I am interested in understanding the effect of prescription drugs (IV) on child maltreatment cases (DV) and I am also interested to find out which drug (opioid, benzo, stimulants) accounts for most variance in the outcome. My unit of analysis is counties in a state and I have access to 5-year data from 2015-2019. First, I thought I will use aggregated data from 2015-2019 and do a cross-sectional analysis, but this way my sample size will be same as the no. of counties which is too small and inappropriate to do any statistical inference. But if I use the data longitudinally, I will have a larger sample size and we can do a repeated measures/within subjects ANOVA. I read up a lot on how to do this in STATA but all the examples is using experimental data. So, I am having difficulty visualizing the data structure and what will I be finding out if I do within subjects' analysis. I would like to know how to do this in STATA.
Given below is how I think I will set up my data, but I could be wrong.
Thanks
Given below is how I think I will set up my data, but I could be wrong.
| County | Opioids | Stimulants | Benzodiazepine | Child Maltreatment Cases (Outcome) | Time |
| County 1 | 283,234 | 106,618 | 181,113 | 40 | 2015 |
| County 1 | 271,837 | 112,979 | 181,999 | 30 | 2016 |
| County 1 | 252,623 | 116,816 | 175,032 | 15 | 2017 |
| County 1 | 227,583 | 119,370 | 169,830 | 10 | 2018 |
| County 1 | 231,590 | 125,521 | 174,990 | 20 | 2019 |
| County 2 | 283,234 | 106,618 | 192,285 | 80 | 2015 |
| County 2 | 271,837 | 112,979 | 189,501 | 23 | 2016 |
| County 2 | 252,623 | 116,816 | 177,669 | 66 | 2017 |
| County 2 | 227,583 | 119,370 | 166,209 | 22 | 2018 |
| County 2 | 231,590 | 125,521 | 157,721 | 11 | 2019 |
