I am currently writing an article about digital repression's impact on the nature (violent/nonviolent) of protests. I draw my digital repression variable from a country-year dataset which assigns a different digital repression score to each country for each year, while my protest data is coming from an event-level dataset. My plan is to employ logistic regression analysis, as my dependent variable is binary (1 for violent, 0 for nonviolent protests).
For a protest which started in 2019 and ended in the same year in the UK, I use the digital repression score of the UK in 2019. But, there are some protests which, for example, started in 2010 and ended in 2019 in the UK. In this case, how should I choose my digital repression score for the UK? Should I take the average of the UK's digital repression score for the years between 2010 and 2019, or is there a better option for this?
