Algorithmic long-term unemployment risk assessment in use: Counselors’ perceptions and use practices
Open accessAbstract
The recent surge of interest in algorithmic decision-making among scholars across disciplines is associated with its potential to resolve the challenges common to administrative decision-making in the public sector, such as greater fairness and equal treatment of each individual, among others. However, algorithmic decision-making combined with human judgment may introduce new complexities with unclear consequences. This article offers evidence that contributes to the ongoing discussion about algorithmic decision-making and governance, contextualizing it within a public employment service. In particular, we discuss the use of a decision support system that employs an algorithm to assess individual risk of becoming long-term unemployed and that informs counselors to assign interventions accordingly. We study the human interaction with algorithms in this context using the lenses of human detachment from and attachment to decision-making. Employing a mixed-method research approach, we show the complexity of enacting the potentials of the data-driven decision-making in the context of a public agency.
Integrity signals
Registry
Cites
Statistics
Claims the paper makes
Claims other papers make about this one
Each is checked against this paper’s own text: does it say what the citing paper says it does?
The IEFP ADS is based on demographic data about the individual, including age, unemployment length, and profession, along with other data on macroeconomic trends in Portugal.
“The system is based on demographic data about the individual, including their age, unemployment length, and profession, along with other data on macroeconomic trends in Portugal.”
From Think about the stakeholders first! Toward an algorithmic transparency playbook for regulatory compliance (2023) · cites this paper as Zejnilović et al., 2020
The IEFP ADS assigns unemployed persons as low, medium, or high risk for remaining unemployed, and job counselors assign them to interventions such as reskilling, resume building, or job search training.
“The ADS assigns unemployed persons as low, medium, or high risk for remaining unemployed, and then job counselors have the responsibility of assigning them to interventions such as reskilling, resume building, or job search training (Zejnilović et al., 2020).”
From Think about the stakeholders first! Toward an algorithmic transparency playbook for regulatory compliance (2023) · cites this paper as Zejnilović et al., 2020
The Portuguese national employment agency IEFP uses an automated decision system to allocate unemployment resources to at-risk unemployed persons.
“The Portuguese national employment agency, the Institute for Employment and Vocational Training (IEFP), uses an ADS to allocate unemployment resources to at-risk unemployed persons.”
From Think about the stakeholders first! Toward an algorithmic transparency playbook for regulatory compliance (2023) · cites this paper as Zejnilović et al., 2020
Researchers found that the explanations improved the confidence of the decisions, but counterintuitively, had a somewhat negative effect on the quality of those decisions.
“Overall, researchers found that the explanations improved the confidence of the decisions, but counterintuitively, had a somewhat negative effect on the quality of those decisions (Zejnilović et al., 2020).”
From Think about the stakeholders first! Toward an algorithmic transparency playbook for regulatory compliance (2023) · cites this paper as Zejnilović et al., 2020
In the IEFP use case, SHAP factors were given to job counselors to show the top factors influencing the score of a candidate both positively and negatively.
“In the IEFP use case, SHAP factors were given to job counselors to show the top factors influencing the score of a candidate both positively and negatively (Zejnilović et al., 2020).”
From Think about the stakeholders first! Toward an algorithmic transparency playbook for regulatory compliance (2023) · cites this paper as Zejnilović et al., 2020
The survey emulates how HITL users interact with ML systems in real-world policy settings.
“We designed the survey to emulate how HITL users interact with ML systems in real-world policy settings [70].”
From It’s Just Not That Simple: An Empirical Study of the Accuracy-Explainability Trade-off in Machine Learning for Public Policy (2022) · cites this paper as [70]
Cites
Untitled work
(n.d.)
UnresolvedUntitled work
(n.d.)
UnresolvedUntitled work
(n.d.)
UnresolvedUntitled work
(n.d.)
UnresolvedUntitled work
(n.d.)
UnresolvedUntitled work
(n.d.)
UnresolvedUntitled work
(n.d.)
UnresolvedUntitled work
(n.d.)
UnresolvedUntitled work
(n.d.)
UnresolvedUntitled work
(n.d.)
UnresolvedUntitled work
(n.d.)
UnresolvedUntitled work
(n.d.)
UnresolvedUntitled work
(n.d.)
UnresolvedUntitled work
(n.d.)
UnresolvedUntitled work
(n.d.)
UnresolvedUntitled work
(n.d.)
UnresolvedUntitled work
(n.d.)
UnresolvedUntitled work
(n.d.)
UnresolvedUntitled work
(n.d.)
UnresolvedUntitled work
(n.d.)
UnresolvedUntitled work
(n.d.)
UnresolvedUntitled work
(n.d.)
UnresolvedUntitled work
(n.d.)
UnresolvedUntitled work
(n.d.)
UnresolvedUntitled work
(n.d.)
UnresolvedUntitled work
(n.d.)
UnresolvedUntitled work
(n.d.)
UnresolvedUntitled work
(n.d.)
UnresolvedUntitled work
(n.d.)
UnresolvedUntitled work
(n.d.)
UnresolvedUntitled work
(n.d.)
UnresolvedUntitled work
(n.d.)
UnresolvedUntitled work
(n.d.)
UnresolvedUntitled work
(n.d.)
UnresolvedUntitled work
(n.d.)
UnresolvedUntitled work
(n.d.)
UnresolvedUntitled work
(n.d.)
UnresolvedUntitled work
(n.d.)
UnresolvedUntitled work
(n.d.)
UnresolvedUntitled work
(n.d.)
UnresolvedUntitled work
(n.d.)
UnresolvedUntitled work
(n.d.)
UnresolvedUntitled work
(n.d.)
UnresolvedUntitled work
(n.d.)
UnresolvedUntitled work
(n.d.)
UnresolvedUntitled work
(n.d.)
UnresolvedUntitled work
(n.d.)
UnresolvedUntitled work
(n.d.)
UnresolvedUntitled work
(n.d.)
UnresolvedUntitled work
(n.d.)
UnresolvedUntitled work
(n.d.)
UnresolvedUntitled work
(n.d.)
UnresolvedUntitled work
(n.d.)
UnresolvedUntitled work
(n.d.)
UnresolvedUntitled work
(n.d.)
UnresolvedUntitled work
(n.d.)
UnresolvedUntitled work
(n.d.)
UnresolvedUntitled work
(n.d.)
UnresolvedUntitled work
(n.d.)
UnresolvedUntitled work
(n.d.)
UnresolvedUntitled work
(n.d.)
UnresolvedUntitled work
(n.d.)
Unresolved
Flagged first, then by citations
62 of the 62 references couldn’t be matched to a known paper.
Cited by
Sorted by citations
Bell, A., Solano-Kamaiko, I., Nov, O., & Stoyanovich, J. (2022) · 2022 ACM Conference on Fairness Accountability and Transparency · doi:10.1145/3531146.3533090 · Cited by 103 · see citing works
Self-citationStreet-Level Algorithms and AI in Bureaucratic Decision-Making
Ammitzbøll Flügge, A., Hildebrandt, T., & Møller, N. H. (2021) · Proceedings of the ACM on Human-Computer Interaction · doi:10.1145/3449114 · Cited by 78 · see citing works
“This is a concrete example of how AI is implemented to support decision-making in job placement [38, 44], which is also being adopted in countries such as Austria [2] and Portugal [60].”
BackgroundExplainable machine learning for public policy: Use cases, gaps, and research directions
Amarasinghe, K., Rodolfa, K. T., Lamba, H., & Ghani, R. (2023) · Data & Policy · doi:10.1017/dap.2023.2 · Cited by 73 · see citing works
Isbanner, S., O’Shaughnessy, P., Steel, D., Wilcock, S., & Carter, S. (2022) · Journal of Medical Internet Research · doi:10.2196/37611 · Cited by 45 · see citing works
Bell, A., Nov, O., & Stoyanovich, J. (2023) · Data & Policy · doi:10.1017/dap.2023.8 · Cited by 44 · see citing works
“Overall, researchers found that the explanations improved the confidence of the decisions, but counter-intuitively, had a somewhat negative effect on the quality of those decisions [80].”
BackgroundMethodInfluentialSelf-citationMind the gap! On the future of AI research
Dahlin, E. (2021) · Humanities and Social Sciences Communications · doi:10.1057/s41599-021-00750-9 · Cited by 36 · see citing works
Gallagher, P., & Griffin, R. (2023) · Social Policy and Society · doi:10.1017/s1474746423000428 · Cited by 34 · see citing works
“Statistical profiling is seen as more accurate than simple rule-based systems or human decision-making (Zejnilovi´c et al. , 2020; Desiere & Struyven, 2021) and is often considered an objective way of segmenting jobseekers and prioritising the rationing of scarce resources (Martin & Grubb, 2001).”
Perspectives from Practice: Algorithmic Decision-Making in Public Employment Services
Flügge, A. W. A. (2021) · ACM Conference on Computer Supported Cooperative Work (CSCW) · doi:10.1145/3462204.3481787 · Cited by 10 · see citing works
Haug, K. B. (2022) · International Journal of Sociology and Social Policy · doi:10.1108/ijssp-03-2022-0085 · Cited by 6 · see citing works
Misabstraction in Sociotechnical Systems
de Troya, Í. M. D. R., Kernahan, J., Doorn, N., Dignum, V., & Dobbe, R. (2025) · ACM Conference on Fairness, Accountability, and Transparency (FAccT) · doi:10.1145/3715275.3732122 · Cited by 6 · see citing works
“…is undermined by their reluctance to disagree with the computer [112]; alienating caseworkers when AI tools don’t fit into existing workflows [116, 123]; operationalising aus-terity politics that marginalise people predicted to be at risk of unemployment [80]; codifying the notion of an ideal…”
InfluentialSelf-citationThe taming of sociodigital anticipations: AI in the digital welfare state
Zenkl, T. (2025) · Frontiers in Sociology · doi:10.3389/fsoc.2025.1556675 · Cited by 5 · see citing works
Dossche, W., Vansteenkiste, S., Baesens, B., & Lemahieu, W. (2024) · SN Computer Science · doi:10.1007/s42979-024-02884-4 · Cited by 3 · see citing works
“Prior studies consensually stipulated that the perceived usefulness of such profiling is primarily compromised by counselors’ distrust in the model arising from a dearth of understanding of the algorithm’s logic and the paucity of actionable information [10, 13].”
A Systems View of Enterprise Data Governance for Artificial Intelligence Applications in Government
Luna‐Reyes, L. F., & Harrison, T. M. (2022) · Annual International Conference on Digital Government Research · doi:10.1145/3543434.3543450 · Cited by 1 · see citing works
“Although there are some exceptions, the EU prohibits entirely automated profile decision-making for individuals; however, in countries such as Austria and Portugal, caseworkers can overrule the judgment of the algorithm [31, 56], thus inserting “humans-inthe-loop,” and maintaining the legality of the AI approach.”
BackgroundMethodInfluentialIsbanner, S., O’Shaughnessy, P., Steel, D. G., Wilcock, S., & Carter, S. M. (2022) · doi:10.2196/preprints.37611
The Performance-Actionability Trade-Off in Retention Prediction at Middle School
Lavado, S., Mateus, M., & Zejnilović, L. (2022) · International Conference on Machine Learning and Applications · doi:10.1109/icmla55696.2022.00087
“In an era of elevated expectations from machine learning models, the results often fail to meet the end-user’s expectations [21] [22], making it ever more important to make the potentialities and limitations of such models transparent to end-users.”
BackgroundSelf-citationZwaenepoel, J., & Zanoni, P. (2025) · International perspectives on equality, diversity and inclusion · doi:10.1108/s2051-233320250000012010
Mapping Risk Work and Designing Technologies to Support it in CSCW Research
Lee, M. A., Mejía, G. M., Warren, R. B., Chen, Y., Siraj, H., Mazmanian, M., Mandhre, R. A., & Pine, K. H. (2025) · ACM Conference on Computer Supported Cooperative Work (CSCW) · doi:10.1145/3715070.3748299
“For example, social workers are tasked with using simple algorithms to identify youth at risk of abuse [14, 43] or individuals who are likely to need long term un-employment benefits [2, 45, 52].”
Demirci, M. R. (2026) · Uluslararası Yönetim Akademisi Dergisi · doi:10.33712/mana.1805560
- Authors
- Zejnilović, L., Lavado, S., Martínez de Rituerto de Troya, Í., Sim, S., & Bell, A.
- Affiliations
- NOVA School of Business and Economics (PT)
- Nova School of Business and Economics, Universidade Nova de Lisboa, Lisbon, Portugal
- Venue
- Global Perspectives
- Publisher
- University of California Press
- Published
- 2020
- Volume
- 1
- Issue
- 1
- ISSN
- 2575-7350 (ISSN-L)
- OpenAlex
- W3037113180
- Cited by
- 18
- References
- 62
- Citations by year
- 2026: 1 · 2025: 4 · 2024: 1 · 2023: 3 · 2022: 6 · 2021: 3
- FWCI
- 4.09
- Type
- journal article
- Language
- en
- Open access
- green
- papers.ssrn.com
- Locations
- Global Perspectives (publishedVersion)
- SSRN Electronic Journal (submittedVersion)
- Portuguese National Funding Agency for Science, Research and Technology (RCAAP Project by FCT) (submittedVersion, other-oa)
- Funders
- Carnegie Mellon University
- Topics
- Social Sciences › Safety Research › Ethics and Social Impacts of AI
- Social Sciences › Political Science and International Relations › E-Government and Public Services
- Social Sciences › Business, Management and Accounting › Organizational Behavior and Human Resource Management › AI and HR Technologies
- Keywords
- algorithmic decision-making, long-term unemployment, risk assessment, decision support system, public employment services, employment counselors, human judgment, human-automation interaction, data-driven decision making, public governance, equal treatment
About these checks
Registry, retraction, Cites and statistics results are computed, not judged, with one exception: a printed reference that no index lists is read once by a model before it counts as a citation error. Claims are extracted by a model and kept only when their quote appears verbatim in the text; each cited source is then checked, with a model’s help, for whether it says what the claim relies on. Those verdicts can be wrong — read the claim’s quote and the cited source before relying on one.
Statistics and arithmetic are recomputed, not judged.
Confidence bands are shown only for fields calibrated on our annotated test set. None is calibrated yet, so no band is shown.
What counts as a claim ›How we measure confidence ›How we check this ›
Are you an author, or do you see a problem with this page? Request a review
