Computational Social Welfare: Applying Data Science in Social Work

by

Cheng Ren, Marla Stuart, Julian Chun-Chung Chow

• LAST REVIEWED: 27 May 2020
• LAST MODIFIED: 27 May 2020
• DOI: 10.1093/obo/9780195389678-0286

Introduction

Computational social welfare, a powerful new science, combines a focal commitment to social justice and equity with adoption of computational modeling as an epistemological paradigm and with advanced data science skills as the methodology. As a science focused on learning from data, it mirrors the values and processes of grounded theory already well established in social welfare and it elevates the use of administrative data, which is so prevalent in the social settings of interest to social welfare scholars. As a science deeply rooted in complexity theory, it promises to produce new insights about the complex and adaptive social environments in which social workers practice and conduct research. As an inherently cross-disciplinary science, it welcomes new perspectives about how to understand and solve social problems. As a science led by innovations and one in use outside of universities with later adoption by academic researchers, it provides a template for social welfare to embrace an action-oriented research agenda led by practitioners and communities. In this way, it aligns well with the participatory paradigm already embraced by many social welfare scholars. As a science that promotes transparency and open access, it facilitates a critical paradigm that can challenge oppressive beliefs and practices embedded in traditional, historical, and legacy research traditions. Computational social welfare is situated within the umbrella of computational social science. It is analogous to computational approaches in other fields, including computational biology, computational linguistics, computational finance, and computational cognition. All computational approaches exist within the broader domain of computational science, understood to be a science that uses networks, computers, software, algorithms, and simulations to create new knowledge. Please refer to the separate Oxford Bibliographies in Philosophy article “Computational Science” for more information. Computational social welfare also benefits from technology development. Technology innovation provides a foundation for computational social welfare. However, computational social welfare focuses more on application and analysis than hardware development. Please refer to the separate Oxford Bibliographies in Social Work articles “Technology in Social Work” and “Technology for Social Work Interventions” for more information about technology in social work. Computational social welfare seems like a science well suited for solving modern social challenges. However, it has not yet been widely embraced and tested by social welfare scholars. Therefore, this article aims to introduce the various facets of computational social welfare to practitioners and scholars dedicated to social well-being with a goal of advancing its use and testing. It is generally focused on the field of social welfare but will be of interest to those involved in, and it draws citations from, fields that share a commitment to improving conditions for people, including but not limited to public policy, sociology, economics, nursing, education, criminal justice, public health, psychology, and political science. Social work practitioners can learn how data science is applied in other disciplines for social well-being, including trends, argument, methods, and analysis, that could inspire social welfare scholars to enhance the social work discipline.

General Overviews

Each of these general overviews discusses a variety of subjects related to computational social welfare. These subjects are considered more fully in following sections. The author of Tukey 1962 is widely credited with creating the field of data analysis and articulated a vision, which has since been largely operationalized. Donoho 2017 provides the broadest overview and is the first work to propose a list of data science activities. Brady 2019 similarly traces the history of and describes current challenges to data science and big data specifically related to the social sciences. Lazer, et al. 2009 notes a failure to embrace the benefits of big data and data science by social academia, while Metzler, et al. 2016 reports a survey of social scientists and their engagement in data science. Mooney and Pejaver 2018 broadly summarizes ethical, epistemological, and pedagogical discussions. Coulton, et al. 2015 notes that social welfare researchers and practitioners have made limited use of big data analytics and makes the call for them to do so by building the necessary knowledge and technology infrastructures. National Association of Social Workers, Association of Social Work Boards, Council on Social Work Education, and Clinical Social Work Association 2017 provides a guide of data usage in social work practice. Cariceo, et al. 2018 summarizes some data science key elements for social work practice. Berzin, et al. 2015 illustrates the adoption of information and communication technology in social work and some potential benefits. Santiago and Smith 2019 explores the application of big data methods and data science in macro social work research and practice.

• Berzin, Stephanie Cosner, Jonathan Singer, and Chitat Chan. 2015. Practice innovation through technology in the digital age: A grand challenge for social work. Working paper 12. Columbia, SC: American Academy of Social Work & Social Welfare.

  The call to adopt information and communication technology (ICT) for social work. Notes that technology integration can shape social work practice into customized services that meet individual needs. Meanwhile, more tools such as mobile devices and social media are available for social work practice. The technology also helps social workers collected data in different dimensions. Moreover, the field needs to include technology in social work and understand how technology is incorporated in practice.

• Brady, Henry E. 2019. The challenge of big data and data science. Annual Review of Political Science 22.1: 297–323.

  DOI: 10.1146/annurev-polisci-090216-023229

  Provides a clear definition of and lists varieties of big data. Adds a new activity (archiving and governance) to the list of activities proposed in Donoho 2017. Summarizes important societal changes driven by data science. Notes that social science needs new courses and is learning new ways of working. Discusses how data science promotes the four basic problems of empirical research: forming and providing measures of concepts, providing reliable descriptive inferences, making causal inferences, and making predictions.

• Cariceo, Oscar, Murali Nair, and Jay Lytton. 2018. Data science for social work practice. Methodological Innovations 11.3

  DOI: 10.1177/2059799118814392

  Provides a brief introduction of data science and key elements for social work practice. Discusses machine learning techniques for some social work fields such as social research and evidence-based practice. Points out several challenges and ethical dilemmas from data science for social work practice.

• Coulton, Claudia J., Robert Goerge, Emily Putnam-Hornstein, and Benjamin de Haan. 2015. Harnessing big data for social good: A grand challenge for social work. Working paper 11. Columbia, SC: American Academy of Social Work & Social Welfare.

  The call to adopt data science for social work. Notes that knowledge of big data analytics among social work scholars is limited, reducing the field’s ability to demonstrate program effectiveness and collaborate with other disciplines. Moreover, the field lacks the infrastructure to adequately address privacy and ethical concerns. The grand challenge is to build the necessary infrastructure to harness the power of data science to benefit society.

• Donoho, David. 2017. 50 years of data science. Journal of Computational and Graphical Statistics 26.4: 745–766.

  DOI: 10.1080/10618600.2017.1384734

  This often-cited history of data science begins with the proposal in Tukey 1962 that data analysis is a science focused on learning from data. Attributes the acceleration of data science to the emergence of big data and advances in computer power. Highlights the importance of “the common task framework”—online predictive competitions. Proposes six activities of data science: data gathering and preparation and exploration, data representation and transformation, computing, modeling, visualization and presentation, and science about data science.

• Lazer, David, Alex Pentland, Lada Adamic, et al. 2009. Computational social science. Science 323.5915: 721–723.

  DOI: 10.1126/science.1167742

  While the private sector has embraced and developed computational methods, academia has been slow to adopt computational social science. This article lists the ways that big data answers social science questions. Warns that progress in knowledge production may be limited if (1) scientists continue to build on existing human behavior theories without exploring new theories suggested by big data, (2) infrastructure and governance standards are slow to develop, and (3) new paradigms for training and evaluation of scientific contribution are not adopted.

• Metzler, Katie, David A. Kim, Nick Allum, and Angella Denman. 2016. Who is doing computational social science? Trends in big data research. A SAGE white paper. London: SAGE.

  DOI: 10.4135/wp160926

  In this SAGE survey of social scientists, a third reported some engagement with computational research, over half use administrative data, and 79 percent engage in cross-discipline collaboration. Only 54 percent have shared code, suggesting a slower adoption of open science strategies. Key challenges include funding, data access, finding collaborators, learning new methods and software, choosing journals, and establishing a career path. Pedagogical problems include students not having appropriate programming and/or statistical knowledge.

• Mooney, Stephen J., and Vikas Pejaver. 2018. Big data in public health: Terminology, machine learning, and privacy. Annual Review of Public Health 39.1: 95–112.

  DOI: 10.1146/annurev-publhealth-040617-014208

  A discussion of big data and machine learning. Summarizes the big data types currently used in public health and the machine learning algorithms commonly applied and provides a glossary of data science terms. Discusses epistemological concerns related to machine learning for hypothesis testing, big data for causal inference, and measurement error in big data. Clearly specifies the privacy problems associated with data linkages. Suggests two necessary core skills to include in education: how to think like a computer and quantitative bias analysis.

• National Association of Social Workers, Association of Social Work Boards, Council on Social Work Education, and Clinical Social Work Association. 2017. NASW, ASWB. CSWE, & CSWA standards for technology in social work practice. Washington, DC: National Association of Social Workers.

  A guide of social work practice and technology. Summarizes how social work practice interacts with technology. With respect to data, the guide informs how social workers could collect, store, protect, and use data to benefit the public.

• Santiago, Anna Maria, and Richard J. Smith. 2019. What can “big data” methods offer human services research on organizations and communities? Human Service Organizations: Management, Leadership & Governance 43.4: 344–356.

  DOI: 10.1080/23303131.2019.1674756

  Provides an overview of the promises and challenges for using big data in human services research through a macro practice and organization and community lens. Outlines specific steps social work scholars and professionals can take to advance knowledge and skills in the mastery of this emerging method.

• Tukey, John W. 1962. The future of data analysis. The Annals of Mathematical Statistics 33.1: 1–67.

  DOI: 10.1214/aoms/1177704711

  Tukey is credited with creating the science of data analysis: the science of learning from data. He describes procedures for planning data collection, analyzing data, and interpreting results. This new science addresses problems of spotty data, abandons expectations of normal distributions, acknowledges multiple-outcome situations, emphasizes judgment and intuition, and requires a lab. Donoho 2017 traces the emergence of data science to this article.