Last week, I received a few further MN to act as control groups in our study. These MN had at least 3 months of data prior to joining DataONE. I ran regressions on them and will compare them to our study group this coming week.
I also changed from a separate analysis of create and read counts to a joined analysis of “rate” counts. Our goal is measure change in read counts, but we must account for the increasing database size. Therefore the new measurement is a “rate” count – which is individual read counts / cumulative create counts on some tests and cumulative read counts / cumulative create counts on other tests. This week, I will clarify which “rate” is preferable for what we are studying. This change has actually made some individual results significant, but it does not make the combined AOV test statistically significance.
This week, I am (hopefully) the last few tests, before beginning to write up the results to they study.