Chart this poll with one simple trick
Welcome to YouGov's weekly newsletter The Surveyor, with new polling data, insights, and charts on politics, life, and other topical issues — from our U.S. News team.
This week, we're spotlighting surveys about Donald Trump, JD Vance, Elon Musk, swearing, AI, abortion, Jeffrey Epstein, immigration, and Coca-Cola.
Non-MAGA Republicans come home to Trump
The share of Americans who disapprove of how Donald Trump is handling his job as president has hit a record high for his second term: 55% strongly or somewhat disapprove, and just 41% approve.
Trump's net job approval — the percent who approve minus the percent who disapprove — of -14 is not a record low, but it's close to the lowest approval Trump received during his first term, too.
But what particularly interested me is how Trump got to this 41% approval — because opinion of Trump hasn't been changing with every group at the same rate.
Trump's approval among Democrats has declined from a low 12% at the start of his second term to an extremely low 3% today; his approval among Independents has fallen from 41% to 29% in that same time. Meanwhile Republicans who identify as MAGA Republicans — half (50%) of all Republicans — have unchanged, near-unanimous approval of Trump: 98%.
What's interesting to me is the shifting views of non-MAGA Republicans. Unsurprisingly, fewer members of this group approve of Trump than do MAGA Republicans. But non-MAGA Republicans have changed, from 90% approval at the start of Trump's term, down to 70% in mid-April, and now back up to 85% today.
A 15-percentage-point increase in Trump's approval among non-MAGA Republicans is equivalent to about a 2-point increase in his approval among all U.S. adult citizens.
How we design a chart
We share lots of charts in this newsletter, breaking down our poll findings along with key crosstabs or trends. But this week I wanted to take some time to talk about the charts themselves, not merely the findings. How do we at YouGov decide the best chart for a job?
This is on the front of my mind this week in particular because this week's Economist / YouGov Poll had some results that didn't present an obvious visualization choice.
What I mean by that is not that the results were difficult to chart, but rather that they were difficult to chart well. A good chart doesn't just convey the data — you can look at a table if all you want is numbers — it conveys the important takeaway from the data in an elegant and precise manner.
Let's get to some specifics. The questions involved here are a straightforward set we ask periodically to gauge how Americans feel about U.S. presidents or potential presidents. In this case, we asked them about both Trump and Vice President JD Vance:
Do you think [Donald Trump / JD Vance] is honest and trustworthy, or not?
Would you say [Donald Trump / JD Vance] is a strong or a weak leader?
Regardless of whether you agree with him, do you like or dislike [Donald Trump / JD Vance] as a person?
Do you think [Donald Trump / JD Vance] has the temperament to be the President?
Are you confident in [Donald Trump's / JD Vance's] ability to deal wisely with an international crisis, or are you uneasy about his approach?
Each of these questions on their own is easy to chart. But in our case, we wanted to look at A) all five questions, B) for both Trump and Vance, and C) both overall and among Democrats and Republicans (and maybe Independents, too). That's 5 times 2 times 3 or 30 different data points. Meanwhile, for each of those questions we have multiple answer options: "Honest and trustworthy" / "Not honest and trustworthy," "Very strong" / "Somewhat strong," "Somewhat weak" / "Very weak," and so on. That's a lot of data.
One way to handle the data is to just break it up. For example, here's a chart for one of those questions, comparing responses for Trump and Vance among U.S. adult citizens and by political party:
It's clean, and gets the job done. But multiply those eight bars by five questions and you're looking at a really, really tall graphic. Not only is that unwieldy, the size of that graphic would make it difficult to identify important comparisons! You'd be constantly scrolling up and down. Ideally we want a chart that conveys all the key information at a glance.
So as data visualizers, the next step is how we can simplify the chart. That is, how can we remove extraneous elements to focus the chart on the most important information?
The trick we use most often to do this with survey data is to present the "net": the percent who give one answer minus the percent who give the other. This takes three data points — the percent who say "honest and trustworthy," the percent who say "not honest and trustworthy," and the percent who aren't sure — and collapses it into one data point. (The "not sures" remain on the sideline, where if they so choose they can think harder about which side they'd choose.)
With this, we take our eight-line chart down to four lines — or three, if we omit Independents:
With those cuts, we've gone from 40 lines on our hypothetical chart of all five questions, down to 15 — still large, but not unworkably so. Here's what that would look like:
The biggest problem I have with this chart isn't so much its unwieldy size as the lack of clarity. There are a lot of data points on here, and it's not immediately clear which ones are important.
If you spend a minute looking over all the data points, though, there is an interesting pattern: For most of the questions, Democrats rank Vance ahead of Trump, while Republicans are more likely to rank Trump ahead of Vance.
So I set out to try to make a good chart that zoomed in on that pattern, ruthlessly purging everything else. I basically used the same mathematical approach as before: Subtracting the net preference for Vance from the net preference for Trump. Only this time it was one more level of abstraction, a net of two net quantities. Here's the chart we ended up with:
Now, this is not a perfect chart by any means. For one thing, it needs a lot of text to explain itself. Also, it's two levels of abstraction away from the actual data. I addressed that a bit by adding some more labels onto the chart to help a reader interpret it.
But this kind of charting is a game of tradeoffs. This chart clearly shows a trend, but only at the cost of showing significantly less information than a less concise chart — or a series of charts — would do. We'll sometimes present a ruthlessly pared-down chart like this — but also maintain our practice of sharing the full data so readers can see for themselves anything we omitted.
Charting opinions
What Republicans think of Musk and Trump after Musk's acrimonious exit from the administration
Musk remains more popular with Republicans than with Democrats, though the share who view him very or somewhat favorably has declined with them in the last month -— since his departure from the administration. Democrats generally have expressed unfavorable views of Musk since the beginning of the year, and now the share who view Musk favorably is below 10%. (Kathy Frankovic)
How do attitudes to swearing differ in Australia, the UK and the US?
A clear majority of Britons (57%) say they swear at least most days, compared to just under half of Australians (47%) and only four in ten Americans (39%). Britain has the most regular swearers, with 37% of the British public saying they swear every day, compared to just a quarter of Australians (23%) or Americans (25%). (Dylan Difford)
Americans are increasingly likely to say AI will negatively affect society
Since March, there have been increases in the shares of Americans who are very concerned about AI resulting in an increase in human dependency on technology (from 45% in March to 50% now), diminishment of human creativity and drive (from 44% to 49%), the spread of misleading video and audio deepfakes (from 58% to 63%), and a decrease of face-to-face social interactions (from 41% to 46%). (Jamie Ballard)
The share of Americans who believe abortion should always be legal and there should be no restrictions on it has increased since August 2024, to 32% from 27%. Among Democrats, 59% believe abortion should always be legal, up from 48% last August. The share of Independents who say abortion should always be legal has increased to 33% from 22%. (Other YouGov polling has found that over a longer time period, views on this question have been mostly stable.) The share of Republicans who believe abortion should be illegal and should never be allowed has risen since last August, to 22% from 13%. (Jamie Ballard)
Quick takes
Citizenship: 10% of Americans say the U.S. president has the power to revoke the citizenship of a U.S.-born citizen, while 65% say the president doesn't
Coke: 43% of Americans say they've tried Mexican Coca-Cola, and 48% say they haven't
Detention: 97% of Americans say immigrants in U.S. immigrant detention centers should have clean drinking water, while 1% say they should not; 51% say they do have it and 19% say they do not
Elimination: 21% of Americans say the U.S. president has the power to eliminate the Department of Education via executive order, and 63% say the president does not
Epstein: 58% of Americans say Donald Trump and Jeffrey Epstein were once friends including 42% who say they were once close friends, while 21% say they were acquaintances, 1% say they were strangers, and 1% say they were enemies
Elsewhere
Polling partnerships
The Economist + YouGov on Trump's declining job approval, Texas flooding, Alligator Alcatraz, Jeffrey Epstein, and JD Vance
Polling abroad
Polling in the press
Donald Trump attacks own supporters over Jeffrey Epstein conspiracies (Financial Times)
Trump wants to dismantle the Education Department. What would that mean? (Washington Post)
Everyone’s Obsessed With True Crime. Even Prisoners Like Me. (New York Times)
Hawaii is, again, Americans’ favorite state (Sherwood News)
Learn more and get in touch
Sign up to participate in our polls and share your opinions on the people, companies, and issues that shape your life and the world
For more polling data and insights, visit our website or follow us on X, Threads, and Bluesky.
Have questions or feedback?
If you have a press or media inquiry, email uspress@yougov.com
If you have an idea for a future survey, share it here
If you are a panelist and have questions, contact support here
For information on privacy, please read our privacy notice
Carl Bialik contributed to this newsletter.