Digging a little

I wondered, after posting the other day, if the drop in the percentage of vote for Ilhan Omar in her district was because of a drop in her support, or because this year there were many Democrat ballots for Biden that voted only for president and didn't fill out the downballot choices.

After looking at the preliminary numbers as of this morning and comparing with past election results, the answer appears to be:  both are true.  Omar received about 12,000 fewer votes than she did in 2018, while Biden in her district gained about 55,000 votes over what Hillary Clinton got in 2016.  Omar did receive about 6,000 votes more in 2020 than Keith Ellison got for that congressional seat in 2016.

I noticed some discrepancies in the vote totals for Omar's District 5, as reported in different places on the state's website:   410,000 votes cast for president, 398,000 cast for Omar's seat, but the votes for Omar, her Republican challenger, and the write-ins in her race only add up to 380,000.  Why these totals are so different, I can't say at this point.  I may be wrongly interpreting the state's text files of results.

Generally, I would say to "Watch the Denominators"; when a percentage is reported, ask:  "X% of what?"

I also looked at how Trump did in Omar's district in 2016 and 2020, compared to the Republican challenger for the congressional seat.  Trump gained almost 6,000 votes in 2020 over 2016, but Omar's 2020 challenger had over 30,000 more votes than Trump did.  Compare that to 2016, where Ellison's opponent had 46,000 12,000 [UPDATE:  Correction.] more votes than Trump did in the district.

So there were lots of Republicans there that didn't vote for Trump this year, but not as many as and more than in 2016.*

Statewide, Biden is up by 233,000 votes.  Trump gained 160,000 votes compared to 2016, but Biden gained 350,000 votes over Hillary Clinton.

Voter participation:  3,279,000 votes out of 3,590,000 registered voters (as of 7 am on Election Day; Minnesota allows same-day registration), out of 4,118,000 eligible to register.  So, 91% turnout compared to registered voters, and 87% 79%** [UPDATE:  Secretary of State corrected their number; see below] compared to eligible voters.  Apparently almost everyone with a pulse and the ability to sign their name voted.

----------------------------------------------------

* In my Ph.D. research, we had to leverage what accuracy we could get out of our computational methods:  "We can calculate A with only moderate accuracy, and B with only moderate accuracy, but we can calculate (A - B) much more accurately, because our errors in calculating A and in calculating B are almost equal."

** I was thinking over the numbers and realized that 3.28 million is not 87% of 4.12 million. Checking the state's website, I found the updated number above.  It looks they miscalculated at first by dividing the number of registered voters by the number of people eligible, instead of dividing actual voters by eligible.  Apparently we have to Watch the Numerators too now.