The November general results are (mostly) in. How do they line up with our Seattle polling?

In ten days, Wash­ing­ton State’s gen­er­al elec­tion will be cer­ti­fied and the results of the qual­i­fy­ing round of this year’s local elec­tions cycle will become final.

With only a hun­dred bal­lots list­ed by King Coun­ty Elec­tions as await­ing tab­u­la­tion, and with almost a month hav­ing now passed since we announced the elec­toral find­ings from our Octo­ber 2021 sur­vey of the Seat­tle elec­torate, it seems like a good time to revis­it our poll find­ings and exam­ine to what extent they antic­i­pat­ed the out­comes in the actu­al returns tal­lied by King Coun­ty Elections.

Our sur­vey, con­duct­ed by Change Research from Octo­ber 12th-15th, 2021, was our sec­ond poll of Seat­tle this cycle and one of just three inde­pen­dent polls con­duct­ed in the Emer­ald City between the cer­ti­fi­ca­tion of the August 2021 Top Two gen­er­al elec­tion and the dead­line to return bal­lots in this gen­er­al election.

617 like­ly vot­ers par­tic­i­pat­ed, all online.

We released the results of all of the elec­toral ques­tions on Tues­day, Octo­ber 19th, via press con­fer­ence and here on NPI’s Cas­ca­dia Advocate.

Our objec­tive in com­mis­sion­ing these sur­veys was to help every­one inter­est­ed in Seat­tle pol­i­tics get a bet­ter sense of the elec­toral land­scape and dynamics.

As I not­ed last sum­mer, polls can’t pre­dict the future. That is not their pur­pose. What polling can do, how­ev­er, is indi­cate how peo­ple (espe­cial­ly like­ly vot­ers!) may be feel­ing about pro­posed leg­is­la­tion, bal­lot mea­sures, an issue, or a can­di­date elec­tion at a par­tic­u­lar junc­ture. It’s cru­cial to note that all polls are snap­shots in time and that polls vary in terms of their qual­i­ty and credibility.

The method­ol­o­gy used to con­duct a poll is of great importance.

If a poll has a prop­er­ly rep­re­sen­ta­tive sam­ple and asks neu­tral ques­tions, the results can be use­ful. But if the inputs are bad, the out­puts will be bad.

That’s why we and Change Research worked hard to design and field a sur­vey that would be cred­i­ble. We want­ed the data to be use­ful. We are firm believ­ers in rig­or­ous, high qual­i­ty research, and we prac­tice what we preach.

Since we’re now at a junc­ture where we can safe­ly com­pare our data to the (most­ly) com­plete unof­fi­cial results, let’s dive in and see how the returns mesh with our polling from about a month ago.

Mayor

What the poll found: Of the two final­ists for May­or, the poll found that for­mer Seat­tle City Coun­cilmem­ber Bruce Har­rell had a six­teen point lead over cur­rent Coun­cil Pres­i­dent M. Lore­na González, with near­ly a major­i­ty of respon­dents express­ing a pref­er­ence for Har­rell and just under a third express­ing a pref­er­ence for Gon­za­lez. 18% said they were not sure and 2% said they would not vote.

What hap­pened in the elec­tion: Har­rell is deci­sive­ly defeat­ing González, who con­ced­ed the race last week. As of today, Har­rell has 58.58% of the vote and González has 41.12%. That’s a mar­gin of vic­to­ry a lit­tle more than sev­en­teen points, which is awful­ly close to the mar­gin in our polling.

Analy­sis: Har­rell entered the gen­er­al elec­tion as the fron­trun­ner, hav­ing won the August Top Two elec­tion by a nar­row mar­gin. There is no evi­dence that Har­rell ever lost his advan­tage. While he ulti­mate­ly pre­vailed over González in August by just a cou­ple points, he was able to jump out to a much big­ger lead before the vot­ing began in this final round, owing in part to the weak­ness of González’s cam­paign, which did­n’t offer a com­pelling mes­sage to voters.

The oth­er two inde­pen­dent polls I men­tioned ear­li­er — con­duct­ed by Elway for Cross­cut and by Strate­gies 360 for KOMO — each found Har­rell ahead of González by a few points at two dif­fer­ent junc­tures in Sep­tem­ber. By the time our sur­vey field­ed, Har­rell had built an even big­ger lead… one he would not relinquish.

The remain­ing unde­cid­ed vot­ers seem­ing­ly went to both can­di­dates in rough­ly equal pro­por­tions, leav­ing Har­rel­l’s already siz­able lead intact. Har­rel­l’s more than two-to-one advan­tage with vot­ers ages 65 and old­er was a key factor.

Seat­tle Times reporter Daniel Beek­man asked the González cam­paign for reac­tion to our poll find­ing as part of his sto­ry on the sur­vey’s findings.

Cam­paign man­ag­er Alex Koren’s response was to cite the cam­paign’s own inter­nal polling from a month pri­or show­ing González and Har­rell tied at 45% apiece, and to ques­tion the verac­i­ty of our sur­vey’s sam­ple. From Beek­man’s arti­cle:

Asked about the NPI poll, González cam­paign man­ag­er Alex Koren said the campaign’s own poll in mid-Sep­tem­ber showed the race tied at 45%.

In a Crosscut/Elway poll con­duct­ed in ear­ly Sep­tem­ber, 42% of respon­dents picked Har­rell, 27% picked González and 24% were unde­cid­ed. In a KOMO/Strategies 360 poll con­duct­ed in mid-Sep­tem­ber, 40% picked Har­rell, 33% González and 27% were undecided.

The NPI poll may have under­es­ti­mat­ed sup­port for pro­gres­sive can­di­dates like González and City Coun­cil Posi­tion 8 incum­bent Tere­sa Mosque­da, Koren said.

The most­ly com­plete results clear­ly demon­strate that our pol­l’s mod­el­ing and sam­pling was sound, where­as the González cam­paign’s poll was an outlier.

González’s inter­nal polling was nev­er cor­rob­o­rat­ed by any oth­er data that we saw, where­as our find­ing in favor of Har­rell was, we under­stand, cor­rob­o­rat­ed by pri­vate polling com­mis­sioned in Octo­ber by some of González’s allies.

Their data may not have been pub­licly avail­able, but ours was. We delib­er­ate­ly chose to release our find­ings in the hopes of pro­vid­ing a ser­vice to the public.

City Attorney

What the poll found: Of the two final­ists for Seat­tle City Attor­ney, our poll found a big lead for Repub­li­can Ann Davi­son, with abo­li­tion­ist Nicole Thomas-Kennedy nine­teen points behind. 43% of respon­dents to the poll said they were vot­ing for Davi­son, while 24% said they were vot­ing for Thomas-Kennedy. A sig­nif­i­cant num­ber, 30% said they were not sure. 2% said they would not vote.

What hap­pened in the elec­tion: Davi­son pre­vailed and will go on to win the elec­tion, but not by any­thing resem­bling the mar­gin in our poll. Davi­son cur­rent­ly has 51.52% of the vote, while Thomas-Kennedy has 47.7%. On Elec­tion Night, Davi­son had a much, much big­ger lead: 58.25% to Thomas-Kennedy’s 40.96%.

Analy­sis: This is a race in which the dynam­ics def­i­nite­ly changed dur­ing the vot­ing peri­od. The late bal­lots prove it. The rea­son Thomas-Kennedy is doing so much bet­ter in the elec­tion than in the sur­vey is that she was able to reel in a lot of those unde­cid­ed vot­ers in the final days and weeks of the 2021 campaign.

Those not sure vot­ers had to go some­where (93% of our sur­vey tak­ers said they were “def­i­nite­ly” plan­ning on vot­ing) and most went to Thomas-Kennedy.

Unlike González, Thomas-Kennedy’s cam­paign offered a strong clos­ing argu­ment in favor of her can­di­da­cy that was wide­ly shared on social media plat­forms. The elec­tion results, com­pared to our polling, sug­gest that argu­ment resonated.

Had Thomas-Kennedy cam­paigned with the brand­ing of a pro­gres­sive Demo­c­rat and not had to con­tend with for­mer Demo­c­ra­t­ic gov­er­nors and Supreme Court jus­tices endors­ing her oppo­nent (who was rat­ed “not qual­i­fied” by a coali­tion of bar asso­ci­a­tions) she might have pre­vailed in the elec­tion, even despite her Twit­ter his­to­ry, which her oppo­nents seized upon in an attempt to dis­cred­it her.

City Council, Position #8 (At-Large)

What the poll found: Of the two final­ists for Seat­tle City Coun­cil Posi­tion #8, one of the coun­cil’s two at large posi­tions, our poll found an eight point lead for Coun­cilmem­ber Tere­sa Mosque­da. Mosque­da, the only incum­bent on the bal­lot in the four city­wide races, got 39% in the sur­vey, while chal­lenger Ken­neth Wil­son got 31%. 26% said they were unde­cid­ed and 3% said they would not vote.

What hap­pened in the elec­tion: Mosque­da is eas­i­ly defeat­ing Wil­son, with a per­cent­age almost iden­ti­cal to her fin­ish in the August 2021 Top Two elec­tion. Mosque­da has 59.36% of the bal­lots cast, while Wil­son has 40.19%. Mosque­da is present­ly the best per­form­ing can­di­date out of all eight can­di­dates in the four city­wide races. On Elec­tion Night, how­ev­er, her lead over Wil­son looked a lot more like the mar­gin in our poll. Then, she had 52.4%, while Wil­son had 47.1%.

Analy­sis: When we released our polling, I not­ed that there were a sig­nif­i­cant num­ber of unde­cid­ed vot­ers in the race and added that I saw mul­ti­ple plau­si­ble sce­nar­ios for how the con­test could play out. One of those was that we’d see a repeat of the Top Two elec­tion dynam­ics, in which Mosque­da did not per­form all that impres­sive­ly in our pre­elec­tion polling, but went on to do just fine in the actu­al elec­tion, where “not sure” isn’t an option.

And that is the sce­nario that came to pass.

Wil­son has as much sup­port as all of Mosqueda’s chal­lengers col­lec­tive­ly had in the sum­mer, which sug­gests that peo­ple who were look­ing for an alter­na­tive to Mosque­da in the gen­er­al elec­tion hap­pi­ly grav­i­tat­ed to his can­di­da­cy, whilst most of the “not sure” vot­ers sim­ply came home to Mosque­da, pro­pelling her to a com­fort­able vic­to­ry and a sec­ond term on the Seat­tle City Council.

It is not at all uncom­mon for vot­ers in local races (espe­cial­ly races that don’t have par­ti­san labels on the bal­lot) to be not sure who they are vot­ing for pret­ty late in the elec­tion cycle, includ­ing through a lot of the home stretch. Change Research’s Ben Green­field and I dis­cussed this phe­nom­e­non in our post-results release Q&A.

City Council, Position #9 (At-Large)

What the poll found: Of the two final­ists for Seat­tle City Coun­cil Posi­tion #9, the coun­cil’s oth­er at-large posi­tion, our poll found a four point lead for Fre­mont Brew­ing cofounder Sara Nel­son. Nel­son received 41% sup­port in the sur­vey, while her oppo­nent, author and activist Nikki­ta Oliv­er, received 37%. 21% said they were unde­cid­ed and 2% said they would not vote.

What hap­pened in the elec­tion: Dur­ing the vot­ing peri­od, Nel­son surged out to a big lead over Oliv­er, mir­ror­ing what hap­pened in the August elec­tion. On Elec­tion Night, Nel­son had 60.31% of the vote and Oliv­er was under forty per­cent, with 39.46%. But as bal­lots were count­ed, Oliv­er closed the gap, and the race is now clos­er… not as close as the City Attor­ney race, but cer­tain­ly clos­er than it was. Nel­son cur­rent­ly stands at 53.87%; Oliv­er has 45.96%.

Analy­sis: In the August Top Two elec­tion, Oliv­er was even­tu­al­ly able to over­come Nel­son’s Elec­tion Night lead and claim the first place spot thanks to late bal­lots swing­ing sev­er­al suc­ces­sive counts in their favor. How­ev­er, they weren’t able to do that in this final round because Nel­son jumped out to a much big­ger lead.

In August, Nel­son’s Elec­tion Night lead over Oliv­er was 7.4%, which was not an insur­mount­able spread. But in this round, Nel­son’s Elec­tion Night lead was over twice as big: 20.85%. And so, even though Oliv­er out­per­formed Nel­son in the late bal­lots again, it was­n’t enough to change the out­come of the race.

For Oliv­er to pull off anoth­er come­back, they would have had to have been clos­er to Nel­son in the ini­tial returns. But that sce­nario did­n’t come to pass.

Our polling found that the high­est geo­graph­ic con­cen­tra­tions of unde­cid­ed vot­ers in this race were in neigh­bor­hoods like West Seat­tle, Mag­no­lia, and down­town. The poll found that Nel­son was already doing real­ly well in North Seat­tle neigh­bor­hoods like Bit­ter Lake while Oliv­er had firm, enthu­si­as­tic sup­port in neigh­bor­hoods like the Rainier Val­ley and Bea­con Hill.

If you look at the Elec­tion Night precinct lev­el data (visu­al­ized here thanks to Jason Weill), you can see that Nel­son won most of the ear­ly vote in those neigh­bor­hoods with the most not sure vot­ers — like West Seat­tle. That accounts, at least in part, for how Nel­son pulled off the gen­er­al elec­tion surge.

Seattle School Board races (Districts #4, #5, and #7)

What the poll found: Of the six final­ists for Seat­tle School Board races, our poll found Vivan Song-Maritz, Michelle Sar­ju, and Bran­don Hersey with leads. Sar­ju and Hersey had large leads of more than twen­ty-five points over oppo­nents Dan Hard­er and Gen­e­sis Williamson, while Song-Maritz had a ten point lead over her oppo­nent Lau­ra Marie Rivera. Majori­ties or near majori­ties of vot­ers said they were unde­cid­ed. Small­er per­cent­ages said they would not vote.

What hap­pened in the elec­tion: Song-Maritz, Sar­ju, and Hersey are all win­ning, and all are win­ning by mar­gins exceed­ing those in the poll, which is to be expect­ed con­sid­er­ing how many unde­cid­ed vot­ers there were. Hersey has the biggest mar­gin of vic­to­ry because his oppo­nent dropped out months ago, with 92.71% of the vote to Williamson’s 6.63%. Michelle Sar­ju has 85.14% of the vote to Dan Hard­er’s 14.63%. And Song-Maritz has 71.97%, while Rivera has 27.67%.

Analy­sis: Our polling showed that there were three clear fron­trun­ners in these Seat­tle School Board races, and all of them were well posi­tioned for victory.

Not sur­pris­ing­ly, the unde­cid­ed vot­ers flocked to the fron­trun­ners and each of them jumped out to big leads in the ini­tial returns that they did not relinquish.

Song-Maritz was ini­tial­ly at 67% and was able to go up to 71% in the late bal­lots. Sar­ju and Hersey, mean­while, start­ed out in the eight­ies and nineties on Elec­tion Night, so their races saw less move­ment, but even they picked up sup­port as the count­ing went along, while Hard­er and Williamson slipped.

Hersey was the only incum­bent on the bal­lot, with Song Mar­tiz and Rivera hav­ing oust­ed appoint­ed board­mem­ber Erin Dury in the August Top Two election.

Final thoughts

Although our poll could­n’t pre­dict what would actu­al­ly hap­pen in the elec­tion, it did have the poten­tial to indi­cate who might win, as explained in the intro­duc­tion above. And those indi­ca­tions turned out to be right across the board.

All sev­en of the can­di­dates who placed first in our sur­vey are fin­ish­ing first in their races, and all sev­en of the can­di­dates who placed sec­ond are fin­ish­ing second.

That con­tin­ues the pat­tern we saw last year in our statewide polling, in which every result matched up with what the polling indi­cat­ed could hap­pen.

Our Octo­ber 2021 gen­er­al elec­tion sur­vey end­ed up being the only pub­lic poll in Seat­tle dur­ing the final month of the 2021 elec­tion cycle. While it was nice to see Elway/Crosscut and Strate­gies 360/KOMO do polls in Sep­tem­ber, we would have liked to have had anoth­er Octo­ber sur­vey to com­pare our data to.

With cred­i­ble pub­lic polling get­ting hard­er and hard­er to find, what we may do next time around, resources per­mit­ting, is sim­ply try to con­duct a sec­ond pre­elec­tion poll right before the vot­ing ends. That way, we won’t have to hope that anoth­er trust­ed enti­ty will come along and pro­vide fresh data.

Polls cost a lot of mon­ey to com­mis­sion, and a lot of thought­ful work goes into design­ing them. If you appre­ci­ate the research NPI is doing, we invite you to become a mem­ber of our orga­ni­za­tion. You can donate annu­al­ly or monthly.

If you belong to an orga­ni­za­tion that would be inter­est­ed in join­ing NPI, we offer Asso­ciate Mem­ber­ships and spon­sor­ship opportunities.

Please get in touch with us for more infor­ma­tion.

Survey methodology

• Change Research, a Pub­lic Ben­e­fit Cor­po­ra­tion based in Cal­i­for­nia, sur­veyed 617 like­ly Novem­ber 2021 Top Two elec­tion vot­ers in Seat­tle from Tues­day, Octo­ber 12th to Fri­day, Octo­ber 15th on behalf of the North­west Pro­gres­sive Insti­tute. All respon­dents par­tic­i­pat­ed online.
• Change used tar­get­ed adver­tise­ments on Face­book, tar­get­ed adver­tise­ments on Insta­gram, and text mes­sages sent via the echo19 and/or Scale To Win plat­forms to cell phone num­bers list­ed on the vot­er file for indi­vid­u­als who qual­i­fied for the survey’s sam­ple uni­verse, based on their vot­er file data.
• Regard­less of which of these sources a respon­dent came from, they were direct­ed to a sur­vey host­ed on SurveyMonkey’s web­site. Ads placed on social media tar­get­ed all adults liv­ing in Seat­tle. Those who indi­cat­ed that they were not reg­is­tered to vote were terminated.
• As the sur­vey field­ed, Change used dynam­ic online sam­pling: adjust­ing ad bud­gets, low­er­ing bud­gets for ads tar­get­ing groups that were over­rep­re­sent­ed and rais­ing bud­gets for ads tar­get­ing groups that were under­rep­re­sent­ed, so that the final sam­ple was rough­ly rep­re­sen­ta­tive of the pop­u­la­tion across dif­fer­ent groups.
• The sur­vey was con­duct­ed in Eng­lish, and has a mod­eled mar­gin of error of 4.1% at the 95% con­fi­dence interval.

Addi­tion­al infor­ma­tion about the method­ol­o­gy is avail­able here. Final­ly, if you missed it, be sure to read this Q&A with Change Research’s Ben Green­field about how to read assess the gen­er­al elec­tion poll find­ings that we released.