Around 25,000 citizens visited the Choose4Cyprus platform during the most recent parliamentary elections. Of the 20,000 completed questionnaires submitted, approximately 17,000 responses were ultimately selected for analysis after the application of strict scientific filters.
The post-election assessment of Choose4Cyprus is not simply a measure of the success of an online platform. It also represents a substantial body of data showing how Cypriots position themselves on some of the country's most controversial political and social issues.
Dr Vicky Triga, Associate Professor of Political Communication and Dean of the School of Communication and Media Studies at the Cyprus University of Technology, who created the project, makes it clear from the outset that the tool is not, and does not aim to become, an opinion poll.
Its primary purpose is different: to allow citizens to compare their own views with those of political parties and understand where genuine political affinity exists.
The distinction matters, particularly because the volume of data collected is enough to create the temptation of using it for election forecasting.
25,000 visits and 17,000 responses
According to Dr Triga, the platform recorded approximately 25,000 visits, while about 20,000 users completed the questionnaire.
For scientific analysis, however, that number was reduced to around 17,000.
The reason lies in the reliability filters.
Responses submitted extremely quickly through continuous clicking were excluded as potentially random. The same applied to questionnaires showing unusually long response times for particular questions, such as cases where users left their devices and returned later.
For the most rigorous analyses, factors such as average completion time and the submission of demographic information were also taken into account.
The objective, she explains, was not simply to generate an impressive participation figure but to create a dataset that could be analysed reliably.
Questions that divide opinion
Choose4Cyprus included 24 questions, and their selection was not based simply on what happened to dominate the news cycle.
As Dr Triga explains, there is one fundamental rule: the question must genuinely distinguish between citizens.
For example, a question such as "Should bureaucracy be abolished?" has little analytical value because almost everyone is likely to agree.
Much more useful are questions involving a genuine dilemma or political cost.
One example is whether there should be greater spending on education if it also requires higher taxation.
Such trade-offs force users to consider not merely what they would ideally like, but what they are willing to support when a policy choice carries a cost.
Some questions remain constant features of political competition and are repeated over time, such as wealth redistribution and privatisation.
Others emerge through ongoing monitoring of public debate, election campaigns and issues that generate genuine political disagreement.
NATO, Israel and migration
According to the research team, several questions proved particularly effective during the latest election cycle.
One concerned Cyprus' relationship with NATO.
The electorate appeared sharply divided, while most participants held clear and strongly expressed views.
A similar pattern emerged regarding the stance the Republic of Cyprus should adopt towards Israel.
Questions with a direct social impact also consistently perform well. Migration, integration of migrants and multiculturalism are issues on which citizens tend to take more defined positions.
For researchers, a good question is not one with a correct answer.
It is one that captures a genuine political divide.
Party alignment
Following the weighting process, Dr Triga says the figures for AKEL and DISY were very close to the final election outcome, while EDEK and DIPA were also reflected accurately.
In other cases, such as Fidias Panayiotou and ALMA, the picture produced by the platform was somewhat more inflated.
Volt remained a borderline case that was ultimately not reflected in the ballot box as the tool suggested.
Dr Triga nevertheless insists that Choose4Cyprus should not be assessed as an electoral forecasting mechanism.
"Our role is the period before the election," she says.
Alignment with none
Perhaps the most interesting finding concerns users for whom no strong alignment emerged with any political party.
The platform was able to produce a "yellow" alignment, indicating a weak match, when a user's responses did not correspond clearly with any political force.
In some cases, users displayed low alignment with all parties.
Political parties themselves could not easily hide behind repeated selections of "neither agree nor disagree".
The algorithm effectively penalised parties that systematically chose the middle option, reducing their chances of showing a high degree of compatibility with voters.
For Dr Triga, this is also one of the educational functions of the platform: anyone seeking power must clearly state their positions and expose them to political scrutiny.
The election outcome itself, she argues, does not necessarily mean that citizens enthusiastically returned to traditional political parties.
Some may have voted for them with considerable reservations, motivated more by distrust of the unknown than genuine identification with the familiar.
And perhaps that is the most interesting story behind those 17,000 responses: not which party the algorithm matched people with, but how many citizens discovered that their views no longer fit comfortably within traditional party labels.
