Reproducing Chakraborty (2021)
Takeaways from Reproducing Chakraborty (2021)
Reproducing this study has highlighted the multitude of decisions made by researchers behind the scenes for any given study that make reproduction difficult or ambiguous, even when the researcher does their best to document their methods.
One improvement to the reproduction of the study that I would love to implement would be to the map visualizing the SpatialEpi Kulldorff COVID-19 Clusters. The current map does a great job of showing how the spatial clusters are formed, and some problems that this may have posed in the study. But I would also be interested to see which “other parts of cluster” belong to which “center of cluster,” by associating each by color. I looked into the data, and unless I modify the spatial scan to keep track of cluster ID number, I do not think there is an easy solution to make the map I am thinking of.
One change I did make was in the color ramp for the map showing Local Relative Risk (RR) Class for each county. Because they are categorized in 6 groups, from low risk to high risk, I thought a diverging color palette would best convey the information. I opted for the RdYlGn color ramp, with green representing the least amount of RR and red the most. This is a small change, but now the few counties at high risk stand out from the map more than they did previously.
Our updated paper can be found at this GitHub Repository. You can view it as a HTML Document here.