I want to estimate a difference in difference regression.
To test the parallel trend hypothesis I used the function
This function, accorind to the manual
Is there any more exact description on how the second plot is produced, so that I could reproduce it manually.
First I want to know it, because I should know what my statistic software is doing.
Second, because instead of plotting the two lines I would like to plot the difference of the two lines. It would allow more easily to assess whether the difference is constant, and I could include several lines in a single graph if I want to estimate the DiD for several subgroups.
To test the parallel trend hypothesis I used the function
Code:
estat trendplots
produces two diagnostic plots for assessing the parallel-trends assumption that
is required for consistent estimation of the ATET using didregress or xtdidregress. The first plot
consists of two lines showing the mean of the outcome over time for the treatment and the control
groups. The second plot augments the DID model to include interactions of time with an indicator
of treatment and plots the predicted values of this augmented model for the treatment and control
groups. Both plots include a vertical line one period before treatment
is required for consistent estimation of the ATET using didregress or xtdidregress. The first plot
consists of two lines showing the mean of the outcome over time for the treatment and the control
groups. The second plot augments the DID model to include interactions of time with an indicator
of treatment and plots the predicted values of this augmented model for the treatment and control
groups. Both plots include a vertical line one period before treatment
First I want to know it, because I should know what my statistic software is doing.
Second, because instead of plotting the two lines I would like to plot the difference of the two lines. It would allow more easily to assess whether the difference is constant, and I could include several lines in a single graph if I want to estimate the DiD for several subgroups.
Comment