As a rule, we at AGS are not overly keen on the approach of talking about competitors negatively. Perhaps it is our polite Canadian heritage, where we are more likely to say that we are sorry for something we didn’t do than point fingers at the ones who did.

Sometimes, however, rules need to be set aside.

As many of you know, we have been sounding the alarm about the 2020 Census data release and how it affects small area forecasting. We just didn’t expect to be the only ones. After months of talking about this, we expected to turn and find our cohorts shouting from their own rooftops, telling their clients about the issues and limitations to the data. And instead? Crickets.

Demographic data at the census block and block group levels are key drivers of locational decisions that often involve millions of dollars of investor capital. Nobody, at least nobody with any hint of intelligence, claims that their data is perfect. We all know better. Error is the nature of the beast here, and the impact of error is to reduce your certainty in decision making. You acquire data from a provider because you trust them to provide good quality, transparent data that will help you reduce uncertainty. That is the essence of the relationship between data provider and data user.

So, when the census bureau deliberately injects massive error into a mission-critical data source, and nobody talks about it, what does that tell you?

We recently published an article at Directions Magazine, which did elicit a comment on LinkedIn from a company that we have personally known and admired for decades. What did they say?

Forgive us for being cynical here, but this says that Claritas has special relationships with the Census Bureau, so therefore they had insider information on the issue. Really? How about this – we knew about it but chose to not inform you, lest you worry about the quality of the data.

But we will give them credit for at least mentioning that there is a problem here. Others have simply ignored it. That quarterly data release that occurred just weeks after the PL-94 release and claimed to have “integrated” the data into their series? This speaks volumes about methodology when data below the county level can’t be trusted and yet it has been injected into a small area forecasting model.

Perhaps it is time to turn to a vendor who doesn’t just pat you on the head and tell you not to worry about it, that we have “processes” to handle it. Trust in this business is everything, and we wonder how you can trust a data source that doesn’t trust you enough to hear truth.