Annotated: Paradigm’s July 2024 Democratic Public Opinion Poll

There may be “reports” (generally from the industry) that candidates and incumbent Congresspeople are “becoming pro-crypto”, but I would definitely like to see some better support for the claim that this is them “following the lead of their voters” and not, oh, I don’t know, related to the hundreds of millions of dollars available to candidates who are deferential to the industry.

Mercury Analytics isn’t listed on 538’s pollster ratings list like DCG’s pollster was, but they also aren’t a completely unknown quantity, having published research on topics ranging from vaping to 2020 presidential election debate sentiment. Their polls have been reported upon in The Wall Street Journal and elsewhere, though one such article noted some flaws with survey design. Their CEO’s June 2024 article about how “AI is revolutionizing the market research industry” made me wonder if AI was involved in interpreting qualitative answers to this survey, as he advocated, but after speaking with Paradigm’s policy team I learned that no qualitative interviews were performed.

I asked Paradigm for more detail about how the poll was conducted, and they informed me that it was conducted all online (unlike their June poll of Republicans, which was conducted by phone and online). They do not publicly note this, nor does their poll summary address the skew that this likely introduces.

A May 2024 online survey by the Federal Reserve notes: “Because the survey is conducted online, the sample population may be more technologically connected than the overall population, which could increase the share of adults reporting use of emerging technologies such as cryptocurrencies.” David Marker, fellow of the American Statistical Association and American Academy for the Advancement of Science and elected member of the International Statistical Institute, criticized the DCG poll for failing to include such a disclaimer, commenting: “Online surveys into crypto ownership can skew or misrepresent true figures, and they’re inherently limited, as they don’t factor in the digitally excluded on a topic that is directly related to digital comfort.”

At the very least, Paradigm should have disclosed the method by which respondents were surveyed, and preferably also proactively noted the potential skew this likely introduces.

For clarity’s sake, I verified with Paradigm that the oversample brought the total number of respondents to around 1,100 people. The 300 additional surveys were in addition to the 804 people surveyed, not a part of that number.

They get right to the point about their goals with this survey. With polling that began just after Harris replaced Biden as the Democratic nominee and immediately began picking up steam, this poll seems to have been commissioned at least in part in hopes of convincing the Harris campaign, and those around her, that “making inroads with crypto owners” is something she needs to prioritize.

It also fits with the timing of their previous poll, which was trying to convince Trump that he needed to attract Republicans who own crypto while he was leading in the polls.

This is a bit of an odd data point, and odder still was the decision to highlight it as the very first finding, given it is also one of the weakest claims from this poll.

First, the math doesn’t really work out. I’ve confirmed with Paradigm that they were calculating from the following graphic the number of “non-Harris voters” (more on that in a second) who responded that the Biden admin was “too hostile” to cryptocurrency, and munging those two together to conclude that this segment thus “may be leaning towards Trump due the Biden Administration’s hostility to crypto”. However, this figure actually comes out to about 2.7%, which is not within the stated “1–2%” range. (13% of 804 respondents is 104.52; 21% of 104.52 is ~22; and 22 out of 804 people is ~2.7%.) I was informed they decided to provide a lower estimate in the summary than their polling suggested, because 2–3% “felt too aggressive”. But that raises the question: why bother polling if you’re just going to change the numbers?

Second, they did not ask any questions about whether the Biden admin’s perceived hostility might actually push these people to vote for Trump, nor do they provide any details about how many of these “non-Harris voters” would consider voting for Trump in general.

The scale on this first bar chart is hilarious, particularly juxtaposed above the chart just below it (which captures the full 100% of respondents, while the same amount of space fits maybe 30% of respondents for this question).

I can help them out a little bit by editing in the bar for the remainder of the “non-Harris voters” who don’t hold crypto:

Or perhaps scaling it down to the same proportions as the other chart, assuming the full width of the space represents 100%:

Ahh, much better.

In the second bar chart, the designers of this report have for some reason opted to place the 52% “don’t know” respondents between “just about right” and “too hostile”, as though it signifies “somewhat too hostile”.

I verified with Paradigm that this 13% “undecided” figure is indeed the proportion of people who indicated they would be voting for Trump, a third-party candidate, or had not yet decided, but that they had not polled on whether these voters would choose not to vote for Harris. In my opinion, it’s misleading to suggest in the summary that these people are all “not voting for Harris” (or refer to them throughout as “non-Harris voters”) given that they did not indicate that, and some of those undecided voters may well end up voting for Harris — as might some of those who indicated they would vote for a third-party candidate who later dropped out.

Paradigm verified that 5% of the self-identified Democrats in this poll indicated they would be voting for Trump.

This figure seems to come from the later question in section 3 that asks “Have you ever invested in, traded, or used a cryptocurrency such as Bitcoin or Ethereum?” I don’t love these types of questions because they include people like myself, who are highly skeptical of cryptocurrencies but own (or previously owned) some for research purposes. It also includes those who purchased crypto and then got rid of (or lost) it all; those who have used crypto for a single transaction; those who have just received crypto in promotional deals; etc. This is not necessarily a great measure of people who have strong and positive feelings about crypto, or of people who would support the kinds of policies generally supported by the industry.

If you were curious like I was: 7% of these “non-Harris voters” responded “much too hostile” and 14% responded “a little too hostile”. We’re really in percent-of-a-percent territory here: we’re talking about 7 people and 15 people (respectively). It’s pretty wild that they’re looking at 7 responses — 22 if we generously count those who don’t feel as strongly — from a poll of more than 800 people, and extrapolating that over the US population to claim that “Democratic crypto voters are up for grabs this November”.

This may be a stated goal, but it’s far from a real-life feature of most popular cryptocurrencies. See my essays “Abuse and harassment on the blockchain” and “Anonymous cryptocurrency wallets are not so simple”, which go into more detail. In short, while banks keep records of financial transactions that they may later share with law enforcement, data on public blockchains is visible to anyone, including to law enforcement without a warrant. Blockchain analysis firms like Chainalysis (which has earned more than $80 million from contracts with US federal government agencies alone, not to mention contracts with state-level organizations and non-US government and law enforcement clients) like to brag about just how easy crypto tracing is for them.

Who is supposing this other than Republican talking heads?

This 15% is going to love blockchains, then. Cryptocurrency’s reputation as a useful tool for tax evasion has primarily been a result of the IRS’s lack of interest in scrutinizing once niche categories of transactions — and that is something that has been changing in recent times.

Notably, this poll did not ask any questions about respondents’ opinions on using cryptocurrencies to achieve financial privacy, and instead is simply assuming that respondents who want financial privacy would support crypto, and vice versa.

Okay, putting the sarcasm aside for a minute here: please for the love of god do not put your abortions on the blockchain! Do not try to pay for your or another person’s abortion using crypto, or encourage others to do so! This is an incredibly dangerous idea, as states are working to criminalize not only abortion, but also punish those who try to provide aid to others who need them. While there are a handful of cryptocurrencies that seek to provide some privacy for those who transact with them, the vast majority of popular cryptocurrencies do not.

Another thing: while it’s good of Paradigm to note the importance of financial privacy in situations where people are seeking medical care that has been criminalized or may be criminalized in the future, they should perhaps consider also not funding the campaign of a candidate who has supported incredibly strict anti-abortion laws, described abortion rights activists as “demonic”, and said that abortion is a “religious sacrifice to these [abortion rights activists].” (Paradigm contributed $500,000 to the cryptocurrency super PACs, which have spent about $583,000 on Blake Masters’ fortunately unsuccessful Republican primary campaign in Arizona’s District 8.)

Okay, who decided it was a good idea to make the affirmative responses go from dark blue to light blue as the respondents’ level of agreement gets less strong, but then use the dark orange to mark “not that important” and light orange for “not at all important”? I just want to talk to them, I swear...

I assume by now you have noticed the pattern in this section, which will continue throughout, where the poll asks a very broad question about a financial topic that few people would object to, then implies the response is an endorsement of cryptocurrencies. This was also a strategy in the DCG poll, which asked respondents to agree or disagree with statements like “The current financial system lacks transparency” and “The cost and rules for participating in the current financial system are not designed to benefit the average person.”

Paradigm seems to be using current or past ownership of cryptocurrencies as the metric by which they determine a person “views crypto positively”, which is flawed for reasons I’ve already described.

However, even if we put that aside and assume that the 18% figure (or 15–27% when broken down by race) actually represents support for cryptocurrencies, it would seem that Democratic politicians who “come out as pro-crypto” would in fact be failing to follow the lead of the remaining 82% (or 73–85%) of their constituents who don’t view crypto positively.

It’s also possible they’re reaching this conclusion from the later question (not captured in a graphic), stating that 13–32% of respondents say that “crypto plays a positive role in the U.S. economy”. If that’s the case, the inverse is roughly the same: 68–87% of respondents don’t view crypto positively.

The March 2024 Paradigm poll cited here claimed that “19% of American registered voters say they’ve bought crypto.” It is generally an interesting exercise to compare crypto industry polling results to polls from outside the industry (though there are variations in question formulation, namely with some polls asking about current crypto ownership vs. ever having owned crypto):

Non-industry-funded polls

• JPMorganChase, 2022: 15% of their American customers have conducted transfers to crypto accounts. (I suppose someone could argue that JPMorgan is an industry pollster, given their blockchain forays, but that is not their primary business. Take their numbers with grains of salt as you deem fit.)
• Pew Research Center, March 2023: 17% of American adults have ever invested in, traded or used a cryptocurrency
• Federal Reserve, 2023: 7% of American adults hold or use cryptocurrency

Industry-funded polls

• Andreessen Horowitz, December 2021: 20% of American registered voters own crypto
• Coinbase, 2022–2024¹: 18.2% of American adults own crypto (varies over time between 16% and 22%)
• Paradigm, March 2024: 19% of American registered voters have ever invested in, traded or used a cryptocurrency such as Bitcoin or Ethereum
• DCG, April 2024: 14% of registered voters in six swing states own crypto; 26% own or have previously owned crypto

To their credit, Paradigm’s March poll is reasonably close to the Pew and JPMorgan surveys, particularly when compared to crypto industry surveys that have polled current crypto ownership and come in way higher than non-industry polls. In this poll, Paradigm asks about the amount of cryptocurrency respondents currently own later on, revealing that 12% of their respondents currently own crypto (leaving 6% in the “former owner” bucket). This is quite high compared to the Federal Reserve’s numbers, but considerably lower than the 20% figure Coinbase and others like to throw around.

Paradigm also claimed to me that they’ve found that providing examples of cryptocurrencies in their question (“have you ever invested in, traded or used a cryptocurrency such as Bitcoin or Ethereum?”) results in higher affirmative responses. This is surprising to me, since it would seem to suggest a significant number of people hold BTC or ETH but don’t view them as cryptocurrencies.

Other recent polls (including Pew’s) have had similar findings to Paradigm’s: that people of color are somewhat more likely to have owned crypto than white respondents. This is generally treated by the cryptocurrency industry as a positive thing, and some influential crypto figures like to describe crypto as a particular opportunity for people of color to “build generational wealth” or access financial services they have historically not been able to access (sometimes using tactics that have been described as “predatory inclusion”).

However, the higher crypto ownership among people of color is a fairly recent development — and one that means that many of these investors bought in at all-time-highs in 2020 and 2021, with many later losing much of it in the 2022 crash. “[R]ecent polls show a sharp decline in the share of Black Americans holding bitcoin, indicating that many folks might have bought high and sold low,” reported Annie Lowrey in The Atlantic in November 2022.

With polls like these, it’s often informative to invert the numbers used in the conclusions they’re trying to draw attention to. 68–87% of respondents do not agree that crypto plays a positive role in the U.S. economy. This is a significant majority, and it’s worth considering why the authors of this report chose to focus on the minority of respondents who do agree.

This is inaccurate. As the poll states in the second bullet point below, 17% of non-crypto-holding Democrats say they plan to purchase crypto in the next year, whereas 18% of Democrats overall have already bought crypto.

I suspect this is an artifact of an earlier error in this poll summary. Bullet point two originally read: “This means about 27% of Democrats who haven’t yet bought crypto are likely to buy in the next year.” That seemed extremely high to me as a percentage of non-crypto-holding respondents, and after I inquired, Paradigm corrected the bullet point to 17%. However, they seem to have missed this conclusory sentence up here when making that correction.

Polling by Morning Consult (the firm that has been doing the polling for both Andreessen Horowitz and Coinbase mentioned above)¹ has claimed that the percentage of respondents who were “likely to purchase cryptocurrency in the next month” has hovered relatively stably between 20% and 26% ever since they began polling in January 2022. However, the same polls also suggest that the percentage of respondents who hold crypto has remained relatively stable, with some increase going into April 2023, and some decline over the past year. The Federal Reserve survey shows declining cryptocurrency ownership from 2021 to 2023.

This suggests that, far from being a “fire bell in the night”, this may just be a poor indicator of actual future adoption. Either that, or people are getting out of crypto in such numbers that they are more than offsetting these supposed new adopters.

This is out of all respondents, including those who do not currently own cryptocurrencies. It’s probably safe to assume that some (though perhaps not many) of those 9% “very likely” respondents do not currently hold crypto, but even if we generously assume every single person in that 9% already holds, it seems noteworthy that at least half of the 18% of respondents who said they currently hold crypto are at best lukewarm about buying more of it going into the next year.

Again, the other numbers hidden away in the graphic here seem notable. A full half of respondents slammed the “hell no” button when asked if they might like to try out crypto in the upcoming year. Almost 3 in 4 were varying degrees of unlikely to get involved. This roughly aligns with Pew’s findings in 2023 that 75% of Americans who have heard of cryptocurrencies are not confident in their safety and reliability.

Even assuming these numbers are good (see later), it is not at all clear to me how 12% current ownership of crypto can be interpreted as “approaching parity” with 33% ownership of stocks. I guess if you had historical data that showed <12% current ownership, then any increase would be “approaching parity” in a sense... but Paradigm doesn’t mention having any historical data besides the polls they conducted on different demographic groups earlier this year. Year-over-year data from other pollsters show crypto ownership remaining roughly flat (crypto industry polls) or decreasing (non-crypto industry polls).

Again, not clear where this claim is coming from, given their lack of historical data. My best guess is they’re just going off of the respondents who said they were likely to purchase crypto going forward — which, as I said, seems to be a dubious indicator of actual growth. Remember at the beginning when I said we would come back to the opening claim that “Democratic voters are increasingly gravitating towards crypto”? This is that moment.

This is significantly lower than the numbers reported by Gallup, which showed that 61% of Americans owned stocks² in 2023, with that number ranging between 50% and 60% for the past decade. However, Paradigm’s number is closer to the results for a question from a 2019 poll by Pew Research, which found that 35% of respondents directly owned stocks outside of retirement accounts, and 55% had IRAs, 401(k)s, or similar retirement accounts.³ This makes me think people may have been interpreting the question by Paradigm in the latter way, but the definition probably should have been made clearer both in the question and in this poll summary.

This is surprisingly high when compared to their finding that 12% own crypto, full stop. This may support the hypothesis I mention above that a substantial number of people who bought in a year or more ago have since sold off those holdings.

I wonder how many of these respondents could feasibly liquidate their crypto holdings for the dollar amounts they’re estimating, and how many just have “$1,000” on paper.

These figures don’t really line up with other estimates, such as from a 2022 JPMorgan survey that claimed that “the median gross amount transferred to crypto accounts over the period 2015 through the first half of 2022 was approximately $620”. They also found that, out of people who hold crypto, about 11–15% of them have transferred one month of income or more into crypto accounts, and 4–6% have transfered three or more months of income. Doing some extremely rough napkin math: the median monthly income is approximately $5,000. JPMorgan found that 15% of their sample had transferred funds to crypto accounts. 11–15% of that 15% is 1.65–2.25% of the total sample who have transferred, based on median income, $5,000 or more. 4–6% of that 15% is 0.6–0.9% who have transferred, based on median income, $15,000 or more.

Is it just me or do these seem like surprisingly round numbers? Usually things like this have a bit of a long tail to them, and it seems unlikely that there are really the same number of people who own between $25,000 and $50,000 of crypto as there are those who own more than $100,000. It makes me wonder if Paradigm is rounding each group up to the nearest whole number, and then compounding the error by adding the rounded numbers together. Paradigm did not immediately respond to a follow-up question about this detail.

This portion of the poll summary was by far the most surprising to me. I asked Paradigm for more details on this question, but they told me that they had not given respondents examples of what this might look like, nor did they collect open-ended responses or do interviews to get a better sense of what people were envisioning when they responded in this way.

While it doesn’t surprise me that there are people who are concerned about losing access to financial services generally speaking, or perhaps about cutbacks of programs like social security, food or housing assistance, or Medicare/Medicaid, I was surprised to see such a high percentage of people who believed that a) that this would come as a direct result of their political views, and that b) it would affect them personally (vs. some hypothetical more radical person — for example, someone running an abortion fund — who might end up in the crosshairs of a politically motivated debanking campaign).

I wonder to what extent people are thinking less of politically motivated termination of financial services and are instead reflecting their broader concerns about financial topics like inflation and housing affordability.

Paradigm seemed surprised by the results as well.

Continuing what I was saying above, I was really surprised at just how many people believe that they personally (or their family) might be cut off from financial services, versus agreeing that Trump is likely to cut of some people’s access to financial services.

But, as I said previously, this is another one of those questions that implies that crypto is one of these “alternative means of protecting their access to financial services” that they mention above, without actually asking respondents about their opinions about crypto’s viability for such a purpose.

It seems to me that if you’re looking to find out if “Democrats want the United States to be a world leader in crypto”, you should ask questions like “How important is it that the United States leads the world in cryptocurrency?”, and not “How important is it that the United States leads the world in high-tech software innovations, such as cryptocurrency and fintech (software for financial services like banking and investing)?”

I, for one, think the US has a long ways to go when it comes to fintech, which could be so much better. I’m not too precious about where that software is developed, but as an American I’d sure love to be able to use it instead of still having to write checks from time to time, so I’d probably answer somewhere in the “important” spectrum on a question like this. Does this mean I think the US should be a “world leader in crypto”? No. Does it mean that I think the way to make US crypto and fintech firms “world leaders” is to slash regulations? Also no.