The Dutch government is facing calls to pay compensation after it emerged that students with foreign-sounding surnames?were disproportionately likely to be checked for grant fraud.
In a , the Ministry of Education, Culture and Science (OCW) admitted that students living in “neighbourhoods with a high number of residents with a migration background” as well as those with a “traditionally non-Dutch-sounding surname” were more likely to receive home visits to ensure they lived at their stated addresses. To qualify for a non-resident grant, students must live away from their family homes.
To select students for checks, the Education Executive Agency (DUO) employed an algorithm?that assigned them a “risk score”. While the algorithm did not directly employ data on migration background, it did attribute higher risk to associated characteristics. For instance, grant recipients who lived closer to their parents were more likely to be chosen for checks, while data from Statistics Netherlands indicated that students with migration backgrounds tended to live closer to home, the ministry said.
Robbert Dijkgraaf, the education minister, said that the “indirect discrimination” should “not have happened” and issued an apology. But Gülsüm ?eki?, a lawyer who represented several students, told?Times Higher Education that the government should go further.
“Hopefully [the government] will compensate the students for the damage they have caused,” Ms ?eki? said. “For now they have only said sorry.”
The discriminatory checks?affected students “immensely”, said Elisa Weehuizen, chair of the National Student Union (LSVb). “They feel like the government is not there for them; that they’re not supporting them as students; that they’re not part of the norm that they want in education; that what makes them different in a cultural sense isn’t appreciated. It leads to a very big sense of not being able to lean on the education system or the government,” she said.
Details of DUO’s approach were who surveyed more than 80 lawyers and determined that, in 97 per cent of cases, the students accused of fraud came from migrant backgrounds.
Dr Dijkgraaf subsequently suspended DUO’s use of the algorithm, with home visits now determined at random.?After an investigation conducted by the consultancy PricewaterhouseCoopers (PwC), the ministry also admitted that it, along with DUO, had failed to address earlier “signals indicating that prejudice was a risk”.
The PwC investigation did not assess individual cases in which students were accused of fraud. But Ms ?eki? said that many of her clients had faced significant financial strain after having to pay back their grants, plus fines.
Asked whether it would pay compensation where students had been wrongly targeted, a government spokesperson said it would “engage in dialogue with the group concerned in such situations. Not acting hastily, but first discussing what has happened and what the experiences and needs of the students are.”
Gwen van Eijk, a researcher and policy officer at Amnesty Netherlands, said DUO’s algorithm was developed “without any checks on possible discriminatory effects. There was no proper evaluation, no oversight.”
Dr van Eijk pointed to a recent??that saw “dual nationality” and “foreign-sounding names” used as markers of fraud.
“There is a lack of transparency and accountability,” she said. “We don’t really know how governmental organisations in the Netherlands are checking people, whether they use algorithms, whether they use risk profiles, what they do to prevent discrimination.”
