1
INTRODUCTION TO THE HANDBOOK OF COMPUTATIONAL SOCIAL SCIENCE
Uwe Engel, Anabel Quan-Haase, Sunny Xun Liu and Lars Lyberg
DOI: 10.4324/9781003025245-1
We write the introduction to the two-volume Handbook of Computational Social Science with excitement and awe. The handbook brings together a considerable corpus of research and insight, with 22 contributions for volume 1 âTheory, Case Studies, and Ethicsâ and 22 contributions for volume 2 âData Science, Statistical Modelling, and Machine Learning Methods.â Over 90 experts contributed from a wide range of academic disciplines and institutions to provide a mosaic of the diversity of computational social science (CSS) scholarship. They lay out the foundation for what CSS is today and where it is heading, outlining key debates in the field, showcasing novel statistical modeling and machine learning methods, and also drawing from specific case studies to demonstrate the opportunities and challenges presented by CSS approaches.
Our goal with the handbook is to reach a wide readership by taking a multidisciplinary and multimethod approach. The handbook includes foundational chapters for up-and-coming scholars and practitioners, who are interested in consolidating their understanding of key terms, methods, and means of data interpretation as well as more advanced analytical approaches that serve as a learning resource for current experts in CSS. The handbook is aimed at a wide range of scholars with backgrounds in technical fields such as statistics, methods, and computer science, as well as scholars in the social and behavioral sciences, the latter notably including psychology, cognitive science, sociology, communication, new media studies, and political science, among others. The handbook also allows practitioners and policymakers who are tasked with analyzing a specific data set to adopt a set of best practices to guide their research efforts.
Computational social science is an exciting field of research that is growing rapidly and in diverse directions. In 2018, Springer started the publication of its dedicated journal, the Journal of Computational Social Science, which is broad in scope and interdisciplinary in nature and aims âto understand social and economic structures and phenomena from a computational point of viewâ (https://www.springer.com/journal/42001, para. 2). For example, its volume 3, issue 2, special issue was on âMisinformation, Manipulation and Abuse in the Era of COVID-19,â bringing together top modeling and simulation experts on the topic. This demonstrates how CSS is having a direct impact on our understanding of pressing societal research questions and advancing novel methodological approaches. Computational social science research has an important role to play in society by providing unique and policy-relevant insights through its capability to apply computational approaches to large data sets. Besides, many special issues with a CSS focus are being proposed, such as the 2021 call for papers by the journal Chinese Sociological Review. The call requests manuscripts that âoffer important theoretical and empirical insights to advance our understanding of the development in CSS and help move forward the field in Chinese societiesâ (https://think.taylorandfrancis.com/special_issues/computational-social-science/, para. 3). This shows that there is an invisible college of the kind that Diana Crane (1972) described in her earlier work forming around the topic, with a rapid âproliferation of conferences, workshops, and summer schools across the globe, across disciplines, and across sources of dataâ (Lazer et al., 2020, p. 1060). As the field grows, CSS collaborations and research clusters are emerging globally. The field is not only growing but also consolidating, and our handbook represents an important step forward in this process.
The chapters in the two-volume handbook provide a deep understanding of theory as well as methodological opportunities and challenges. The main aims include:
- Theoretical debates: Key theoretical debates are presented from different perspectives to show how the field is gaining strength and evolving.
- Showcasing novel statistical modeling and machine learning methods: The chapters describe cutting-edge methodological developments and their application to a range of data sets.
- Ethical debates and guidelines: The chapters demonstrate the dimensions of data ethics and the need for guidelines and appropriate pedagogy.
- State-of-the-art artificial intelligence: The chapters highlight the use of various AI systems as methodological tools, data sets, and a combination of both.
- Demonstrating cross-disciplinary applications of CSS: The chapters promote the development of new interdisciplinary research approaches to answer new and pressing research questions.
- Computational methods of data collection and data management: The chapters support our understanding and application of such methods in the field.
1 Computational social science and the digitization of everyday life
Computational social science is an interdisciplinary field of study at the intersection of data science and social science that pursues causal and predictive inference as its main objective. With historical roots in mathematical modeling and social simulation, the recent digitization of all aspects of everyday life has turned CSS into a dynamically developing and rapidly growing research field. With the seamless integration of digital technology, from mobile phones to AI, into the rhythms of everyday life, there is also a greater generation and accumulation of related behavioral data of prime interest to the social sciences. Because most of these data are digital, CSS calls for computational methods of data collection, data management, data processing, and data analysis (Lazer et al., 2020). Computational social science is, thus, an evolving field with a mix of big-data, computational-methods, and data-science facets, as will be further detailed in this handbook (Engel, volume 1: chapter 9).
While it is true that computational methods are becoming increasingly relevant to the social sciences, it is equally true that the social sciences have participated in the development of a range of computational methods from the outset: some examples include statistical data analysis, social simulation, and mathematical modeling. For example, Colemanâs (1964) seminal book Introduction to Mathematical Sociology constitutes one of the precursors of contemporary analytical sociology by explaining the unique challenges in attempting to quantify social behavior: âBecause behavior is usually expressed in qualitative terms, any mathematical language which can serve for social science must in some fashion mirror this discrete, nonquantitative behaviorâ (p. 102). Also included in these early developments was Columbia Universityâs Bureau of Applied Social Research (Barton, 1979) with its pioneering work on multivariate analysis, measurement, the analysis of change, and multilevel methodology. Additionally, we can also highlight the contributions to CSS made in the area of social simulation and its multilevel approach toward social complexity (Cioffi-Revilla, 2017). Hox (2017, p. 3) similarly identifies the social simulation branch as critical in the development of CSS as a unique field of study when reviewing âthree important elementsâ of CSS, listing âbig data, analytics, and simulation.â
The very interdisciplinary nature of CSS and its diverse historical roots make this field difficult to define and delimit. While some scholars narrow the scope of CSS to âbig data,â harvested through computational methods from social media platforms â often referred to as digital trace data â this narrow approach is perhaps questionable (Cioffi-Revilla, volume 1: chapter 2; Lorenz, volume 1: chapter 10). âBig dataâ are not only available through social media data extraction; many other sources of âbig dataâ are available, including location-based data collected via mobile phones, bank transaction data, e-health records, and e-commerce transactions. Another important point is that the development of artificial intelligence (AI) and its integration into daily life through robotics with smart speaker assistants (SSA) is another source of data for CSS. As SSAs like Google Home and Amazon Echo (Alexa) support a range of activities from e-commerce to information provision, they also generate increasing amounts of data (Brause & Blank, 2020). As our society advances toward sensor-rich computational environments in the future, smartphone, smart office, and smart city devices will also be primary data sources.
Another approach to delimiting the field of CSS is by looking at its relation to data science. Hox (2017, p. 3), for instance, regards CSS as an interdisciplinary field that combines mathematics, statistics, data science, and social science. In this view, data science is a central tool, among others, in CSS. Some proponents of data science go as far as to subsume CSS under data science, in which case it is regarded as a ânew interdisciplinary field that synthesizes and builds on statistics, informatics, computing, communication, management, and sociology ⌠to transform data to insights and decisionsâ (Cao, 2017, p. 43:8). In contrast, Kelleher and Tierney (2018) move away from any kind of taxonomy and instead stress the commonalities that exist between data science and CSS such as the focus on improving decision-making through a reliance on data. For Kelleher and Tierney, all these concepts are used in the literature interchangeably â data science, machine learning, and data mining â although data science can be broader in scope. While machine learning âfocuses on the design and evaluation of algorithms f...