ArticlePDF Available

Abstract and Figures

Firms utilizing Big Data combine information from diverse sources to create knowledge, make better predictions, and tailor services. Yet Big Data has also been criticized as a breach of privacy, as potentially discriminatory, and as just creepy. Part of the ambiguity in examining Big Data is choosing what to analyze: Big Data has been framed as an ability, a process, a tactic, and an asset. Here, I analyze Big Data as the Big Data Industry (BDI) to examine the systemic risks in current business practices. Such an approach situates Big Data within a larger system of firms, organizations, processes, and norms for the analysis of the information supply chains and industry standards. The Big Data industry is at a critical tipping point where industry leaders – with intimate knowledge of the systemic risks as well as the power from a unique position – could enact meaningful change. Remedies for these potential issues are examined with the goal of fostering a sustainable Big Data industry.
Content may be subject to copyright.
June 2015 (14:2) | MIS Quarterly Executive 67
The Big Data Industry1 2
Big Data receives a lot of press and attention—and rightly so. Big Data, the combination of
greater size and complexity of data with advanced analytics,3 has been effective in improving
national security, making marketing more effective, reducing credit risk, improving medical
research and facilitating urban planning. In leveraging easily observable characteristics and
events, Big Data combines information from diverse sources in new ways to create knowledge,
make better predictions or tailor services. Governments serve their citizens better, hospitals
more criminals and nations are safer.
Yet Big Data (also known in academic circles as “data analytics”) has also been criticized as a
breach of privacy, as potentially discriminatory, as distorting the power relationship and as just
“creepy.4 In generating large, complex data sets and using new predictions and generalizations,
               
needed, ignored citizens when repairing streets, informed friends and family that someone
is pregnant or engaged, and charged consumers more based on their computer type. Table 1
             
1 Dorothy Leidner is the accepting senior editor for this article.
2 This work has been funded by National Science Foundation Grant #1311823 supporting a three-year study of privacy online. I
wish to thank the participants at the American Statistical Association annual meeting (2014), American Association of Public Opin-
ion Researchers (2014) and the Philosophy of Management conference (2014), as well as Mary Culnan, Chris Hoofnagle and Katie
Shilton for their thoughtful comments on an earlier version of this article.
3 Both the size of the data set, due to the volume, variety and velocity of the data, as well as the advanced analytics, combine to
create Big Data. Key to denitions of Big Data are that the amount of data and the software used to analyze it have changed and
combine to support new insights and new uses. See also Ohm, P. “Fourth Amendment in a World without Privacy,” Mississippi.
Law Journal (81), 2011, pp. 1309-1356; Boyd, D. and Crawford, K. “Critical Questions for Big Data: Provocations for a Cultural,
Technological, and Scholarly Phenomenon,” Information, Communication & Society (15:5), 2012, pp. 662-679; Rubinstein, I. S.
“Big Data: The End of Privacy or a New Beginning?,” International Data Privacy Law (3:2), 2012, pp. 74-87; and Hartzog, W. and
Selinger, E. “Big Data in Small Hands,” Stanford Law Review Online (66), 2013, pp. 81-87.
4 Ur, B. et al. “Smart, Useful, Scary, Creepy: Perceptions of Online Behavioral Advertising,” presented at the Symposium On
Usable Privacy and Security, July 11-13, 2012, Washington, D.C. See also Barocas, S. and Selbst, A. D. “Big Data’s Disparate
Impact,” 2015, draft available at SSRN 2477899; and Richards, N. M. and King, J. H. “Three Paradoxes of Big Data,” Stanford Law
Review Online (66), 2013 pp. 41-46.
Ethical Issues in the Big Data Industry
Big Data combines information from diverse sources to create knowledge, make better
predictions and tailor services. This article analyzes Big Data as an industry, not a
consumers’ data to the secondary market for Big Data. Remedies for the issues are
proposed, with the goal of fostering a sustainable Big Data Industry.1,2
Kirsten E. Martin
George Washington University
68 MIS Quarterly Executive | June 2015 (14:2) | © 2015 University of Minnesota
Ethical Issues in the Big Data Industry
Table 1: Examples of Benecial and Questionable Uses of Big Data
Benecial Uses Quesonable Uses
By Technology
License Plate
Reading passing cars for tolls on highway;
police locang stolen car
Used by private detecves; placed on
trucks to gather license plate data broadly
Finding potenal terrorists at large
sporng events
Used by social networking sites to idenfy
members in pictures
GPS Locaon-based coupons; trac
predicons; direcons on map
Locaon-based stalking; iPhone as a
homing beacon
By Context
Healthcare Treatment of cancer; health of pregnancy;
Google Flu Trends
Insights into interacon between medi-
caons from search terms; insights into
hospital spread of infecons
Idenfying veterans’ potenal suicidal
Discriminaon in healthcare and
insurance; app knows how t you are
Development of a health score from pur-
chase habits and from search terms
Educaon Personalizing student instrucon
Accountability for performance by school
Idenfying students at risk of dropping out
Using data for possible admissions
Electricity Turning on/o home electricity Allowing criminals to know if you are
home; smart homes hacked
Machine learning to idenfy burglar;
accessing phone records to idenfy
potenal suspects in a mugging
New York Fire Department using data min-
ing to predict problems
Accessing smartphone without a warrant;
idenfying suspects by web browsing
Individuals under scruny for not parci-
pang in tracking
Retail Improving layout of store based on typical
movements of customers
Beer coupons, suggested items
WalMart’s use of RetailLink to integrate
suppliers with onsite supplier inventory
Tracking movements/shopping habits of
spectators at a stadium using Verizon’s
Precision Markeng Insight program
Price discriminaon (e.g., Amazon, Orbitz)
Target sending noce of pregnancy to
unsuspecng teen’s parents
Trac management; smart grid technology
Use of popular app by compeve cyclists
and runners for road planning
Idenfying areas for road improvement
Idenfying who is listening to which
radio staon; EZ Pass responder tracked
Possibility of hackers changing trac
lights and creang trac jams
Idenfying areas for road improvement
but focusing only on those with mobile
June 2015 (14:2) | MIS Quarterly Executive 69
Ethical Issues in the Big Data Industry
        
community—if at all.
Part of the ambiguity in researching Big Data is
choosing what to study. Big Data has been framed
as: (1) the ability to process huge “treasure
troves” of data and predict future outcomes, (2)
a process that “leverages massive data sets and
algorithmic analysis” to extract new information
and meaning, (3) an asset, (4) a moment where
use of traditional tools and (5) a tactic to operate
at a large scale not possible at a smaller scale.5
Framing Big Data as an asset, ability or
    
discussion. Big Data is mistakenly framed as
     
any costs. Grand statements such as “Big Data
itself, like all technology, is ethically neutral”6 are
implicit in reports that focus on the strategic and
operational challenges of Big Data, but which
largely ignore the ethical and social implications.7
      
ethical analysis in both practice and academia. Yet
creating, aggregating and selling data can change
     
rethinking information governance strategies—
including issues concerning ethics and privacy.8
I suggest Big Data should be analyzed as the
Big Data Industry (BDI) in order to identify the
5 In order of reference: Barocas, S. and Selbst, A. D., op. cit.,
2014; Hartzog, W. and Selinger, E., op. cit., 2013; Big Data
Management & Analytics, Gartner, 2014, available at http://www.; Mayer-Schönberger, V.
and Cukier, K. Big Data: A Revolution That Will Transform How We
Live, Work, and Think, Houghton Mifin Harcourt, 2013; Richards,
N. M. and King, J. H., op. cit., 2013.
6 Wen, H. “Big ethics for big data,” O’Reilly Radar, June 11, 2012,
available at
7 For example, Gartner notes that there are three strategic and
operational challenges: information strategy, data analytics and
enterprise information management, but makes no mention of
ethical challenges. See also Big Data and Privacy: A Technological
Perspective, Report to the President, 2014, available at https://www.les/microsites/ostp/PCAST/pcast_big_
data_and_privacy_-_may_2014.pdf; Big Data Platform - Bringing
Big Data to the Enterprise, IBM ; and Manyika, J. et al. Big Data:
The next Frontier for Innovation, Competition, and Productivity,
McKinsey & Company, 2011, available at
8 The shift to creating value through monetizing data impacts
relationships with stakeholders as well as policies internal to the
organization—see Tallon, P. P., Short, J. E and Harkins, M. W. “The
Evolution of Information Governance at Intel,” MIS Quarterly
Executive (12:4), 2013, pp. 189-198; Najjar, M. S. and Kettinger,
W. J., “Data Monetization: Lessons from a Retailer’s Journey,” MIS
Quarterly Executive (12:4), 2013, pp. 213-225.
systemic risks in current Big Data practices. Such
an approach situates Big Data within a larger
     
norms for analysis. The volume, variety and
velocity9 of the data, plus the novel analytics
    
renders Big Data a difference in kind rather than
degree. To create and use these large data sets to
      
create a new “whole” and sell access to this new
data set.
Industry work through agreements to produce
a product (Big Data) for customers—similar to
any other industry.10 In response, CIOs and CDOs
       
strategic focus in leveraging Big Data rather than
the inward, service focus used for traditional
data. At present, however, there are not yet any
industry norms or supply chain best practices
that can guide them.11
This article examines the ethical issues in the
nascent Big Data Industry. Industries are the
      
and distribution of a product—e.g., the software
industry, the ERP industry, the automobile
industry, etc. Importantly, if a market exists for
a product, then a corresponding industry exists
to meet that demand. And, as the market for Big
Data continues to grow and be measured, the
corresponding Big Data Industry, comprised of
      
and use of Big Data, begins to coalesce around
standard industry practices. (Note that this article
focuses on privacy issues in the U.S. Big Data
Industry; as described in the panel below, the
privacy regulatory environments in the U.S. and
9 The 3Vs of Big Data—volume, variety and velocity—were
originally dened in a META/Gartner report but have subsequently
been expanded with veracity, value, validity, variability and even
visualization, leading to the term “V confusion”—see Grimes, S.
“Big Data: Avoid ‘Wanna V’ Confusion,” InformationWeek, August
7, 2013, available at
10 Firms monetizing the value of data require new tactics and
strategies as well as, perhaps, accounting rules to capture the value
(and risk) created in new transactions. See Monga, V. “The Big
Mystery: What’s Big Data Really Worth?,” Wall Street Journal,
October 13, 2014, available at
11 Lee, Y. et al., “A Cubic Framework for the Chief Data Ofcer:
Succeeding in a World of Big Data,” MIS Quarterly Executive (13:1),
2014, pp. 1-13.
70 MIS Quarterly Executive | June 2015 (14:2) | © 2015 University of Minnesota
Ethical Issues in the Big Data Industry
     
supply chain within the Big Data Industry,
including upstream sources of data and
downstream uses of data. Next, it examines
two crucial consumer-related ethical issues
created by systemic norms and practices of the
Big Data Industry: (1) the negative externality
of surveillance and (2) destructive demand.
Remedies for these potential issues are proposed,
with the goal of fostering a sustainable Big Data
Industry. 1213
An industry-level analysis extends the
examination of Big Data in three ways. First,
12 See also Lomas, N. “Facebook’s Data Protection Practices
Under Fresh Fire In Europe,” TechCrunch, available at http://social.
13 Scott, M. “Where Tech Giants Protect Privacy,” The New
York Times, December 13, 2014, available at http://www.nytimes.
framing Big Data as an industry highlights the
participants, power relationships and systemic
issues that arise within the production and
use of Big Data, insights that are not available
when Big Data is isolated as a technology.
Second, an industry-level analysis captures
pervasive industry practices that are missed
when considering single uses of Big Data. These
systemic issues can be resolved with the industry-
     
Finally, an industry-level analysis broadens the
number of interested parties to all who have a
stake in creating a sustainable Big Data Industry.
All companies in controversial industries have
      
interest in creating sustainable industry norms.
In other words, the recognition that bad behavior
may delegitimize the entire industry provides
an incentive for industry leaders to curb such
practices.14       
in the Big Data Industry is given in the left panel
on the next page.
The Big Data Industry’s
Supply Chain
Within the Big Data Industry, data, such as
online consumer data or location data from an
        
within an information supply chain, comparable
to supply chains in traditional industries (see text
panel on the next page). Within this supply chain,
     
then pass it to tracking companies, which may
also pass it to data aggregators. Data aggregators
14 The BDI requires not only information brokers to aggregate
data, but also hardware, software and professional services rms to
support the collection, storage and use of the data. Leaders include
rms focused on analytics solutions (e.g., SAS, IBM, SAP) as well
as industry specialists (e.g., Amazon Web Services) and service
providers (Accenture). For more information, see Robb, D. “Top 20
Big Data Companies,” Datamation, November 20, 2014, available
nies-1.html. Importantly, many rms combine products and services
that support the BDI—e.g., IBM (hardware, software and services),
HP (cloud and storage), Dell (storage), SAP (analytics), Teradata and
Oracle (hardware, software, services), SAS and Palantir (analytics
and software) and Accenture (software and services). See Leopold,
G. “Big Data Rankings: Leaders Generated $6B in Revenues,”
Datanami, December 4, 2014, available at http://www.datanami.
While hardware, software, analytics and even technology consulting
rms make most of the industry leader lists, missing are the data bro-
kers and data aggregators that make up the information supply chain
discussed in the next section.
Privacy: U.S. Versus EU
The use of Big Data in Europe faces a disnct
set of regulatory constraints governed by the
EU’s Data Protecon Direcve (95/46/EC) and,
for example, the United Kingdom’s Data Protec-
on Act 1998. Regulaons require those using
“personal data” to abide by the direcve’s re-
quirements to being fair, to be clear as to the
purpose of gathered informaon and, problem-
ac for Big Data, to strive for minimizaon. See
also the Bureau of Naonal Aairs’ World Data
protecon Report 14(9) as well as the U.K.’s In-
formaon Commissioner’s Oce Big Data and
Data Protecon (2014).
For example, Facebook recently was unable
to comply with the stricter EU regulaons be-
cause of a lack of adequate consent and control
for users: Facebook users have no true opt-out
mechanism, no valid consent for the transfer of
data to third pares and a general lack of con-
trol over their data. In other words, Facebook’s
“take it or leave it” approach to choice is not
sucient for European law.12 Generally, privacy
is taken more seriously by regulators in the EU
(and by U.S. companies doing business in Eu-
rope), with “data subjects” having a right to be
forgoen, authenc user consent and a general
leaning toward “opt-in” as the default.13
June 2015 (14:2) | MIS Quarterly Executive 71
Ethical Issues in the Big Data Industry
act as distributors by holding consolidated
information of many users across many contexts.
Data aggregators or data brokers may sell the
information to researchers, government agencies
or polling companies, or an ad network may use
the information from an aggregator or broker
to place an advertisement on a website when a
user returns to browse or shop online. Survey
    
      
a data broker directly to use data to supplement
survey research, make employment decisions
and investigate possible criminal activity. An
information supply chain is thus created with
    
adding value to the data.
As with traditional supply chains, the
information supply chain can be analyzed both by
the downstream distribution and use of Big Data
as well as by the upstream sourcing (see Figure
The issues arising from the downstream use
of Big Data and upstream sourcing of information
are summarized in Figure 2 and described in
detail below.
Issues with Downstream Customers
and Uses of Big Data
As shown in Table 1, downstream uses
of Big Data can be perceived as producing
    
and harmful) outcomes. However, the potential
harm that can result from using Big Data should
     
diseases to identifying fraud. Nonetheless, selling
information increases the risk of secondary
misuse of the data, with eventual harmful impacts
on users. While the potential harm from incorrect
information or false conclusions merits attention,
harm downstream in the supply chain includes
harm from the correct conclusions. For instance,
     
teenager based on her purchase history and sent
a congratulatory letter to her house, which was
seen by her parents who were unaware that their
daughter was pregnant.15
The harmful effects of using Big Data can be
extended to include:
Value destruction (rather than creation) for
Diminished rights (rather than realized) for
Disrespectful to someone involved in the
process (rather than supporting them).
Such effects are not possible without
information provided upstream, thereby linking
15 Duhigg, C. “How Companies Learn Your Secrets,” The New
York Times, February 16, 2012, available at http://www.nytimes.
Supply Chains
In a tradional business model, supply chains
comprise a series of rms working together to
deliver value by transforming raw material into
a nished product. Trees are harvested in the
forest, traded to the pulp manufacturer and
eventually become the paper used to print an
arcle; tomatoes are picked, packed, shipped
and crushed into sauce to be used on a deliv-
ered pizza. The gure below illustrates a generic
supply chain: each rm adds value to the prod-
uct or service to transform the raw materials in
one locaon and deliver a nished product to
the end customer through value creaon and
All supply chains carry ethical issues both down-
stream and upstream. Soware companies
must ensure that their products are not eventu-
ally sold in Syria through a distribuon center in
Dubai; Apple is held accountable for the work-
ing condions of its upstream suppliers, such as
Foxconn. Supply chain researchers examine up-
stream sourcing issues, looking at how supplier
selecon takes account of, for example, the way
forests are harvested in the paper industry or
how apparel is manufactured overseas, as well
as following products downstream through lo-
giscs and eventual sale and use.
72 MIS Quarterly Executive | June 2015 (14:2) | © 2015 University of Minnesota
Ethical Issues in the Big Data Industry
Figure 1: Example of Information Supply Chain Within the Big Data Industry
Figure 2: Issues within the BDI Supply Chain
June 2015 (14:2) | MIS Quarterly Executive 73
Ethical Issues in the Big Data Industry
all supply chain members to the eventual uses of
First, data uses can be analyzed based on the
consequences to the individual. More obvious
    
credit, losing a job, having secrets outed to
your family, paying more for insurance, etc. For
example, information may be used downstream
to modify insurance premiums or mortgage rates.
      
search results for a travel site.17 Table 1 focuses on
       
use of Big Data.
      
what law scholar Ryan Calo conceptualizes as
     
more information about consumers with an
      
   
at a personal level and to trigger vulnerability in
consumers in their marketing.18 Calo’s argument
suggests that Target, for example, would not
only identify a consumer who is pregnant, but
could also engineer food cravings in her through
increasingly be in the position to create ‘suckers’
rather than waiting for one to be born every
The harm resulting from the use of Big Data
        
16 As stated by Bambauer, “estimating harm is a wearisome
task”—see Bambauer, J. “Other People’s Papers,” draft paper,
2014, p. 15, available atles/
Bambauer_Other_Peoples_Papers_GMU.pdf. Bambauer categorizes
privacy harms as arising from collection, risk of misuse, aggregation,
obstruction and hassle; Richards lists sorting, discrimination,
persuasion and blackmail as potential harms—Richards, N. M. “The
Dangers of Surveillance,” Harvard Law Review, 2013, available at;
Calo focuses more broadly on objective and subjective harms—Calo,
M. R. “Boundaries of Privacy Harm,” Indiana Law Journal (86),
2011, pp. 1131-1162.
17 For example, car insurance companies are moving toward
usage-based premiums based on driving data collected in real
time—see Boulton, C. “Auto Insurers Bank on Big Data to Drive
New Business,” Wall Street Journal, February 20, 2013, available
data-to-drive-new-business/. Similarly, health insurance companies
can deny services and increase premiums through accessing data
online—see Gittelson, K. “How Big Data Is Changing Insurance,”
BBC News, November 15, 2013, available at
18 Calo, M. R. “Digital Market Manipulation,” The George
Washington Law Review (82:4), 2013, pp. 995-1051.
value is created or destroyed for individuals,
but also whether individuals’ rights are being
realized in the process of using the data. Barocos
and Selbst nicely illustrate the harm that can
arise not only from the information supply chain,
but also from the process followed in using Big
Data. Big Data may develop learned prejudice
algorithms based on pre-existing information.
By basing predictive algorithms on previous data
patterns, learned prejudice builds on previously
institutionalized prejudice—for example, in
areas such as college admissions or when a
Google search on black-sounding names brings
up arrest records. Such algorithms can produce
objectionable outcomes, as with accidental or
intentional discrimination.19
Finally, categorizing individuals under certain
headings can be disrespectful to them—for
example, the categorization of individuals based
on their personal history, such as rape victim
status, becomes an exercise in objectifying
individuals as a mere category. Big Data
aggregators have been known to list individuals
     
dysfunction sufferers and even as “daughter
killed in car crash.20 Even without value being
destroyed, individuals can be disrespected
through objectifying them as a mere category—
particularly a category that overwhelms in
        
struggling with an addiction or coping with a
Issues with Upstream Sources
In addition to the possible downstream
        
information supply chain must also contend with
issues concerned with upstream suppliers of
data, in particular the possibility of partnering
with bad suppliers. The ability to develop an
ever-greater volume, velocity and variety of data
19 For the concept of objectionable classication and biases,
see Barocas, S. and Selbst, A. D., op. cit., 2015; Sweeney, L.
“Discrimination in Online Ad Delivery,” acmqueue (11:3), 2013,
available at; and Cohen,
J. E. “What Privacy Is for,” Harvard Law Review (126), 2013, pp.
20 For examples of objectionable categorizations, see Hill, K.
“Data Broker Was Selling Lists Of Rape Victims, Alcoholics, and
‘Erectile Dysfunction Sufferers’,” Forbes, September 19, 2013,
available at
74 MIS Quarterly Executive | June 2015 (14:2) | © 2015 University of Minnesota
Ethical Issues in the Big Data Industry
   
from many sources. Sources of data within the
Big Data Industry include consumers, products,
location, machines and transactions (and all
combinations of these). In fact, the variety of
combined data differentiates Big Data from
traditional data analysis: many data sources
combine data types or use data in novel ways.
This pooling of diverse, sometimes innocuous,
pieces of data contributes to a greater potential
       
new knowledge.21
Within the Big Data Industry, upstream
sources may be undesirable because of the
       
privacy issues in the collection and sharing of
     
inaccuracies in the data or a lack of coverage.22
Inaccuracies may arise from the manner in which
the data was collected, the degree of imputed23
data within the data source or from deliberate
obfuscation by users.24    
       
of upstream sources in a manufacturing supply
     
      
using upstream information further down the
information supply chain will be held accountable
Data may also have biases that skew it toward
       
race, ethnicity, gender, socioeconomic status
or location. Using upstream data further down
      
the level of bias in the data—skewed data will
21 Groves has previously categorized data sources as organic vs.
designed—Groves, R. M. “Three Eras of Survey Research,” Public
Opinion Quarterly (75:5), 2011, pp. 861-871. Sources have also
been categorized as analog vs. digital in Big Data and Privacy: A
Technological Perspective, Report to the President, 2014. However,
the differences in these categories are not always clear or meaningful
in determining the appropriateness of the supplier.
22 For an analysis of quality and bias issues in Big Data
sources, see Boyd, D. and Crawford, K. Critical Questions for Big
Data: Provocations for a Cultural, Technological, and Scholarly
Phenomenon, Microsoft Research, 2012; Lerman, J. “Big Data and
Its Exclusions,” Stanford Law Review Online (66), 2013, pp. 55-
63; and Crawford, K. “The Hidden Biases in Big Data,” Harvard
Business Review, April 1, 2013.
23 Imputation is the process of replacing missing data with
substituted values.
24 The role of obfuscation in protecting privacy is examined
in Brunton, F. and Nissenbaum, H. “Vernacular resistance to data
collection and analysis: A political theory of obfuscation,” First
Monday (16:5), 2011.
skew the results and limit the generalization
      
       
for transit scheduling; however, if one group is
systematically ignored in the source data (e.g.,
groups with less access to mobile devices used to
       
from the improved transit system or may have
Finally, and importantly for the ethical
       
supplying data should be assessed on how it
respects privacy in the collection of information.
Consumers disclose information within a set of
privacy rules, and sharing that information with
   
privacy expectations. In other words, information
always has “terms of use” or norms governing
when, how, why and where it can be used.26 For
example, information shared with Orbitz, a travel
website, has a distinct set of privacy expectations
based on the individual’s relationship with
the website and the context of the interaction.
Individuals may expect location information to
be used to offer hotel or restaurant discounts
for their destination, but they do not expect that
information be passed to data aggregators and
used a year later to make pricing decisions. Users
disclose information with a purpose in mind and
Privacy law scholar Woodrow Hartzog
    
      
receive or gather the information within a
    27 The
expectations present at initial disclosure—
who should receive information, how it can be
used, how long it will be stored—should persist
throughout the online information supply chain.
25 O’Leary, D. E. “Exploiting Big Data from Mobile Device
Sensor-Based Apps: Challenges and Benets,” MIS Quarterly
Executive (12:4), 2013, pp. 179-187.
26 Nissenbaum, H. Privacy in Context: Technology, Policy, and the
Integrity of Social Life, Stanford University Press, 2009; Martin, K.
“Understanding Privacy Online: Development of a Social Contract
Approach to Privacy,” Journal of Business Ethics, 2015, pp. 1-19;
and Richards, N. M. and King, J. H. “Big Data Ethics,” Wake Forest
Law Review (23), 2014.
27 Hartzog, W. “Chain-Link Condentiality,” Georgia Law Review
(46), 2011, pp. 657-704.
June 2015 (14:2) | MIS Quarterly Executive 75
Ethical Issues in the Big Data Industry
Role of Firms in the Information
Supply Chain
In conventional supply chains, upstream
      
In the 1990s, for example, Wal-Mart and Nike
infamously relied on overseas manufacturers
that used child labor and unsafe working
conditions. More recently, Apple has grappled
with the reputational problems arising from
using Foxconn, a supplier with harsh working
conditions. Firms that willingly enter a supply
chain have an obligation to ensure that the
       
their own. Similarly, organizations within the
information supply chain are held responsible for
the data stewardship practices of both upstream
and downstream partners.
An organization’s responsibility within a
       
receives from the practices of the supply chain. In
up to the practices of the supply chain—including
    
       
suppliers even though the working conditions of
those suppliers leave a lot to be desired.
  
and willingly takes on part of the responsibility
for actions and practices within that chain. For
example, when Facebook seeks to use information
from upstream data brokers such as Acxiom,
Epsilon, Datalogix and BlueKai,28 it must not only
worry about its own collection methods, but also
the upstream sources’ data collection methods.
Choosing and creating supply chains means
       
treatment of users throughout the chain. Thus
Nike is held responsible for how its products are
sourced, and coffee retailers are held responsible
for how their coffee is farmed.
28 Hill, K. “Facebook Joins Forces With Data Brokers To Gather
More Intel About Users For Ads,” Forbes, February 27, 2013,
available at
Systemic Issues in the
Big Data Industry
       
their information supply chain should be
analyzed, but the Big Data Industry includes
     
practices. In effect, the systemic participation in
the Big Data Industry gives rise to “everyone does
it” ethical issues—where norms of practice are
      
chains, as illustrated in Figure 3. Quadrants A and
B capture the ethical issues within a single supply
chain, as described above.
This section examines the ethical issues
captured by Quadrants C and D, and links them
to parallel, more traditional industries. The
      
surveillance as pollution), where surveillance
is a byproduct of the systematic collection,
aggregation and use of individual data (Quadrant
D). The second is the growing problem of
destructive demand within the Big Data Industry
(Quadrant C), where the need for consumer data
     
and sell increasing amounts of information with
lower standards. Both sets of ethical issues stem
from the systemic norms and practices within the
industry. In addition, both are more consumer- or
individual-focused and may apply to a particular
The ethical issues that have to be faced at
both the supply-chain level and the industry level
are summarized in Table 2 (For comparison, the
table provides corresponding examples from
traditional industries; it also describes how CIOs
and CDOs will have to deal with the issues.)
Figure 3: Current Ethical Issues within
the Big Data Industry
76 MIS Quarterly Executive | June 2015 (14:2) | © 2015 University of Minnesota
Ethical Issues in the Big Data Industry
Creating Negative Externalities
(or Surveillance as Pollution)
In all markets, costs regularly accrue to parties
not directly involved in an immediate decision
       
can create harm to the community in the form
of the pollution it produces. The steel company
may contract with a customer—which does not
feel the effects of pollution—without including
the “cost” of pollution. This is an example of a
negative externality, which exists when the harm
done to others is not taken into account in the
immediate transaction.29
There are also negative externalities in the Big
Data Industry arising from the aggressive focus
on collecting consumer data. The danger is that
disclosing personal data can become the default,
29 Coase illustrated negative externalities with the example
of a spark from a train that causes harm to farmers along the
tracks—Coase, R. H. “Problem of Social Cost,” Journal of Law
and Economics (3), 1960, pp. 1-44. Importantly for Coase, negative
externalities do not necessarily require government intervention,
which carries its own cost, but may be resolved through private
ordering between parties.
Table 2: Ethical Issues in the Big Data Industry
Ethical Issues Big Data Industry Examples Tradional Industry
As Faced by CIOs
and CDOs
Supply Chain Level
Unfair or
objeconable harms
from using Big Data
Harms from downstream
use, such as using Big Data
to discriminate in consumer
credit decisions or college
Sale of computer
systems in Iran or
Syria; use of product
in crime
How do downstream
users of your data
protect the consumer
data or impact
Gathering of data
as an intrusion or
violaon of privacy
Quesonable upstream
sourcing, such as purchasing
locaon data surrepously
gathered from mobile
applicaons or using data
from invisible web beacons
unknown to user
Apple and
Foxconn; Nike and
What quesons do
you ask about using
data from unknown or
quesonable sources?
Industry Level
Harm to those
not involved in
the immediate
decision or
transacon caused
by broad tracking
of consumers
and collecon of
Negave externality of
surveillance, such as the
hidden and systemac
aggregaon of data about
Steel industry and
How is your company
possibly contribung
to surveillance by
parcipang in broad
user tracking—or
partnering within
someone who does?
Focus on resale of
consumer data;
treang consumers
simply as a means
to supply the
secondary market of
informaon traders
Destrucve demand, such
as creang a ashlight
applicaon just to gather
user contact or locaon data
Demand for
mortgages created
by the mortgage-
backed securies
industry; websites
and applicaons
used as bait
How is your company
creang destrucve
demand by using data
of quesonable quality
or that was collected
by breaching privacy
June 2015 (14:2) | MIS Quarterly Executive 77
Ethical Issues in the Big Data Industry
and individuals who choose not to disclose can
be harmed. For example, individuals who attempt
to opt out of aggressive data collection by using
TOR30 or other obfuscation technologies may
be targeted by the National Security Agency as
suspicious.31 The harm to individuals who do not
share their data is a result of the decisions of the
majority who do share.
More complicated is when the harmful
effect is compounded by many parties in an
industry acting in a similar way. For example, a
       
harmful effects on the local community of the
pollution it produces. However, the aggregated
harm of pollution from manufacturers worldwide
becomes a problem for society in general through
global warming. Aggregated negative externalities
       
harm results from the fact that the practice
is pervasive in an industry. The harm from
aggregated actions across an industry is more
than the sum of the harms caused by individual
Firms within the Big Data Industry create an
aggregated negative externality because they
contribute to a larger system of surveillance
through the breadth of information gathered and
      
are invisible to users. In general, surveillance
       
       
and a sense of self. An individual’s personal
space permits “unconstrained, unobserved
physical and intellectual movement” to develop
as an individual and to cultivate relationships.32
Surveillance can cause harm by violating
the personal space—both physical and
metaphorical—that is important to develop as an
individual and within relationships. Importantly,
the fear of being watched and judged by others
causes “spaces exposed by surveillance [to]
function differently than spaces that are not so
30 TOR—The Onion Router—is a service to make accessing
websites anonymous. Users’ requests are routed among many other
TOR users’ requests and are bounced throughout the TOR network of
client computers to remain hidden to outsiders. For more information,
31 Zetter, K. “The NSA Is Targeting Users of Privacy Services,
Leaked Code Shows,” WIRED, July 3, 2014.
32 Fried, F. An Anatomy of Values: Problems of Personal and
Social Choice, Harvard University Press, 1970; and Rachels, J. “Why
Privacy Is Important,” Philosophy & Public Affairs, 1975, pp. 323-
exposed” by changing how individuals behave and
Surveillance works by affecting not only
those who are being watched, but also those
who are not actually being watched. In fact, the
mere belief that someone is being watched is
enough for individuals to act as though they are
under surveillance. Prisons are designed so that
only some of the prisoners are watched, but the
       
watched at any one time. Individuals do not need
to know they are under surveillance to act as
though they are under surveillance. Importantly
for the Big Data Industry, the negative externality
of surveillance means the industry can rely on
those individuals not currently being watched
to believe and act as though they are under
Surveillance is particularly effective in
changing behavior and thoughts when individuals
(1) cannot avoid the gaze of the watcher and (2)
cannot identify the watchers.34 By aggregating
data across disparate contexts online, the Big
Data Industry contributes to the perception
that surveillance is impossible to avoid yet
also creates a data record that tells a richer,
more personalized story than individual data
points.35 Broad data aggregators summarize
highly diverse data (the “variety” in Big Data)
so they can analyze individualized behavior. In
addition, most data aggregators are invisible to
the user and thereby aggravate the surveillance
problem by being not only unknown, but also
     
33 Cohen, J. E. “Privacy, Visibility, Transparency, and Exposure,”
The University of Chicago Law Review (75:1), 2008, pp. 181-201.
The inability to escape online surveillance is illustrated in Brunton,
F. and Nissenbaum, H., op. cit., 2011, and Strandburg, K. J. “Home,
Home on the Web and Other Fourth Amendment Implications of
Technosocial Change,” Maryland Law Review, (70:3), 2011. In
the words of Cohen, “Pervasive monitoring of every rst move or
false start will, at the margin, incline choices toward the bland and
mainstream” thereby causing “a blunting and blurring of rough edges
and sharp lines.” —Cohen, J. E. “Examined lives: Informational
privacy and the subject as object,” Stanford Law Review, (52), 2000,
pp. 1373-1438.
34 Cohen, J. E., op. cit., 2008.
35 The Mosaic Theory of privacy explains why privacy scholars
are concerned with all elements of tracking, including transaction
surveillance and purchasing behavior. This theory suggests that
the whole of one’s movements reveal far more than the individual
movements—where the aggregation of small movements across
contexts is a difference in kind and not in degree. See Kerr, O.S. “The
Mosaic Theory of the Fourth Amendment,” Michigan Law Review
(111:3), 2012; and United States v. Jones, Supreme Court of United
States, January 23, 2012, available at
78 MIS Quarterly Executive | June 2015 (14:2) | © 2015 University of Minnesota
Ethical Issues in the Big Data Industry
gather and store data contribute to the perception
of omnipresent and omniscient surveillance and
exacerbate the power imbalance between the
watched and the watcher.36
Currently, the Big Data Industry does
not consider or take account of the negative
externality of surveillance. Firms that capture,
aggregate or use Big Data create a cost to the
larger community in the form of surveillance.
Contributing to Destructive Demand
In addition to the aggregate harm of
surveillance, the Big Data Industry has the
potential to foster destructive demand for
      
consumer-facing organizations to collect more
information. As described below, consumers
unknowingly can become suppliers to a
secondary Big Data market.
The main source of information for the
Big Data Industry is a byproduct of legitimate
    
is collected from a transaction in the primary
market—e.g., checking the weather, buying
groceries, using a phone, paying bills, etc.—and
is then aggregated and merged to create a large
robust data set. In effect, that data is seen as
Data market—such as a data broker or tracking
company—creates value through the secondary
use of the data. The consumer data from the
initial transaction, such as buying books on
Amazon or reading news on The New York Times,
can be sold or repurposed in a secondary market
without losing value. Examples of destructive
demand created by secondary markets are
described in the panel on the following page.
A tipping point exists where the product—
whether residential mortgages as described
in the panel or consumer information—is no
longer pushed into the secondary market, but
rather the secondary market becomes a pull for
the product of the primary, consumer-targeted
market. In this situation, the secondary market
creates a destructive demand by exerting
      
unethical practices to meet the demands of the
     
residential mortgage originators) then treat
36 Richards, N. M., op. cit., 2013.
customers as a mere means37 to the secondary
market (for mortgage-backed securities). The
demand becomes particularly destructive when
the service in the primary market serves as a lure
(or bait) for the supply of the secondary market—
as when mortgage originators became a lure to
produce mortgages for the mortgage-backed
securities market.
Within the Big Data Industry, websites and
applications with trusted relationships with
consumers can become the bait for Big Data,
      
your location or when a website with numerous
tracking beacons38 stores consumer information.
The primary market promises a customer-
   
when it is actually attempting to sell customers’
information to a secondary market.
The attributes of the mortgage-backed
securities market, and the destructive demand
it created, provide a warning for the secondary
market for consumer information in the Big Data
Industry. The demand for the primary market
becomes destructive:
1. Where the secondary market becomes as
or more lucrative than the primary market.
For example, the fee charged to consumers
from the sale of mortgages into the
secondary market. Mortgage originators
could lose money on a mortgage but still
the secondary market. Within the Big Data
Industry, problems will arise when the sale
of consumer information is more lucrative
primary market activities, such as selling
an application or providing a service.
2. When the quality in the secondary market
is less than in the primary market—i.e.,
brokers or data aggregators do not match
the expectations of consumers who
disclose information. For example, the
mortgage-backed securities market was
37 The Mere Means Principle is an ethical principle that posits that
you should never treat people merely as a means to your own ends.
38 A tracking is an often-transparent graphic image, usually no
larger than 1 pixel x 1 pixel, that is placed on a website that is used to
monitor the behavior of the user visiting the site.
June 2015 (14:2) | MIS Quarterly Executive 79
Ethical Issues in the Big Data Industry
residential mortgages they purchased from
3. 
limited accountability to consumers for
their transactions in the secondary market.
bad behavior when they sell into the
secondary market because their activity
in the secondary market is not visible
or incorporated in the primary market.
The term “moral hazard” refers to when
individuals or institutions do not bear the
Examples of Destrucve Demand from Secondary Markets
Secondary markets can be benecial. A secondary market for bicycles and cars can increase
the life of the product. In fact, customers may be more willing to invest in a car in the primary
“new car” market knowing that the robust secondary market for used cars exists to sell the
car when necessary. Other secondary markets create value from items that would otherwise
be thrown away—e.g., the byproduct from cale ranching (wax) or from steel-making (scrap
metal). The secondary market allows rms to capture value from seemingly waste products,
such as ranchers selling the byproduct of cow fat used for candles.
However, secondary markets can apply perverse pressures to distort the demand, quality or
price in the primary market. An example is the market for carbon credits. Firms who create
HFC-23, a super greenhouse gas, as a byproduct of their manufacturing are paid to destroy it
to prevent the gas causing environmental damage. However, the secondary market for HFC-
23 became too lucrave: some rms had an incenve to create HFC-23 so they would be paid
to destroy it. In fact, the World Bank paid $1 billion to two chemical factories in China to de-
stroy HFC-23, and later evidence suggested these rms may have deliberately overproduced
the gas so they could be paid to destroy it in the secondary market.
More problemac is when the secondary market begins to systemacally distort the pri-
mary market, as in the well-known case of mortgage-backed securies and the residenal
mortgage market. The primary market for mortgages is between a lender and home-buyer.
Financial instuons lend money to qualied individuals to buy a home at a rate that takes
into account the potenal risk of the individual defaulng on the loan.
A secondary market for residenal mortgages uses consumer mortgages as the inventory for
a new nancial instrument: mortgage-backed securies (MBS). The MBS market increased
dramacally between 2000 and 2008, and the associated demand for consumer mortgages
to feed the MBS market led to lax sourcing in the primary mortgage market. Interesngly, the
price did not change in the primary market; rates and interest rate spreads remained steady
throughout the growth in the MBS market. However, the quality standards for consumer
mortgages required in the primary market dropped to match the (lower) requirements in the
secondary market. More mortgage originaons and fewer denials led to a greater number of
high-risk borrowers through lax sourcing for the MBS market.
This mismatch between the quality required in the secondary and primary markets proved
parcularly hazardous. The interests of rms in the secondary market did not align with those
of consumers, and without a relaonship with consumers there were higher default rates for
the mortgages included in their MBS. However, when incenves of the secondary market
were aligned with the primary market of the consumer, as in the case of aliated investors,
economists found no change in the mortgage default rates. The increase in private securiza-
on by non-commercial bank nancial rms, with lower requirements for quality, created a
destrucve demand for lower quality mortgages in the primary market.
80 MIS Quarterly Executive | June 2015 (14:2) | © 2015 University of Minnesota
Ethical Issues in the Big Data Industry
the case of mortgage originators selling
bad loans into the MBS secondary market.
In the Big Data Industry, consumer-facing
organizations are currently not held
accountable for selling access to consumer
data even by market forces, and their
activities in the secondary market are
invisible to the primary consumer market.
Guidelines for a Sustainable
Big Data Industry
  
     
risks but without clear industry leaders to
develop cooperative strategies. Moreover, the
power of Big Data is generated by non-consumer-
      
       
     39
       
      
their industry and have a stake in a sustainable
        
Industry are of particular importance in creating
sustainable industry practices:
1. Possible leaders in the industry, which
as gatekeepers, such as consumer-facing
companies, website operators and
application providers. These companies
control how information is initially
2. 
and knowledge in the area of Big Data
analytics, such as the American Statistical
Association and the Census Bureau, as well
as HHS and the National Research Council
(which govern academics’ Institutional
Review Boards). These organizations
have the stature and deep knowledge
of research, data sets, analytics and
39 For a comparison of regulating the credit reporting industry
with regulating Big Data, see Hoofnagle, C. J. How the Fair Credit
Reporting Act Regulates Big Data, paper presented at Future of
Privacy Forum Workshop on Big Data and Privacy: Making Ends
Meet, September 10, 2013, available at
3. Providers of key products within the Big
Data Industry, such as Palantir, Microsoft,
SAP, IBM, etc. These companies have few
analytic products and services, and can
analyze and use Big Data.
As this article has shown, the ethical issues
and problems facing the Big Data Industry
are similar to those faced by other industries.
and sustainable relationships within the industry
include visible data stewardship practices,
greater data due process internally and using the
services of a data integrity professional. These
solutions, which are summarized in Table 3 and
      
in this article. (Table 3 also describes how CIOs
and CDOs can address the problems.) Despite the
potential to create harm, the Big Data Industry
has the potential to be a force for good and the
focus therefore should be on implementing the
solutions described below to create value for all
1. Identify and Communicate
Data Stewardship Practices
Current information supply chains are not
visible, putting consumers at a disadvantage
in choosing preferred supply chains or holding
        
particular supply chain. Such information
asymmetries could be minimized by clearly
illustrating the upstream sourcing information
and downstream use in order to report the data
stewardship practices. Data stewardship includes
the rules about internal treatment and external
sharing of information for different types of data.
Industry groups can develop data stewardship
      
coalesce around a format for communicating data
stewardship practices.
Making the supply chain visible will clearly
       
       
      
different upstream sources of information, the
type of information collected, its internal uses
40 For a balanced view on solutions that both optimize the use
of technology and respect privacy and ethics, see Mayer, J. and
Narayanan, A. “Privacy Substitutes,” Stanford Law Review Online
(66), 2013, pp. 89-96; and Bambauer, J., op. cit., 2014.
June 2015 (14:2) | MIS Quarterly Executive 81
Ethical Issues in the Big Data Industry
     
customers and recipients are all important
for understanding the entirety of the supply
     
An illustrative example is shown in Figure
5. The data sources, type of data and level of
    
     
primary use, secondary use and storage explains
the purpose and vulnerability of the data; and the
types of data, recipients and level of trust in the
recipients explains the downstream uses of the
data collected.
While the information supply chain may look
complicated, a similar problem has been resolved
in areas such as free-trade coffee, organic food
   
customer groups. The information supply chain of
      
Table 3: Possible Solutions to the Big Data Industry’s Ethical Issues
Type of Issue Cause of Problem Potenal Soluon As Faced by CIOs and CDOs
Supply Chain
Sourcing and
Use Issues
Data Stewardship
Firms not accountable
for conduct of
upstream sources and
downstream customers
Illustrate role of rm in
larger supply chain
Make machine read-
able nocaon of
supply chain informa-
on available to policy
makers, reports and
privacy advocates
Idenfy and take ownership
of upstream sources and
downstream customers/uses
of data
Ensure informaon about data
stewardship pracces is avail-
able to experts and novices
Supply chain not visible Make data stewardship
pracces of supply
chain visible
Do not enter into
condenality agreements
that preclude explaining your
data partners, either upstream
sources or downstream users
as Negave
Data Due Process
Harm to others not
captured by rms
collecng, storing
or using personally
idenable informaon
Minimize surveillance Make tracking visible to
Internalize cost of
surveillance with
increased data due
(Industry) Require addional
data due process for rms
acquiring and retaining PII
Demand for
Data Integrity
Secondary market of
data trading has lower
quality requirements
than primary
Use a data integrity
professional when
handling or selling PII
(Industry) Instute data
integrity professional or board
for projects partnering with Big
Data sources and customers
Secondary market is
not visible to primary
market (consumers)
Make acvity in
secondary market
visible to regulators
and consumers
Account for and communicate
addional risk from partnering
in secondary market for Big
Data through disclosure
82 MIS Quarterly Executive | June 2015 (14:2) | © 2015 University of Minnesota
Ethical Issues in the Big Data Industry
Figure 4: Guidelines for a Sustainable Big Data Industry
Prob lems Supply Chain Sourcing
and Use Issues
Surveillance as
Negative Externality
Destruct ive Demand for
Consumer Information
General Solutions Make supply chain vi sible to
technologists, researchers,
consume rs and regulators
Decrease survei llance harm
and internali ze costs when
contributing to surveillance
Make secondary market
for consume r data visible
Make data sources and uses
of information visible and
searchable (see Figure 2)
Clearly identify firms
within an information
supply chain
Internalize surveillance cost:
additional data due process
required when retaining PII
Communicate to consumers,
regulators and investors the
valu e create d and associ ated
risk from activity in secondary
marke t for inf ormation
Explain % of data sol d and
% of sales from selling
information in se condary
marke t for inf ormation
Ensure adherence to and
compliance with stewardshi p
norms through professional
data integrity
Make primary dat a
collectors responsible for
quality of information
Internalize surveillance cost:
require data integrity
professional or board when
using personal data (PII)
Guidelines for the
Big Data Industry
1. Identify and
communicate data
stewardship practices
2. Differentiate data due
process model for PII
and non-PII*
3. Quantify activity in
secondary market for
Big Data
4. Institute data inte grity
professional or board
for Big Data Anal ytics
*The ability to fully differentiate between personally identifiable information (PII) and non-PI I is d eb atab le, a s arg ue d by N arayanan, A.
and Shmatikov, V. “Myths and Fallacies of Personally Identifiable Information,” Communications of the ACM (53: 6), 2010), pp. 24-26.
Figure 5: Example of a Firm’s Information Supply Chain Diagram
Fi rm’s
Fi rm’s
Use and
Upstream Information Flow Downstream Information Flow
Indivi dual ized
Informati on
Informati on
Member of Verified
Supply Chain
Confirmed Trusted
Unverified Recipient
(Data Stewardship
Practice Unknown)
Arrows des ignate type of information
Boxes designate type/name of recipient
June 2015 (14:2) | MIS Quarterly Executive 83
Ethical Issues in the Big Data Industry
industry groups, customer groups and regulators
that have the knowledge necessary to certify a
level of data stewardship within the supply chain.
Making information supply chains available in
a machine-readable form would support the
illustration in Figure 5, as has been developed
and effectively called for by Cranor.41 Users would
then be able to identify and choose trusted and
     
supply chain and data stewardship practices
in a uniform way is critical not only for helping
users directly, but also for allowing researchers,
reporters, technologists and academics to easily
diagram and analyze the many different supply
chains and provide an audit trail for the data.
2. Differentiate Data Due Process
Requirements for Personal Data
Two approaches can be used to manage
surveillance as a negative externality: (1)
      
contributing to surveillance and (2) the industry
can implement policies to internalize the cost of
      
effective (and therefore most harmful) when the
watcher is hidden yet omnipresent.42 Firms can
reduce their role in consumer surveillance by
becoming more visible to the consumers and by
limiting data collection. The negative externality
      
invisible to users, such as data aggregators and
data brokers, have a special role in the online
surveillance system. Both data aggregators
and data brokers are invisible to users while
aggregating data across diverse sources. Making
the tracking of individuals obvious at the time
41 Cranor, L. F. “Necessary but Not Sufcient: Standardized
Mechanisms for Privacy Notice and Choice,” Journal on
Telecommunications and High Technology Law (10), 2012, pp. 273-
307. Rather than focus on the type of information, the rm’s storage,
information use or third-party access to data would be highlighted
if such tactics diverge from commonly accepted practices. Research
demonstrates that users care most about the possible secondary
use or third-party access to information both online and with
mobile devices, as noted by Martin, K., op. cit., 2015; Shilton,
K. and Martin, K. E. “Mobile Privacy Expectations in Context,”
Telecommunications Policy Research Conference (41), 2013; and
Martin, K. E. “Privacy Notices as Tabula Rasa: An Empirical
Investigation into How Complying with a Privacy Notice Is Related
to Meeting Privacy Expectations Online,” Journal of Public Policy
and Marketing, 2015.
42 Cohen, J. E., op. cit., 2008.
of data collection can diminish the harm of
In addition to decreasing the effectiveness
and related harm of surveillance, internalizing
        
tool to diminish this negative externality. For
example, data brokers and aggregators that store
and distribute information within the Big Data
Industry could have additional data due process
     
    
information (PII). While some have claimed
that PII is not clearly distinguishable,43 
that retain information that can be linked back
to an individual so it can be fused with other
information about the same individual should
have an additional obligation of data due process.
     
      
instructive moving forward: (1) identifying
audit trails, (2) offering interactive modeling
and (3) supporting user objections.44 In addition
       
trail for how information is sourced, used and
distributed similar to that shown in Figure 5,
       
modeling of the use of information and a process
to enable individuals to examine and object
to the information stored. These additional
       
     
The additional obligations would increase the
cost of retaining the information, internalize the
previously externalized harm (surveillance) and
      
retaining PII.
     
data stewardship practices and additional data
due process procedures would increase the cost
of holding individualized yet comprehensive
data and internalize the cost of contributing to
surveillance. Many negative externalities are
         
43 Ohm, P. “Broken Promises of Privacy: Responding to the
Surprising Failure of Anonymization,” UCLA Law Review (57), 2009,
pp. 1701-1777; and Narayanan, A. and Shmatikov, V. “Myths and
Fallacies of Personally Identiable Information,” Communications of
the ACM (53:6), 2010, pp. 24-26.
44 Citron, D. K. and Pasquale, F. “The Scored Society: Due
Process for Automated Predictions,” Washington Law Review (89),
2014, pp. 1-33; see also Crawford, K. and Schultz, J. “Big Data and
Due Process: Toward a Framework to Redress Predictive Privacy
Harms,” Boston College Law Review (55:1), 2014, pp. 93-129.
84 MIS Quarterly Executive | June 2015 (14:2) | © 2015 University of Minnesota
Ethical Issues in the Big Data Industry
cost of reigning in surveillance is too much for
          
changing its data practices would be minimal.
For those that wish to use large samples of
    
governance together with the services of a data
     
       
would ensure that data stewardship practices
and data due process are followed. Similarly, an
internal consumer review board, as advocated by
Calo and cited in the draft White House Consumer
Privacy Bill of Rights Act of 2015, would
similarly internalize the cost of storing and using
3. Quantify Activity in the
Secondary Market for Big Data
    
interests of the secondary market for consumer
information are not aligned with the primary
market and when the secondary market is not
visible to the primary market. By linking all
     
       
in the game” and thus an incentive to align
interests.46 In other words, by framing themselves
        
a vested interest in ensuring others in the chain
uphold data stewardship and data due process
practices. Otherwise, their reputation would be at
In addition, by making the secondary market
more visible to the primary market, the primary
market can take into consideration secondary
      
        
in the primary market depending on its type of
involvement in the secondary market for selling
information. Importantly, the current approach,
where the secondary market for Big Data is
invisible to the primary consumer-facing market,
does not allow for such feedback.
     
       
      
45 Calo, R. “Consumer Subject Review Boards: A Thought
Experiment,” Stanford Law Review Online (66), 2013, pp. 97-102.
46 For the mortgage-backed securities market, skin in the game—
and aligning interests—was effective to avoid losses—James,
C.M. “Mortgage-Backed Securities: How Important Is ‘Skin in the
Game’?,” FRBSF Economic Letter, December 13, 2010.
example, within the mortgage-backed securities
    
did not have interests aligned with the primary
market, were not able to sell their securities
at the same rate as those companies that were
      
recognized the inherent risk of trading with
      
with the consumer market. For the Big Data
Industry, history suggests there would be a
market for Big Data.
4. Institute Data Integrity Professional
or Board for Big Data Analytics
The practical implications of these guidelines
call for renewed attention to the training and
development of data integrity professionals. The
focus of their training should be on incorporating
an ethical analysis, which is consistent with FTC
Commissioner Julie Brill’s focus on the role of
technologists in protecting privacy in the age
of Big Data, as well as Mayer and Narayanan’s
call for engineers to develop privacy substitutes
within their design.47
First, professional data scientists are needed to
implement the solutions outlined above to curtail
surveillance and destructive demand, as well as
to ensure data stewardship practices. Currently,
advice for Big Data professionals, including
data scientists, data analytics specialists, and
business intelligence and analytics specialists,
focuses on the challenges in using Big Data, such
as leadership, talent management, technology,
decision making and company culture. There is
little advice on ensuring data integrity.48
Second, consumer review boards, made up
partly of professional data scientists, would
oversee and authorize research on human beings
within the commercial space. As Calo notes,
     
to conduct research from their Institutional
Review Board and undertake associated
training, even when the research is for societal
     
47 Brill, J. A Call to Arms: The Role of Technologists in Protecting
Privacy in the Age of Big Data, Sloan Cyber Security Lecture by
Commissioner Julie Brill, Polytechnic Institute of NYU, October 23,
2013; Mayer, J. and Narayanan, A., op. cit., 2013.
48 Chen, H., Chiang, R. H. L. and Storey, V. C. “Business
Intelligence and Analytics: From Big Data to Big Impact,” MIS
Quarterly (36:4), 2012, pp. 1165-1188.
June 2015 (14:2) | MIS Quarterly Executive 85
Ethical Issues in the Big Data Industry
without oversight, even when at the expense
of the consumer.49 Revelations that OKCupid
and Facebook50 had conducted experiments
on users without their knowledge only show
how prescient Calo was in the call for consumer
review boards; and effective consumer review
Finally, academic institutions continue to
develop degree courses in business analytics,
business intelligence and data analytics to train
      
students to take a course in ethics. A survey of
the top 15 such programs shows the intense
       
privacy, ethics or corporate and professional
responsibility.51 Accreditation for such programs
       
professionals who graduate with a degree in data
science, data analytics or business intelligence,
and to support the solutions proposed in these
Concluding Comments
This article has examined Big Data within the
       
persistent issues and points of weakness in
current market practices. In doing so, it has
examined the industry’s information supply chain
of upstream suppliers and downstream uses of
data, the ethical issues arising from the negative
externality of surveillance caused by persistent
tracking, aggregation and the use of consumer-
level data, and the potential destructive demand
driven by the secondary market for consumer
information. Importantly, the article has
       
economic and ethical issues at the individual
    
has suggested associated solutions to preserve
sustainable industry practices.
49 McAfee, A. and Brynjolfsson, E. “Big Data: The Management
Revolution,” Harvard Business Review, October 2012.
50 Stampler, L. “Facebook Isn’t the Only Website Running
Experiments on Human Beings,” Time, July 28, 2014, available at
51 The survey includes both bachelor’s and master’s programs
from across schools/programs such as business and engineering.
id/1108042?page_number=3 and
id=4184. Programs reviewed include Bentley, Columbia, LSU,
NYU, GWSB, Northwestern, Rutgers, CMU, Harvard, MIT, NCSU,
Stanford, UT Austin and UC Berkeley. Both Information Week’s and
NCSU’s lists focus on U.S. universities.
About the Author
Kirsten E. Martin
Kirsten Martin ( is an
assistant professor of Strategic Management and
Public Policy at the George Washington University
School of Business. She is principle investigator
on a three-year grant from the National Science
Foundation to study online privacy. Martin is a
member of the advisory board for the Future
Privacy Forum and the Census Bureau’s National
Advisory Committee, and is a fellow at the
Business Roundtable Institute for Corporate
Ethics. Her research interests include online
privacy, the ethics of Big Data, privacy, corporate
responsibility and stakeholder theory.
... Explainability and transparency provide a platform for evaluating fairness because if the AI process is explainable, then organizations are more apt to address bias removal [24,74,104]. Other key principles concern data privacy and data security encompassed by data management and data quality, which are paramount to the integrity of the AI process [80,97]. There are also RAI principles that are related to management, accountability, and governance, however, an overarching and cohesive principle to measure the significance of these items is missing. ...
... Martin [80] CDO Involvement Leadership support is key not only in the organizational commitment category, but critical to the success of responsible data management with CDO involvement Helmy et al. [56] Use of Synthetic Data The notion of synthetic data also exists where the corporation will infuse manufactured data into the environment to influence the models Campbell [26] Use of DataOps More mature data management operations that practice RAI have a few tools in place known as DataOps, which is based on DevOps for programming, but applied to data management Rodriguez et al. [110] Right to be Forgotten In addition to data privacy, Banks also enable the 'right to be forgotten', where a customer would like their data purged Tjong Tjin Tai [122] Content courtesy of Springer Nature, terms of use apply. Rights reserved. ...
... Managing privacy while performing ML techniques that do not expose personally identifiable information (PII) is present in differential privacy capabilities [42]. Banks concerned with good data management ethics and governance will focus on data privacy as well as differential privacy, leverage data pipeline tools as well as data Ops, EDA for pre-modeling, knowledge graphs for visual representation of data, big data lakes, generating synthetic data (though nascent and sometimes debated) [26], enabling the right to be forgotten, and have Chief Data Officer (CDO) involvement [80]. In terms of other references to this principle and category, there was partial alignment with Jobin et al. [62] and Fjeld et al. [43] in terms of data privacy, partial alignment with Myers and Nejkov [95] who focused on data validation, and no clear alignment with Hagendorff [54]. ...
Full-text available
Widespread use of artificial intelligence (AI) and machine learning (ML) in the US banking industry raises red flags with regulators and social groups due to potential risk of data-driven algorithmic bias in credit lending decisions. The absence of a valid and reliable measure of responsible AI (RAI) has stunted the growth of organizational research on RAI (i.e., the organizational balancing act to optimize efficiency and equity). To address this void, we develop a novel measurement instrument to assess RAI maturity in firms. A review of the nascent literature reveals that there is a wide distribution of RAI capabilities. The RAI instrument that we advance is based on the exhaustive review of this dispersed literature. Analyses of data from large US banks show strong evidence of validity and reliability of the RAI maturity instrument.
... Second, these perspectives offer a bifurcated view of agency, privileging either organizational or environmental factors. With algorithmic decision-making, it becomes particularly difficult and tricky to unpack who is responsible for what, as observed by scholars who have explored the ethical implications of algorithmic decision-making (e.g., Martin, 2015Martin, , 2019Metz, 2023). Third, existing metaphors fail to account for the deep technological and material infrastructure that facilitates algorithmic decision-making (e.g., Orlikowski, 2007;Orlikowski and Scott, 2016). ...
Full-text available
According to the ‘Point’ essay, management research’s reliance on corporate data threatens to replace objective theory with profit-biased ‘corporate empiricism,’ undermining the scientific and ethical integrity of the field. In this ‘Counterpoint’ essay, we offer a more expansive understanding of big data and algorithmic processing and, by extension, see promising applications to management theory. Specifically, we propose a novel management metaphor: organizations as algorithms. This metaphor offers three insights for developing innovative, relevant, and grounded organization theory. First, agency is distributed in assemblages rather than being solely attributed to individuals, algorithms, or data. Second, machine-readability serves as the immutable and mobile base for organizing and decision-making. Third, prompting and programming transform the role of professional expertise and organizational relationships with technologies. Contrary to the ‘Point’ essay, we see no theoretical ‘end’ in sight; the organization as algorithm metaphor enables scholars to build innovative theories that account for the intricacies of algorithmic decision-making.
... Nevertheless, the related fiscal allocation for corrective institutions is partially constrained by social norms and issues of privacy. Big data implementation leads to ethical concerns in the context of surveillance operations, potentially exacerbating power imbalances between the watched and the watcher (Martin, 2015). Governors must consider the adverse externalities associated with surveillance. ...
Full-text available
The construction of smart cities and smart government is in full swing. Public sectors emphasize the advancement of electronic processes and applications to foster innovation. Hence, accelerating the digital reform of the correctional governance system and capabilities is the general trend. This article utilizes the good governance theory as its central framework to elucidate the integration of the rule of law, accountability, transparency, efficiency, equity, responsiveness, participation, and consensus within the correctional services in Macao. Additionally, the article seeks to analyse the impact of implementing big data on the governance system through a multidisciplinary policy analysis. Drawing from the perspectives and values associated with good governance, this study examines the various components and assessments of capabilities. Specifically, it identifies leadership, innovation, execution, responsiveness, and readiness as the determinants that collectively shape the path of capabilities reform. The incorporation of relevant implications and practices offers a scientific foundation and orientation for the digital reform of governance system and capabilities within correctional services.
... Harakka pohtii osin varsin ansiokkaasti sitä, kuinka verotuksen uudistaminen voisi vastata datakapitalismin haasteisiin. Tosin yllättävää on, että hän ei näyttäisi mitenkään vahvasti rinnastavan datakapitalismiin liittyvää kansalaisten valvontaa negatiiviseen ulkoisvaikutukseen (Martin 2015), jota Big Tech ei toiminnastaan maksa, vaan joka jää kansalaisten ja yhteiskunnan maksettavaksi. Tämän negatiivisen ulkoisvaikutuksen haittaverotuksen mahdollisuuksia olisi myös mahdollista pohtia laajemmin. ...
Full-text available
Sanna Marinin hallituksen liikenne-ja viestintäministeri, kansanedustaja, toimittaja sekä tietokirjailija Timo Harakka on kirjoittanut mukaansatempaavan teoksen kapitalismin 2000-luvun alkua luonnehtivasta aikakaudesta, jota hän kutsuu datakapitalismiksi. Tässä suhteellisen nopeasti kehittyneessä kapitalismin muodossa datan kerääminen, käyttö ja hallinta ovat Harakan mukaan kaiken liiketoiminnan ytimessä. Samalla datasta on tullut aikamme arvokkain pääoma, jonka hallinnan avulla taloudellinen, sosiaalinen ja poliittinen valta nykyisin hankitaan ja ylläpidetään. Kiinnostavasti Harakka paljastaa samalla, kuinka alipolitisoitunutta tämä kapitalismin uusimman version tarkastelu Suomessa on. Esimerkiksi kotimaisilla vaalikentillä juuri kukaan ei tule kysymään häneltä datapolitiikasta, vaikka varmaan moni ymmärtää hänen näköalapaikkansa siihen ja kyvykkään datapolitiikan merkityksen Suomelle. Jos ei ymmärrä, niin Harakan teos on syytä lukea mitä pikimmiten. Niillekin, jotka tätä teemaa tutkivat ja teoretisoivat, kirja avaa runsaasti uusia ja arvokkaita näkymiä kentän ajankohtaisiin ja historiallisiin tapahtumiin ja kehityskulkuihin. Alan entisenä ministerinä Harakka on tehnyt arvokkaan työn avatessaan tarkastelemaansa aihetta iskevällä kielellä, huomattavalla analyyttisellä terävyydellä ja asiantuntemuksella. Datakapitalismin alipolitisoituneisuuden konketisoiminen Suomessa ja laajemminkin on jo itsessään tärkeä yhteiskunnallinen panos. Datakapitalismi ei näyttäydy Harakalle yhtenäisenä ilmiönä vaan sillä on eri muotoja ajassamme. Harakka nimeää datakapitalismin amerikkalaisen version, jossa 'voittaja vie koko potin' kasinoksi ja erottaa sen kiinalaisesta datan avulla kansalaisia kontrolloivasta versiosta, jota hän kutsuu kasarmiksi sekä eurooppalaisesta yhdenvertaiseen kansalaisuuteen tähtäävästä mallista, jota hän nimittää kansankodiksi. Allekirjoittaneelle kansankodista ponnistavalle tutkijalle varsinainen akateeminen kohtaaminen amerikkalaisen version kanssa liittyy kahteen tutkimusvierailuun Yhdysvalloissa (vuonna 2013 Stanfordin yliopistossa ja vuonna 2019 Harvardin yliopistossa). Piilaakson ytimessä Stanfordissa törmäsin vahvaan teknologisen optimismin sävyttämään keskusteluun aiheesta. Tässä yhteydessä amerikkalaisten digijättien uskottiin demokratisoivan vaiheittain koko planeetan ja pidemmällä tähtäimellä ehkä myös avaruuden. Lieneekö taustalla ollut arabikevään jälkimaininkeja vai pelkkää teknologiahybristä vai molempia? Tutkimusvuoteni Harvardissa vuonna 2019 toi puolestaan eteeni varsin erilaisen kuvan samasta yksityisestä digivallasta. Nyt Big Tech nähtiin USA:n itärannikolla merkittävänä uhkana amerikkalaiselle liberaalidemokratialle ja laajemminkin ihmisoikeuksille. Huomattava poliittinen muutos keskustelun luonteessa. Tämä muutos johti kohdallani Harvardin vuoden jälkeen siihen, että
... (Fogg, 2003;Matz et al., 2017). However, there are some high-level suggestions that organisations could follow, such as taking ownership of their data sources, not entering into confidentiality agreements that prevent explaining who their data partners are and making the data supply chain visible so that an organisation can ensure there is no misuse of data (Martin, 2015). ...
Full-text available
The use of machine learning (ML) to create more innovative services is a growing trend. However, there is a risk that ML models can create biased services, leading to discrimination against certain groups of users. This structured literature review explores the role of service design in preventing machine learning bias. Our results show ways designers share the responsibility of identifying potential bias in ML-based services, methods to mitigate bias and ensure that data is collected and used in a fair and equitable way to all service users. By being attentive to the topic, service designers can support creating more inclusive and innovative services.
Data science promises to revolutionize healthcare, providing insight into disease mechanisms, enabling a more personalized approach to care, improving public health surveillance and the ability to predict trends. However, there are challenges and barriers to its application and implementation, requiring careful attention to data governance and ethics. In this chapter, we present examples of the application of data science in global health and briefly discuss the challenges that should be addressed for its equitable implementation.KeywordsData scienceBig dataEquitable data scienceData sharingData transparency
This chapter provides an overview of how the Caribbean can leverage real-time big data sources and appropriate analytical techniques. This is done through three case studies to assess the efficacy of Shelter in Place Order (SIPO) policies in Trinidad and Tobago using Facebook Data for Good; to gauge the mental health and coping strategies in Caribbean countries during COVID-19 using Google Community reports; and to assess the impact of COVID-19 on stock market returns. We highlight the strengths and limitations of the data and the empirical methods which can be used to strengthen the accuracy of estimates as well as test the robustness of the results. The chapter concludes by drawing lessons learnt to demonstrate that once correctly sourced and applied, near real-time data can be a viable alternative, and in some instances can be combined with official data to support decision-making and policy development in the Caribbean. Ultimately near real-time data can be particularly useful not only during COVID-19 but during other crises and natural disasters that occur in the region such as hurricanes, earthquakes, and volcanic eruptions.KeywordsCOVID-19Big dataArtificial IntelligenceCaribbean
Full-text available
Artificial Intelligence (AI) is revolutionizing businesses and daily life, with AI-powered technologies like personal assistants and medical diagnostic systems transforming how we interact and make decisions. However, the ethical implications of these technologies cannot be ignored. AI systems can produce biased results and decisions if not designed to be fair and unbiased. Corporate Digital Responsibility (CDR) provides a valuable framework for addressing these ethical dilemmas. Service organizations need to navigate CDR issues across the data and technology life-cycle stages (i.e. their creation, operation, refinement, and retention) and across its digital service ecosystem (incl. its external business partners). Despite the risks associated with poor CDR practices, companies may adopt them to benefit from data monetization, enhanced customer experience, and productivity improvement. To mitigate these risks and build a strong CDR culture, organizations need to establish ethical norms, prioritize customer privacy, and ensure equitable power dynamics with business partners. The emergence of generative AI poses enhanced CDR challenges, such as AI complexity, monitoring, accountability, and workforce changes. Going forward, CDR is a crucial framework for firms to address the needs of their multiple stakeholders and to ensure sustainable business practices in the increasingly digital service world.
Full-text available
Business intelligence and analytics (BI&A) has emerged as an important area of study for both practitioners and researchers, reflecting the magnitude and impact of data-related problems to be solved in contemporary business organizations. This introduction to the MIS Quarterly Special Issue on Business Intelligence Research first provides a framework that identifies the evolution, applications, and emerging research areas of BI&A. BI&A 1.0, BI&A 2.0, and BI&A 3.0 are defined and described in terms of their key characteristics and capabilities. Current research in BI&A is analyzed and challenges and opportunities associated with BI&A research and education are identified. We also report a bibliometric study of critical BI&A publications, researchers, and research topics based on more than a decade of related academic and industry publications. Finally, the six articles that comprise this special issue are introduced and characterized in terms of the proposed BI&A research framework.
Full-text available
A new breed of executive, the chief data officer (CDO), is emerging as a key leader in the organization. We provide a three-dimensional cubic framework that describes the role of the CDO. The three dimensions are: (1) Collaboration Direction (inwards vs. outwards), (2) Data Space (traditional data vs. big data) and (3) Value Impact (service vs. strategy). We illustrate the framework with examples from early adopters of the CDO role and provide recommendations to help organizations assess and strategize the establishment of their own CDOs.
Full-text available
Recent privacy scholarship has focused on the failure of adequate notice and consumer choice as a tool to address consumers' privacy expectations online. However, a direct examination of how complying with privacy notice is related to meeting privacy expectations online has not been performed. This paper reports the findings of two factorial vignette studies describing online consumer tracking, where respondents rated the degree online scenarios met consumers' privacy expectations or complied with a privacy notice. The results suggest respondents perceived the privacy notice as offering greater protections than the actual privacy notice. Perhaps most problematic, respondents projected the important factors of their privacy expectations onto the privacy notice. In other words, privacy notices became a tabula rasa for users' privacy expectations. The findings provide guidance for policy makers and firms to avoid unnecessary privacy violations caused by an over reliance on privacy notices. Considering the importance of privacy notices in managing privacy online, more work should extend this study to understand how consumers understand notices and how consumers' perceptions of privacy notices map to their privacy expectations – if at all.
Full-text available
Recent scholarship in philosophy, law, and information systems suggests that respecting privacy entails understanding the implicit privacy norms about what, why, and to whom information is shared within specific relationships. These social contracts are important to understand if firms are to adequately manage the privacy expectations of stakeholders. This paper explores a social contract approach to developing, acknowledging, and protecting privacy norms within specific contexts. While privacy as a social contract – a mutually beneficial agreement within a community about sharing and using information – has been introduced theoretically and empirically, the full impact on firms of an alternative framework to respecting the privacy expectations of stakeholders has not been examined. The goal of this paper is to examine how privacy norms develop through social contract’s narrative, to redescribe privacy violations given the social contract approach, and to critically examine the role of business as a contractor in developing privacy norms. A social contract narrative dealing specifically with issues of privacy is an important next step in exploring a social contract approach to privacy. Here, the narrative is used to explain to analyze the dynamic process of privacy norm generation within particular communities. Based on this narrative, individuals within a given community discriminately share information with a particular set of obligations in mind as to who has access to the information and how it will be used – rather than giving away privacy, individuals discriminately share information within a particular community and with norms governing the use of their information. Similar to contractual business ethics’ impact on global commerce in explaining how and why norms vary across global contexts, the social contract approach to privacy explains how and why norms vary across communities of actors. Focusing on agreements around privacy expectations shifts the responsibility of firms from adequate notification to the responsibility of firms as contractors to maintain a mutually beneficial and sustainable solution.
Full-text available
We use the Business Roundtable’s challenge to the SEC’s 2010 proxy access rule as a natural experiment to measure the value of shareholder proxy access. We find that firms that would have been most vulnerable to proxy access, as measured by institutional ownership and activist institutional ownership in particular, lost value on October 4, 2010, when the SEC unexpectedly announced that it would delay implementation of the Rule in response to the Business Roundtable challenge. We also examine intra-day returns and find that the value loss occurred just after the SEC’s announcement on October 4. We find similar results on July 22, 2011, when the D.C. Circuit ruled in favor of the Business Roundtable. These findings are consistent with the view that financial markets placed a positive value on shareholder access, as implemented in the SEC’s 2010 Rule.
Sensor data gathered from mobile devices generates big data, which can help organizations continuously monitor a wide range of processes in ways that prompt actions and generate value. However, generating reliable data from such devices is not straightforward and presents a variety of challenges. This article describes the challenges and provides guidelines based on a case study of "Street Bump," a mobile device app that the City of Boston uses to facilitate road infrastructure management.
Big Data is increasingly mined to rank and rate individuals. Predictive algorithms assess whether we are good credit risks, desirable employees, reliable tenants, valuable customers-or deadbeats, shirkers, menaces, and "wastes of time." Crucial opportunities are on the line, including the ability to obtain loans, work, housing, and insurance. Though automated scoring is pervasive and consequential, it is also opaque and lacking oversight. In one area where regulation does prevail-credit-the law focuses on credit history, not the derivation of scores from data. Procedural regularity is essential for those stigmatized by "artificially intelligent" scoring systems. The American due process tradition should inform basic safeguards. Regulators should be able to test scoring systems to ensure their fairness and accuracy. Individuals should be granted meaningful opportunities to challenge adverse decisions based on scores miscategorizing them. Without such protections in place, systems could launder biased and arbitrary data into powerfully stigmatizing scores.
Although survey research is a young field relative to many scientific domains, it has already experienced three distinct stages of development. In the first era (1930-1960), the founders of the field invented the basic components of the design of data collection and the tools to produce the statistical information from surveys. As they were inventing the method, they were also building the institutions that conduct surveys in the private, academic, and government sectors. The second era (1960-1990) witnessed a vast growth in the use of the survey method. This growth was aided by the needs of the U.S. federal government to monitor the effects of investments in human and physical infrastructure, the growth of the quantitative social sciences, and the use of quantitative information to study consumer behaviors. The third era (1990 and forward) witnessed the declines in survey participation rates, the growth of alternative modes of data collection, the weakening of sampling frames, and the growth of continuously produced process data from digital systems in all sectors, but especially those emanating from the Internet. Throughout each era, survey research methods adapted to changes in society and exploited new technologies when they proved valuable to the field.