Johns hopkins nursing evidence based practice research evidence appraisal tool

Singapore Management University Institutional Knowledge at Singapore Management University Research Collection Lee Kong Chian School Of Business

Lee Kong Chian School of Business

10-2016

Big data and data science methods for management research: From the Editors Gerard GEORGE Singapore Management University, ggeorge@smu.edu.sg

Ernst C. OSINGA Singapore Management University, ecosinga@smu.edu.sg

Dovev LAVIE Technion

Brent A. SCOTT Michigan State University DOI: https://doi.org/10.5465/amj.2016.4005

Follow this and additional works at: https://ink.library.smu.edu.sg/lkcsb_research

Part of the Management Sciences and Quantitative Methods Commons, and the Strategic Management Policy Commons

This Editorial is brought to you for free and open access by the Lee Kong Chian School of Business at Institutional Knowledge at Singapore Management University. It has been accepted for inclusion in Research Collection Lee Kong Chian School Of Business by an authorized administrator of Institutional Knowledge at Singapore Management University. For more information, please email libIR@smu.edu.sg.

Citation GEORGE, Gerard; Ernst C. OSINGA; LAVIE, Dovev; and SCOTT, Brent A.. Big data and data science methods for management research: From the Editors. (2016). Academy of Management Journal. 59, (5), 1493-1507. Research Collection Lee Kong Chian School Of Business. Available at: https://ink.library.smu.edu.sg/lkcsb_research/4964

https://ink.library.smu.edu.sg?utm_source=ink.library.smu.edu.sg%2Flkcsb_research%2F4964&utm_medium=PDF&utm_campaign=PDFCoverPages
https://ink.library.smu.edu.sg/lkcsb_research?utm_source=ink.library.smu.edu.sg%2Flkcsb_research%2F4964&utm_medium=PDF&utm_campaign=PDFCoverPages
https://ink.library.smu.edu.sg/lkcsb_research?utm_source=ink.library.smu.edu.sg%2Flkcsb_research%2F4964&utm_medium=PDF&utm_campaign=PDFCoverPages
https://ink.library.smu.edu.sg/lkcsb?utm_source=ink.library.smu.edu.sg%2Flkcsb_research%2F4964&utm_medium=PDF&utm_campaign=PDFCoverPages
https://doi.org/10.5465/amj.2016.4005
https://ink.library.smu.edu.sg/lkcsb_research?utm_source=ink.library.smu.edu.sg%2Flkcsb_research%2F4964&utm_medium=PDF&utm_campaign=PDFCoverPages
http://network.bepress.com/hgg/discipline/637?utm_source=ink.library.smu.edu.sg%2Flkcsb_research%2F4964&utm_medium=PDF&utm_campaign=PDFCoverPages
http://network.bepress.com/hgg/discipline/642?utm_source=ink.library.smu.edu.sg%2Flkcsb_research%2F4964&utm_medium=PDF&utm_campaign=PDFCoverPages
http://network.bepress.com/hgg/discipline/642?utm_source=ink.library.smu.edu.sg%2Flkcsb_research%2F4964&utm_medium=PDF&utm_campaign=PDFCoverPages
mailto:libIR@smu.edu.sg
1

FROM THE EDITORS

BIG DATA AND DATA SCIENCE METHODS FOR MANAGEMENT RESEARCH

Published in Academy of Management Journal, October 2016, 59 (5), pp. 1493-1507.

http://doi.org/10.5465/amj.2016.4005

The recent advent of remote sensing, mobile technologies, novel transaction systems, and

high performance computing offers opportunities to understand trends, behaviors, and actions in

a manner that has not been previously possible. Researchers can thus leverage 'big data' that are

generated from a plurality of sources including mobile transactions, wearable technologies,

social media, ambient networks, and business transactions. An earlier AMJ editorial explored the

potential implications for data science in management research and highlighted questions for

management scholarship, and the attendant challenges of data sharing and privacy (George, Haas

& Pentland, 2014). This nascent field is evolving rapidly and at a speed that leaves scholars and

practitioners alike attempting to make sense of the emergent opportunities that big data holds.

With the promise of big data come questions about the analytical value and thus relevance of this

data for theory development -- including concerns over the context-specific relevance, its

reliability and its validity.

To address this challenge, data science is emerging as an interdisciplinary field that

combines statistics, data mining, machine learning, and analytics to understand and explain how

we can generate analytical insights and prediction models from structured and unstructured big

data. Data science emphasizes the systematic study of the organization, properties, and analysis

of data and its role in inference, including our confidence in the inference (Dhar, 2013). Whereas

both big data and data science terms are often used interchangeably, big data is about collecting

and managing large, varied data while data science develops models that capture, visualize, and

http://doi.org/10.5465/amj.2016.4005
2

analyze the underlying patterns to develop business applications. In this editorial, we address

both the collection and handling of big data and the analytical tools provided by data science for

management scholars.

At the current time, practitioners suggest that data science applications tackle the three

core elements of big data: volume, velocity, and variety (McAfee & Brynjolfsson, 2012;

Zikopoulos & Eaton, 2011). Volume represents the sheer size of the dataset due to the

aggregation of a large number of variables and an even larger set of observations for each

variable. Velocity reflects the speed at which these data are collected and analyzed, whether real-

time or near real-time from sensors, sales transactions, social media posts and sentiment data for

breaking news and social trends. Variety in big data comes from the plurality of structured and

unstructured data sources such as text, videos, networks, and graphics among others. The

combinations of volume, velocity and variety reveal the complex task of generating knowledge

from big data, which often runs into millions of observations, and deriving theoretical

contributions from such data. In this editorial, we provide a primer or a “starter kit” for potential

data science applications in management research. We do so with a caveat that emerging fields

outdate and improve upon methodologies while often supplanting them with new applications.

Nevertheless, this primer can guide management scholars who wish to use data science

techniques to reach better answers to existing questions or explore completely new research

questions.

BIG DATA, DATA SCIENCE, AND MANAGEMENT THEORY

Big data and data science have potential as new tools for developing management theory,

but given the differences from existing data collection and analytical techniques to which

3

scholars are socialized in doctoral training it will take more effort and skill in adapting new

practices. The current model of management research is post hoc analysis, wherein scholars

analyze data collected after the temporal occurrence of the event – a manuscript is drafted

months or years after the original data are collected. Therefore, velocity or the real-time

applications important for management practice is not a critical concern for management

scholars in the current research paradigm. However, data volume and data variety hold potential

for scholarly research. Particularly, these two elements of data science can be transposed as data

scope and data granularity for management research.

Data Scope. Building on the notion of volume, data scope refers to the

comprehensiveness of data by which a phenomenon can be examined. Scope could imply a wide

range of variables, holistic populations rather than sampling, or numerous observations on each

participant. By increasing the number of observations, a higher data scope can shift the analysis

from samples to populations. Thus, instead of focusing on sample selection and biases,

researchers could potentially collect data on the complete population. Within organizations,

many employers collect data on their employees, and more data is being digitized and made

accessible. This includes email communication, office entry and exit, RFID tagging, wearable

sociometric sensors, web browsers, and phone calls, which enable researchers to tap into large

databases on employee behavior on a continuous basis. Researchers have begun to examine the

utility and psychometric properties of such data collection methods, which is critical if they are

to be incorporated into and tied to existing theories and literatures. For example, Chaffin et al. (in

press) examined the feasibility of using wearable sociometric sensors, which use a Bluetooth

sensor to measure physical proximity, an infrared detector to assess face-to-face positioning

between actors, and a microphone to capture verbal activity, to detect structure within a social

4

network. As another example, researchers have begun to analyze large samples of language (e.g.,

individuals’ posts on social media) as a non-obtrusive way to assess personality (Park et al.,

2015). With changes in workplace design, communication patterns, and performance feedback

mechanisms, we have called for research on how businesses are harnessing technologies and data

to shape employee experience and talent management systems (Colbert, Yee & George, 2016;

Gruber, Leon, Thompson & George, 2015).

Data Granularity. Following the notion of variety, we refer to data granularity as the

most theoretically proximal measurement of a phenomenon or unit of analysis. Granularity

implies direct measurement of constituent characteristics of a construct rather than distal

inferences from data. For example, in a study of employee stress, granular data would include

emotions through facial recognition patterns or biometrics such as elevated heart rates during

every minute on the job or task rather than surveys or respondent interviews. In experience-

sampling studies on well-being, for example, researchers have begun to incorporate portable

blood pressure monitors. For instance, in a 3-week experience-sampling study, Bono, Glomb,

Shen, Kim, and Koch (2013) had employees wear ambulatory blood pressure monitors that

recorded measurements every 30 minutes for two hours in the morning, afternoon, and evening.

Similarly, Ilies, Dimotakis, and DePater (2010) used blood pressure monitors in a field setting to

record employees’ blood pressure at the end of each workday over a two-week period. Haas,

Criscuolo and George (2015) studied message posts and derived meaning in words to predict

whether individuals are likely to contribute to problem solving and knowledge sharing across

organizational units. Researchers in other areas could also increase granularity in other ways. In

network analysis for instance, researchers can monitor communication patterns across employees

instead of asking employees with whom they interact or seek advice from retrospectively.

5

Equivalent data were earlier collected using surveys and indirect observation, but with big data

the unit of analysis shifts from individual employees to messages and physical interactions.

Though such efforts are already seen in smaller samples of emails or messages posted on a

network (e.g., Haas, Criscuolo & George, 2015), organization-wide efforts are likely to provide

clearer and holistic representations of networks, communications, friendships, advice-giving and

taking, and information flows (van Knippenberg, Dahlander, Haas & George, 2015).

Better Answers and New Questions

Together, data scope and data granularity allow management scholars to develop new

questions and new theories, and to potentially generate better answers to established questions.

In Figure 1, we portray a stylistic model of how data scope and data granularity could

productively inform management research.

6

Better Answers to Existing Questions. Data science techniques enable researchers to get more

immediate and accurate results for testing existing theories. In doing so, we hope to get more

accurate estimations of effect sizes and their contingencies. Over the past decade, management

theories have begun emphasizing effect sizes. This emphasis on precision is typically observed in

strategy research rather than in behavioral studies. With data science techniques, the precision of

effect sizes and their confidence intervals will likely be higher and can reveal nuances in

moderating effects that have hitherto not been possible to identify or estimate effectively.

Better answers could also come from establishing clearer causal mechanisms. For

instance, network studies rely on surveys of informants to assess friendship and advice ties, but

in these studies, the temporal dimension is missing, and therefore it is difficult to determine

whether network structure drives behavior or vice versa. Instead, collecting email

communications or other forms of exchange on a continuous level would enable researchers to

measure networks and behavior dynamically, and thus assess more systematically cause and

effect.

Although rare event modeling is uncommon in management research, data science

techniques could potentially shed more light on, for example, organizational responses to

disasters, modeling and estimating probability of failure, at risk behavior, and systemic resilience

(van der Vegt, Essens & Wahlstrom, 2015). Research on rare events can use motor car accident

data, for instance, to analyze the role of driver experience in seconds leading up to an accident

and how previous behaviors could be modeled to predict reaction times and responses. Insurance

companies now routinely use such data to price insurance coverage, but this type of data could

also be useful for modeling individual-level risk propensity, aggressiveness, or even avoidance

behaviors. At an aggregate level, data science approaches such as collecting driver behavior

7

using sensors to gauge actions like speeding and sudden stopping, allow more than observing

accidents, and therefore generate a better understanding of their occurrence. Such data allows

cities to plan traffic flows, map road rage or accident hot spots, and avert congestion, and

researchers to connect such data to timeliness at work, and negative or positive effects of

commuting sentiment on workplace behaviors.

Additionally, data science techniques such as monitoring call center calls can enable

researchers to identify specific triggers to certain behaviors as opposed to simply measuring

those behaviors. This can help better understand phenomena such as employee attrition. Studying

misbehavior is problematic due to sensitivity, privacy and availability of data. Yet, banks are

now introducing tighter behavioral monitoring and compliance systems that are tracked in real-

time to predict and deter misbehaviors. Scholars already examine lawsuits, fraud, and collusion,

but by using data science techniques, they can search electronic communication or press data

using keywords that characterize misbehavior in order to identify the likelihood of misbehavior

before its occurrence. As these techniques become prevalent, it will be important to tie the new

measures, and the constructs they purportedly assess, to existing theories and knowledge bases;

otherwise, we risk the emergence of separate literatures using “big” and “little” data that have the

capacity to inform each other.

New Questions. With higher scope and granularity of data, it becomes possible to explore

new questions that have not been examined in the past. This could arise because data science

allows us to introduce new constructs, but it could also arise because data science allows us to

operationalize existing constructs in a novel way. Web scraping and sentiment data from social

media posts are now being seen in the management literature, but they have yet to push scholars

to ask new questions. Granular data with high scope could open questions in new areas of

8

mobility and communications, physical space, and collaboration patterns where we could delve

deeper into causal mechanisms underlying collaboration and team dynamics, decision-making

and the physical environment, workplace design and virtual collaborations. Tracking phone

usage and physical proximity cues could provide insight into whether individuals spend too

much time on communications technology and attention allocation to social situations at work or

at home. Studies suggest that time spent on email increases anger and conflict at work and at

home (Butts, Becker & Boswell, 2015). But such work could then be extended to physical and

social contingencies, nature of work, work performance outcomes, and their quality of life

implications.

Data on customer purchase decisions and social feedback mechanisms can be

complemented with digital payments and transaction data to delve deeper into innovation and

product adoption as well as behavioral dynamics of specific customer segments. The United

Nations’ Global Pulse is harnessing data science for humanitarian action. Digital money and

transactions through mobile platforms provide a window into social and financial inclusion, such

as access to credit, energy and water purchase through phone credits, transfer of money for

goods and services, create spending profiles, identify indebtedness or wealth accumulation, and

promote entrepreneurship (Dodgson et al., 2015). Data science applications allow the delivery

and coordination of public services such as treatment for disease outbreaks, coordination across

grassroots agencies for emergency management, and provision of fundamental services such as

energy and transport. Data on carbon emissions and mobility can be superimposed for tackling

issues of climate change and optimizing transport services or traffic management systems. Such

technological advances that promote social wellbeing can also raise new questions for scholars in

9

identifying ways of organizing and ask fundamentally new questions on organization design,

social inclusion, and the delivery of services to disenfranchised communities.

New questions could emerge from existing theories. For example, once researchers can

observe and analyze email communication or online search data, they can ask questions

concerning the processes by which executives make decisions as opposed to studying the

individual/TMT characteristics that affect managerial decisions. There is room for using

unstructured data such as video and graphic data, and face recognition for emotions. Together,

these data could expand conversations beyond roles, experience, and homogeneity to political

coalitions, public or corporate sentiment, decision dynamics, message framing, issue selling,

negotiations, persuasion, and decision outcomes.

Text mining can be used when seeking to answer questions such as where do ideas or

innovations come from -- as opposed to testing whether certain conditions generate ground-

breaking innovations. This requires data mining of patent citations that can track the sources of

knowledge embedded in a given patent and its relationships with the entire population of patents.

In addition, analytics allow inference of meaning, rather than word co-occurrence, which could

be helpful in understanding cumulativeness, evolution and emergence of ideas and knowledge.

A new repertoire of capabilities is required for scholars to explore these questions and to

handle challenges posed by data scope and granularity. Data are now more easily available from

corporates and “Open Data” warehouses such as the London DataStore. These data initiatives

encourage citizens to access platforms and develop solutions using big data on public services,

mobility and geophysical mapping among others data sources. Hence, as new data sources and

analytics become available to researchers, the field of management can evolve by raising

questions that have not received attention as a result of data acce