Advan Neighborhood Patterns: Anomaly in Total Device Count

We are working with Advan neighborhood pattern data between 2019-2023 across all of the US for a research project. We noticed a pattern in the total device count that we were hoping could be explained. Below I have included two images aggregated at the state-level (I did not label the states as the majority seem to follow the general trend) and you can see that the total device count and the device count per person (using ACS state population) Is much higher the first 3 months of 2019 than the rest of the study period. We evaluated this at the Census Block level and found multiple examples were the device count exceeded the CBG population, and in the last image (looking at the CBG data) you can see that the 95th percentile is close to one for the first quarter of 2019. We wanted to know if anyone from Advan could explain why there is such variation in total devices, especially only in the beginning of the dataset. We think it could just be some census blocks that are skewing the data and need to be cleaned and not a uniform increase in devices across all census blocks, but we were wondering if this pattern can be explained.

One specific example I can provide is this CBG has 710,625 devices in 1/2019 but a population of 2,702 people. It has similarly high values for the next two months.

|435px;x353px;

|500px;x385px;

1 Like

Hello! The best thing we would recommend in this case is to ignore Q1 2019 as it’s showing inaccurate - the rest should be good to go. I’ve passed this along to the team and they’re aware of the discrepancies.