1 item has been added to your cart.
| Duration: | 1 hour |
|---|---|
| Where: | Join us from anywhere! |
| Cost: | Free—but registrations are limited |
When errors are not independent but correlated within clusters, conventional confidence intervals can lead us to overreject null hypotheses and produce false-positive results. This problem is pervasive in applied research, showing up in school-level data, industry-level data, experimental designs, and panel-data settings.
To address this problem, we introduce you to cluster–robust
inference in Stata. We will begin with the option vce(cluster clustervar)
and the different commands for which it is
available. We will see how, if the number of clusters is large,
this option can provide accurate confidence intervals when the
errors are correlated within clusters. However, we will also see
that this is not the case when the number of clusters is small
or their sizes are very heterogeneous. To obtain more accurate
confidence intervals in these cases, we will introduce the
option
vce(hc2 clustervar, dfadjust) and the command wildbootstrap.
The webinar is free, but you must register to attend. Registrations are limited so register soon.
We will send you an email prior to the start with instructions on how to access the webinar.
Eduardo García Echeverri is a Senior Econometrician at StataCorp LLC. He holds a PhD in Economics from the University of Rochester and a master’s degree from Universidad de los Andes in Colombia. His research focuses on nonparametric and semiparametric methods in econometrics. At Stata, he produces documentation, develops webinars, and contributes to the development of new statistical features.
