Screenshot from an Echelon Insights slide deck
A slide provided by Echelon Insights from a deck created to promote findings from a skewed survey it conducted for Concerned Taxpayers of Washington State

Wash­ing­ton State’s right wing move­ment is once again cir­cu­lat­ing skewed sur­vey data to make the argu­ment that one of its top can­di­dates this cycle has far bet­ter prospects than they real­ly have, in what appears to be a reprise of their failed attempt in 2022 to put the Ever­green State in play for Demo­c­ra­t­ic Sen­a­tor Pat­ty Mur­ray’s Repub­li­can chal­lenger Tiffany Smi­ley, who Mur­ray dis­patched with rel­a­tive ease in the midterms.

This time around, the poll­ster deploy­ing bad data to try to shape the nar­ra­tive is Ech­e­lon Insights, rather than Trafal­gar, and the intend­ed ben­e­fi­cia­ry is Repub­li­can guber­na­to­r­i­al hope­ful Dave Reichert, rather than Smiley. 

Ech­e­lon’s March 2024 sur­vey for “Con­cerned Tax­pay­ers of Wash­ing­ton State” has Reichert up nine points over Demo­c­ra­t­ic Attor­ney Gen­er­al Bob Fer­gu­son in a head-to-head matchup. Ech­e­lon also claims that a plu­ral­i­ty of vot­ers favor Repub­li­cans in a gener­ic leg­isla­tive bal­lot ques­tion. Those find­ings are total­ly at odds with our own recent pub­lic opin­ion research and oth­er cred­i­ble pri­vate polling that we’ve seen.

Ech­e­lon obtained these find­ings pri­mar­i­ly by con­struct­ing a skewed statewide sam­ple that does­n’t have enough Demo­c­ra­t­ic and pro­gres­sive vot­ers in it to prop­er­ly resem­ble the like­ly Wash­ing­ton State Novem­ber elec­torate, just as it did last year when it polled for the right wing group Future42 and Bran­di Kruse’s pod­cast.

This is a gam­bit that our team has seen before. 

Pub­lish­ing skewed data in an effort to cre­ate a self-ful­fill­ing prophe­cy did­n’t work last cycle for Tiffany Smi­ley and we don’t think it’ll work this time around, either. 

But Repub­li­cans — who are tired of being on the los­ing end of the longest cur­rent streak of con­sec­u­tive guber­na­to­r­i­al elec­tion vic­to­ries by one polit­i­cal par­ty any­where in the coun­try — are desperate. 

They crave rel­e­vance and dream of flip­ping the Ever­green State despite Don­ald Trump’s stran­gle­hold on their par­ty, which nowa­days func­tions more like a cult. And so they’re pay­ing Ech­e­lon Insights to peri­od­i­cal­ly pub­lish infor­ma­tion that sug­gests a favor­able elec­toral land­scape for Repub­li­cans that sim­ply does­n’t exist in reality.

If last cycle’s pat­tern holds, then we can expect a few oth­er Repub­li­can-aligned firms to even­tu­al­ly mate­ri­al­ize and drop sur­veys that show Reichert ahead or tied with Demo­c­ra­t­ic fron­trun­ner Bob Fer­gu­son. That’s what we saw in the midterms — it was the final act in their data pol­lu­tion gam­bit. First, Trafal­gar pro­duced a series of peri­od­ic sur­veys that were favor­able to NRSC recruit Smi­ley, and then in the home stretch, firms like Moore Infor­ma­tion, co/efficient, and KACon­sult­ing joined the par­ty, con­tribut­ing their own flawed data to make it look like Smi­ley was com­pet­i­tive when she real­ly was­n’t.

Cred­i­ble pre­elec­tion polling from non­par­ti­san and Demo­c­ra­t­ic-aligned poll­sters had found Mur­ray well ahead, but Repub­li­cans were able to “flood the zone” and obtain a lot of favor­able media cov­er­age with their bad polling. NPI worked before, dur­ing, and after the elec­tion to hold them account­able, and we are doing so again now. 

In a moment, I’ll share the deck that Ech­e­lon released and invite you to scru­ti­nize it. Then we’ll dis­cuss why the find­ings lack cred­i­bil­i­ty and should not be tak­en seriously.

How we evaluate polls

First, a few words about how we eval­u­ate polls.

We believe that sub­jec­tive orga­ni­za­tions are per­fect­ly capa­ble of con­duct­ing objec­tive research, so we won’t reflex­ive­ly dis­miss data from a sur­vey mere­ly because it comes from right wing spon­sor. How­ev­er, we are stick­lers for the sci­en­tif­ic method, so we will look to see if that spon­sor and their poll­ster adhered to best prac­tices for obtain­ing sound data, or whether they cut corners.

The key to accu­rate, cred­i­ble polling is neu­tral ques­tions asked of a rep­re­sen­ta­tive sam­ple. You’ve got to have both, or your data is going to be skewed. You can’t find out what peo­ple think if you tell them what to think first, which is why neu­tral ques­tions are impor­tant. And your sam­ple needs to prop­er­ly resem­ble the elec­torate or the pop­u­la­tion of the area you’re sur­vey­ing, or the find­ings sim­ply aren’t going to accu­rate­ly reflect what pub­lic opin­ion is.

It’s extreme­ly dif­fi­cult to write neu­tral ques­tions and it can also be tough to build rep­re­sen­ta­tive sam­ples. But it can be done, as we’ve demon­strat­ed over the past decade in our work with Pub­lic Pol­i­cy Polling, a well-regard­ed poll­ster, and more recent­ly also with Change Research and Civiqs, two oth­er trust­ed firms we reg­u­lar­ly work with. 

Survey details

Accord­ing to Ech­e­lon Insights, this was a sur­vey of 600 reg­is­tered vot­ers in Wash­ing­ton State “aged 18 or old­er”, con­duct­ed for Con­cerned Tax­pay­ers of Wash­ing­ton State. They were inter­viewed “from March 18–21, 2024 using a mix of live tele­phone calls (30% land­line) and text to online (70%), matched to the L2 Wash­ing­ton vot­er file.” 

The brief method­ol­o­gy state­ment goes on to say: “The sam­ple was weight­ed to reflect pop­u­la­tion bench­marks for Novem­ber 2024 reg­is­tered vot­ers for gen­der, age, race/ethnicity, edu­ca­tion, region, par­ty, turnout prob­a­bil­i­ty, and 2020 vote. The mar­gin of error for a ran­dom sam­ple of 600 vot­ers is ± 4.7%.”

Ech­e­lon is based in Vir­ginia, and is Republican-aligned.

(As you can prob­a­bly guess, a Repub­li­can-aligned firm is par­ti­san and works main­ly with right wing clients. A Demo­c­ra­t­ic-aligned firm like one of our poll­sters is the equiv­a­lent, also par­ti­san and works with pro­gres­sive clients.)

Iron­i­cal­ly, accord­ing to Buz­zFeed, the firm was cre­at­ed to “fix the chron­ic Repub­li­can prob­lem of bad polling.” Co-founders Patrick Ruffi­ni and poll­ster Kris­ten Soltis Ander­son were frus­trat­ed that many Repub­li­can cam­paigns were rely­ing on inter­nal polls with skewed sam­ples that weren’t accu­rate­ly gaug­ing cur­rent elec­toral dynam­ics, which was lead­ing to Elec­tion Night sur­pris­es in a num­ber of high-pro­file Repub­li­can campaigns.

And yet, here they are, gen­er­at­ing the sort of bad data that they were call­ing on Repub­li­cans to stop wast­ing mon­ey cre­at­ing around a decade ago. 

A copy of their deck sum­ma­riz­ing select­ed results is below.


Evaluating the sample

As was the case last year, the sam­ple suf­fers from a fatal flaw: it does­n’t have enough Demo­c­ra­t­ic and pro­gres­sive vot­ers in it. On the “Sur­vey Demo­graph­ics” slide, you can see that Ech­e­lon iden­ti­fies 28% of the respon­dents as Repub­li­can, 34% as Inde­pen­dent, 34% as Demo­c­ra­t­ic, and 3% as “some­thing else.” 

With respect to ide­ol­o­gy, 35% are iden­ti­fied as “con­ser­v­a­tive”, 32% as “mod­er­ate” (which is a prob­lem­at­ic label, not an ide­ol­o­gy), and 29% as “lib­er­al”. Ech­e­lon claims that a total of 42% of its sam­ple are Repub­li­can or lean Repub­li­can, 10% are pure inde­pen­dent, and 38% are Demo­c­ra­t­ic or lean Democratic.

Wash­ing­ton State does­n’t have vot­er reg­is­tra­tion by par­ty. Nev­er­the­less, we can see from elec­tion results that a major­i­ty of Wash­ing­to­ni­ans are Demo­c­ra­t­ic or lean Demo­c­ra­t­ic — not a plu­ral­i­ty. A prop­er­ly rep­re­sen­ta­tive sam­ple needs to match Wash­ing­ton’s elec­toral pro­file. Data going back many cycles shows that most Wash­ing­ton vot­ers reli­ably vote for Democ­rats at the top of the ticket. 

For exam­ple, in the 2020 pres­i­den­tial elec­tion, 57.97% of Wash­ing­ton vot­ers sup­port­ed Joe Biden, and in the 2022 U.S. Sen­ate elec­tion, 57.15% of Wash­ing­ton vot­ers sup­port­ed Pat­ty Mur­ray. Repub­li­cans haven’t won a guber­na­to­r­i­al elec­tion since 1980, the state’s pres­i­den­tial elec­tors since 1984, or a U.S. Sen­ate race since 1994.

A prop­er­ly rep­re­sen­ta­tive statewide sam­ple for Wash­ing­ton should con­sist of at least 54% or so of Biden vot­ers, and ide­al­ly clos­er to 57%. Our Feb­ru­ary 2024 statewide sur­vey, which was con­duct­ed by Pub­lic Pol­i­cy Polling, con­sist­ed of 56% Biden vot­ers, 37% Trump vot­ers, and 7% who report­ed vot­ing for some­one else or did­n’t vote. 

41% of the respon­dents in that poll iden­ti­fied as Democ­rats, 23% as Repub­li­cans, and 36% as independents. 

Con­trast that with Ech­e­lon’s sam­ple. They have too few Demo­c­ra­t­ic vot­ers, which means their sam­ple resem­bles a Wash­ing­ton elec­torate that does­n’t exist.

Though Ech­e­lon omit­ted the 2020 vote break­down from its slide deck, Ech­e­lon’s Patrick Ruffi­ni did pro­vide, in response to a tweet we post­ed, the fol­low­ing screen­shot, which shows the 2020 vote break­down in the sur­vey… 55% Biden, 37% Trump. 

A screen­shot of toplines pro­vid­ed by Ech­e­lon’s Patrick Ruffini

So that at least clears that up. But who are those Biden vot­ers? That’s impor­tant: mere­ly hav­ing align­ment with pres­i­den­tial vote results does­n’t mean a sam­ple is representative. 

Across the entire­ty of the Ech­e­lon sur­vey, Demo­c­ra­t­ic can­di­dates look much weak­er than they do in our polling, which uti­lizes prop­er­ly rep­re­sen­ta­tive sam­pling. Fail­ing to get the par­ti­san bal­ance of the elec­torate right is pub­lic opin­ion research mal­prac­tice, and giv­en that Ech­e­lon has con­tin­ued to under­sam­ple Democ­rats even after we called them out for doing so last sum­mer and urged them to do bet­ter next time, we can only con­clude that they’re doing this delib­er­ate­ly to try to dri­ve a pro-Reichert narrative.

Comparing Echelon’s findings to NPI’s most recent findings

NPI’s most recent sur­vey field­ed in Feb­ru­ary. Each link below goes to the find­ing for that con­test, with the full ques­tion text and toplines and accom­pa­ny­ing analysis. 

Gov­er­nor (Head to head) 

Gov­er­nor (Top Two field)


U.S. Sen­ate

Gener­ic leg­isla­tive ballot

You can see that in each con­test, Ech­e­lon found the Demo­c­ra­t­ic can­di­date / Demo­c­ra­t­ic tick­et in a weak­er posi­tion than we did. The deficit ranges from six to six­teen points com­pared to our Feb­ru­ary 2024 statewide poll. And even though the num­ber of “not sure” vot­ers is also high­er in Ech­e­lon’s poll, Repub­li­can can­di­dates and the gener­ic Repub­li­can leg­isla­tive tick­et nev­er­the­less per­formed at lev­els com­pa­ra­ble to our survey. 

Ech­e­lon also asked about Bri­an Hey­wood and Jim Wal­sh’s slate of ini­tia­tives, but curi­ous­ly, did not test the actu­al bal­lot titles. Instead, respon­dents were asked some­what biased ques­tions which don’t meet our cri­te­ria for neu­tral­i­ty. Con­sid­er­ing that the sam­ple is also skewed, those find­ings aren’t use­ful. Amus­ing­ly, despite the skew of the sam­ple and the some­what biased ques­tions, the respons­es still weren’t all that great for the slate. 

Credible polling versus bad data

Cred­i­ble polling like NPI’s adheres to the sci­en­tif­ic method. While it can­not pre­dict elec­tion results, it can sug­gest what is most like­ly to hap­pen with a rea­son­ably strong degree of con­fi­dence. It’s very impor­tant to under­stand that distinction. 

Our pub­lic opin­ion research has a track record of excel­lence that dates back to when we began com­mis­sion­ing sur­veys to gauge vot­er sen­ti­ment in upcom­ing Wash­ing­ton State elec­tions. At every lev­el that we’ve polled — statewide, con­gres­sion­al, leg­isla­tive, local — the can­di­dates who’ve led in our sur­veys have gone on to win.

The most recent exam­ple is our sur­vey of the Wash­ing­ton Repub­li­can pres­i­den­tial pri­ma­ry. Work­ing with Civiqs, we found Don­ald Trump at 77% and Nik­ki Haley at 8%, with x% unde­cid­ed. In the pri­ma­ry, which was recent­ly cer­ti­fied, Trump received about 76% and Haley got 19%. That’s about as spot-on as elec­toral polling can be! 

Observers from across the ide­o­log­i­cal spec­trum know that NPI’s polling has con­sis­tent­ly been on the mark; some have called it the gold stan­dard. There’s a sim­ple rea­son for this suc­cess: we are stick­lers for fol­low­ing the sci­en­tif­ic method. We com­mis­sion research because we want to know what peo­ple think, not because we want to gen­er­ate favor­able num­bers for pro­gres­sive caus­es or candidates.

Those who put their faith in bad data, mean­while, end up look­ing pret­ty out of touch.

Take Jason Rantz, a local right wing talk radio host employed by Bon­neville. He went all in on Trafal­gar­’s 2022 WA-Sen polling and was hyp­ing it excit­ed­ly on the air and on posts for right up until the elec­tion demon­strat­ed it was total bunk. Here’s one from Octo­ber 30th, titled WA Sen­ate race sta­tis­ti­cal­ly tied as Tiffany Smi­ley surges:

The Wash­ing­ton Sen­ate race is now sta­tis­ti­cal­ly tied, accord­ing to a new Trafal­gar Poll.

Repub­li­can new­com­er Tiffany Smi­ley earned 48.2% sup­port to 30-year incum­bent Pat­ty Murray’s 49.4%. The mar­gin of error is 2.9%, mak­ing this a sta­tis­ti­cal dead heat. 2.4% of vot­ers are undecided.

It appears the surge is com­ing from inde­pen­dent vot­ers, in line with the sig­nif­i­cant swing expressed in a recent Seat­tle Times poll. Trafal­gar over­sam­pled Democ­rats at 44.2% ver­sus 33.4% of Repub­li­can respon­dents. Inde­pen­dents account­ed for 22.4% of the respon­dents. And there were more female (53.1%) respon­dents than male (46.9%).

Smi­ley is on a statewide bus tour and her stops have been stand­ing-room only. Con­trast that with Mur­ray who, even with guests like Sen­a­tor Eliz­a­beth War­ren, has sparse participation.

Rantz kept the drum­beat going a few days lat­er with this piece — Big names step up sup­port for Tiffany Smi­ley as polls show ‘dead heat’ in sen­ate race.

Of course, cred­i­ble polling did not show a dead heat in Wash­ing­ton’s U.S. Sen­ate race. Rantz ignored the cred­i­ble polling and put his faith — at least pub­licly — in bad data. A few weeks lat­er, he was forced to lament Tiffany Smi­ley’s con­ces­sion and bemoan that “too few peo­ple chose to take this seri­ous­ly and let her [Smi­ley] down.” 

He did not apol­o­gize for hav­ing played a lead­ing role in pro­mot­ing garbage right wing polling that was feed­ing a fab­ri­cat­ed nar­ra­tive about Wash­ing­ton’s U.S. Sen­ate race.

An opportunity squandered

Wash­ing­ton State and the Pacif­ic North­west could ben­e­fit from hav­ing a Repub­li­can-aligned poll­ster that con­tributes to our region’s body of cred­i­ble pub­lic opin­ion research, com­ple­ment­ing the work that NPI and its poll­sters do from the pro­gres­sive side and the work that media orga­ni­za­tions like The Seat­tle Times and Cross­cut do with their poll­sters (usu­al­ly Sur­veyUSA and Elway Research, respectively.) 

Ech­e­lon Insights could have been that firm. 

Sad­ly, instead, it has cho­sen to behave like Trafal­gar or Moore and use its infra­struc­ture and rep­u­ta­tion to ped­dle bad data. That’s extreme­ly dis­ap­point­ing. And unfor­tu­nate­ly, this is the sort of behav­ior that goes total­ly unpun­ished nowa­days in Don­ald Trump’s Repub­li­can Par­ty, which pre­tends to care a lot about the truth, but in prac­tice has become a fire­hose of mis­in­for­ma­tion and disinformation. 

About the author

Andrew Villeneuve is the founder and executive director of the Northwest Progressive Institute, as well as the founder of NPI's sibling, the Northwest Progressive Foundation. He has worked to advance progressive causes for over two decades as a strategist, speaker, author, and organizer. Andrew is also a cybersecurity expert, a veteran facilitator, a delegate to the Washington State Democratic Central Committee, and a member of the Climate Reality Leadership Corps.

Adjacent posts

2 replies on “Poll Watch: Right wing firm Echelon Insights drops another skewed survey to boost Dave Reichert’s gubernatorial ambitions”

  1. Andrew, as always an inter­est­ing post. I’ve been around the inter­sec­tion of media and pol­i­tics a while now, and when this poll appeared I thought (being right of cen­ter) that it was too good to be true. But after exam­in­ing some of the same data that caught your atten­tion, I’m now think­ing the poll might be catch­ing a trend: that as Seat­tle vot­ers pulled away from an out of touch city coun­cil and swung toward the cen­ter in last year’s city coun­cil elec­tions, the state’s vot­ers seem poised to do the same thing. I ini­tial­ly thought the poll under­count­ed Democ­rats, but I noticed that when asked who they vot­ed for in 2020, the respon­dents favored Biden by 18 per­cent­age points. Biden actu­al­ly won by.….19 points. Pret­ty close. So what’s hap­pen­ing? My guess is that just as many self described inde­pen­dents moved into the Demo­c­ra­t­ic camp dur­ing the Trump pres­i­den­cy, they are now migrat­ing back to the inde­pen­dent camp. We’ll see if this is indeed a trend or a blip, but the fact that the Dems in Olympia vot­ed to repeal their own law restrict­ing police chas­es sug­gests that they real­ize they over­played their hand, and are try­ing to con­tain the dam­age, lest they lose the three “mon­ey” ini­tia­tives in Novem­ber. Thanks for the chance to offer feedback.

    1. Hi John. We aren’t see­ing any evi­dence across our research to sup­port your the­o­ry that “self described inde­pen­dents” are “migrat­ing” towards Repub­li­can can­di­dates and caus­es. When we ask Biden vot­ers if there is any­thing the Repub­li­cans can do to earn their vote in the next elec­tion, most say no. Those who say yes are asked to com­ment. It turns out that the vast major­i­ty of these yes folks are also nos, how­ev­er. The sen­ti­ments we hear from them go like this: Yeah, if the Repub­li­can Par­ty repu­di­at­ed Trump and every­thing he stands for, and found a good can­di­date who was­n’t an extrem­ist, I’d con­sid­er vot­ing Repub­li­can in a future elec­tion, but not this next one. No way, no how.

      The Repub­li­can Par­ty’s trans­for­ma­tion into a polit­i­cal cult cen­tered around Don­ald Trump has been cat­a­stroph­ic for sev­er­al con­sec­u­tive elec­tion cycles. Even in the midterms, when Trump was­n’t on the bal­lot, Trump­ism was… and Wash­ing­ton State Repub­li­cans did very poorly. 

      The Leg­is­la­ture’s deci­sion to vote I‑2113 into law was a strate­gic choice to keep it off the bal­lot to deny Repub­li­cans a mes­sag­ing tool and guar­an­tee the next Leg­is­la­ture’s abil­i­ty to mod­i­fy the law again (which they first did last ses­sion, before the I‑2113 sig­na­ture dri­ve). It was not an admis­sion of over­reach. And the out­comes of last year’s Seat­tle city coun­cil races are not a fore­shad­ow­ing of any­thing. By that same log­ic, Adam Fort­ney’s loss in Sno­homish Coun­ty and Nadine Wood­ward’s loss in Spokane are a very bad omen for Republicans. 

      Can­di­date elec­tions turn on iden­ti­ty and trust, not ide­ol­o­gy. Way too much polit­i­cal analy­sis is ide­ol­o­gy-dri­ven or ide­ol­o­gy-cen­tric. What research shows us, though, is that ide­ol­o­gy is only one facet of peo­ple’s polit­i­cal iden­ti­ty. It is impor­tant, to be sure, but there are oth­er facets, oth­er scales, to be con­sid­ered, like means and ends or the desired rate and speed of change.

Comments are closed.