Big Data for Qualitative Research

Kathy A. Mills

• Publisher: Taylor & Francis
• Year: 2019
• eBook ISBN: 9780429509292
• Edition: 1
• Topic: Psychology
• Subtopic: Research & Methodology in Psychology
• Index: Psychology

Big Data for Qualitative Research covers everything small data researchers need to know about big data, from the potentials of big data analytics to its methodological and ethical challenges. The data that we generate in everyday life is now digitally mediated, stored, and analyzed by web sites, companies, institutions, and governments. Big data is large volume, rapidly generated, digitally encoded information that is often related to other networked data, and can provide valuable evidence for study of phenomena.

This book explores the potentials of qualitative methods and analysis for big data, including text mining, sentiment analysis, information and data visualization, netnography, follow-the-thing methods, mobile research methods, multimodal analysis, and rhythmanalysis. It debates new concerns about ethics, privacy, and dataveillance for big data qualitative researchers.

This book is essential reading for those who do qualitative and mixed methods research, and are curious, excited, or even skeptical about big data and what it means for future research. Now is the time for researchers to understand, debate, and envisage the new possibilities and challenges of the rapidly developing and dynamic field of big data from the vantage point of the qualitative researcher.

1 What is big data?

We are living in an era of massive digital change that is transforming the fundamental nature of communication and the way we research. The prevalence and potential of big data has captured the imagination of public media since early articles in The Economist (Cukier, 2010) and The New York Times (Lohr, 2012). Premier scientific journals, such as Nature (“Community cleverness required”, 2008) and Science (“Special online collection: Dealing with data”, 2011), launched special columns dedicated to the discussion of big data. Big data across a broad range of fields is collected, analyzed, stored, and disseminated digitally. These changes have contributed to a variety of responses, ranging from the amplification of big data promise, to digital angst about ethical and productive ways to exist amid digital change and disruption. What constitutes appropriate use of big data in contexts of increasing global multiplicity of texts, information, and facts?

The digitalization of data is becoming ever more apparent, functioning in transformational ways in everyday life for individuals and organizations. Digital data is particularly salient given the broadened range of communication channels, media, and social media. Data is now generated rapidly from a variety of sources, such as social media and video feeds, mobile devices, radio frequency identification readers, genome sequences, medical databases, software logs, and wireless sensory networks (Fuller, Buote, & Stanley, 2017). The main big data types commonly cited include social media data, transactional data, administrative data, sensor data, and personal data, such as from tracking devices (Shlomo & Goldstein, 2015).

Big data is generating immense interest from researchers, research grant funding bodies, industries, marketing companies, and beyond. Researchers may find the terminology and analytic methods associated with big data confusing, such as data mining, machine learning, sentiment analysis, and deep learning. In addition, there are new methods of text, image, audio, and video analytics (Fuller et al., 2017). Across many disciplines, researchers can no longer ignore the big data hype, as many seek to know how big data practice can transform the future of fields that use both quantitative and qualitative research methods, and to determine whether new analytic methods for big data can help to answer different kinds of research questions in innovative ways.

Even if qualitative researchers may not need real-time analytics to answer questions urgently, there is an unprecedented quantity of textual data beyond the internet that is potentially useful to qualitative researchers, such as digital copies of almost every book published (e.g. Google Books), news archives and newscasts (e.g. Proquest), legislative reports and discourses (e.g. National Archives of the United States and Great Britain), podcasts, historical audio sources, local town hall meetings, interviews, and field notes [e.g. Dataverse Network or Economic and Social Data Service (ESDS) Qualidata]. There are increasingly fewer types of text, image, and speech-based data that are not being archived by someone (Bail, 2014). Qualitative researchers typically invest large amounts of time manually organizing and coding textual, image, and video data, while qualitative data may be transformed into quantitative data to apply data mining and visualization techniques (O’Halloran, Tan, Pham, Bateman, & Vande Moere, 2018).

Aims of this book

This book addresses the roles of both big data and qualitative research in a world in which there are data of massive breadth across so many fields and spheres of human activity. Within the academic community, some have argued that big data renders small-scale research, commonly used in the social sciences and humanities, potentially at risk (Alberts, 2012; Berlekamp, 2012; Mayer-Schönberger & Cukier, 2013). If social and behavioral data that were previously the locus of much qualitative research have been “datified” (Strong, 2013), is the role of the qualitative researcher losing a significant foothold? Before researchers blindly follow big data trends, questions need to be asked about the accessibility, ethics, utility, costs, and limits of big data. What is the scale of analysis necessary to understand phenomena in the particular area of research interest?

Is the use of big data incommensurate or diametrically opposed to the values of the qualitative researcher? Should big data and qualitative research be seen as complementary and suited to particular types of social questions and problems? What exemplars do we have to support the integration of qualitative methods with big data in mixed methods research investigations, which integrate the combined strengths of quantitative and qualitative interpretations to address research problems?

This small book on big data debates both the limits and potentials of big data for qualitative researchers. After historically contextualizing big data in Chapter 2, it outlines a range of big data methods and analytic tools that are increasingly of interest to qualitative researchers in Chapter 3. The potentials and pitfalls of big data are examined, including assumptions about who has access to big data and who misses out. Chapter 4 explores big data methods and analytics, with an eye towards new developments for qualitative researchers. Later chapters discuss issues of ethics and privacy in a risk society, such as surveillance and data ownership. Together, the chapters explore some of the potentials of combining the strengths of big data with those of qualitative research, such as combining automated tools for the analysis of big data with the interpretative theories or cultural frames of reference generated through qualitative research. Such approaches can use qualitative research to contextualize the social milieu in which the data were produced (Mills, 2017).

Throughout the volume, I address critical questions, debates, and perspectives of big data trends, unearthing both the promises and challenges. This is vital given current and future opportunities for big data applications, and conversely, new concerns about ethics, privacy, surveillance, and secondhand digital data use, on the other. The use of data, whether big or small, must account for the burgeoning array of data forms that extend well beyond numerical data to include multimodal data—data that includes combinations of words, images, audio (e.g. sound effects, music, voice recordings), moving images (e.g. video), and three-dimensional designs. The notion of big data has inspired, excited, confused, frustrated, and provoked researchers worldwide as digital media render voluminous data sets more readily discoverable, distributable, open to scrutiny, and more efficiently able to generate answers to the questions that researchers ask (Mills, 2017).

Chapter snapshot

This chapter will define and debate the role of big data for researchers, while considering the questions that big data poses for qualitative researchers. Too often, data experts do not acknowledge the different roles and values of researchers who work in fields with little data, rare data, or even with no data (Borgman, 2015). To qualitative researchers, data can include vastly diverse traces of human activity across a variety of modes and media—data which is often not primarily numerical (e.g. video, photographs, conversations, 3D printing, drawings, tapestries). Thus, to the qualitative researcher, big data might appear on the surface to be somewhat irrelevant. However, qualitative researchers are increasingly conscious that scholarship occurs in a digitally networked world (Borgman, 2018), and most qualitative data can be rendered, analyzed, archived, and shared digitally. The digital transformation of information and communication technologies means that qualitative researchers are not insulated from the changes that are opening up new efficiencies, larger data sets, quicker analysis, and new ways of answering important questions. Some disciplines have been slower than others to recognize and harness the potentials of new big data analytics, such as sentiment analysis, machine learning, and techniques for big data visualization.

Defining big data

What exactly constitutes big data, and how do qualitative researchers come to terms with this phenomenon? There are almost as many definitions of big data as there are theorists. For example, some simply define big data as “enormous datasets” that may be structured, semi-structured, or unstructured (Chen, Mao, & Liu, 2014), while others argue that the main emphasis to date in much of the literature is on unstructured data (Dedić & Stanier, 2017). The size of big data is constantly changing with technological advances, rendering the use of terabytes or exabytes to define big data as an inexact science. Regardless, there is agreement that big data are diverse, complex, and massive in scale, requiring specialized analytic tools and technologies to capture, process, and reveal insights in a timely way (Hashem et al., 2015).

Many specify the characteristics of big data using the three V’s: volume, velocity, and variety (Chandler, 2015; Laney, 2001). Volume refers to the mass scale of data, although theorists acknowledge that the relative scale varies across different disciplines. The current volume of big data is historically unprecedented, and increasing every year (Fuller et al., 2017).

Velocity refers to the rapid generation of data or timeliness. Examples of high-velocity applications include machine sensory monitoring in production lines, satellite imagery, credit card fraud detection, wearable data, parcel tracking and other global positioning applications, social media feeds, customer web browsing, power usage data, and clickstream analysis of web user data to recommend products and target advertising.

Variety addresses the various modalities or types of structured, semi-structured, and unstructured data, which may include numerical data, textual data, audio, images, and so on (Chandler, 2015). Integrating different types of data from multiple sources can potentially address new questions. Using multiple data types is called data fusion—this may produce more accurate, complete, and contextualized information than what can be achieved through analyzing as single type of data.

Others have added a fourth V—value—which refers to the need to extract the often-covert value from enormous, rapidly generated data sets of various types (Chen et al., 2014). Still others have added a fifth V—veracity—meaning the quality of the data or low “noise” in the data. A shift in any of these dimensions can influence the scale of research and scholarship (Fuller et al., 2017).

Distinguishing between big and not big is problematic, because data can be big in a multitude of ways—such as what can be done with them, the insights that they reveal, and the scale of extraction and analysis required to make them useful (Borgman, 2015). The challenges of managing data of significant volume, velocity, and variety have actually been discussed since the turn of the twenty-first century, including debates about 3D data (Laney, 2001). The term “big data”, however, was only added to the Oxford Dictionary in 2013, defined as “data of a very large size, typically to the extent that its manipulation and management present significant logistical challenges” (cited in Borgman, 2015, p. 6).

I define big data as rapidly generated, digitally encoded information of significant volume, velocity, variety, value, and veracity: Data that is used as valued evidence for a phenomenon, and which often has relationality with other networked data (Clarke, 2016; Mills, 2017). Big data needs analyses to be useful for scientific work or knowledge production, and a common feature of big data is the associated logistical challenges of its analysis, manipulation, reduction, and management, due to its enormity.

It is also important to consider what is meant by the contested term “data” in scholarship. The term “data” refers to “entities that are used as evidence of phenomena for the purpose of research” (Borgman, 2015, p. 29). Entities include any variety of forms of information suitable for interpretation—from material objects, such as sculptures, artwork, and drawings, to digital data, such as digital images, textual data, and numerical information. When referring to entities, data is used in the plural form, except when referring to data as a concept. It is important to recognize that data are not truths in themselves, but rather, are facts or sources of evidence that are used to support a claim about reality (Borgman, 2015).

Big data examples

Digital data, such as third-party cookies, web and marketing analytics, on-site web engagement analytics, GPS tracking, and other tracking technologies, are like footprints marking time and place, creating ongoing records of social communication, and location activity (Mills, 2017). Common types of big data can include traces of human activity captured on social media, learning analytics, business and operational data, web and mobile analytics, streaming data, commercial and government data, and sensory data from the Internet of Things or IoT. Other applications of big data include visualizations of large data sets, machine learning, sentiment analysis, opinion mining, computer-assisted content analysis, natural language processing, and automated data aggregation and mining (Lohmeier, 2014; Parks, 2014). For example, Google manages one of the largest sources of big data, enabling analysis by the public through the open access tool, Google Insights. Apple, Twitter, and Facebook also keep big data, with some companies granting researchers access to subsets of data, such as iScience Maps™ for Twitter (Reips & Garaizar, 2013).

Social media sites generate large bursts of data of current relevance about a significant, but inexhaustive number of users. Screen-scraping refers to the extraction of information from internet sites, and data is collected and used for social purposes that range from gene sequencing to consumer behavior, and from learning analytics to predictive analytics (Bail, 2014; Siegel, 2013). The spread of mobile technologies has assisted the scope of these and other kinds of big data, with higher numbers of devices owned by family units and users throughout most parts of the world (Borgman, 2015).

Digital data are becoming computation intensive and data intensive, and its manipulation often requires significant logistical challenges (Meyer, 2009). The IoT is increasingly becoming an important source of big data, as sensory technologies become more consistently used to collect usage and environmental, geographical, logical, and astronomical data. Mobile devices, transportation facilities, public facilities, and home appliances are becoming data acquisition technologies that are connected to the IoT. Currently, internal data owned by enterprises are the main sources of big data (Chen et al., 2014).

Impact of big data on research fields

So which research fields have taken up big data? The research potentials of big data have been explored in a growing number of fields that include political science (Clarke & Margettes, 2014), global league tables in education research (Crossley, 2014), learning analytics (Rockwell & Berendt, 2016), immigration control and border security (Ajana, 2015), business scholarship (Frizzo-Barker, Chow-White, Mozafari, & Ha, 2016), and civil strife management (Nardulli, Althaus, & Hayes, 2015). Stock market shifts are traced in communication scholarship (Bollen, Mao, & Zeng, 2011), while patterns in children’s media cultures have also been observed (Montgomery, 2015).

Big data has been used for forensic social science in sociology (McFarland, Lewis, & Goldberg, 2016), applications in human geography (Kitchin, 2013), disaster response and recovery (Ragini, Anand, & Bhaskar, 2018), monitoring disease trends in public health (Paul & Dredze, 2011), and e-cometrics (O’Brien, Sampson, & Winship, 2015), to name a few. New research fields, such as digital humanities, have burgeoned in the big data era (Bail, 2014). Fields such as astronomy, genomics, physics, macroeconomics, and digital humanities tend to work with very large volumes of data, while a large number of scholars in some fields conduct research with minimal amounts of data (Borgman et al., 2016; Sawyer, 2008). This is by no means an exhaustive list, as newer forms of big data analysis, such as text, video, image, and learning analytics, are emerging globally.

Big data and digital life

As a consequence of the internet and the associated mobile technologies, big data is networked, connected, and traceable, but more difficult to analyze with conventional statistical analysis software (Snijders, Matzat, & Reips, 2012). Big data researchers aim to harness the potentials of computational capability and algorithmic accuracy to mine, examine, engineer, and employ extensive digital data sets to discover new findings about phenomena. In some quarters, the use of big data is undergirded by the epistemic assumption that big is better, offering increased sophistication, power, and forms of intelligence— a claim that has been challenged by qualitative researchers and other existing literature (see, e.g., boyd & Crawford, 2011).

While the world appears to be overflowing with big data, or the so-called “data deluge” highlighted by the Economist in the article “Data Everywhere” (Cukier, 2010), what is significant is the assumption that big data-driven knowledge is changing the way in which knowledge is generated, analyzed, shared, or governed (Chandler, 2015). In areas such as international relations, discussions have centered on the possibilities for solving problems of international scale through applications of knowledge generated by access to big data. These include prevention and timely responses to natural disasters, global conflict, climate change, disease management, and other societal problems that transcend national boundaries in globalized societies. For example, research published in Nature used the frequency of Google search engine queries to track outbreaks of influenza (Ginsberg et al., 2009). More recently, researchers have argued that along with epistemological and ontological assumptions of big data, we need to engage with more fundamental concerns regarding privacy and surveillance, data acc...

Table of contents

• Cover
• Half Title
• Title Page
• Copyright Page
• Dedication Page
• Table of Contents
• Foreword
• Acknowledgments
• Introduction
• 1 What is big data?
• 2 Big data in historical context
• 3 Challenges of big data for qualitative researchers
• 4 Potentials of big data analytics for qualitative researchers
• 5 Big data ethics, privacy, and dataveillance
• 6 Anticipating big data futures for qualitative researchers
• References
• Index