Thank you so much Chuck Huber and STATA's team for this interesting video. How can I find any dataset used in the DID analysis to see how variables are presented?
Thank you for this great video. But I do not find any ptrends or granger estimation (estat ptrends and estat granger) for Repeated cross-sectional data. It shows treatment assignment times vary; not allowed with estat ptrend/estat granger. Do you have any solution?
*estat ptrends* and *estat granger* work with repeated cross-sections and panel data. However, they do not work when the groups are treated at different periods in time. Contact us at tech-support@stata.com if you have any further questions about it.
I am a doctoral student at Harbin Institute of Technology, Harbin, China and I have STATA 13. Kindly guide me how I perform DiD regression in this version?
Difference-in-differences (DID) and DDD models were introduced in Stata 17 (www.stata.com/stata17/difference-in-differences-DID-DDD). You would need the current version of Stata. Contact tech-support@stata.com for more details.
The observed-means graph is a graph of the mean outcome for each group, at each time point. The linear-trends graph augments the original model to include the difference in trends before and after the intervention period. One is a graph of means and one is a line of trends over time. Both graphs are diagnostics of the parallel-trends assumption. The means and also the trends should be parallel before the treatment/intervention. Otherwise, the parallel-trends assumption comes into question. For a discussion of these diagnostic plots, take a look at example 4 in [TE] didregress (www.stata.com/manuals/te.pdf). You can also look at the Methods and formulas section in [TE] didregress postestimation to see the formula used for the linear-trends model. Also of interest is the Statalist post www.statalist.org/forums/forum/general-stata-discussion/general/1629338-reproducing-estat-trendplots-linear-trends-model And, you can always contact tech-support@stata.com with questions.