Do you have any advice on what to do with these census block groups?

I am analyzing monthly patterns data to see number of visitors to SF Giants games by census block, and have questions about correcting for sample bias / adjusting the visitor number. I’m wondering what I should do about block groups that are listed as having 4 visitors (which I understand could be 2-4 visitors). For some block groups, the difference between 2 vs. 4 “real” visitors changes the adjusted visitor count by thousands of people. Do you have any advice on what to do with these block groups? I thought about filtering them out, but that would mean eliminating many block groups

This topic was automatically generated from Slack. You can find the original thread here.

Hey @Nami_Sumida ! It ultimately depends on your needs and use case. However, have you checked out this resource. Might be a good place to start!

If your application can take it, it could also be reasonable to use visitor_home_aggregation which is the aggregation by census tract instead of census block group. The larger geographic area typically has fewer 4s (and therefore fewer ambiguous situations).

I didn’t realize a tract grouping was available - thank you! I think this will work for us

this variable is showing the number of unique visitors from a tract that visited at any point during the month? not total number of visits?

if someone visited three times in the month, that person would only appear once in their tract?

~visitor_home_aggregation is the number of visitors to the POI from each census tract, not unique visitors. If someone visited three times, it would appear three times in that tract.~

@Niki_Kaz interesting… is there a reason why you don’t label it as number of visits then?

Hey @Nami_Sumida - apologies for the confusion. Going to strikethrough my previous comment. Received confirmation that it is indeed unique visitors. Hope that clarifies things!

H/T to @Jeff_Ho_SafeGraph for catching!