Fact-check: Did Bemidji's American Indian population lead to Persell's victory in MN5A?

Guest post by Max Hailperin, Guest Author

Minneapolis Star Tribune reporter J. Patrick Coolican framed his article on American Indian participation in Minnesota's 2018 general election with the claim that "If not for the American Indian population, which comprises 13 percent of Bemidji, John Persell probably would not have won his Minnesota House seat." Given that Persell only won the District 5A race by 11 votes, just about anything could have made a difference, but Coolican is right to point to American Indian participation as a factor—there's hard evidence to support that claim.

Unfortunately, Coolican doesn't report any of that evidence. And his mention of Bemidji is a red herring. Persell did well in areas where American Indians predominate and also, independently, did well in Bemidji.

 

The first point to clarify is Coolican's reference to American Indians as making up 13% of Bemidji's population. That number corresponds to the Census Bureau's 2013–2017 estimate of those identifying as American Indian (or Alaska Native) alone. Presumably he grabbed that number just because it was easily available, not to support the problematic idea of purity. However, a better number (used in the rest of this article) is the percent that identify as American Indian (or Alaska Native) either alone or together with one or more other races or ethnicities. That number for Bemidji is 17%. But here's the thing: while either 13% or 17% is impressive by Minneapolis standards (1.2% and 2.7%), they really aren't a lot compared with some other parts of District 5A.

 

Moreover, 17% of the population can't in itself explain how Bemidji so overwhelmingly went for Persell. He earned 862 more votes there than his opponent, Matt Bliss. That kind of a landslide victory must be a fact as much about the other 83% of the population as about the 17% that are American Indian. But it isn't hard to find other demographic factors that could account for Bemidji's lopsided vote. The city is home to a university, so the statistical distribution of educational attainment and age are surely both quite different from the district as a whole.

 

Looking beyond Bemidji, however, we find that all of Persell's other strongholds (places where he had more than a 10-vote margin) have a population in which American Indians constitute an outright majority:

 

County	Place	Persell's Margin	American Indian
Beltrami	City of Bemidji	862	17.1%
Cass	Pike Bay Township	306	75.4%
Cass	City of Cass Lake	156	73.4%
Beltrami	Ten Lake Township	140	69.6%
Cass	Wilkinson Township	56	55.8%
Itasca	Inger Unorganized	33	82.4%
Cass	City of Bena	19	85.8%

 

By contrast, Bliss's strength was spread over a larger number of places—but none of them had a majority of American Indians, and most didn't come anywhere close:

 

County	Place	Bliss's Margin	American Indian
Beltrami	Northern Township	313	7.6%
Beltrami	Bemidji Township	153	7.0%
Cass	Birch Lake Township	113	1.8%
Cass	Shingobee Township	88	20.2%
Itasca	Morse Township	83	14.2%
Beltrami	Turtle River Township	71	7.3%
Cass	Kego Township	67	24.8%
Beltrami	Frohn Township	60	13.5%
Cass	Turtle Lake Township	58	42.4%
Cass	Leech Lake Township	55	26.3%
Cass	City of Longville	48	0.0%
Beltrami	Moose Lake Township	44	4.0%
Cass	Wahnena Unorganized	42	1.7%
Cass	Inguadona Township	36	4.9%
Itasca	Ardenhurst Township	36	0.0%
Itasca	Oteneagen Township	35	6.3%
Itasca	Good Hope Township	26	0.0%
Itasca	Kinghurst Township	20	0.0%
Itasca	Sand Lake Township	20	13.9%
Cass	City of Walker	16	18.9%
Cass	Winnie Portage Lakes Unorganized	16	#N/A
Itasca	Moose Park Township	15	3.0%
Itasca	Third River Township	14	0.0%
Cass	Otter Tail Peninsula Township	13	1.9%
Cass	Pine Lake Township	13	4.3%
Itasca	Liberty Unorganized	13	0.0%
Cass	Hardwood South Unorganized	12	0.0%
Hubbard	Hart Lake Township	12	21.1%
Itasca	Alvwood Township	11	0.0%
Itasca	Grattan Township	11	0.0%

 

These statistics need to be qualified with a number of caveats, but the big picture holds: Persell did well in areas with American Indian majorities, suggesting that group's participation in the election was important. And he also did well in Bemidji.

 

And now for those caveats:

1. The decision to focus on places with a margin of more than 10 is arbitrary, though nothing in the more-closely-divided places seems to change the story. 
2. The percent of American Indian (or Alaska Native) is as estimated for 2013–2017. The Census Bureau reports these to the tenth of a percent but provides sampling margins of error that are much larger, especially for the less populous places. The City of Bemdji, for example, is 17.1% plus or minus 2.5%, whereas Grattan Township (population 65) is 0.0%, plus or minus (sic) 23.3%. This doesn't seem to be enough to confuse which places are majority American Indian.
3. The percent of American Indians is of the full population, not the citizen voting age population, which may be somewhat different.
4. Unorganized territories are difficult to cross reference from the election results to the census data. In one case this resulted in a #N/A (not available) percentage. The data for the other unorganized territories may be less reliable than the cities and townships.
5. Census data represents entire cities and townships. In those that have multiple precincts, these were aggregated. In cases (particularly in Hubbard County) where only a small portion of a township is in District 5A, no allowance was made for the fact that this portion might have different demographics than the township as a whole.

About the guest author

GAC Emerti Max Hailperin is the author of a textbook on the principles of operating systems and middleware and together with colleagues Barbara Kaiser and Karl Knight is a co-author of another textbook that is an introduction to computer science. In 2002, he received Gustavus's Edgar M. Carlson Award for Distinguished Teaching. He has also published journal articles on aspects of computer program optimization by compilers, as well as conference presentations on computer science education. He has served as a consultant and volunteer lobbyist regarding technology-related aspects of election law.

He is now walking through All Of Minneapolis and writing about his discoveries at the site of the same name.

Map: Minnesota House District 5A.

If you appreciate our posts and original analysis, you can mail contributions (payable to Sally Jo Sorensen, 600 Maple Street, Summit SD 57266) or use the paypal button in the upper right hand corner of this post. Those wishing to make a small ongoing monthly contribution should click on the  paypal subscription button.

Or you can contribute via this link to paypal; use email [email protected] as recipient.

Comments

[Mike Worcester]

I grew up in what is now HD5A, still have family there, and an interesting exercise for you would be to compare turnout numbers for those precincts that gave Persell such large percentages to both 2016 and 2014. Like many areas, those precincts see lowered turnout in non-presidential years. I'd be curious to see if the mid-term numbers for this year were elevated when compared to other years.