Including Stata (alternative)

As mentioned before, the nbstata approach to incorporating Stata code and output in Quarto documents does not permit parallel execution of other languages. In this section, I will demonstrate a solution that uses the R-package Statamarkdown in place of the Python-package nbstata.

An alternative solution is offered by Dr Tom Palmer in the R-bloggers post. It exploits the fact that you can compile .qmd files using different engines/kernels and then embed the output into the same file. My reason for skipping this solution is that you would need to write a separate file for each language and codeblock.

Instead, I am following a solution alluded to in the above blog using Statamarkdown: a R-package designed to execute Stata code within R-markdown documents.

Setup

You will need to install the following package in R.

install.packages("Statamarkdown")

Then, in your .qmd file you will need to add the following:

```{r}
#| include: false
library(Statamarkdown)
```

This will load the package needed to execute the .qmd file.

Implementation

Once, the package is loaded the implementation is relatively straight foward. As before, you can setup tabs using tabset.

:::{.panel-tabset group="languages"}

## R

```{r}
print("Hello world!")
```

## Stata

```{stata}
dis "Hello world!"
```
:::

The output will look as follows:

R

print("Hello world!")

[1] "Hello world!"

Stata

dis "Hello world!"

Hello world!

Limitation

A key limitation of Statamarkdown is that it each codeblock is executed as a separate .do file. This means that memory is lost between codeblocks.

R

test <- "Am I still defined in R?"
test

[1] "Am I still defined in R?"

Stata

global test "Am I still defined in Stata?"
display "$test"

Am I still defined in Stata?

Now, in a separate codeblock we try to display the same objects:

R

test

[1] "Am I still defined in R?"

Stata

dis "$test No"

 No

Note, this problem is specific to Stata and not R.

Solution

You need to add collectcode=TRUE to ensure that the Stata codes are not run as separate files.

:::{.panel-tabset group="languages"}

## R

```{r}
print("Hello world!")
```

## Stata

```{stata, collectcode=TRUE}
dis "Hello world!"
```
:::

Alternatively, you can set this globally at the top of the .qmd file by adding the following code in R:

```{r}
#| include: false
knitr::opts_chunk$set(collectcode = TRUE)
```

In the next section I will implement this change. Switching collectcode on part way through a file causes trouble in the rendering.

NoteStatamarkdown vs nbstata

Statamarkdown’s implementation is not as neat as nbstata. The key limitation is that each codeblock appears to be executed as a separate (temporary) do-file, which means local macros are lost across codeblocks, even when collectcode=TRUE (see next page for example).

That said, Statamarkdown works with tabset, which is a very nice feature for resource design.