This week: https://nieuwscheckersleiden.substack.com/

Fact-checks help undecided voters

In the past week, the election campaign gathered real momentum  with debates broadcast on radio and television. This week, further debates will follow on RTL (Sunday) and SBS (Thursday). Various media outlets fact-checked the claims made by politicians last week. In this newsletter, we’ll summarise their findings. 

Also in this newsletter: AI-generated photos and videos have become part of the political PR arsenal, especially for the PVV. Weekly magazine De Groene Amsterdammer revealed that a  member of parliament from that party is responsible for photorealistic images of blonde women being chased by men with beards and Palestinian scarves..Lastly, a study has shown that – particularly among undecided voters – live fact-checks during TV debates can influence how people vote.

Election fact-checks

Many of this week’s fact-checks focus on money. Ten political parties have had their election platforms assessed by the Netherlands Bureau for Economic Policy Analysis (CPB); the PVV, FvD and SP chose not to take part.

Financial fact-checks

• De Volkskrant criticises parties’ plans to cut healthcare spending. According to experts, these plans are primarily intended to present favourable budget figures on paper, whilst in reality they are unfeasible.
• Dilan Yeşilgöz and the VVD are campaigning to defend the mortgage interest deduction. First-time homebuyers would lose €400 per month if it were to be abolished. But this is only the case for first-time homebuyers who both earn the average income and currently buy a house for the average price of €385,000, and if the mortgage interest deduction is abolished immediately, and if this loss is not compensated for by other measures. No party supports this, as shown by de Volkskrant and RTL Nieuws.
• Under PVV Minister Reinette Klever, there have been significant cuts to development aid, which, according to international agreements, should amount to 0.7 percent of gross national income. GroenLinks-PvdA and the ChristenUnie want to return to this standard, but according to de Volkskrant, the CPB’s calculations show they will not achieve this. The same applies to all other parties that have had their platforms assessed.
• JA21 made headlines because the party submitted different figures and policy plans to the CPB than those in its election manifesto. NRC revealed that JA21 is not the only party doing this: the CDA, Volt, VVD, D66, ChristenUnie, and SGP are also guilty of this. In some cases, these discrepancies arise because the CPB requires more concrete measures for its calculations than those provided in the election manifestos, but sometimes the parties use this to obscure less popular measures.

Election fact-checks on religious disaffiliation and the Council of State

• In a tirade about Islam, Geert Wilders claimed that it imposes the death penalty on Muslims who leave their faith. This does happen, but certainly not in all Islamic countries. There is disagreement about this within Islam itself; while some countries apply the death penalty for religious disaffiliation, others impose different punishments – or none at all.  Read the full fact-check at Nieuwscheckers.
• According to various BBB politicians, a majority of the members of the Council of State are either members of D66 or otherwise lean progressive. LUBACH listed the political affiliations of the members: of the eighteen, six are members of D66 and GroenLinks-PvdA. The remainder are members of the SGP, CDA, the Arubaanse Volkspartij (Aruban People’s Party), or have no political affiliation.

Chatbots as a voting aid and other election news

• In 2023, research was conducted into the interaction between social media and parliamentary debates. The study concluded that the increasingly harsh tone online and in parliament reinforce each other in a vicious circle. The researchers told iBestuur that follow-up studies on platforms such as X are no longer possible, as Elon Musk now holds full control there.
• Can a chatbot be used as a voting aid? RTL Z tested ChatGPT, DeepSeek, and Grok with statements from the Kieskompas (an online voter guide). The chatbots consistently aligned most closely with the  positions of  D66, GroenLinks-PvdA, and Volt. Since these chatbots are not politically neutral, the experts interviewed by RTL Z recommended that voters use them instead to summarise election manifestos, for example.
• Parliamentary groups wanted to meet with representatives of Facebook, TikTok, X and Google to discuss algorithms and interference during the election campaign. The parties are concerned about improper influence — from which the platforms also profit. The companies’ representatives either declined the invitation or said they were not given enough notice.
• That tech companies have been slow to intervene has also become evident in an article by Follow the Money, which opens with a report on death threats received by Rob Jetten during a livestream on TikTok.
• EWMagazine interviews political scientist and founder of Peilingwijzer Tom Louwerse about the value of political polls.
• From simple data manipulation to sophisticated deepfakes: Nieuwscheckers’ Peter Burger spoke at the Leiden Public Library about deception in election campaigns.
• De Volkskrant published an extensive analysis of the role TV debates play in the election campaign, examining how decisions are made about which politicians to invite, how the debates are produced editorially, their potential impact on the election outcome, and briefly, the value of live fact-checking.
• If you want to  know who is appearing on screen, where, and for how long, consult De Groene’s Schermtijdteller (Screen Time Tracker). It also shows that, despite his absence from the TV debates so far, Wilders has still been the most talked-about politician.

Political image-makers
Generating realistic-looking photos using AI has never been easier for political parties. In Europe, radical right-wing parties in particular are using these images, and the Netherlands is no exception. Last year weekly magazine De Groene Amsterdammer revealed that  PVV MP Maikel Boon is doing this as well. The magazine reports that he even creates such images himself: the prompts he provided to OpenAI were found online alongside  the images, which can be linked to Boon via the nickname he also used on GeenStijl. One of the prompts read: “Take a hyper-realistic photo of a beautiful, innocent blonde woman walking on the beach. Behind her walks a group of darker-skinned young men. They’re calling her name, but she’s not responding.” Combined with slogans like “Keep your hands off our daughters!”, these images can be found on the Facebook page “Wij doen GEEN aangifte tegen Geert Wilders” (We are NOT filing charges against Geert Wilders). According to De Groene, this page is managed by PVV MP Patrick Crijns, although this is not visible to Facebook users; the administrators are anonymous. The page is remarkably successful. With 126,000 followers and a record number of likes, comments, and shared posts, it was number two in the top 20 most popular political Facebook pages at the end of September. At number one was Geert Wilders’ own page.

CampAIgntracker monitors political AI images

The data on the reach of the PVV’s “GEEN aangift” Facebook page comes from CampAIgntracker, which has been monitoring AI images in political social media posts on Facebook, Instagram, X, and TikTok since September. Researchers Fabio Votta (University of Amsterdam) and Simon Kruschinski (University of Mainz) previously used their tracker to monitorthe German federal elections in February. CampAIgntracker displays trends on a dashboard. For example, it shows that the PVV is the biggest image creator, and that the majority of the detected AI images appear to be realistic but are not labelled with a warning. The most frequently depicted politician is Geert Wilders, and the largest category of people is “ordinary people”.

Research
What if election debates were fact-checked live?
During election debates on television, politicians regularly make factual claims. As a viewer, you can’t help but wonder: are these statements true? Is it really correct that abolishing the mortgage interest deduction would make a difference of €400 per month? Or that a large portion of rental properties go to people with refugee status? In de Volkskrant, one reader suggests the introduction of a fact-check button:

“Give every participant a red button. During each debate, they can press the red button a certain number of times to fact-check an opponent’s claim.”

It would also be helpful if a fact-check were displayed immediately for every factual claim. This way, viewers would immediately know if a politician was trying to convince you of something that is not factually true. But what effect do such live fact-checks have on the audience? Surprisingly, very little research has been conducted on this. The only study we found was from 2017, by American researcher Amanda Wintersieck. In her experiment, Wintersieck showed students clips from an election debate in New Jersey between a Republican and a Democratic politician. In some versions of the debate, a bar at the bottom of the screen provided fact-checks, indicating confirmation (‘correct’), correction (‘incorrect’), or a mixed assessment (‘partly incorrect’).

The results show that the live fact-checks influence how viewers judge the candidates. When the fact-checks confirmed that a candidate was telling the truth, viewers’ ratings for their performance and willingness to vote for them increased. However, when the fact-checks corrected the candidate, viewers gave them lower ratings. Interestingly, the fact-check’s source (Fox, MSNBC, or PolitiFact) mattered less than the content itself. The effect is not uniform for all viewers. People with strong party preferences are less likely to switch to a candidate from another party, while people without a strong preference are more likely to change their support. Live fact-checks can therefore play a decisive role in determining viewers’ party preference and voting behaviour, especially for undecided voters. And we currently have a particularly large number of such undecided voters in the Netherlands.

Header picture: European Parliament, CC BY 2.0 <https://creativecommons.org/licenses/by/2.0>, via Wikimedia Commons