Q&A – A Census Redo?
Each day I feature a listener question sent by one of these methods.
Email: brianmudd@iheartmedia.com
Social: @brianmuddradio
iHeartRadio: Use the Talkback feature – the microphone button on our station page in the iHeart app.
Today’s entry: @brianmuddradio Please explain “differential privacy” w/census & how that factored into the rigged outcome.
Bottom Line: Right, so today’s question is on back of Florida Attorney General James Uthmeier’s recent letter to the U.S. Commerce Department calling for a correction to the 2020 Census. As Uthmeier said: In addition to skewing the data by utilizing dubuious “differential privacy” and irreguarly delaying the publication of state population counts the 2020 national headcount erronesously overcounted or undercounted the populations of fourteen States. Florida was among those egregiously undercounted, which deprived Florida of an additonal House seat that its citizens deserved to ensure fair representation in Congress and the Electoral College. Not only that, but this under-enumeration also cost Florida financially, given the census’s role in “allocating federal funds to the States”. So what is differential privacy and how was it used in the 2020 Census?
Differential privacy is effectively a form of “new math”. It was first formally introduced in 2006 by Cynthia Dwork, Frank McSherry, Kobbi Nissim, and Adam Smith in their paper, "Calibrating Noise to Sensitivity in Private Data Analysis. The premise behind the complicated concept is to modernize data collection and analysis while protecting individual privacy. Used appropriately the concept has several applications and is regularly used by major tech companies when collecting data on users of products. It’s also regularly used in health studies to protect the automomy of patients. It’s defined like this:
Differential privacy is a mathematically rigorous framework designed to release statistical information about datasets while ensuring the privacy of individual data subjects. It guarantees that the inclusion or exclusion of any single data point does not significantly impact the outcome of a query or analysis, thereby protecting individual privacy within large datasets.
Notably, the “new math” had not been applied to census data until the questionable 2020 census results.
Now, here’s the not-so-insignificant thing about it’s application for census data. These are known trade-offs of using this mathateical method:
- Privacy vs. Accuracy: More noise (stronger privacy) can reduce result accuracy
- Complexity: Designing differentially private algorithms complicates the process
- Computational Cost: Adding additional privacy is more expensive and resource-intensive.
As you can tell simply trying to explain what the method was that was used to calculate the 2020 Census is complicated and not easily understood. That certainly lends itself to potentially errors, intentionally or otherwise. It’s notable that the first time this new method was used for calculating census data the results were different than literally every expert’s estimates heading into the Census.
As was noted in a recent report by Florida TaxWatch: The Census Undercount Hurts Florida’s Political Influence demonstrates that the 2020 Census missed about 750,000 Floridians — 3.48 % of the population. Correcting that error with U.S. Census Bureau methodology shows the undercount shifted three U.S. House seats nationally: Colorado, Minnesota, and Rhode Island would each lose a seat, while Florida, Tennessee, and Texas would each gain one — raising Florida’s delegation to 29 seats instead of 28.
Every House seat equals one electoral vote, so the miscount also diverted three electoral votes away from Florida‑leaning states; in the razor‑thin 2024 presidential contest, that swing alone could have altered the Electoral College result. Inside Congress, where landmark legislation has passed by a single vote (215‑214), Florida’s missing representative could likewise decide national policy.
The question isn’t about whether the previous census was accurate. It’s about whether the errors were intentional or unintentional. But either way Florida’s Attorney General James Uthmeier has stated it’s time for a correction. Uthmeier suggested it’s possible and nessasary for the Commerce Department to correct the previous census. Perhaps without even conducting a new census. If, for example, the 2020 Census data still exists – would it be possible to apply the traditional method of census calculations to it to determine a more accurate outcome? There’s nothing in the constitution that prohibits that from taking place and it would surely be the most efficient method of attempting to right the wrongs of the previous calculation.
Florida would stand to benefit most from a newly conducted census that would take into account it’s post-pandemic population growth which was the largest in the nation – though given the need for congress to authorize a mid-decade census – that remains very unlikely. It will be interesting to see what happens from here.