I'm running a time series analysis to determine the effect of a policy intervention over time. Visually, I can see there is much less variability in outcome measurements following the intervention (i.e. individual data points are tighter to the regression line). I believe this could be a direct impact of the policy itself, so rather than transforming the data so the variance is consistent, I would like to model this change in variance as a variable of interest. Can anyone provide any ideas or resources in this area? Thanks!
Take a look at the attached screenshots and see if they help you. The book is available in the z-library.. Best wishes David Booth
are there any other factors or covariates that could have resulted in this phenomena you are assuming cause
The simplest approach would be to divide your overall timeseries into shorter sections (even just 2, but probably more than that if you have enough data-points), and compute the variance or standard deviation about the regression line for each section separately, or even just the difference between the minimum and maximum values within each section.
- Badges
- Science topic