The Dutch Voting Advice Application StemWijzer is an unreliable recommendation system and should be drastically changed

Danielle Melger (2649158) and Freek Cool (2666833) 


In the modern democratic society, every few years some sort of election is held. This can be regional elections or, for example, elections to choose new members of parliament. During elections, there are people who know upfront, who or which party they will vote for. This can either be because of sufficient knowledge about politics, or a strong opinion. The remaining part of the population, who do not know already what or who they will vote for, often make use of a Voting Advice Application (VAA) or vote recommendation system. One of the most popular VAA in the Netherlands is called “StemWijzer” and is created by the company ProDemos. Such a VAA presents the user with a range of statements, to which they either can agree, be neutral or disagree. In the end, the VAA will present a list of parties, and how much those candidates agree to the personal preferences of the user. According to ProDemos, their voting guide is used, in 2021 for the elections of the House of Representatives, by 7.8 million people (ProDemos, 2023). Which is a large percentage of the total Dutch inhabitants today, namely 43.76% (Centraal Bureau voor de Statistiek, 2023). Hence, VAA’s stimulate the public to vote, but also give citizens a better-informed view of the positions of parties to a specific statement (Munzert et al., 2020). 

Nevertheless, the result from a voting guide can influence the final decision of a person in such a way that he or she will be voting for the party or person resulting from this test (Boudreau et al., 2018). Therefore, VAAs like the Stemwijzer have the ability to steer people in their voting behavior. This can have a big impact when VAA services are biased or corrupt. In 2010, for example, there was already a fuss about the Stemwijzer. The Dutch public broadcaster BNN VARA (2010), had published a news article, in which they wrote that the “StemWijzer” is unreliable. This was due to parties delivering standpoints that were not in alignment with their election program. A negative consequence of such a deception, whether it is consciously or unconsciously, is a diminished democracy. This article, therefore, states that the Dutch VAA “StemWijzer” should be adjusted as it is an unreliable system for making recommendations. In addition, because “StemWijzer” is used by many people, it is important for democracy to create a VAA that is in any way unbiased. 

The first part of this article provides the cons against this statement and the second part discusses to what extent the Dutch VAA must be changed to prevent vote manipulation and save democracy. Moreover, this article proposes alternatives that might be more reliable than the existing design of the Dutch VAA. Ultimately, a conclusion is stated in which an overview is given of both sides of the standpoint and why this article aims to adjust the Dutch VAA. 

The cons for adjustments to the Dutch voting guide “StemWijzer”

The StemWijzer is created by an organization called ProDemos. ProDemos has the mission to explain the rules of law and democracy and inform people about what they themselves can do to exert influence in the municipality, the province, the country and Europe. This is done, among others, by guest lectures and educational activities at schools, courts and provincial houses throughout the Netherlands (Over ProDemos – Huis voor democratie en rechtsstaat, n.d.). The organization is subsidized by the Ministry of the Interior and Kingdom Relations and the House of Representatives. Moreover, the corporate structure contains a Supervisory Board, which means that the interests of stakeholders, in this case, the candidates, are sufficiently safeguarded by the management. Furthermore, StemWijzer serves as a digital VAA in which the positions of candidates or parties are compared to the personal preferences of citizens. To ensure reliability and impartiality ProDemos values the simplicity of the voting advice application, referring to not generating any confusion regarding the standpoints, with the aim of not losing transparency (Fossen et al., 2012). 

The pros for adjustments to the Dutch voting guide “Stemwijzer”

The previous section has so far shown that Stemwijzer should not be adjusted as the organization that has created the VAA first of all, shows to have great knowledge about the rules of law and democracy. Secondly, the institution contains a supervisory board and finally, the organization is aware of transparency issues and therefore, values the simplicity of the tool. However, this article argues that VAAs like Stemwijzer need to be changed urgently, as it has flaws that make StemWijzer dangerous because of the relatively many people that use it each election. Namely, some recommendations made for the voter are unreliable and inaccurate, causing voters to vote for a party they do not necessarily agree with. A relevant problem is that parties can change their standpoints on the basis of the propositions stated in a VAA, or even have an influence on which propositions are presented to the user, and what the standpoint of the party is (Garzia & Marschall, 2012). This can cause the propositions and the corresponding opinions of parties to be unreliable, as there is a high chance that the parties change their standpoints after the elections. The system can thus return to the user that their opinions correspond a lot to a certain party, while this party changed their standpoints, or even submitted a faulty standpoint. This is very misleading for voters. Therefore, is proposed that political parties must be prohibited from submitting their own statements and their standpoints. This is to make the online VAA system more reliable, and true to the voters’ opinions. To add is suggested that the propositions and standpoints are designed by multiple people, not connected with any political party, to make the designed propositions as unbiased as possible. Biased data would, namely, lead to faulty recommendations.

Also, the spatial model used to calculate which party is most in line with the user’s opinion has an influence on which party is agreed most upon by the user. As research concluded, the order of agreement of users changes when the VAA changes its spatial model (Louwerse & Rosema, 2013). Specifically, the StemWijzer makes use of an agreement model, which increments points when a user agrees with a certain party, and decreases points if not. However, the paper by Louwerse and Rosema (2013) concluded that 90% of users would have received a different recommendation from the system when a different spatial model was used. As this percentage suggests, this is a huge number considering that the StemWijzer had 4.9 million usages already in 2012 (Garzia et al., 2014). This makes a VAA like the StemWijzer highly inaccurate, as the user gets a different recommendation just by changing the spatial model, and nothing on the user his/her opinion itself. This is a big problem, and changes should be made to cope with this, as users can be making a wrong vote just because of the spatial model. Ultimately, from an Artificial Intelligence perspective it would be suggested to try to use a decision tree algorithm to determine the recommendations to public voters. The benefit of a decision tree is that it is explainable and transparent to the user (Xu et al., 2019). This in combination with an independent team that determines the statements and parties that fairly show their standpoints to these statements might be a solution to the unreliability of the current Dutch VAA. 


This article has presented the pros and cons of the statement: The Dutch voting advice application “StemWijzer” is an unreliable recommendation system and should be drastically changed. Initially, this article showed that the “StemWijzer” should not necessarily be adjusted as ProDemos, the company that created it has well-grounded knowledge about the rules of law and democracy and moreover, the organization values transparency and the interests of the candidates by having a supervisory board and a simple tool to understand.  

Nevertheless, the pros section has shown that the standpoints provided by the parties do not always agree with their election program. This has the consequence that biases can occur in VAA, which means that VAA can consciously or unconsciously manipulate voting. Moreover, it is shown that the spatial model used by StemWijzer is not accurate, as another spatial model returns different recommendations. For both of these issues, a possible solution is presented in the article, that proposes the use of a decision-tree model in combination with an independent team that composes the statements to which the parties can deliver their standpoints.


Boudreau, C., Elmendorf, C.S. and MacKenzie, S.A. (2018) “Roadmaps to representation: An experimental study of how voter education tools affect citizen decision making,” Political Behavior, 41(4), pp. 1001–1024. Available at: 

Centraal Bureau voor de Statistiek. (2023, 30 januari). Bevolkingsteller.

Fossen, T., Anderson, J. H., Tiemeijer, W., van’t Hof, C., Timmer, J., & van Est, R. (2012). Wijzer stemmen? StemWijzer, Kieskompas en het voorgeprogrammeerde electoraat.

Garzia, D., Trechsel, A. H., Vassil, K., & Dinas, E. (2014). Indirect campaigning: past, present and future of voting advice applications (pp. 25-41). Springer International Publishing.

Garzia, D., & Marschall, S. (2012). Voting advice applications under review: The State of Research. International Journal of Electronic Governance, 5(3/4), 203. 

Louwerse, T., & Rosema, M. (2013). The design effects of voting advice applications: Comparing methods of calculating matches. Acta Politica, 49(3), 286–312. 

Munzert, S., BarberÁ, P., Guess, A., & Yang, J. (2020). Do online voter guides empower citizens? Evidence from a field experiment with digital trace data. Public Opinion Quarterly, 84(3), 675-698.

Over ProDemos – Huis voor democratie en rechtsstaat. (z.d.). ProDemos. 

ProDemos. (2023, January 25). Recordaantal Gebruikers voor Stemwijzer: 7,8 Miljoen. ProDemos. Retrieved February 3, 2023, from 

“Stemwijzer niet betrouwbaar” – Joop – BNNVARA. (2010, 8 mei). BNNVARA.
Xu, F., Uszkoreit, H., Du, Y., Fan, W., Zhao, D., & Zhu, J. (2019). Explainable AI: A brief survey on history, research areas, approaches and challenges. In Natural Language Processing and Chinese Computing: 8th CCF International Conference, NLPCC 2019, Dunhuang, China, October 9–14, 2019, Proceedings, Part II 8 (pp. 563-574). Springer International Publishing.

Leave a Reply

Your email address will not be published. Required fields are marked *

Human & Machine Power & Democracy

AI and Personal Privacy: Navigating the Fine Line Between Convenience and Surveillance

As Artificial Intelligence continues to evolve, it is integrated into almost every aspect of our lives, bringing a new level of convenience and efficiency. From smart assistants and chatbots that perform a range of tasks on our command to facial recognition software and predictive policing, AI has undoubtedly made our lives easier. But, this convenience […]

Read More
Power & Democracy Power & Inequality

Navigating the AI Era: The Imperative for Internet Digital IDs

The rapid advancement of Artificial Intelligence (AI) presents a dual-edged sword, offering unprecedented opportunities while introducing complex challenges, particularly in the realm of digital security. At the heart of these challenges is the pressing need for effective internet identification systems capable of distinguishing between human and AI interactions. We will explore the vital importance of […]

Read More
Power & Democracy Power & Inequality

Data privacy: Why it should be the next step in AI regulation

In the current stage of development in Artificial Intelligence (AI), there is nothing more important than data. It’s the fuel of any statistical-based AI method. The most popular classes of models ingest enormous amounts of data to be trained, such as ChatGPT, Google Bard, PaLM. However, in many models, the users do not explicitly give […]

Read More