Type and press Enter.

How a tool called Pairity is using data to gauge community support for refugees

Written by Craig Damian Smith, Ryerson University and Anja Kilibarda, Columbia University. Photo credit: Darryl Dyck/THE CANADIAN PRESS. Originally published in The Conversation.

A record 26.4 million refugees are displaced globally. Typically, less than one per cent are resettled annually. Resettlement figures, in fact, dropped to a 10-year low in 2020 as a result of efforts in the United States by former president Donald Trump to slash quotas and due to global travel restrictions from the COVID-19 pandemic.

Pointing to Canada’s Global Refugee Sponsorship Initiative, some countries and non-governmental organizations increasingly see community sponsorship as a means of increasing resettlement. U.S. President Joe Biden signed an executive order to explore private sponsorship, the European Commission has proposed an EU-wide model and several state and city-level initiatives exist throughout Europe, Latin America and Australia. Particularly in Europe, there’s a belief that the Canadian system of privately sponsoring refugees leads to better integration.

A 2020 book on the Canadian model argues that our resettlement policies, integration successes and large-scale immigration are mutually reinforcing dynamics. Favourable public opinion helps create political will to increase resettlement and alleviate the global refugee crisis. Both are largely absent in Europe.

Evidence about the impact of private sponsorship on refugee integration also remains mostly anecdotal or reliant on national-level data in Europe. European policy-makers want reliable data and project-based monitoring and evaluation instead of good news stories or data skewed by a range of factors, including the notion that resettling privately sponsored refugees often acts as de facto family reunification.

People take part in a rally calling on the federal government to expand the permanent status program to include all refugees, international students, undocumented migrants and temporary foreign workers in Montréal in May 2021. Graham Hughes/ THE CANADIAN PRESS.

‘Pairity’ paints a more accurate picture

In order to fill the evidentiary gap, our team of social and data scientists developed a unique matching and monitoring tool called Pairity. Pairity is a data-driven, real-world intervention that tests long-standing speculation that giving refugees access to social networks leads to faster and better integration, while also improving social cohesion with communities that open their doors to them.

We partnered with Justice & Peace Netherlands through a pilot initiative called Samen Hier, which means “together here” in Dutch, to match 35 groups of volunteers with refugee newcomers in four municipalities for a year. The pilot was recently covered in detail by the BBC World Service program People Fixing the World.

While Pairity is based partly on the private refugee sponsorship model, it matches volunteers and refugees residing in the same community and eliminates monetary or legal liability for the first year of settlement. That helps avoid the paternalism that financial dependence can foster.

It works by administering biographical and preference-ranking surveys to groups of volunteers and refugees. Surveys collect baseline, mid-point and exit data on quantitative metrics like language, education and employment, and on less tangible measures like social attitudes and a sense of belonging. Matches are made by running baseline data through a preference-matching algorithm similar to those used by dating services and medical residency placements.

For the pilot, matching protocols used only three inclusion and exclusion criteria: preferences for household size and composition, geographical distance and a vulnerability index for newcomers and a capacity score for volunteers. The pilot recruited a larger pool of newcomers who were randomly sorted into treatment (matched with volunteer groups) and control pools (not matched) once the algorithm identified at least two viable matches. It made best matches across the whole population.

Improving refugee outcomes

We recognize there are problems associated with using algorithms and artificial intelligence in migration governance, particularly as a tool for immigrant and refugee selection or as a means for predicting mobility and controlling borders. That said, ongoing work by colleagues at Stanford University shows significant potential for improving refugee employment and labour market opportunities. The Pairity algorithm has no impact on immigrant selection or settlement services.

The Samen Hier mid-term evaluation, based on treatment group surveys, showed matching is associated with employment opportunities, improved language competence, better social cohesion, cultural empathy, awareness of government services and a general improvement in the life experiences of refugees.

As with a similar matching initiative we helped launch in Canada in 2016, Samen Hier was able to recruit volunteers in the 30-year-old to 60-year-old age range. Most had not previously worked with refugees. This is remarkable for three reasons.

A person holds up a sign at a pro-immigration rally in Montréal in May 2021. Graham Hughes/ THE CANADIAN PRESS.

First, volunteers in developed countries are most often students or retired people. While they bring vibrancy and experience, mid-career professionals more closely match the demographics of refugee newcomers. Second, people in their working years have more active social and professional networks. Third, it shows community support programs can include new people in integration.

Random assignment to treatment or control pools is the key to the ongoing evaluation of final data. Because we collect rich baseline data, we can also evaluate whether the effects of matching are conditional on newcomers’ characteristics, those of volunteers or some combination of the two. Results will be submitted to scientific journals for peer review, aiming to offer robust data for evidence-based policy-making.

Next steps

Expanding from the Pairity pilot will provide the statistical power to compare outcomes between treatment and control groups and analyze causal relationships between community sponsorship and integration outcomes. Broader recruitment can help identify the impact of social networks on a broad range of integration indicators, and measure whether differences are caused by social connections or other factors.

To our knowledge, Pairity is the first tool to use a data-driven framework to match and measure the effects of community sponsorship for refugees. It started with the recognition that European policy-makers are more concerned with integrating refugees already in Europe than increasing resettlement.

Pairity was designed to be adaptable to different countries. Our hope is that policy-makers and civil society organizations will adopt the model to empower everyday people to play a role in integrating resettled refugees or new asylum-seekers. In turn, these relationships can help build the experience and evidence necessary to promote further global resettlement for displaced people around the world.