We are analyzing park visits and trying to decide on normalization. The device panel declined in the weeks following the US response to the pandemic (let me know if this is not true). Do we know why? The why determines the interpretation of the trends. Raw park visit counts fell in the weeks following the start of the pandemic, but not as much as the panel declined. I see two possible interpretations: 1. if the devices that dropped out of the panel stayed home, then park visits fell, 2. if the drop out devices visited at the same rate as those that remained in the panel, then normalizing by the number of devices in the panel is appropriate. I know this is a widely discussed topic (Workspace Deleted | Slack), and there is a lot of info on best practices, but I didn’t see discussions of why the panel fell in the weeks after the start of the pandemic. Thanks as always!
The device panel declined in the weeks following the US response to the pandemic (let me know if this is not true). Do we know why?
Hi @Jude_Bayham_Colorado_State_U , the current understanding is that more devices staying home more often resulted in less devices showing up in the panel. We are working to understand the relationship between raw counts, normalization, and the truth across a major panel change like the one following the US response to the pandemic.
Coincidentally, I am working on some analysis right now that may be helpful for you. I may even have it ready later today—I’ll follow up today or tomorrow
Thanks for the reply, @Ryan_Kruse_MN_State. Good timing! I’ll keep an eye out.
Hi @Jude_Bayham_Colorado_State_U, I’m still working on the project, but hopefully I can give you something helpful. My belief is that (1) is more accurate, but (2) may have some validity too.
First, do you happen to have “truth” counts for any of the parks over that period of time?
I’ve only really tried to ground truth rocky mountain national park, which closed over this period when the panel declined. I seem to recall some group truthing beach visitation around this time. Can’t recall the authors or title.
Hi @Jude_Bayham_Colorado_State_U, I would expect the drop in raw visit counts to indicate a drop in real visits, particularly at a single POI across time. So I believe interpretation (1) is most correct. But of course any truth data you have over that span of time would be great to validate that interpretation before generalizing