I am studying patterns of attention in government social media accounts and could use some help with the quantitative side of the analysis. I am looking at the number of posts in a particular time period and examining both trends over time and distribution. Because the data are autocorrelated, I am using the change in total number of posts per time period (which are not autocorrelated). The distribution of the data follows a high peak, long tailed distribution that mirrors punctuated equilibrium theory of Jones and Baumgartner's _Politics of Attention_. So I have a few questions:
- Is it possible to examine two time series using Granger in this scenario, or do I have to modify the data/use logarithms instead?
- Is it possible to test a change in kurtosis given time series x versus time series x+y? (the question I want to ask is if two online conversations are following the same attention patterns).
- Any other advice? I am somewhat a beginner of this kind of quantitative analysis.
I think you may calculate the Kurtosis and use a t test to see if they are same or different!!
- Badges
- Science topic
- Similar topics
- Computer Science
- Computer Communications (Networks)
- Internet