I am running an analysis to assess the impact of a stewardship policy on the consumption of antibiotics.
I have a time series panel of quarterly use per capita for eight countries from 2013-2018 - the policy was introduced by q3 2017 for all the countries. The panel variable is the country-antibiotic (panelvar) & l have ~17,000 observations (~4,000 observations over 6 periods post-policy).
The antibiotics are split into 3 categories & I would like to see whether there is any impact of the policy on changing consumption in any of the three categories. The data is set up according to the dataex extract below.
I initially looked at using itsa (ssc install itsa), but I don't think it is suitable for the multiple treated panels.
I am now looking at running fixed or random effects regression (xtreg):
Code:
tsset panelvar qdate, q xtreg percapita policy##category, fe robust xtreg percapitapolicy##category i.antibiotic i.country, re robust
Does anyone have any advice on how to run this model?
Thank you for any help.
Code:
* Example generated by -dataex-. To install: ssc install dataex clear input long panelvar float(qdate category percapita policy) long(antibiotic country) 1 212 1 8.491942 0 3 1 1 213 1 6.455289 0 3 1 1 214 1 5.392517 0 3 1 1 215 1 7.843199 0 3 1 1 216 1 7.863443 0 3 1 1 217 1 6.440968 0 3 1 1 218 1 5.475216 0 3 1 1 219 1 8.582564 0 3 1 1 220 1 8.241756 0 3 1 1 221 1 5.972879 0 3 1 1 222 1 5.031816 0 3 1 1 223 1 7.213692 0 3 1 1 224 1 7.903334 0 3 1 1 225 1 5.495628 0 3 1 1 226 1 4.689399 0 3 1 1 227 1 7.141853 0 3 1 1 228 1 6.740578 0 3 1 1 229 1 4.962944 0 3 1 1 230 1 4.5171933 1 3 1 1 231 1 6.471659 1 3 1 1 232 1 6.534578 1 3 1 1 233 1 4.4081707 1 3 1 1 234 1 3.778891 1 3 1 1 235 1 5.652427 1 3 1 2 212 2 .20528096 0 12 1 2 213 2 .20077975 0 12 1 2 214 2 .2002376 0 12 1 2 215 2 .22712195 0 12 1 2 216 2 .2269007 0 12 1 2 217 2 .2264132 0 12 1 2 218 2 .2296442 0 12 1 2 219 2 .25586352 0 12 1 2 220 2 .24679537 0 12 1 2 221 2 .24202643 0 12 1 2 222 2 .25290585 0 12 1 2 223 2 .27849534 0 12 1 2 224 2 .27684602 0 12 1 2 225 2 .26330328 0 12 1 2 226 2 .2578124 0 12 1 2 227 2 .28231534 0 12 1 2 228 2 .28142148 0 12 1 2 229 2 .27017698 0 12 1 2 230 2 .26818603 1 12 1 2 231 2 .2890595 1 12 1 2 232 2 .28878802 1 12 1 2 233 2 .27159473 1 12 1 2 234 2 .27031007 1 12 1 2 235 2 .30036235 1 12 1 3 212 1 7.357988 0 3 2 3 213 1 5.517321 0 3 2 3 214 1 4.871842 0 3 2 3 215 1 6.50174 0 3 2 3 216 1 6.900932 0 3 2 3 217 1 5.58475 0 3 2 3 218 1 5.113198 0 3 2 3 219 1 7.01998 0 3 2 3 220 1 7.539312 0 3 2 3 221 1 5.874916 0 3 2 3 222 1 5.165118 0 3 2 3 223 1 6.804155 0 3 2 3 224 1 7.475902 0 3 2 3 225 1 5.733838 0 3 2 3 226 1 4.826259 0 3 2 3 227 1 6.752472 0 3 2 3 228 1 7.602476 0 3 2 3 229 1 5.641543 0 3 2 3 230 1 4.832673 1 3 2 3 231 1 6.864495 1 3 2 3 232 1 6.832723 1 3 2 3 233 1 5.269779 1 3 2 3 234 1 4.7140703 1 3 2 3 235 1 6.157481 1 3 2 4 212 2 1.8451626 0 12 2 4 213 2 .9347179 0 12 2 4 214 2 .9068737 0 12 2 4 215 2 1.3392904 0 12 2 4 216 2 1.2946705 0 12 2 4 217 2 .910511 0 12 2 4 218 2 .9301074 0 12 2 4 219 2 1.3665724 0 12 2 4 220 2 1.453778 0 12 2 4 221 2 1.0201169 0 12 2 4 222 2 .8277714 0 12 2 4 223 2 1.31947 0 12 2 4 224 2 1.455298 0 12 2 4 225 2 .9360857 0 12 2 4 226 2 .815813 0 12 2 4 227 2 1.31421 0 12 2 4 228 2 1.458465 0 12 2 4 229 2 .9216899 0 12 2 4 230 2 .7625733 1 12 2 4 231 2 1.5324233 1 12 2 4 232 2 1.5997287 1 12 2 4 233 2 .9900257 1 12 2 4 234 2 .8145158 1 12 2 4 235 2 1.4061882 1 12 2 end format %tq qdate label values panelvar panelvar label def panelvar 1 "China_amoxicillin", modify label def panelvar 2 "China_azithromycin", modify label def panelvar 3 "India_amoxicillin", modify label def panelvar 4 "India_azithromycin", modify label values antibiotic product label def product 3 "amoxicillin", modify label def product 12 "azithromycin", modify label values country co label def co 1 "China", modify label def co 2 "India", modify
0 Response to Advice on time series analysis to evaluate the impact of a policy
Post a Comment