33 Boxplot comparing domestic and foreign.

Examples



Graphs

Graph: PDF via LATEX

10

20

Mileage (mpg) 30

40

Now lets look at a boxplot comparing mpg between domestic and foreign.

Domestic

Foreign

Figure 1: Boxplot comparing domestic and foreign.

reStructuredText

Examples



reStructuredText

Examples

Substitutions

Substitution: do-file * Using a t-test to compare mpg between foreign and domestic * cars yields a p-value of |s2d_ttp|. s2d ttp=(string(r(p),"%05.4f")): ttest mpg, by(foreign)



reStructuredText

Substitutions

Substitution: reStructuredText Using a t-test to compare mpg between foreign and domestic cars yields a p-value of |s2d_ttp|. .. s2d ttp=(string(r(p),"%05.4f")): ttest mpg, by(foreign) .. |s2d_ttp| replace:: 0.0005

Examples



reStructuredText

Examples

Substitutions

Substitution: PDF via LATEX

Using a t-test to compare mpg between foreign and domestic cars yields a p-value of 0.0005.



reStructuredText

Examples

Tables

Table: do-file * Finally, we’ll try regressing ‘‘mpg‘‘ on ‘‘weight‘‘, ‘‘weightsq‘‘, * and ‘‘foreign‘‘. * Regression of mpg on several covariates. s2d, t: reg mpg weight weightsq foreign



reStructuredText

Examples

Tables

Table: reStructuredText Finally, we’ll try regressing ‘‘mpg‘‘ on ‘‘weight‘‘, ‘‘weightsq‘‘, and ‘‘foreign‘‘. .. s2d, t: reg mpg weight weightsq foreign .. stata-table:: Regression of mpg on several covariates. -----------------------------------------------------------------------------mpg | Coef. Std. Err. t P>|t| [95% Conf. Interval] -------------+---------------------------------------------------------------weight | -.0165729 .0039692 -4.18 0.000 -.0244892 -.0086567 weightsq | 1.59e-06 6.25e-07 2.55 0.013 3.45e-07 2.84e-06 foreign | -2.2035 1.059246 -2.08 0.041 -4.3161 -.0909002 _cons | 56.53884 6.197383 9.12 0.000 44.17855 68.89913 ------------------------------------------------------------------------------



reStructuredText

Examples

Tables

Table: PDF via LATEX

Finally, we’ll try regressing mpg on weight, weightsq, and foreign. Table 1: Regression of mpg on several covariates. mpg weight weightsq foreign cons

Coef. -.0165729 1.59e-06 -2.2035 56.53884

Std. Err. .0039692 6.25e-07 1.059246 6.197383

t -4.18 2.55 -2.08 9.12

P>|t| 0.000 0.013 0.041 0.000

[95% Conf. Interval] -.0244892 3.45e-07 -4.3161 44.17855

-.0086567 2.84e-06 -.0909002 68.89913