The Guardian presented a surprising international poll conducted by YouGov stating the willingness to support Ukraine significantly declining in all the polled countries: France, Germany, Italy, Spain, Sweden, Denmark and the UK

“Support for Ukraine ‘until it wins’ falls sharply in western Europe, poll finds” Link: Support for Ukraine ‘until it wins’ falls sharply in western Europe, poll finds | Ukraine | The Guardian

When this poll is replicated using a credible company and scientifically proven methods, the results are widely different.

75 % of the swedes support Ukraine until Russia withdraws, only 25 % support a negotiated end to the fighting.

Not 50 % for Ukraine, and 24 % for a negotiated end that Guardian/Yougov state.

From Gallup Nordic/Novus poll, using proper methods. There is no way you can say that the support for Ukraine have “fallen sharply” in Sweden. And it should be questioned if the same error is in the other 6 countries. Or is the support for Ukraine massively underestimating in those countries as well?

Looking at the YouGov results being presented in the Guardian article, it is likely they are not presenting a “don’t know” answer, the sum is not 100 %. In Novus we did not give that alternative. We wanted to force the respondents to chose, and more importantly using a scientifically proven method.

In our poll 75 % of the swedes giving support for supporting Ukraine to fight against Russia, and only 25 % chose to negotiate an end to the fight.

(Looking at the article in YouGov Feb 2023 states that the support in Sweden was 57 %. Although the Guardian chart indicate a bit over 60% which is weird, and that also may result in the “encourage” response 2023 not being really correct in the chart I took the numbers from.)

Novus poll give no support for the statement that the support in Sweden have fallen sharply. And with these widely different results in Sweden is the error the same in the other 6 measured countries? Maybe the support for Ukraine is much higher in the other countries as well?

Novus results is in line with all other research being done in Sweden and come as no surprise. YouGov’s results are a big surprise, and you would be wise to question the results as well as the methodology being used…

For the Guardian I would like to ask this question:

If it is not true, can it really be news? What are the reasons for using nonscientific polls as a news source?

Two reliable polls will with roughly 95 % certainty give the same or at least similar results. Widely different results however often happen in our studies when we look at nonscientific polls.

When comparing open acces panels like Yougov, to proper ones in Sweden, difference of this magnitude is often the norm. The reason why almost no serious Swedish news outlet use nonscientific (open access etc.) polls as news source. It simply can not be news if it is not known to be true.

Novus is the best known market research firm in Sweden. Our average error in ALL elections in Sweden since we started is under 1 percentage point. Our Covid research was not noly highly accurate, but also the only reliable data source of the pandemic in Sweden during the breakout and the first 6 months of the . Novus is also the provider official statistics of several metrics in Sweden. All our polls use scientifically proven methods. Even when we use online panels.

From a scientific and a practical standpoint Yougov polls are very problematic. The polls are a so called “open access panel” allowing anyone who are willing to contribute to join their panel.

To my knowledge there are some anecdotal evidence to trust open access panels. But in those cases they do not differ from those based on scientifical methods, adding no value. There to my knowledge no evidence showing that scientific polls are wrong and Yougov is right. But a lot of examples like the one we seen on Ukraine, giving widely different result using nonscientific methods.

Because Yougov polls are not using the scientifically proven method, they can not state that the results properly give the view of the population they claim to research. They also can not present any form of margin of error.

There are some that argue that no polls work, not only are they wrong, it is not a working argument to trust nonscientific polls.

Novus studies have time after time showed a lot of bias in the open access panels. Bias that severely impact the reliability of the results. However, because they are so cheap to conduct, they are wrongfully often seen as “good enough”.

These kind of open access panels are wide open for bad actors to skew the research results, using AI and registering thousands of false respondents its now very easy to manipulate the results, if all you need is an email address… As an example CINT one of the biggest panel provider in the world saw their market value plummeting when they had to go public about the huge problems they have with fraud.

Geopolitically and for the future of Europe, how the war in Ukraine ends have a huge impact. This issue cannot be taken lightly, and presenting a false view of Europes population is really damaging. Misleading polls and articles like the one in the Guardian may very well have an impact on the support Ukraine gets. Politicians listen to their voters. Showing a false view of the voters is therefor something that must be avoided.

As stated above this is not an isolated issue, Novus have several times seen this in Sweden, but also PEW in US, point out the same problem on a very serious topic: Online opt-in polls can produce misleading results about young adults’, Hispanics’ views | Pew Research Center

Also YouGov have been involved in very serious issues in relations to the Brexit referendum, that Bloomberg reported on, a misuse on the UK laws that actually led to new legistations around polling being created. A horrible example of how a pollster can be used for market manipulation. The results from YouGov was incorrect, but Sky News presented it as the truth and that affected the market in a way hedge funds knowing “the truth” berforehand could profit on: Brexit Big Short: How Pollsters Helped Hedge Funds Beat the Crash – Bloomberg

There are so many more examples, but these are some of the most serious that are known. Most problems with polling is unknown. Because often there is only one poll, and noone is checking the validity, like I did in relations to Ukraine, or PEW did in that example.

More serious reserachers need to step up and dare to show how important proper research is. What we do affect almost everything in a western society, that means it need to be made in the best way possible.

I have revisited this topic once more, new findings can be read in this post: https://gallupnordic.com/2025/01/23/revisiting-guardian-poll-scientific-insight-on-swedens-ukraine-support/

Torbjörn Sjöström

Chairman Gallup Nordic

CEO Novus

Novus poll was conducted with 1001 interviews in Novus probability based panel during the period 8-9 January 2025.

More about Novus panel, see this page (in Swedish) https://novus.se/om-novus/vanliga-fragor/