diff --git a/vignettes/mindfulness.Rmd b/vignettes/mindfulness.Rmd
index 9579055..5943260 100644
--- a/vignettes/mindfulness.Rmd
+++ b/vignettes/mindfulness.Rmd
@@ -75,6 +75,7 @@ ways. The point of mediation analysis is to provide evidence for one or the
 other of these possibilities.
 
 ```{r libraries}
+library(cmdstanr)
 library(ggdist)
 library(ggrepel)
 library(phyloseq)
@@ -105,6 +106,12 @@ exper <- mediation_data(
 exper
 ```
 
+```{r check_install}
+if (is.null(cmdstan_version(error_on_NA = FALSE))) {
+    install_cmdstan()
+}
+```
+
 This block applies a logistic-normal multinomial model to the microbiome
 outcome. This is an example where the response variables are all modeled
 jointly. We'll use a ridge regression model for each of the mediators. This is a