The Census Bureau released voting and registration toplines from the Current Population Survey today.

[Serious data nerd voice] I’ll confess to a longstanding pet peeve with the way that these results are inevitably reported, including by the Bureau itself in the post linked above. 

The Bureau’s survey asks whether Americans voted and whether they were registered.  In the 2022 results, 52.2% of voting-age citizens said they voted, 31.3% said they didn’t, and 16.5% didn’t respond.  Similarly, 69.1% of voting-age citizens said they were registered, 13.9% said they weren’t, and 17% didn’t respond.  (All of these are subject to the usual caveats about survey data.)

The Census Bureau reports this as 52.2% voter turnout and a 69.1% registration rate.  But for that to be right, all of the nonresponses would have to be “no”s.  More precisely, 52.2% of voting-age citizens said they voted, and some portion of the 16.5% who didn’t answer probably also voted.  (Or to put it another way, 62.5% of those answering the question said they voted, and 83.2% of those answering the question said they were registered.)

None of this matters if you’re just using the Census Bureau reports to compare rates year over year, to evaluate 2022 in light of 2018 or 2014, etc.  But people misuse Census data All. The. Time.  Including claims of fraud based on local results that exceed the turnout and registration rates reported by the Census.  And the fact that these results are inevitably reported as something other than what they actually show doesn’t help.