Biased statistics: Woke-washing the I-5 Boondoggle
By Joe Cortright
The Oregon and Washington transportation departments are using a biased, unscientific survey to market their $5 billion I-5 freeway widening project.
The survey over-represents daily bridge users by a factor of 10 compared to the general population.
The IBR survey undercounts lower and middle income households and people of color and overstates the opinions of White non-hispanics, higher income households, and Clark County residents
As we’ve noted, highway builders are increasingly engaging in woke-washing, claiming—after decades of experience in which freeway projects have devastated communities of color and destroyed city neighborhoods across the country—that wider freeways will somehow be a good thing for low income people and people of color.
The latest example of this comes from the sales campaign to promote the $5 billion I-5 freeway widening between Portland and Vancouver Washington, misleadingly branded as the “Interstate Bridge Replacement” (IBR) project. The reality is pretty simple: the primary beneficiaries of a wider roadway would be higher income, overwhelmingly white commuters who drive daily from suburbs in Washington State to jobs in Oregon. As we documented last month, the peak hour drive-alone car commuters who cross the I-5 and I-205 bridges from Washington State to jobs in Oregon are whiter and wealthier than the region’s population, with median incomes of $106,000, and 86 percent non-Hispanic whites.
But the IBR project has carefully constructed an alternate reality in which this car-centric freeway widening project is really something that benefits low income people and people of color. The project’s promotional materials—which actually don’t show the project, or acknowledge its price tag, or the fact that it will charge tolls to bridge users—prominently features stock images of people of color.
Here’s what we mean by “woke-washing.” The project’s home page featured this image . . .
In addition to its woke imagery, the IBR project supplements this messaging with a pseudo-scientific web-based survey which purports to show that the project is really for lower income people of color.
Selling a $5 billion freeway widening with a woke-washed fable
The IBR staff have developed a fictional “just so” story of how the freeway widening project is needed to help low income households and people of color, who’ve moved to Clark County for cheaper housing, but have to travel to jobs and other opportunities in Oregon. The survey is grounded, not in actual scientific data, but the project’s own unscientific and biased web-based survey.
Here is IBR staff person Jake Warr, making this false claim to the January 20, 2022 Executive Steering Committee meeting:
One thing that really came out through this survey that I want to highlight is when we . . . asked how often people drive across the bridge, we found a higher percentage of folks who identified with a race or ethnicity besides white or or in addition to white/Caucasian, the non-white respondents really reported more frequently traveling across the bridge.
So that 53 percent‑that’s listed there, 53 percent‑of our of our BIPOC survey respondents reported traveling across the bridge either daily or a few times a week. That’s compared to closer to 40 percent for the white respondents.
IBR’s unscientific and biased web-based survey.
So just something that that really kind of drives home a point that we’ve suspected. It provides further data that you, we’ve seen a trend in our region of folks of color being pushed to further areas of the region, being pushed north of the river, or seeking out more affordable housing north of the Columbia River, but still relying on services jobs etc, in Multnomah County.
And so, there’s that piece that I think this speaks to. We also suspect that related to Covid, as people were answering this question in the context this pandemic, there might be some explanation there, as we know that BIPOC individuals tend to be, disproportionately rely needing to work still in a location and not be able to work from home.
That might have contributed to this but just something that we really found was was a poignant data piece to point out.
The trouble is, this claim is easily disproved by referring to valid survey data from the Census Bureau which shows that commuters across the I-5 and I-205 bridges are actually disproportionately white, and higher income. Low income workers, and those of color, are dramatically under-represented among bridge commuters.
A biased, unscientific survey from the IBR
The trouble with web-based surveys is they suffer from self-selection bias. Only highly motivated people take such surveys, and the opinions, experience and demographics of these people differ substantially, and systematically, from the general population. As a result, it’s simply invalid to make statistical claims (such as people of color are more likely to use the bridge frequently). That’s especially true when there’s valid scientific data from the American Community Survey, which shows exactly the opposite: peak hour users (for whom the bridge is being expanded) are 86 percent non-Hispanic white and have average incomes of $106,000).
To see just how biased the unscientific IBR web-survey is, we can compare it to other surveys conducted with more valid methodologies. The correct way to do surveys is with an random selection methodology; the IBR actually commissioned such a survey in 2020. In its random survey of more than 900 Portland area voters, 13 percent of respondents reported never crossing the I-5 bridge over the Columbia, compared to just 1 percent in the unscientific online survey. The random survey of voters showed only 5 percent of respondents crossed the I-5 bridge every day, compared to 19 percent in the unscientific online survey. As a result, the unscientific online survey implies the ratio of daily users to non users is 19 to 1 (there are 19 times as many daily users as never users), while the random survey shows that there are two and a half times as many non-users as daily users of the I-5 bridge. That means that the unscientific survey overweights the role—and opinions—of daily users relative to non users by more than an order of magnitude relative their share of the overall population of the Portland metropolitan area.
Source: IBR Community Opinion Survey, 2020
Demographic bias in the IBR unscientific web survey
A quick look at the American Community Survey, which is conducted annually by the Census Bureau, shows that the demographics of the IBR’s unscientific web-based survey are dramatically different from the metro area.
One essential for surveys is that participants should be randomly selected. If they’re not randomly selected, there’s little guarantee that the results will be representative of the larger population. One of the sure tells of a non-random survey is that the characteristics of survey participants don’t match up well with the characteristics of the overall population of the area being surveyed. That’s the case here. The IBR survey systematically over-represents some groups, and systematically underrepresents others, which should cast doubt on the validity of its results. The survey systematically over-represents white, non-Hispanic people, higher income households, and residents of Washington State, and systematically under-represents people of color, low and moderate income households, and Oregon residents. Here are the details.
Income: Higher incomes over-represented. The respondents to the unscientific web-survey are much higher income than the overall population. Some 44 percent of survey respondents had household incomes over $100,000; only 38 percent of the region’s households had incomes that high.
Race and Ethnicity: People of color under-represented. The respondents to the unscientific web-based survey are much more likely to be non-Hispanic white than the overall population; some 85 percent of survey respondents were non-Hispanic white compared to 72 percent of the region’s population. People of color were 28 percent of the region’s population, but only 15 percent of survey respondents. People of color were undercounted by almost half in this unscientific survey.
Residence: Clark County over-represented. The respondents to the unscientific web-based survey are disproportionately residents of Clark County. Clark County accounts for less than 20 percent (488,000 of the region’s 2.5 million residents) but accounts for 43 percent of those taking the survey. Clark County resident views are given more than double the weight of view of other of the region’s residents in this unscientific survey.
Age: Young people significantly under-represented. There’s also a strong generational bias: only 5 percent of survey respondents are under 25, compared to nearly 30 percent of the population. And these people will be the ones who have to live with the environmental consequences of the project.
No doubt the highway agencies will point with pride to the large number of completed surveys–more than 9,000 to date. But large numbers are irrelevant if you don’t have a random sample. For a metropolitan area the size of Portland, you need only about 400 to 800 survey participants to come up with statistically valid results, if you have a random sample. If you don’t have a random sample, then even very large numbers (and IBR surveyed only about one-third of one percent of the region’s residents) just aren’t meaningful. The underlying problem that invalidates the survey is called “Self-Selection Bias.” Because this isn’t a true random survey, and because respondents choose whether to participate, there’s no guarantee that the survey data reflect the views (and experiences) of the larger population. Because those who are predisposed to care about this issue are likely to differ systematically from the rest of the population, the survey produces results that are biased.
Not asking the most important question: Who wants to pay a toll?
There’s a lot more to dislike about the survey beyond its poor quality sampling strategy and biased sample. The questions posed in the survey don’t get at the real issues raised by the freeway widening project. The project’s financial plan shows that it won’t be built without tolls—something you’d be hard-pressed to learn from any of the “public information” work. The last estimates prepared for the Columbia River Crossing showed I-5 tolls would be a minimum of $2.30 during off peak hours, rising to $3.25 during rush hour, with additional surcharges for those who didn’t buy transponders for their cars in advance. The survey didn’t reveal these toll rates, or ask people whether they might prefer a smaller, less expensive bridge with lower tolls, to a larger one with these high tolls, or whether they’d really rather keep the existing bridge if it meant they could avoid tolling altogether. Despite the fact that the survey avoided talking about tolls, many survey respondents raised the question in answering open-ended questions.
It’s rather like a taste test survey that asks people whether they’d prefer filet mignon to a hot dog, without revealing the price tag of either alternative. For a project that claims so prominently to care about “centering equity,” failing to reveal that people might have to pay on the order of $1,600 per month to commute daily across this bridge is a monumental omission. But it’s no accident: the project’s “public information” campaign is designed is an intentionally misleading way to manufacture consent, not to accurately measure public attitudes.
Surveys can be a useful way to gauge public opinion, if they’re undertaken in a scientifically valid fashion. But if you aren’t careful, you end up with a classic, garbage-in, garbage-out exercise. That appears to be the case with survey work commissioned by the “Interstate Bridge Replacement” project, a thinly veiled marketing campaign for freeway widening funded by the Oregon and Washington transportation departments—with “communications” consultants reaping more more than $4 million for their services in the past few years.
Biased statistics: Woke-washing the I-5 Boondoggle
The Oregon and Washington transportation departments are using a biased, unscientific survey to market their $5 billion I-5 freeway widening project.
The survey over-represents daily bridge users by a factor of 10 compared to the general population.
The IBR survey undercounts lower and middle income households and people of color and overstates the opinions of White non-hispanics, higher income households, and Clark County residents
As we’ve noted, highway builders are increasingly engaging in woke-washing, claiming—after decades of experience in which freeway projects have devastated communities of color and destroyed city neighborhoods across the country—that wider freeways will somehow be a good thing for low income people and people of color.
The latest example of this comes from the sales campaign to promote the $5 billion I-5 freeway widening between Portland and Vancouver Washington, misleadingly branded as the “Interstate Bridge Replacement” (IBR) project. The reality is pretty simple: the primary beneficiaries of a wider roadway would be higher income, overwhelmingly white commuters who drive daily from suburbs in Washington State to jobs in Oregon. As we documented last month, the peak hour drive-alone car commuters who cross the I-5 and I-205 bridges from Washington State to jobs in Oregon are whiter and wealthier than the region’s population, with median incomes of $106,000, and 86 percent non-Hispanic whites.
But the IBR project has carefully constructed an alternate reality in which this car-centric freeway widening project is really something that benefits low income people and people of color. The project’s promotional materials—which actually don’t show the project, or acknowledge its price tag, or the fact that it will charge tolls to bridge users—prominently features stock images of people of color.
Here’s what we mean by “woke-washing.” The project’s home page featured this image . . .
. . . which is a stock photograph used by hundreds of websites, mostly those focusing on women’s health. (Just an aside: A true health-oriented and equity focused project wouldn’t build a 12-lane wide, 5 mile long freeway guaranteed to increase air pollution and with a long history of destroying neighborhoods.)
In addition to its woke imagery, the IBR project supplements this messaging with a pseudo-scientific web-based survey which purports to show that the project is really for lower income people of color.
Selling a $5 billion freeway widening with a woke-washed fable
The IBR staff have developed a fictional “just so” story of how the freeway widening project is needed to help low income households and people of color, who’ve moved to Clark County for cheaper housing, but have to travel to jobs and other opportunities in Oregon. The survey is grounded, not in actual scientific data, but the project’s own unscientific and biased web-based survey.
Here is IBR staff person Jake Warr, making this false claim to the January 20, 2022 Executive Steering Committee meeting:
The trouble is, this claim is easily disproved by referring to valid survey data from the Census Bureau which shows that commuters across the I-5 and I-205 bridges are actually disproportionately white, and higher income. Low income workers, and those of color, are dramatically under-represented among bridge commuters.
A biased, unscientific survey from the IBR
The trouble with web-based surveys is they suffer from self-selection bias. Only highly motivated people take such surveys, and the opinions, experience and demographics of these people differ substantially, and systematically, from the general population. As a result, it’s simply invalid to make statistical claims (such as people of color are more likely to use the bridge frequently). That’s especially true when there’s valid scientific data from the American Community Survey, which shows exactly the opposite: peak hour users (for whom the bridge is being expanded) are 86 percent non-Hispanic white and have average incomes of $106,000).
To see just how biased the unscientific IBR web-survey is, we can compare it to other surveys conducted with more valid methodologies. The correct way to do surveys is with an random selection methodology; the IBR actually commissioned such a survey in 2020. In its random survey of more than 900 Portland area voters, 13 percent of respondents reported never crossing the I-5 bridge over the Columbia, compared to just 1 percent in the unscientific online survey. The random survey of voters showed only 5 percent of respondents crossed the I-5 bridge every day, compared to 19 percent in the unscientific online survey. As a result, the unscientific online survey implies the ratio of daily users to non users is 19 to 1 (there are 19 times as many daily users as never users), while the random survey shows that there are two and a half times as many non-users as daily users of the I-5 bridge. That means that the unscientific survey overweights the role—and opinions—of daily users relative to non users by more than an order of magnitude relative their share of the overall population of the Portland metropolitan area.
Demographic bias in the IBR unscientific web survey
A quick look at the American Community Survey, which is conducted annually by the Census Bureau, shows that the demographics of the IBR’s unscientific web-based survey are dramatically different from the metro area.
One essential for surveys is that participants should be randomly selected. If they’re not randomly selected, there’s little guarantee that the results will be representative of the larger population. One of the sure tells of a non-random survey is that the characteristics of survey participants don’t match up well with the characteristics of the overall population of the area being surveyed. That’s the case here. The IBR survey systematically over-represents some groups, and systematically underrepresents others, which should cast doubt on the validity of its results. The survey systematically over-represents white, non-Hispanic people, higher income households, and residents of Washington State, and systematically under-represents people of color, low and moderate income households, and Oregon residents. Here are the details.
Income: Higher incomes over-represented. The respondents to the unscientific web-survey are much higher income than the overall population. Some 44 percent of survey respondents had household incomes over $100,000; only 38 percent of the region’s households had incomes that high.
Race and Ethnicity: People of color under-represented. The respondents to the unscientific web-based survey are much more likely to be non-Hispanic white than the overall population; some 85 percent of survey respondents were non-Hispanic white compared to 72 percent of the region’s population. People of color were 28 percent of the region’s population, but only 15 percent of survey respondents. People of color were undercounted by almost half in this unscientific survey.
Residence: Clark County over-represented. The respondents to the unscientific web-based survey are disproportionately residents of Clark County. Clark County accounts for less than 20 percent (488,000 of the region’s 2.5 million residents) but accounts for 43 percent of those taking the survey. Clark County resident views are given more than double the weight of view of other of the region’s residents in this unscientific survey.
Age: Young people significantly under-represented. There’s also a strong generational bias: only 5 percent of survey respondents are under 25, compared to nearly 30 percent of the population. And these people will be the ones who have to live with the environmental consequences of the project.
No doubt the highway agencies will point with pride to the large number of completed surveys–more than 9,000 to date. But large numbers are irrelevant if you don’t have a random sample. For a metropolitan area the size of Portland, you need only about 400 to 800 survey participants to come up with statistically valid results, if you have a random sample. If you don’t have a random sample, then even very large numbers (and IBR surveyed only about one-third of one percent of the region’s residents) just aren’t meaningful. The underlying problem that invalidates the survey is called “Self-Selection Bias.” Because this isn’t a true random survey, and because respondents choose whether to participate, there’s no guarantee that the survey data reflect the views (and experiences) of the larger population. Because those who are predisposed to care about this issue are likely to differ systematically from the rest of the population, the survey produces results that are biased.
Not asking the most important question: Who wants to pay a toll?
There’s a lot more to dislike about the survey beyond its poor quality sampling strategy and biased sample. The questions posed in the survey don’t get at the real issues raised by the freeway widening project. The project’s financial plan shows that it won’t be built without tolls—something you’d be hard-pressed to learn from any of the “public information” work. The last estimates prepared for the Columbia River Crossing showed I-5 tolls would be a minimum of $2.30 during off peak hours, rising to $3.25 during rush hour, with additional surcharges for those who didn’t buy transponders for their cars in advance. The survey didn’t reveal these toll rates, or ask people whether they might prefer a smaller, less expensive bridge with lower tolls, to a larger one with these high tolls, or whether they’d really rather keep the existing bridge if it meant they could avoid tolling altogether. Despite the fact that the survey avoided talking about tolls, many survey respondents raised the question in answering open-ended questions.
It’s rather like a taste test survey that asks people whether they’d prefer filet mignon to a hot dog, without revealing the price tag of either alternative. For a project that claims so prominently to care about “centering equity,” failing to reveal that people might have to pay on the order of $1,600 per month to commute daily across this bridge is a monumental omission. But it’s no accident: the project’s “public information” campaign is designed is an intentionally misleading way to manufacture consent, not to accurately measure public attitudes.
Surveys can be a useful way to gauge public opinion, if they’re undertaken in a scientifically valid fashion. But if you aren’t careful, you end up with a classic, garbage-in, garbage-out exercise. That appears to be the case with survey work commissioned by the “Interstate Bridge Replacement” project, a thinly veiled marketing campaign for freeway widening funded by the Oregon and Washington transportation departments—with “communications” consultants reaping more more than $4 million for their services in the past few years.
Related Commentary