Developing a Public Interest Mandate for the Governance and Use of Administrative Data in the United Kingdom

DOIhttp://doi.org/10.1111/j.1467-6478.2016.00759.x
Published date01 September 2016
Date01 September 2016
JOURNAL OF LAW AND SOCIETY
VOLUME 43, NUMBER 3, SEPTEMBER 2016
ISSN: 0263-323X, pp. 360±92
Developing a Public Interest Mandate for the Governance
and Use of Administrative Data in the United Kingdom
Graeme Laurie* and Leslie Stevens*
This article addresses the legal and ethical uncertainties surrounding
the use of administrative data for research. Drawing upon best prac-
tices developed by the authors in previous data initiatives and engage-
ment with research communities, the article suggests a problematic
organizational culture as the most significant barrier to proportionate
and good governance of administrative data. Accordingly, it offers a
novel means for data custodians to identify key considerations by
introducing a decision-making template that supports public authori-
ties' assessment of preparedness for data reuse through identification
of challenges faced, related to sector-specific practices. As a catalyst
for change, the authors advocate a public interest mandate ± com-
mitment to safely and ethically use administrative data when it is in the
public interest to do so. This is delivered through implementation of the
decision-making template, overt commitment to principles of public
interest and proportionality, and engagement with stakeholders to
address remaining areas of uncertainty.
INTRODUCTION
Governments of the twenty-first century are in the business of big data. This
includes the United Kingdom government, which handles 1.5 billion trans-
actions with businesses and citizens annually.
1
Each involves administrative
data collected for the delivery of public services, conveying a rich story
360
*Mason Institute, University of Edinburgh School of Law, Old College,
South Bridge, Edinburgh EH6 4DS, Scotland
graeme.laurie@ed.ac.uk leslie.a.stevens@ed.ac.uk
This work was supported by the Economic and Social Research Council grant number
ES/L007487/1 (Administrative Data Research Centre ± Scotland).
1 R. Sargeant, `Digital Marches on: Rising Take-Up, Falling Costs' (2014), at
costs/>.
ß2016 The Author. Journal of Law and Society ß2016 Cardiff University Law School
This is an open access article under the terms of the Creative Commons Attribution License, which permits use, distribution
and reproduction in any medium, provided the original work is properly cited.
about British society on education, health and welfare, income, employment,
migration patterns, and crime. The potential for administrative data to
deepen un derstan dings of so ciety to b etter add ress soci o-econo mic
challenges is without question.
2
Applying big data techniques
3
to administrative data reveals previously
undiscovered patterns, relationships, and associations. It remains unknown,
however, what impact such new associations have on citizens' privacy and
their relationships with their governments.
4
The crucial importance of having
social licence to use personal data (or even de-identified data which was
once `personal')
5
is context-specific and, as we discuss below, salutary
lessons can be learned from specific sectors, such as health, where progress
in governance has been achieved. Notwithstanding, within the amorphous
contexts of administrative data ± which implicates all data collected by
public authorities in the United Kingdom (not just health) ± the concerns are
distinctive and must be recognized as such. These experiences make the
United Kingdom a suitable case study for analysis of wider issues affecting
many countries.
The government has made significant investment in the development of
streamlined, safe, and ethical processes for drawing upon the existing
resource of administrative data held across the United Kingdom's four
countries. The Economic and Social Research Council (ESRC) invested £34
361
2 S. Koonin and M. Holland, `The Value of Big Data for Urban Science' and R.
Goerge, `Data for the Public Good: Challenges and Barriers in the Context of
Cities', both in Privacy, Big Data, and the Public Good: Frameworks for
Engagement, eds. J. Lane et al. (2014); N. Kshetri, `The Emerging Role of Big Data
in Key Development Issues: Opportunities, Challenges, and Concerns' (2014) 1 Big
Data & Soci ety, at
abstract>; Involve, `Summary of Civil Society and Public Sector Policy Discussions
on Data Use in Government' (2014), at
summa ry-of- civil -soci ety-a nd-pu blic- sector -poli cy-di scuss ions- on-da ta-use -in-
government/>
3 Big data are often considered in terms of volume, velocity, and variety ±
organizations collect vast sums of data from a myriad of sources (volume); big data
`techniques' refer to the technologies underpinning big data such as sensors,
database technologies, search engines, data mining, machine learning, statistics, and
so on, which allow `extremely large data sets [to] be analysed computationally to
reveal patterns, trends, and associations, especially relating to human behaviour and
interactions': SAS, `What Is Big Data?', at
data/what-is-big-data.html>; Koonin and Holland, op. cit., n. 2, p. 137.
4 Information Commissioner's Office (ICO), `Big Data and Data Protection' (2014),
at k/media/for-organi sations/documents/1 541/big-data-and-da ta-
protection.pdf>; Law Commission, `Data Sharing Between Public Bodies ± A
Scoping Report' (2014), at
sharing.pdf>; D. Cameron et al., `Dialogue on Data: Exploring the Public's Views
on Using Administrative Data for Research Purposes', Ipsos MORI report (2014);
Goerge, op. cit., n. 2.
5 Anonymization or methods of de-identification are technical solutions and not
ethical ones.
ß2016 The Author. Journal of Law and Society ß2016 Cardiff University Law School
million in October 2013 to establish the United Kingdom Administrative Data
Research Network (ADRN).
6
This established Administrative Data Research
Centres (ADRC) in Scotland, Wales, England, and Northern Ireland, each of
which would `. .. facilitate linkage of routinely collected administrative data,
th er eb y st im ul at i ng o pp or tu ni ti es f o r in no va ti ve r es ea rc h a nd
policymaking.'
7
A key feature of the ADRN is the creation of a single
governance structure to streamline robust decision making to access, use, and
link administrative data for publicly funded research ± a governance structure
that approves projects based on their being feasible, viable, and ethical, with a
clear potential public benefit.
8
To date, however, there is no gold standard for
the governance of administrative data in the United Kingdom.
9
Indeed, the public sector operates within a widely documented `culture of
caution' surrounding the retention and use of administrative data, where
concerns are fuelled not by the law or actual procedures of data sharing but
rather `. . . the perceptions of risk by all parties that will come from actually
attempting to do so'.
10
The culture of caution can be understood as a culture
of `indecision' within the public sector, with the parameters for lawfully and
ethically engaging in data sharing being shrouded in confusion.
11
Two
studies, in particular, provide a strong body of evidence on the extent and
pervasiveness of this culture of caution.
12
The first study is Thomas and
Waldport's 2008 `Data Sharing Review Report' whose recommendations are
based on wide consultation and extensive engagement with data custodians
across public and private sectors, and with professionals and groups
(including researchers, healthcare providers, academics, and so on) with
direct experience of facilitating data sharing.
13
Thomas and Walport found
362
6 Administrative Data Research Network (ADRN), `About Us' (2015), at
adrn.ac.uk/about>
7 ESRC, `The Big Data Family Is Born ± David Willetts MP Announces the ESRC
Big Da ta Ne two rk' ( 201 3), at htt p:/ /ww w.es rc. ac. uk/n ews -ev ent s-an d-
publ icat ions /new s/ne ws-i tems /the -big -dat a-fa mily -is- born -dav id-w ille tts- mp-
announces-the-esrc-big-data-network/>.
8 ADRN, `Administrative Data Research Network ± FAQs ± Is the Service Provided
for Commercial Use?', at .
9 Compare successes in the health sector such as the Scottish Health Informatics
Programme. G. Laurie and N. Sethi, `Information Governance of Use of Health-
Related Data in Medical Research in Scotland: Towards a Good Governance
Framework' (2012) 1, at p.ac.uk/sites/defa ult/files/Reports/
Working_Paper_2.pdf>; Edinburgh Law School, `Research Case Study: Good
Governance for the Scottish Health Informatics Programme (SHIP)' (2015), at
.
10 R. Thomas and M. Walport, `Data Sharing Review Report' (2008) para. 6.21, at
tp: //w ww. con nec tin gfo rhe alt h.n hs. uk/ sy ste msa nds erv ice s/i nfo gov /li nks /
datasharingreview.pdf>.
11 id., p. 54.
12 id.; Law Commission, op. cit., n. 4.
13 Thomas and Walport, id., pp. 11±12.
ß2016 The Author. Journal of Law and Society ß2016 Cardiff University Law School
the confusion surrounding data sharing, and in particular around legal re-
quirements, particular ly destabil izing in the pub lic sector co ntext.
14
However, the consultation revealed few examples where the legal framework
actually prohibited data sharing, rather:
The barriers, therefore, are most often cultural or institutional ± an aversion to
risk, a lack of funds or proper IT, poor legal advice, an unwillingness to put the
required safeguards in place or to seek people's consent.
15
The chief recommendation from this report was `to transform the culture that
influences how personal information is viewed and handled',
16
further
emphasizing the `[failings] within institutions themselves [that] often stand
in the way of appropriate information sharing'
17
as opposed to the law.
More recently, the English Law Commission undertook a similarly large-
scale consultation of public authorities'
18
experience and understandings of
data sharing in practice and their interpretation of the law. The objective was
to identify existing impediments to data sharing which could be addressed by
legal reform.
19
The Law Commission found the laws governing data sharing
as overly complex and difficult to understand, perpetuating legal myths
about what is prohibited or allowed and resulting in inconsistent inter-
pretations between public authorities. A primary recommendation was
therefore to begin a full law reform project to `. . . map, modernise, simplify
and clarify the statutory provisions that permit and control data sharing and
review the common law.'
20
However, like Thomas and Walport, the Law
Commission recognized the impact of organizational cultures around data
and the limits of law reform:
Law reform alone will not provide the necessary solutions, but law reform can
work together with and assist changes in culture and practice, for example by
developing structures which facilitate good and flexible working relationships
in local areas.
Within this `data' landscape dominated by a culture of caution, even if
public authorities want to share data for research facilitated by the ADRN,
they may lack the necessary decision-making tools to distinguish between
actual barriers (for example, where data sharing would be unlawful) and
perceived barriers which can be resolved through internal changes to
governance and/or dialogue with stakeholders such as staff, managers, and
the public. While clarification and simplification of the law would enhance
currently confused misunderstandings of what is or is not legally required to
363
14 id., p. 38.
15 id., p. 46.
16 id., p. 53.
17 id., p. 46.
18 As did the Thomas and Walport report, the Law Commission's report received
consultation responses from private bodies engaged in public service delivery.
19 Law Commission, op. cit., n. 4, para. 1.12.
20 id., para. 1.6.
ß2016 The Author. Journal of Law and Society ß2016 Cardiff University Law School
share data, this article argues strongly that additional laws cannot resolve the
residual cultural and ethical issues which are unique to each public authority.
These might include:
·individual and organizational reluctance to share data due to perceptions
of `ownership' of data;
·fears of public backlash over lawful but novel and ethically controversial
uses of data;
·fears over reputation damage to the public authority;
·a lack of incentives, or understanding of incentives, to share data;
·a lack of clear accountability structures for data handling or senior
leadership involvement; or
·
a lack of formal arrangements between public authorities for sharing data.
21
The first step towards removing unnecessary obstacles to data sharing and
transforming the culture around data in public authorities is to identify real
versus perceived barriers, directly confronting ethical dilemmas and pre-
conceived notions as to the risks, incentives, and disincentives involved with
data sharing. Public authorities in the United Kingdom are currently lacking
any uniform or objective approach to distinguishing in order to undertake
these tasks.
To address this unmet need, and as a crucial first step towards the
development of a good governance framework for administrative data, this
article introduces a tool ± a decision-making template ± that consolidates the
complex legal and ethical considerations at stake when data custodians must
decide whether to permit use of administrative data for research purposes.
The template provides a concise and generalizable illustration of the key
legal, ethical, and cultural scenarios data custodians may face when con-
sidering granting access to administrative data for research, or indeed for any
other purpose not contemplated at the time of data collection. The template
facilitates an assessment of preparedness for data sharing vis-aÁ-vis the
identification of real (versus perceived) barriers operating within their
organization. In this respect, it is a form of maturity model. Importantly, the
template allows data custodians to distinguish between different types of
considerations, helping to reveal when ± or whether ± law is the `problem'.
As explained below, existing research and our engagement with stakeholders
to date reveal that most often data custodians are impeded by cultural-
organizational barriers that prevent proportionate data use.
22
By exposing
364
21 An amalgamation of `cultural' barriers taken from the Law Commission and
Thomas and Walport report. Thomas and Walport, op. cit., n. 10, pp. 41±42, 46±48;
Law Commission, op. cit., n. 4, paras. 7.1±7.9.
22 Thomas and Walport, id., pp. 46±8; The Law Commission, id.; M. Oswald, `Share
and Share Alike? An Examination of Trust, Anonymisation and Data Sharing With
Particular Reference to an Exploratory Research Project Investigating Attitudes to
Sharing Personal Data With the Public Sector' (2014) 11 SCRIPTed, at
script-ed.org/wp-content/uploads/2014/12/oswald.pdf>; R. Wilson and A. Gray,
ß2016 The Author. Journal of Law and Society ß2016 Cardiff University Law School
this key issue, we aim to shift the focus from partial legal solutions to
emphasize where real and meaningful changes can be made: to organiza-
tional culture through commitment to a public interest mandate.
The next section provides an overview of the piecemeal legal landscape
governing administrative data in the United Kingdom. It reveals the irony
that any `sharing stasis' might be driven by an absence of law as much as by
an inability to navigate existing law. The third section offers a deeper
examination of likely concerns, leading to the development of the decision-
making template that sits at the heart of this contribution. The fourth and
final section concludes with the suggestion that the adoption of a public
interest mandate through a series of key steps can help to initiate the neces-
sary change, not only in the United Kingdom's public sector but potentially
also beyond. We advocate that such a mandate ought to be reflected in an
organiz ation's m ission st atemen t, underp inned by pr inciple s of the
promotion of the public interest and proportionality and supported by
meaningful engagement with relevant stakeholders, including publics.
DEALING WITH ADMINISTRATIVE DATA IN A DATA-DRIVEN
WORLD
Administrative data have no single, let alone overarching, legal definition.
Absent this, and without agreement on which bodies perform `functions of a
public nature',
23
for this article we adopt the definition provided by the ADRN:
Administrative data refer to the vast range of information collected by public
authorities in the course of their routine operations. This includes delivery of
services such as census taking, managing income tax payments, providing
child protection, performing health services, running schools, monitoring
crime, allocating benefits, mediating property owner and tenant disputes,
administering council housing and so forth.
24
365
`Information Sharing: Easy to Say Harder to Do Well' (2015), at
inform ation shari ng.org .uk/w p-cont ent/u pload s/2015 /06/P 0248-C oE-Ac ademi c-
report.pdf>.
23 The meanings of a `public authority' or `functions of a public nature' remain unclear
where many services are now outsourced to the private sector. The implications of
being considered a hybrid-public authority in performing functions of a public
nature is a critical consideration given the resulting obligations under the European
Convention on Human Rights, most notably, as to Article 8 and the Right to Respect
for Private and Family Life. The precedent remains from Y.L. v. Birmingham City
Council and others [2007] U.K.H.L. 27, in which a bright-line rule between public/
private was rejected in favour of a list of considerations on a case-by-case basis.
However, for the purposes of this article, the focus remains on the category of public
authorities as defined under the Data Protection Act 1998, s. (1)(1) in reference to
the Freedom of Information Act 2000 Schedule 1 and Freedom of Information
(Scotland) Act 2002.
24 Adapted from the ADRN's definition, `Administrative Data' (2015), at
adrn.ac.uk/admin-data>.
ß2016 The Author. Journal of Law and Society ß2016 Cardiff University Law School
Therefore, the scope of our contribution applies to the myriad of local,
regional, and national public authorities (as defined under the Data Protec-
tion Act 1998 (DPA) s. 1(1)) which collect administrative data from their
regular interactions with citizens. This includes any public authority as
provided by Schedule 1 of the Freedom of Information Act 2000 (FOIA
2000) or a Scottish public authority as defined by the Freedom of
Information (Scotland) Act 2002 (FOISA 2002),
25
including local authorit-
ies, govern ment depar tments, un iversitie s, and so on. Fu rthermore ,
reflecting our work with the ADRN, the focus is on the use of admini-
strative data for research purposes where currently data are not being
accessed from private sector organizations nor being made accessible to
such organizations for their own research. In contrast, if private entities are
acting on behalf of public authorities, then further distinctions must be
made. For example, if private sector organizations merely process and
`hold' data on behalf of public authorities, they will not incur data controller
obligations;
26
as such, they would be excluded from considerations here.
27
Contrariwise, if private organizations have been explicitly declared `public
authorities', for example, for the purposes of freedom of information, then
the following analysis would apply. As an example, in 2015, Scotland
extended their category of public authorities to include organizations funded
wholly or partly by local authorities which carry out functions of a public
nature ± such organizations would clearly come within the remit of this
article and analysis.
28
366
25 Recently the FOISA 2002 definition of public authorities was extended to include
organizations funded in part or wholly by local authorities in Scotland, which carry
out functions of a `public nature' including, for example, tourism-related activities,
museums and art galleries, recreational facilities, and so on: Freedom of Information
(Scotland) Act 2002 (Designation of Persons as Scottish Public Authorities) Order
20 13 N o .2 78 , a t tt p :/ /w w w. le g is la t io n. g ov .u k /s si / 20 13 / 27 8/ p df s/
ssi_20130278_en.pdf>.
26
FOIA, s. 3(2); FOISA, s. 3(2); ICO, `Outsourcing and Freedom of Information ±
Guidance Document ± Freedom of Information Act' (2015) 17±19, at
ico.or g.uk/ media/ 104353 0/out sourci ng-an d-free dom-o f-info rmatio n.pdf >; ICO,
`Data Controllers and Data Processors: What the Difference Is and What the
Governance Implications Are: Data Protection Act' (2014) 8±9, at
media/for- organisati ons/docume nts/1546/d ata-contro llers-and- data-proc essors-dp-
guidance.pdf>.
27 The ICO gives an example where a public authority (created by statute) outsources
complaints handling to another public authority. The ICO advises that the public
authority with the statutory obligation to carry out the processing remains the data
controller's as to data processed by the outsourced party. Importantly, `[a]n
organisation cannot be both data controller and processor for the same data
processing activity; it must be one or the other.' ICO, id. (2014), pp. 8±9, 14±15.
28 The Freedom of Information (Scotland) Act 2002 (Designation of Persons as
Scottish Public Authorities) Order 2013 No. 278. Scotland is considering further
expanding this category to include `various organisations, including private prison
contra ctors, pr ovider s of secur e accomm odation , grant- aided sc hools an d
ß2016 The Author. Journal of Law and Society ß2016 Cardiff University Law School
Mirroring the myriad types of public authorities, there are many distinct
categories of administrative data. Examples include health data, which are
defined by reference to a particular health status of an individual and often
tied to a particular health service provider, and benefits and tax data,
whereby individuals may be traced by reference to a National Insurance
number and are connected to a wider set of interactions with various public
authorities including Her Majesty's Revenue and Customs (HMRC) and the
Departme nt of Work and P ensions (D WP). This ar ticle appl ies to
administrative data in its widest sense covering all potential aspects of
people's lives and their interactions with public authorities.
1. The legal landscape governing administrative data in the United Kingdom
Normally, the p otential priv acy impact on cit izens is the prin cipal
consideration in any decision to use personal data. To this end, the law
regulates the processing of identifiable personal data under the Data
Protection Act 1998 (DPA).
29
Personal data are:
. . . data which relate to a living individual who can be identified Ð
(a) from those data, or
(b) from those data and other information which is in the possession of, or is
likely to come into the possession of, the data controller, and includes any
expression of opinion about the individual and any indication of the intentions
of the data controller or any other person in respect of the individual.
30
The objective of the legislation is to regulate how personal data are pro-
cessed, which in practice means any storage or use of personal data whatso-
ever. Moreover, if personal data are `sensitive' according to the law then
additional justifications are required, and further restrictions on processing
can apply. Sensitive personal data include data relating to an individual's
health, religious beliefs, political opinions, and/or racial/ethnic origin and
thus can encapsulate many forms of administrative data. Personal data and
sensitive personal data are distinguishable from de-identified data, which are
by definition no longer identifiable to an individual, either directly or
indirectly, and as such are not regulated under the DPA. The crucial
consideration, therefore, is whether data remain identifiable, directly or
indirectly. Importantly, this is a question of judgement and degree. In the
United Kingdom, it is recognized best practice to consider data which are de-
identified to the point where the risk of re-identification is only a remote
possibility. However, what is c onsidered sufficiently de-ide ntified or
anonymized remains a contentious issue within the United Kingdom and
367
independent special schools': Scottish Government, `Further Extension of Coverage
of the Freedom of Information (Scotland) Act 2002 to More Organisations' (2015),
at .
29 The Data Protection Act 1998. (Hereinafter `DPA')
30 DPA, s. 1(1).
ß2016 The Author. Journal of Law and Society ß2016 Cardiff University Law School
Europe, particularly with new perspectives being offered in the General Data
Protection Regulation that will replace existing data protection legislation in
2018.
31
Moreover, and irrespective of any legal reform, re-identification is
always a future possibility when existing data are shared or linked.
32
This
requires that the status of data must always be kept under review.
Within this regulatory framework, and given the crucial consideration of
whether data are de-identifiable, it is important to consider how administra-
tive data fit within this scheme. This article focuses on the reuse of de-
identified administrative data for research , particularly social science
research under the auspices of the ADRN.
33
The ADRN only facilitates access to administrative data available in
digitized form given the security arrangements for access in secure facili-
ties.
34
Under ADRN arrangements, data are de-identified and made available
for research via a `trusted third party' mechanism.
35
The ADRN details the
robust process through which trusted third party mechanisms make de-
identified data available to researchers.
36
While acknowledging wider and
on-going debates regarding de-identification,
37
it is submitted that ADRN's
368
31 Data Protection Directive 95/46/EC, recital 26 (hereinafter `DPD'); P. Ohm,
`Brok en Prom ises o f Priv acy: Re spond ing to t he Sur prisi ng Fail ure of
Anonymization' (2009) 57 UCLA Law Rev. 1701; Article 29 Data Protection
Working Party, `Opinion 05/2014 on Anonymisation Techniques' (2014), at
ec .e ur op a. eu / ju st ic e/ da ta - pr ot ec ti on /a r ti cl e- 29 /d oc u me nt at io n/ op i ni on -
recommendation/files/2014/wp216_en.pdf>; ICO, `Anonymisation: Managing Data
Protection Risk Code of Practice' (2012) 6, at
anonymisation-code.pdf>; L. Stevens, `The Proposed Data Protection Regulation
and Its Potential Impact on Social Sciences Research in the UK' (2015) 1 European
Data Protection Law Rev. 100.
32 Ohm, id.; A. Narayanan and V. Shmatikov, `De-Anonymizing Social Networks'
(2009) 30th IEEE Symposium on Security & Privacy, at
~shmat/shmat_oak09.pdf>; P. Schwartz and D. Solove, `The PII Problem: Privacy
and a New Concept of Personally Identifiable Information' (2011) 86 New York
University Law Rev. 1814; M. Gymrek et al., `Identifying Personal Genomes by
Surname Inference' (2013) 339 Science 321, at
content/339/6117/321.abstract>.
33 For ADRN, see `Approved Projects' (2015), at
approved-projects>.
34 The DPA applies to manual and digitized records, so long as they are part of a
`relevant filing system' and structured `. . . either by reference to individuals or by
reference to criteria relating to individuals, in such a way that specific information
relating to a particular individual is readily accessible' ( s. 1).
35 As provided by the ADRN: `A trusted third party is an organisation with secure
facilities for matching data': ADRN, `Trusted Third Parties', at
protecting-privacy/de-identified-data/trusted-third-parties>; Stevens, op. cit., n. 31,
p. 100.
36 ADRN, id. (emphasis added).
37 Ohm, op. cit., n. 31; Narayanan and Shmatikov, op. cit., Schwartz and Solove, op.
cit., Gymrek et al., op. cit., all at n. 32; Wellcome Trust, `Research Funders Outline
Steps to Prevent Re-Identification of Anonymised Study Participants' (2014), at
ß2016 The Author. Journal of Law and Society ß2016 Cardiff University Law School
arrangements would meet standard best practice whereby effective de-
identification need not be risk free (which is arguably unattainable) but
instead ensure that re-identification is beyond a remote possibility.
38
The on-going challenge for data controllers, however, is whether they are
willing and ready to subject their data to such processes. ADRN provides a
means to navigate data protection law safely and responsibly, but the
infrastructure requires the confidence and the data of public authorities in
order to do so.
(a) Beyond data protection: flexibilities and inflexibilities in the law
governing administrative data
Beyond data protection, public authorities must also consider a broad range
of laws governing their activities.
39
Such administrative laws prescribe the
purposes to which administrative data can be put either expressly by statute
(laws often termed as `gateways'), impliedly, or under common law. A
public authority created by statute, such as the HMRC, does not have
authority outside what is provided in its governing legislation ± it must have
implied or explicit statutory authority to act and therefore to use or share
data.
40
This is in contrast to government departments, such as DWP, which
has authority from both applicable legislation and from common law. The
DWP, in fact, has over 63 legislative `gateways' to share data with others.
Importantly, for both statutorily created public authorities and government
departments,
41
these administrative laws apply regardless of whether or not
data are `identifiable' under the DPA.
42
Thus, when considering whether
administrative data can be shared, it is common practice first to consider the
class or kind of entity that holds the data because this determines the legal
basis upon which the public authority may act. In this way, the standard set
of considerations is inverted within the public sector: here it is public
369
tt p :/ /w w w. we l lc om e .a c. u k/ Ne w s/ Me d ia -o f fi ce / Pr es s -r el e as es / 20 14 /
WTP055974.htm>; S. Barocas and H. Nissenbaum, `Big Data's End Run Around
Anonymity and Consent' in Privacy, Big Data, and the Public Good: Frameworks
for Engagement, eds. J. Lane et al. (2014).
38 ICO, op. cit., n. 31, p. 6; Stevens, op. cit., n. 31, p. 100.
39 The notion of `public authorities' vis-aÁ -vis quasi-public bodies (that carry out
`functions of a public nature') is subject to the precedent in Y.L. v. Birmingham City
Council and others, op. cit., n. 23.
40 The Law Commission provides a case study analysing the difference between
statutorily-created public authorities such as HMRC and government departments
such as the DWP using common law powers: Law Commission, op. cit., n. 4, pp.
118±46, 148±62.
41 Other public authorities with powers under common law include Ministers of the
Crown and Parliament, and in Scotland, Scottish Ministers via the Scotland Act
1998, s. 53.
42 Ministry of Justice, `The Data Sharing Protocol: Annex H, Legal Guidance on Data
Sharing' (201 2), at ww.justice.g ov.uk/downlo ads/informa tion-access-
rights/data-sharing/annex-h-data-sharing.pdf>.
ß2016 The Author. Journal of Law and Society ß2016 Cardiff University Law School
permission first, privacy second. This has important cultural-organizational
consequences.
Statutory powers, whether explicit or implicit, should be considered in
contrast to common law powers, which are reserved for Ministers of the
Crown, the Westminster parliament and government departments in the
United Kingdom. Often seen as the `third source' of power, common law is
understood `[a]t its widest . . . as the Crown having all the capacities and
powers of a natural person, subject to the ordinary law and limited to the
extent that there is express statutory provision.'
43
However, in practice, this
concept, in part promulgated by the `Ram Doctrine',
44
is limited by
uncertainty as to the extent of common law powers and the lack of precedent
(and thus confidence) to rely upon them to justify data sharing. In contrast,
common law powers may be more flexible because they obviate the need to
point to express statutory powers to legitimate use of data. In some hybrid
cases, such as the DWP, legal complexity arises because of the confusing
mix of statutory and common law powers, with the added dimension of
whether there are explicit, implicit or silent measures addressing admini-
strative data use. Importantly, the Law Commission concluded from its
consultation with various public authorities that the absence of express (and
obligatory) authority makes it unlikely administrative data will be used.
45
It follows from the above that interpretation of statutory and common-law
powers has a crucial impact upon whether authority to use administrative
data is understood ± or perceived ± to be present. Yet more complexity is
added when considering the role of guidance provided to public authorities
on data sharing. This might not hold the force of law but nevertheless has
crucial influence over data practices. Consider the government's guidance on
the use and sharing of social security data between the DWP and local
authorities: this provides that local authorities `. .. must not act outside their
statutory powers' and `[a]ccount must be taken of any specific data sharing
legislation that applies to the function being undertaken'.
46
This guidance
generates uncertainty as to what extent administrative data linkage can be
permitted under an implied statutory power as opposed to an explicit
statutory power.
Crucially, it is a general principle of law that silence is permissive: that
which is not prohibited is not unlawful. Practically, however, silence and
non-explicit frameworks (or co-existing frameworks) create uncertainty and,
often, stasis. This is compounded by the juxtaposition of examples where
express legal power to share is given. It is understandable, therefore, that this
370
43 Law Commission, op. cit., n. 4, p. 85.
44 400 H.C. Debs., col. WA12 (25 February 2003).
45 Law Commission, op. cit., n. 4, p. 76.
46 DWP, `Data Sharing Guidance for Local Authorities on the Use of Social Security
Data' (20 10), at ttps:/ /www.g ov.uk/g overnm ent/pub licati ons/da ta-shar ing-
guidance-for-local-authorities>.
ß2016 The Author. Journal of Law and Society ß2016 Cardiff University Law School
might lead to an inference of illegality or impermissibility in all other
contexts. But it is important to understand that this is a matter of perception
and cultural resistance, not necessarily an issue of law. Nevertheless,
uncertainty surrounding the law in this area is inescapable and it perpetuates
cautious decision making and inhibits organizational cultural practices, as
pointed out by numerous notable sources.
47
Furthermore, evidence also
suggests that even with the weight of express statutory powers, mere legal
permission to reuse data is not enough to outweigh data custodians' hesit-
ancy to undertake any perceived risks or other costs associated with admini-
strative data use or sharing.
48
This highlights a key distinction between law
operating to mandate sharing, and law leaving open an opportunity to
facilitate sharing.
Nevertheless, a general and permissive legal gateway to share data by the
government has been mooted in three discrete areas (including for research
and statistics),
49
but this remains at the development stage.
50
The problem in
the meantime is two-fold: (i) how should data custodians proceed in
navigating the current landscape? and (ii) more broadly, should we continue
to imagine that a legal solution could deliver a complete solution to the
challenges that decision-makers face? In the next section, we briefly answer
`no' to the second of these questions, before proceeding to offer a more
detailed answer to the first.
FACTORS INFLUENCING ADMINISTRATIVE DATA DECISION
MAKING
We have recently argued elsewhere in the context of health data (as a sub-set
of administrative data), and in the particular context of the care.data debacle
in England, that complexities surrounding decision making to support any
371
47 Thomas and Walport, op. cit., n. 10; Academy of Medical Sciences, `A New
Pathway for the Regulation and Governance of Health Research' (2011), at
www.a cmeds ci.ac .uk/d ownlo ad.p hp?fi le=/i mages /proj ect/1 3073 49574 23.pd f>;
Scottish Government, `Joined up Data for Better Decisions: A Strategy for
Improving Data Access and Analysis' (2012), at
Publications/2012/11/4166/0>; Law Commission, op. cit., n. 4
48 Law Commission, id., pp. 79±81.
49 The proposed legislation would only affect data sharing by United Kingdom
government authorities and would not affect data sharing implicating devolved areas
such as health, education or crime in Scotland or Northern Ireland. See Digital
Government Review, `Making Digital Government Work for Everyone' (2014) 38,
at ://dig italgo vernm entrev iew.re adandc omment .com/ wp-con tent/u ploads /
2014/1 1/EMBAR GOED_CO NFIDEN TIAL_M ASTER-F inal-R eport- 2014112 4_
CLEAN.pdf>.
50 Contrast the effort of Involve, `Data Sharing: Updates from Civil Socie ty
Engagement with the UK Government on Data Sharing' (2016), at
datasharing.org.uk/latest/>.
ß2016 The Author. Journal of Law and Society ß2016 Cardiff University Law School
particular data initiative relate as much to addressing the social legitimacy of
the enterprise as to giving it a basis in law.
51
Care.data was perfectly lawful
under the Health and Social Care Act 2012. This made no difference to the
social reaction that ensued. In other words, experience tells us that while law
might be necessary it is unlikely to be sufficient to address the complexities
adequately.
Rather, in determining how to proceed in the realm of data reuse, we posit
that a crucial first step is to identify the kinds of challenges being faced. To
assist data custodians in doing so, we introduce a novel decision-making
template as an analytical framing device to help understand the legal, ethical,
and organizational concerns at stake with such decisions (see Figure 1).
Understanding the nature and full scope of concerns is integral to developing
a proportionate good governance framework. Equally, it is vital to distin-
guish between real and perceived challenges and to assess the state of
preparedness of an institution to engage in robust data reuse. More par-
ticularly, this template offers three important contributions to current
discourse in information governance of administrative data:
i) It clarifies which considerations are purely legal, a mixture of legal
and ethical, and/or related to organizational culture;
ii) It exposes those concerns which are perpetuated by legal myths or
perceived controversy as opposed to those that can be adequately
addressed within a proportionate governance framework for admini-
strative data;
iii) It expressly acknowledges an often-neglected element, namely, the
problematic organizational and behavioural dynamic around data.
The template works by assisting a decision maker to reflect on and
categorize underlying reasons that might underpin any decision not to use
and/or share administrative data.
1. Approach
The origins of the template are found in our current research undertaken for
ADRC Scotland, involving engagement with researchers and data custodians
to understand existing barriers to administrative data usage. We lead the
legal work package of ADRC Scotland with the aim and objective of
exploring `. . . possible sector and country-specific solutions for delivering
[interoperable information governance] involving administrative data at the
local, national, European and international level.'
52
372
51 P. Carter et al., `The Social Licence for Research: Why Care.data Ran into Trouble'
(2015) 41 J. of Medical Ethics, at
medethics-2014-102374.abstract>.
52 ADRN, `Our Internal Work Packages' (2015), at
centre-scotland/work-packages> (emphasis added).
ß2016 The Author. Journal of Law and Society ß2016 Cardiff University Law School
From the outset, the nascent nature of the administrative data research field
was acknowledged and we were initially guided by their previous work under
the Information Governance Work Package to the Scottish Health Informatics
Programme (SHIP).
53
This focused on the reuse of health data for research in
the public interest and was taken as a legitimate example of a sub-set of
administrative data. The original SHIP research and engagement with data
controllers and the researcher community
54
provided an experiential and
informed basis for analysing the broader culture surrounding administrative
data use across public sectors in Scotland and the United Kingdom,
55
as well
373
Figure 1. Administrative data decision-making template
53 Laurie and Sethi, op. cit., n. 9; Scottish Government, `Joined-Up Data For Better
Dec isi ons : Gu idi ng Pr inc ip les F or Da ta L ink age ' (20 12 ), at htt p: //
www.scotland.gov.uk/R esource/0040/00407739 .pdf>; N. Sethi and G. Laurie ,
`Delivering Proportionate Governance in the Era of eHealth: Making Linkage and
Privacy Work Together' (2013) 13 Medical Law International 168, at
mli.sagepub.com/content/13/2-3/168.abstract>.
54 M. Aitken et al., `Public responses to the Scottish health informatics programme:
preferences and concerns around the use of personal medical records in research'
(2011) 65 J. of Epidemiology and Community Health A27; M. Aitken, `What makes
research/researchers trustworthy? Workshop Report' (2011), at
ship.ac.uk/sites/def ault/files/Reports/Wha t_makes_researchers _ trustworthy.pdf>;
M. Aitken, `Your Data and Health Research: SHIP Public Workshops' (2012), at
/www.sc ot-ship .ac.uk/ sites/d efault/ files/ Reports /Your_D ata_and _Health _
Research.pdf>.
55 Information Governance Working Group to Scottish Health Informatics Programme
(SHIP), `SHIP Guiding Principles and Best Practices' (2010), at
ship. ac.uk /sit es/de fault /fil es/Re ports /Guid ing_P rinc iples _and_ Best _Prac tices _
221010.pdf>; G. Laurie and N. Sethi, `Information Governance Of Use Of Health-
Related Data In Medical Research In Scotland: Current Practices And Future
Scenar ios' (201 1), at ttp://w ww.sco t-ship .ac.uk /sites /defaul t/file s/Repo rts/
Working_Paper_1.pdf>; Scottish Government, op. cit., n. 53; Laurie and Sethi,
ß2016 The Author. Journal of Law and Society ß2016 Cardiff University Law School
as providing theoretical
56
and practical grounding
57
in a principles-based
approach to governance.
This was complemented by work outwith the health context and through
novel doctrinal research situated within data protection, administrative law,
and information governance literatures.
58
Important contributions include
the Thomas and Walport Data Sharing Review (2008), the Administrative
Data Taskforce (2012), Law Commission (2014), Ipsos Mori (2014), and the
Centre of Excellence for Information Sharing (2015).
59
To enrich this analysis, in 2014, we conducted a series of nine informal,
semi-structured interviews with ADRC Scotland researchers across social
science disciplines with diverse experiences in obtaining access to admini-
strative data in Scotland and the United Kingdom. Interviewees were
selected based on purposive or selective sampling, as we identified the
ADRC Scotland co-investigators as an accessible constituency with relevant
experience, and thus approached the interviews with this particular purpose
in mind.
60
Interviewees were broadly asked: (i) what were their experiences
in seeking approval for accessing administrative data in their respective
research areas?; (ii) which safeguards were put in place to protect the
privacy/confidentiality of data?; (iii) how data were to be accessed (remotely
or in a secure setting)?; and (iv) whether they expected or experienced any
particular legal or ethical obstacles in obtaining access in light of previous
experiences?
Out of a pool of thirteen co-investigators we conducted nine informal
interviews including researchers involved in: computer sciences research;
geographical research; informatics; social health and welfare research; social
374
op. cit., n. 9; Sethi and Laurie, op. cit., n. 53; `ScottisH Informatics Programme
(SHIP)', at .
56 R. Jackson, `Principles versus Rules' (2004) 61 Internal Auditor 58; S. Arjoon,
`Striking a Balance Between Rules and Principles-Based Approaches for Effective
Governance: A Risks-Based Approach' (2006) 68 J. of Business Ethics 53, at
; J. Black, `The Rise, Fall and Fate
of Principles Ba sed Regulatio n' (2010) 17, at tp://papers. ssrn.com/sol 3/
papers. cfm?abst ract_id =171286 2&rec=1& srcabs= 1267722 &alg=1&p os=1>; D .
Bambauer, `Rules, Standards, and Geeks' (2011) 5 Brooklyn J. of Corporate
Finance & Commercial Law 49.
57 Based on original research undertaken for SHIP. Laurie and Sethi, op. cit., n. 9 and
op. cit., n. 53.
58 The key literature cited in G. Laurie and L. Stevens, `The Administrative Data
Research Centre Scotland: A Scoping Report on the Legal & Ethical Issues Arising
from Access & Linkage of Administrative Data' (2014), at
sol3/papers.cfm?abstract_id=2487971>.
59 Thomas and Walport, op. cit., n. 10; Administrative Data Taskforce, `The UK
Administrative Data Research Network: Improving Access for Research and Policy'
(2012), at
and-Policy_tcm8-24462.pdf>; Law Commission, op. cit., n. 4; Cameron et al., op.
cit., n. 4; Wilson and Gray, op. cit., n. 22.
60 ADRN, op. cit., n. 52.
ß2016 The Author. Journal of Law and Society ß2016 Cardiff University Law School
work research; informal care research; and transport research. The responses
can be divided into three themes: (i) those with previous (and successful)
experience in obtaining access to administrative data; (ii) those focused on
developing the technology to allow the linkage of administrative data for
research; (iii) those with previous, unsuccessful attempts at accessing
administrative data.
For one group of co-investigators, previous access had been relatively
simple, so long as informed consent was obtained, therefore their concerns
were more practical ± for example, how would they navigate access from
each local authority in Scotland (each with its own process) and how would
this be different from access in the health sector? Their previous experiences
were, in fact within the Scottish health sector, which has more prescribed and
established frameworks for facilitating access to personal data for research.
One other co-investigator had similar and straightforward experiences in the
health context but would be seeking new forms of administrative data (for
example, census data) but raised no particular concerns regarding access. A
third co-investigator had international experience in successfully accessing
administrative data and again raised no particular concerns regarding access
to such data in the United Kingdom.
A group of four co-investigators were separately involved in technological-
focused projects that would develop innovative data-linkage techniques for
research purposes. Two out of four raised concerns as to commercial and
proprietary rights to administrative data, which could raise a cost barrier to
access. One raised concerns over physical access to the required data, that is,
would they be able to run their required software on data from their own
computers or would they have to visit the public authority in question? This
same co-investigator raised issues as to legal and ethical obstacles ± would data
they receive already be sufficiently anonymized so as to be outwith the scope
of data protection, or would they need to `clean' the data from the public
authority? If the latter, what would their liability be? A fourth and final
`technical' co-investigator raised concerns over the legal position on accessing
historical administrative data, for example, where consent may or may not have
been obtained at the time of collection and the future use of such data for
research was not contemplated or thus communicated to individual research
participants.
Two co-investigators had previous, unsuccessful experience in obtaining
access to administrative data. One had experienced this at the research
approvals phase, whereby the application was not approved due to concerns
over the potential effect and impact on individuals whose data were to be
accessed, even though data would be anonymized and re-identification or
contact with individuals would be prohibited. The prospect of undergoing yet
another application and approvals process was considered daunting and the
research project would have to be scaled back. The second co-investigator
had experienced protracted negotiations with United Kingdom government
departments whereby it was unclear on what legal basis the public authorities
375
ß2016 The Author. Journal of Law and Society ß2016 Cardiff University Law School
could share data for research purposes if there was no direct benefit or
impact to the public authority.
The responses of ADRC Scotland researchers demonstrate the complexity
and uncertainty surrounding access to administrative data, particularly to
data outwith the health sector. Uncertainty arises in terms of the practicalities
of access, the legal requirements for access, the public authority's basis for
sharing, and researchers' own obligations as to data. We sought to supple-
ment understandings of the researcher experience with administrative data
with the experiences of a Scottish public authority being asked to grant
access to data. Therefore we met with a team of staff at a Scottish public
authority whose administrative data were to be requested by various ADRC
Scotland researchers.
61
The public authority expressed hesitance to share
data even amongst the separate divisions of its organization (which operated
in distinct departmental silos) and further indicated it had little (to no)
experience in granting/managing access to data to external individuals or
organizations. If ADRC Scotland researchers were to apply for access to its
data, the public authority was concerned as to the type of ethical approval
body that would authorize use of its data and thus how a legal and ethical
basis for sharing would be supported; in doing so it acknowledged the lack
formal and established processes for seeking such approval so outwith the
health sector.
Our engagement with researchers and public authorities in Scotland (and
the United Kin gdom) contin ues, providin g current exam ples of the
difficulties faced in obtaining access to data and related observations of
cautious behaviours displayed by data custodians. Furthermore, we are privy
to, and active participants in, ADRN-wide policy discussions on data access
and security arrangements. We work closely with ADRN executives who
lead negotiations with data custodians over access to administrative data and
are regularly consulted on current issues as they arise. These observations
are supplemented by our direct engagement with Scottish data custodians
across the public sector. This engagement notably included a funded
workshop `Sharing Data Across Sectors for the Public Good' with data
custodians, regulators, researchers, and other stakeholders. This event
focused on the interoperability of governance arrangements as a means to
facilitating data sharing for the public good.
62
376
61 Data which are predominantly unrelated to the `health' of individuals.
62 Using a snowball sampling technique to obtain participants, the workshop was held
in Edinburgh in June 2015 and was funded by the University of Edinburgh's
Knowledge Exchange and Impact Funding Scheme. The workshop was carried out
with colleague Nayha Sethi from The Farr Institute @ Scotland, an organization
intended to `harness health data for patient and public benefit by setting the
international standard for the safe and secure use of electronic patient records and
other population-based datasets for research purposes.' The aim of the workshop
was to facilitate open discussion between stakeholders on the challenges and best
practices in sharing and using data across different sectors (for research that serves
ß2016 The Author. Journal of Law and Society ß2016 Cardiff University Law School
Overall our analysis and conclusions, as illustrated in the decision-making
template below, are informed on this theoretical and experiential basis. The
engagement has driven the approach and is an example of co-production of a
maturity model that reflects the challenges that were revealed. Consistent
findings are that the culture of caution is real and remains a persistent barrier
to proportionate governance.
63
It is from this basis that the need for a
decision-making tool arose, one that could identify more clearly the root of
the cautious behaviour exhibited by data custodians across the United
Kingdom's public sector.
2. Decisions, decisions, decisions
As the decision-making template reveals, administrative data custodians are
faced with at least five possible scenarios in deciding whether (or not) to use
data. In the analysis that follows, we unpack what each of these five
considerations means in practice for data custodians.
(a) Uses of data that are simply unlawful
We contend that contrary to the predominant focus of data custodians in the
United Kingdom, represented in a vast body of literature,
64
the apparently
impenetrable legal landscape is not the greatest barrier to the lawful and
ethical use of administrative data.
65
In fact, the law is often quite clear on
whether a proposed use of administrative data is lawful or unlawful. More-
377
the public good). Workshop findings will be published after a follow-up workshop
in June 2016.
63 See previous discussion on the culture of caution in the `Introduction' above.
64
The emphasis on legal barriers is particularly prevalent in public health literature,
although there are clear indications of the former's impact on data sharing more
generally. For example see: C. Warlow, `Over-Regulation of Clinical Research: A
Threat to Public Health' (2005) 5 Clinical Medicine 33; A. Iversen et al., `Consent,
Confidentiality, and the Data Protection Act' (2006) 332 British Medical J. 165;
Thomas and Walport, op. cit., n. 10; H. Snooks et al., `Bureaucracy Stifles Medical
Research in Britain: A Tale of Three Trials' (2012) 12 BMC Medical Research
Methodology 1; R. Al-Shahi Salman et al., `Increasing Value and Reducing Waste in
Biomedical Research Regulation and Management' (2014) 383 Lancet 176; Oswald,
op. cit., n. 22; M. Oswald, `Law Commission Consultation No 214 Data Sharing
Between Publi c Bodies: Respon se to Consultat ion' (2014), at ttp://www.
winch ester .ac.u k/aca demic depar tment s/Law /Cent re%20 for%2 0Info rmati on%20
Rights/Publications/Documents/Law%20Commission%20Consultation%20No%20
214%20Da ta%20Sha ring%20b etween%2 0Public% 20Bodies %20respo nse%20fr o-
m%20Marion%20Oswald.pdf>; Law Commission, op. cit., n. 4; J. Bamford, `Data
Sharing: Dispelling the Myths to Achieve Multi-Agency Information Sharing' (2015),
at ps:/ /ico. org. uk/me dia/ about -the -ico/ even ts-an d-web inar s/104 3650 /data -
sharing-dispelling-the-myths-jonathan-bamford-20150323.pdf>; Wilson and Gray,
op. cit., n. 22.
65 It is important to contrast concerns about data use and data access. The respective
literatures reveal a similar prevalence of concern about the blocking effect of law.
ß2016 The Author. Journal of Law and Society ß2016 Cardiff University Law School
over, complexity should not be confused with lack of clarity. The key
question here is: are there specific legal provisions that expressly prohibit
data uses?
This requires, at a minimum, identification of the relevant legal provisions
affecting the particular use in question. Absent an explicit prohibition, how-
ever, the decision to share or not to share becomes both a matter of
interpretative flexibility, that is, how facilitatively will the legal framework
be read, and a question of institutional values, that is, by references to which
objectives and risks will the framework be read. If responsible data sharing is
not part of these value preferences, then the implications are obvious.
Moreover, establishing the basis for sharing gives way to the equally
important considerations of individual privacy and data protection, which
can further fuel cautious behaviour about sharing and access.
66
These initial considerations obviously help to map out the precise legal
parameters within which data custodians must operate. Considered in these
terms, however, it further helps to distinguish clear legal authority versus
prohibition and greyer areas, for example, revealing where the law is silent.
If this last scenario prevails, data custodians are encouraged to deploy the
template further to understand other explanations (and potential responses)
for resistance to use.
(b) Uses of data that are lawful but clearly unethical (and therefore should
not be undertaken)
A proportionate good governance framework for administrative data would
require that each data use be not only lawful but also ethical. How, then, can
a data custodian know when a proposed use of data is unethical? This
question involves answering what it means to act ethically in the context of
administrative data, for which we posit that the public interest plays a central
role.
The ethical suitability of any proposed use of data rests with its ability to
adhere to the core values that underpin collection of those data, which are in
turn dictated by the values underpinning action within the relevant sector.
67
Administrative data are first and foremost a public resource; it is information
obtained only by virtue of individuals' interaction with their government.
Use of data by public authorities is underpinned by values focusing on
378
66 See section `The legal landscape governing administrative data in the United
Kingdom' above for discussion around the regulatory importance and implications
of establishing (de)identifiability of data.
67 This references the principle of purpose limitation reflected in the DPA, Sch. 1,
para. 2: `Personal data shall be obtained only for one or more specified and lawful
purposes, and shall not be further processed in any manner incompatible with that
purpose or those purposes.' Notwithstanding data being de-identified, it remains an
ethical imperative to consider at the outset how new proposed uses of administrative
data align (or conflict) with the purposes for which data were originally collected.
ß2016 The Author. Journal of Law and Society ß2016 Cardiff University Law School
citizens' care and a defensible commitment to public service. These core
values coincide with considerations of what is (or is not) in the public
interest in context of a particular data initiative.
As mentioned above, the extent to which a proposed administrative data
initiative furthers the public interest is determined in part by reference to the
underlying values legitimizing action. We suggest, therefore, that deter-
mining whether a data initiative coincides with the public interest in this
regard will offer answers to questions of ethical suitability. That is, the model
of good governance we propose finds an important connection between the
ethics of a data initiative and its social legitimacy, the latter contributing to a
wider public interest grounding of the initiative. As we contend later,
determining the public interests at stake within the context of a particular data
initiative and supporting such initiatives with a clear and articulated public
interest mandate, is crucial to securing the lawful and ethical use of
administrative data that entails the important element of social licence.
This being said, specific examples where a proposed use would be clearly
unethical and should not be undertaken would include uses that conflict with
the public interests that underpin the operations of that particular public
authority. This would entail initiatives that sacrifice the protection of
individual privacy (a public interest in itself
68
) for the sake of organizational
efficiencies or mere convenience related to a specific method (or amount) of
data use. It would also categorize uses that only served private interests,
rather than `public interests' as unethical ± say if data were shared with and
used only for commercial gain by a private actor. Finally, uses that may
serve public interests but require disproportionate intrusions into the privacy
of individuals would also be considered unethical. This is not to suggest that
there is a one-size-fits-all view of public interest; rather, it requires ethical
reflection in each setting as to what counts as legitimate public interests that
can support data use.
(c) Uses of data that are lawful but met with disproportionate regulatory
burden (and thus avoided)
Regulatory burdens are often cited in the literature.
69
Initiatives involved in
areas deemed sensitive, particularly controversial, or simply novel, will often be
379
68 Recognized by the Courts in England in W. v. Egdell [1990] Ch. 359; [1990] 1 All
E.R. 835 and discussed further in: D. Townend, `Overriding Data Subjects' Rights
in the Public Interest' in The Data Protection Directive and Medical Research
Across Europe, eds. D. Beyleveld et al. (2004); C. Raab, `Privacy as a Security
Value' in Jon Bing: En Hyllest/A Tribute (2014), at
wp-content/uploads/2014/08/Raab_PrivacySecurityValue.pdf>; C. Raab, `Privacy
as a Social Value and as a Security Value' in Privacy and Security in an Age of
Surveilla nce (2015) 4 Da gstuhl Pers pectives Wo rkshop 1440 1, at /
drops.dagstuhl.de/opus/volltexte/2015/4888/>.
69 For example, see Law Commission, op. cit., n. 4, pp. 49, 56, 89, 98.
ß2016 The Author. Journal of Law and Society ß2016 Cardiff University Law School
faced with enhanced regulatory scrutiny or are perceived to require enhanced
scrutiny and thus may be avoided on this basis as being disproportionate.
70
There can be various reasons for this, and once again, law is often seen to be a
problem. This should be routinely questioned. For example, data protection law
contains many legitimate use pathways and flexibilities. These include the
argument that data use is necessary to the exercise of a function of the Crown, a
Minister of the Crown or a government department (DPA, Schedule 2, para. 1
5(c)). Still, more directly the viability of consent could be explored (DPA,
Schedule 2, para. 1; Schedule 3, para. 1 requires it to be `explicit').
Far less common is the consideration that the proposed data initiative may
weigh heavily in terms of the public interests it serves and therefore is
necessary to undertake a function of the public authority in question.
71
There is,
however, often a failure for such public interests to be clearly articulated to the
relevant approval/authorizing bodies, or indeed within the bodies themselves.
In other words, a clear commitment to sharing in the public interest is rarely
found. The default too often is, rather, to find reasons not to share.
72
Instances of reluctance to share data (for example, for research) are
widely documented as being subject to delays for access, especially in the
health sector.
73
The evidence of disproportionate regulatory burden is often
linked to a failure to identify and take full advantage of the regulatory
flexibilities already present.
74
Other examples include the disproportionate
repeat of approval mechanisms, such as access and ethics committees. The
380
70 In context of local authorities: E. Copeland, Small Pieces Loosely Joined: How
Smarter Use of Technology and Data Can Deliver Real Reform of Local Govern-
ment (2015) 31±2, at /www.policyexchange .org.uk/images/public ations/
small%20pieces%20loosely%20joined.pdf>.
71 Referring to a basis in data protection law that legitimizes the sharing of
administrative data under the DPA, Sch. 2, para. 5(d).
72 Law Commission, op. cit., n. 4, pp. 39±40; Copeland, op. cit., n. 70, p. 31.
73 T. Walley, `Using Personal Health Information in Medical Research: Overzealous
Interpretation of UK Laws Is Stifling Epidemiological Research' (2006) 332 Brit.
Medical J. 130, at ih.gov/pmc/articles/PMC13367 50/>;
Laurie and Sethi, op. cit., n. 55; K. Pollock, `Procedure versus Process: Ethical
Paradigms and the Conduct of Qualitative Research' (2012) 13 BMC Medical Ethics
1; A. Rid, `How Should We Regulate Risk in Biomedical Research?: An Ethical
Analysis of Recent Policy Proposals and Initiatives' (2014) 117 Health Policy 409;
Department of Health (DH), Jeremy Hunt MP, and National Information Board,
`National Data Guardian Appointed to Safeguard Patients' Healthcare Information'
(20 14), a t tps :// www .go v.u k/g ove rnm ent/ new s/n ati ona l-d ata -gu ard ian -
appointed-to-safeguard-patients-healthcare-information>
74 Academy of Medical Sciences, `Personal Data for Public Good: Using Health
Informa tion in Me dical Res earch' ( 2006), at http:/ /www.a cmedsci .ac.uk /
download.php?f=file&i=13206>; C. Haynes et al., `Legal and Ethical Considera-
tions in Processing Patient-Identifiable Data without Patient Consent: Lessons
Learnt from Developing a Disease Register' (2007) 33 J. of Medical Ethics 302, at
; Snooks et al. and Al-Shahi Salman
et al., op. cit., n. 64.
ß2016 The Author. Journal of Law and Society ß2016 Cardiff University Law School
resulting delays, costs, and other constraints can result in the complete
abandonment of projects even if this is counter to clearly articulated public
interests. Equivalent evidence is lacking in the administrative data context
about the nature and extent of regulatory burdens experienced. However,
valuable and recent contributions confirming a problem have been made by
Policy Exchange on the experience of data sharing in local authorities in
England; the Digital Government Review Team's 2014 report with recom-
mendations on mobilizing data sharing as a national priority; the Law
Commission's 2014 report on data sharing across public bodies; and the
Administrative Data Taskforce's report in 2012 on the use of administrative
data for research.
75
Furthermore, albeit anecdotally, our own work engaging with the ADRN
research community reveals that a disproportionate regulatory burden does
affect access to administrative data. Discussions with researchers involved in
ADRC Scotland revealed that access to administrative data in areas
traditionally conceived as more `sensitive', such as health research involving
children, will often face significant hurdles to obtain data. This has resulted
in entire projects and initiatives being abandoned. In others areas such as
crime and justice, even where a researcher has received the relevant
approvals from a wider governmental organization and/or university, each
police force must be applied to individually whereby `access is limited and is
usually subject to detailed contractual negotiation', the onus being placed on
the individual researcher to manage the approvals.
76
Further research and
evidence of the knock-on effects of such regulatory burden is needed.
77
Without a full understanding of the potential harm caused by risk-averse
behaviour, data custodians will inevitably act more cautiously than may be
warranted under the circumstances.
78
Moreover, without the identification
and sharing of best practices of working within regulatory frameworks and
of experiences of attempting to strike appropriate sectoral balance of inter-
ests, there is little prospect of improving public authorities' confidence in
evolving innovative uses of data.
79
The deployment of the decision-making
template offered here can help to smooth the path through approval
mechanisms by supporting clearer identification of the precise issues at stake
for all parties.
381
75 Copeland, op. cit., n. 70;Digital Government Review, op. cit., n. 49; Law
Commission, op. cit., n. 4; Administrative Data Taskforce, op. cit., n. 59.
76 Administrative Data Liaison Service, `ADLS ± Administrative Data Liaison
Service: Police Force Crime Records Datasets', at
forces/police-force-crime-records-dataset/?detail#ds_jump_access>.
77 Identified as a priority area for engagement and research by the Labour Party as a
result of their Digital Government Review Consultation and Report process, op. cit.,
n. 49, p. 39.
78 Law Commission, op. cit., n. 4, pp. 39±40; Copeland, op. cit., n. 70, p. 31.
79 Digital Government Review, op. cit., n. 49; Copeland, id.
ß2016 The Author. Journal of Law and Society ß2016 Cardiff University Law School
(d) Uses of data that are lawful but met with institutional caution (and thus
not undertaken)
Related, but distinguishable, scenarios arise where the law supports a
particular data initiative but a confluence of institutional factors impact
negatively. This scenario addresses factors specific to the sector or particular
organizational culture that contribute to sub-optimal decision making.
The most recent and robust study undertaken to date on the culture of
caution surrounding administrative data was conducted by the Law Com-
mission in 2014.
80
The report discusses various elements of the said culture in
terms of both perceived (for example, misinterpretations about legal reper-
cussions) and actual barriers (for example, resources) to sharing admini-
strative data. The report suggests that even where the law might impliedly
permit a particular data initiative, data custodians can exhibit extreme caution
for a variety of reasons that are institutionally embedded and therefore would
not be amenable to change without significant cost and/or a paradigmatic
cultur al shift.
81
The Law Co mmissi on ident ified ma ny reaso ns for
institutional caution outwith the law:
·A lack of incentives (or understanding of already present incentives) to
investing necessary resources to facilitate the use of data;
82
·Institutional shortage of expertise, experience, and resources in the area of
data management and, in particular, to dedicate towards ensuring data
quality;
83
·(Unsubstantiated) fear of individual reprisal over decisions taken as to
new data initiatives.
84
From our engagement with public authorities and work with the ADRN, we
would add to this list:
·A lack of a shared organizational vision that understands the benefits of
using administrative data;
·Fear over legal action against the institution if engagement in a particular
initiative is `novel' and thus legally `untested';
·Fear over reputational damage and thus to public trust/faith in the
institution if weaknesses are exposed by means of participation in a
particular data initiative;
85
382
80 However, Thomas and Walport's review on data sharing practices in 2008 also
provide a robust body of evidence: Thomas and Walport, op. cit., n. 10; Law
Commission, op. cit., n. 4.
81 Law Commission, id., p. 113.
82 id., pp. 105±6
83 id., pp. 105, 108±10, 115.
84 id., pp. 106±8.
85 The public backlash experienced post-care remains a constant concern and topic of
debate with stakeholders.
ß2016 The Author. Journal of Law and Society ß2016 Cardiff University Law School
·An absence of beacon examples to lead the way: why should an institution
expose itself in being the first?
86
A common theme between these various instigators of caution is the
overwhelming sense of fear stemming from uncertainty ± uncertainty as to
incentives, as to the real risks involved, as to the consequences for `getting it
wrong', as to the public interest benefits to accrue (to either the organization
itself or the relevant public). Unlike the previous scenario where regulatory
burdens are likely to be (perceived to be) imposed from above, these insti-
gators of caution are perpetuated internally by the data custodians themselves.
Institutional concerns, such as resources, risks to reputational damage,
and a lack of clear incentives, can cause organizations and/or sectors to focus
inwards in the name of caution when the focus could just as easily ± and
perhaps more appropriately ± be placed outward, on the public nature of
administrative data as a public resource. To move beyond a culture of
caution, questions surrounding the reuse of administrative data, such as for
research, could be reframed from the perspective of the public interest and
the public mandate of public authorities to act accordingly. How does a
particular data initiative support or detract from the public interest at stake
for wider society in terms of potential and realizable benefits to particular
groups, for the protection of privacy, for instilling confidence in a public
authority, or for the sector's ability to deliver services to the citizenry? This,
once again, requires data reuse to be seen as a core feature of the operations
of a public authority ± as a central priority that cannot be divorced from its
obligations to deliver services as optimally as possible.
(e) Uses of data that are lawful but ethically controversial (and thus not
undertaken in the absence of an ethical decision-making tool)
Finally, we come to scenarios where a data initiative may be lawful but is
ethically controversial and ultimately not undertaken due to the absence of
robust ethical decision-making tools. This scenario is helpfully illustrated by the
suspended data initiative sector, care.data, mentioned above. This was a data
initiative proposed by England's National Health Service (NHS), among other
things, to `. .. extract data from NHS primary care medical records in England
unless patients have purposefully opted out, in part to facilitate research'.
87
383
86 See, also, Digital Government Review, op. cit., n. 49, pp. 38±41; Copeland, op. cit.,
n. 70, pp. 31±2.
87 NHS England, `Frequently asked questions: care.data guide for gp practices' (2014),
at tps ://w ww.g oog le.c o.uk /se arch ?q=T o+s uppo rt+p ati ents %E2% 80% 99+
choice%3B+to+advance+custo mer+services%3B+to+promote+ greater+transparen-
cy%3B+to+improve+outcomes%3B+to+increase+accountability%3B+to+drive+ec-
on om ic +g ro wth +b y+ ma ki ng +E ng la nd +t he +d ef au lt +l oc at io n+ fo r+ wo rl d-
class+heal th+services +research. %E2%80%99+& ie=utf-8&a q=t&rls=org .mozilla:
e n - U S : o f f i c i a l & c l i e n t = f i r e f o x - a & c h a n n e l = s b & g f e _ r d = c r & e i =
xX8AVYOHFouK4Ab72oCYDw>; Carter et al., op. cit., n. 51.
ß2016 The Author. Journal of Law and Society ß2016 Cardiff University Law School
Legislation (the Health and Social Care Act 2012) was enacted to make lawful
the transfer of primary care patient records to the newly created Health and
Social Care Information Centre for six specific reasons.
88
Furthermore, the
NHS Constitution was amended
89
to include a pledge to:
. .. inform English NHS patients about research studies in which they might be
eligible to participate, and also, crucially, an expectation that patients would
be willing to share their medical information for healthcare planning and for
research purposes.
90
In recent analysis of care.data,
91
one of us has contended that while the
administrative and legal structure was created to carry out the care.data
scheme, social licence was not obtained and this contributed to the resulting
public controversy and postponement of the scheme. Not only are the purpose
of care.data and its public benefit unclear, the attempts made to have
dialogue a with the public were entirely insufficient (information leaflets sent
out to NHS England patients were addressed to households, not individuals
and often mistaken for junk mail).
92
The care.data initiative was (and
remains) controversial and as a result was suspended: this was initially for six
months and it has been further postponed for a more indefinite period.
93
We
would further suggest that this initiative was flawed for a lack of the
384
88 `To support patients' choice; to advance customer services; to promote greater
transparency; to improve outcomes; to increase accountability; to drive economic
growth by making England the default location for world-class health services
research': NHS England, `Care Episode Statistics: Technical Specification of the GP
Extract' (2013) 6, at
ces-tech-spec.pdf>.
89 NHS England, `The NHS Constitution: The NHS Belongs to Us All' (2015) 8, at
//www. gov.uk/ govern ment/up loads/ system/ upload s/atta chment_ data/fi le/
170656/NHS_Constitution.pdf>.
90 Carter et al., op. cit., n. 51, p. 406.
91 id.
92 O. Solon, `The Communication of Care.data to Patients Has Been an Absolute
Shambles' (2014), at
terrible-communication>.
93 As of November 2015, the launch date of care.data has yet to be confirmed and its
status uncertain. The scheme was to be trialled in four surgeries in England;
however, in September 2015, this was postponed by Jeremy Hunt who asked
National Data Guardian, Dame Fiona Caldicott to develop a new model for
obtaining patient consent. The Department of Health appointed Caldicott as the
country's first National Data Guardian in November 2014 on the back of care.data's
controversial and stalled launch earlier in the year. Her role as National Data
Guardian is to `champion on security of personal medical information': DH, op. cit.,
n. 73; A. Matthews-King, `GPs Prepare to Contact Patients Individually as Care.data
Is Relaunched in Some Areas' Pulse (2015), at
practice/practice-topics/i t/gps-prepare-to-contact- patients-individually-as-c aredata-
is-rela unched-i n-some- areas/2 0010215. article #.VX768 RNViko> ; L. Evenst ad,
`Public Trust Is Key to Sharing Health Data Successfully, Says Fiona Caldicott'
(2015), at
to-sharing-health-data-successfully-says-Fiona-Caldicott>.
ß2016 The Author. Journal of Law and Society ß2016 Cardiff University Law School
necessary up-front assessment of preparedness for data reuse and a lack of
ethical tools to successfully support its delivery pathway; the crucial tool
being here, meaningful public and stakeholder (GPs) engagement.
The most obvious lesson is that clearing a legal path for ethically con-
troversial data initiatives is not, in itself, sufficient. This should lead us to
question whether legal solutions in the administrative data context would
fare any better. Second, assistance in identifying and addressing potential
ethical controversy is itself crucial for decision makers. Frameworks for
ethical decision making, by definition, do not prescribe what ought to
happen; rather, they facilitate reflection and justifiable courses of action,
based on commonly recognized values, through meaningful and effective
engagement with relevant stakeholders.
94
Often, a principles-based approach
is used, by which we mean the identification, dissemination, and deployment
of commonly agreed principles that embody the core values at stake and
provide a uniform framework and language for deliberation on how to
proceed.
95
In the context of administrative data, this turns on the elusive
meaning of public interest. `Public interest' is a term imbued with inter-
pretative flexibility. It is neither possible nor desirable to say how it applies
across a range of diverse (public) sectors. The term ± and its legitimating role
± must be defined and crafted by the authorities themselves in a given
context. Equally, and as Taylor has argued, the citizens interacting with the
sector must have good reasons to expect that public interest, and their data,
will be deployed in defensible and accountable ways.
96
The lack of and/or non-use of appropriate ethical tools can create a perfect
storm of controversy even where it may be lawful and in the public interest
to undertake a particular initiative. Our template suggests that, in any given
administrative data-use context, if no ethical framework is in place, then one
ought to be developed as a matter of urgency.
385
94 Public engagement played a crucial role in the development of a good governance
framework for health data under SHIP and plays a similarly critical role in the
research undertaken for ADRC Scotland: SHIP, op. cit., n. 55; S. Davidson et al.,
`Public Acceptability of Data Sharing Between the Public, Private and Third Sectors
for Research Purposes' (2013), at
10/1304>; Laurie and Stevens, op. cit., n. 58.
95 SHIP, id.; Scottish Government, op. cit., n. 53; G. Laurie and N. Sethi, `Towards
Principles-Based Approaches to Governance of Health-Related Research Using
Personal Da ta' (2013) 4 Eur opean J. of Ri sk Regulati on 43, at ://
www.ncbi.nlm.nih.gov/pmc/articles/PMC3885861/>.
96 M. Taylor, `Health Rese arch, Data Prot ection, and the P ublic Interes t in
Notification' (2011) 19 Medical Law Rev. 267.
ß2016 The Author. Journal of Law and Society ß2016 Cardiff University Law School
INFLUENCING A SUB-OPTIMAL DATA CULTURE
In light of the foregoing analysis, the legal, ethical, and organizational
complexities involved combine to create a culture that:
·neither routinely identifies nor communicates the public interest and other
benefits of using administrative data;
·does not incentivize the lawful, ethical, and safe and proportionate use of
administrative data;
·does not have mechanisms to disabuse people of myths, perceived and
unsubstantiated controversies, and liabilities arising from use of admini-
strative data;
·does not invest in the resources required to support use of administrative
data;
·does not provide decision makers with effective support to deliver
responsible sharing as an embedded part of public sector practice
While the prospect of a general legislative gateway to share public authority
data remains a possibility, the organizational culture and resulting practices
that have become associated with `data' cannot be changed with additional
laws. What is required is a cultural paradigm shift within public authorities,
in how data are valued and acted upon. While accepting that cultural change
is a notoriously tall order, we offer some bold suggestions below that, if
adopted, we believe will at least initiate the necessary paradigm shift.
1. A public interest mandate to support the use, sharing, and linkage of
administrative data
To affect the necessary cultural shift within public authorities, something
radical and experience-led is required. However, change need not be in the
nature of extreme rupture; progressive institutional shifts can be initiated and
driven by the adoption of a public interest mandate. Indeed, in the govern-
ment's recent Digital Review, a key recommendation was to develop through
consultation a set of key public interest principles to guide public authorities'
decisions regarding data.
97
The public interest mandate we suggest, requires:
(i) deployment of our decision-making template to isolate those issues
that require attention (and to demystify others);
(ii) organization-wide, overt commitment to ensuring that administrative
data will be put to lawful and ethical use in order to serve the public
interest (to be discussed and agreed with relevant sectors);
(iii) meaningful engagement with relevant publics and stakeholders,
reflected in actual governance practice;
(iv) commitment to proportionality in all aspects of decision-making;
386
97 Digital Government Review, op. cit., n. 49, p. 39.
ß2016 The Author. Journal of Law and Society ß2016 Cardiff University Law School
(v) identifying and addressing areas of remaining uncertainty and
potential conflict.
Importantly, the proposed public interest mandate does not create an addi-
tional regulatory framework or administrative hurdle to the use of admini-
strative data. Rather, it offers a means to assess readiness for responsible data
reuse and to address the most problematic aspects of existing, dispropor-
tionate approaches to administrative data decision making by refocusing
attention to those issues whose resolution are crucial to securing the public
interests at stake. We address each of the mandate's components below.
(a) Adoption of the decision-making template
Deployment of the novel decision-making template is crucial to determining,
in any given context, what are the real or perceived barriers to using
administrative data. Public authorities must commit to dispelling false per-
ceptions as they relate to law and to apparent individual and organizational
repercussions for decisions taken in this area. Identifying and assuaging
concerns will help to encourage more confident decision making in
proportion to actual risks and benefits of any proposed use of data. In
identifying scenarios where problematic cultural practices or attitudes to data
are impeding proportionate decision making and separating those situations
from actual legal barrie rs, new approaches can aris e. These crucial
distinctions, between real and perceived barriers, should be reflected in the
training and development activities across a public authority.
(b) Organizational commitment to the principle of the public interest
In isolating the key considerations, data custodians may re-focus on the
`public interest' of any proposed administrative data initiative. As a public
resource, each proposed use of data should be considered from this
standpoint. Data custodians should embrace a transparent public interest
mandate to effect change to how their data are valued and used ± this
requires publicly committing to the principle of the public interest.
The public interest must not be understood as an `either/or' proposition
between considerations of protecting individuals' privacy and broader
societal interests. Rather, the public interest lies in both the robust protection
of individual privacy and uses of data that stand to result in wider public
benefit. The latter would include research uses of administrative data that
can result in positive outcomes for society (for example, better public
services in health, education, transportation, enhanced understanding of
factors that contribute to the health and overall wellness of individuals in
society, and so on).
98
An example of this kind would be for data custodians
387
98 Administrative Data Taskforce, op. cit., n. 59, p. 1; `Benefits of the Data' (2015)
.
ß2016 The Author. Journal of Law and Society ß2016 Cardiff University Law School
of administrative data to commit to:
Scientifically sound and ethically robust research based on use, linkage and
reuse of administrative data is in the public interest in promoting and
improving economic growth, personal and social well-being, and maximising
the interests of current and future generations of citizens in the UK.
99
A mandate might equally provide that:
Citizens' rights of privacy will be safeguarded by robust and proportionate
safeguards, in recognition of the public interests served by protecting
individual privacy.
This mandate could be given substance in the mission statement of data
custodians and would necessarily be different in each case, according to the
remit of each public authority and their interactions with the public.
Crucially, as recognized above, there is no definitive public interest; it is
always to be determined on a case-by-case basis in line with the values and
specific context in question. By committing to, and being seen to commit to
the principle of the public interest, however, a shared and permeated vision
can be cultivated internally and simultaneously initiate a meaningful point of
engagement with the public and other stakeholders. A shared, organization-
wide commitment to realizing the public benefits of the safe and ethical use
of administrative data is crucial to initiating change. By publicly committing
to serving the public interest in this way, data custodians can seek the crucial
social licence by assuaging individuals' concerns about the likelihood of any
data initiative being used for anything other than projects that will positively
benefit individuals and society.
100
This might be further supported by clear
statements and commitments about uses to which administrative data will
not be put.
(c) Meaningful engagement with likely users and relevant publics
Law cannot prescribe social licence. It cannot be manufactured; it must be
earned. This crucially involves the need to substantiate the public interests
served by all uses of data.
101
Recent public attitudes research has empha-
sized the importance of engaging with relevant publics early in the process
of data initiatives, as the Ipsos MORI study did for ADRN when it was
established in 2013.
102
Meaningful engagement requires ongoing dialogue,
388
99 As originally provided in our working paper: Laurie and Stevens, op. cit., n. 58, p.
39.
100 Cameron et al., op. cit., n. 4, pp. 7, 17.
101 A key finding of the 2014 Ipsos MORI study was that participants were consistently
concerned whether social research involving administrative data would actually lead
to social value: id.
102 Engagement that is ongoing throughout the ADRN, such as through citizens' panels
in Scotland and Wales. `Public Engagement: Become Involved in Social Science
Research ' (2015), at http://a drn.ac.u k/about/r esearch- centre-s cotland/ public-
engagement>.
ß2016 The Author. Journal of Law and Society ß2016 Cardiff University Law School
consideration of the risks and benefits involved with processing admini-
strative data and of how such concerns would be addressed through a
principled, proportionate governance framework in which the public interest
plays a vital part.
Importantly, the Ipsos MORI study found that:
. . . the public would be broadly happy with administrative data-linking for
research projects provided (i) those projects have social value, broadly defined
(ii) data is de-identified, (iii) data is kept secure, and (iv) businesses are not
able to access the data for profit.
103
To translate these findings into meaningful engagement, the concerns raised
must be transparently taken into account. As a good practice example, the
ADRN uses a documented approvals process, which hinges upon the con-
cerns raised by the participants in the Ipsos MORI study.
104
It is important
that engagement not be tokenistic, either in practice or in perception.
Engagement must have at least a reasonable prospect of having a real
bearing on how data initiatives operate. All decisions, including conflicting
policies and procedures, must be robustly justified.
(d) Commit to proportionate governance
Proportionality is a cornerstone principle for decision making and regulation
that requires a balance of interests, relative to an assessment of real risks and
likely benefits.
105
Determining whether a propose d data initiative is
proportionate to the public interest aims sought is indicative of whether a
data initiative is `in the public interest'. The principle of proportionality is
supported by a clear framework of assessment provided within context of the
European Convention on Human Rights and decisions taken by the European
Court of Human Rights.
106
Adapted to the current context, proportionality
points towards the following questions:
·Is there a clear and knowable public interest served by the data initiative?
·Has the necessity of data use been demonstrated relative to the public
interest?
·Has the relative intrusiveness of data use been minimized ± is the least
intrusive method being used?
389
103 Cameron et al., op. cit., n. 4, p. 57.
104 ADRN, `Protecting Privacy: Project Approval' (2015), at ://adrn.ac.uk/
protecting-privacy/project-approval>.
105 Proportionality plays a crucial role in the good governance frameworks one of us
developed in the health data context: SHIP, op. cit., n. 55; Laurie and Sethi, op. cit.,
n. 94; Sethi and Laurie, op. cit., n. 54.
106 Handyside v. UK [1976] 1 E.H.R.R. 737; Sunday Times v. United Kingdom (1979) 2
E.H.R.R. 245; Sunday Times v. United Kingdom (No 2) [1991] E.C.H.R. 50; A., B.
& C. v. Ireland (2010) 53 E.H.R.R. 13.
ß2016 The Author. Journal of Law and Society ß2016 Cardiff University Law School
·Are there safeguards against abuse and misuse of the data, including
sanctions?
107
A commitment to proportionality, including the assessment and management
of real risks reduces undue regulatory burdens and helps to ensure employees
across the public authority are not only trained in the risks involved and
precautions to be taken, but are also made aware of the benefits of using data
and the harm that might arise out of not using data.
108
(e) Identify and address remaining areas of uncertainty
Some areas of uncertainty still remain. First, uncertainty remains as to the
exact nature and scope of the risks involved. Articulating public interest
benefits is only one-half of what is required in committing to a public
interest mandate ± any use of data must be proportionately assessed for its
propensity to cause harm to individuals and other public interests. Yet, this
assessment is currently extremely difficult for data custodians and this might
further forestall proportionate decision making. The development of a sound
evidence base across sectors must therefore also be an imperative. More
research and sharing of best practices is needed. In the absence of evidence
of harms, however, good-faith assessments and sound ethical reflection to
promote a public interest mandate can promote the beginning of the cultural
shift advocated in this article.
390
107 U. Kilkelly, The Right to Respect for Private and Family Life: A Guide to the
Implementation of Article 8 of the European Convention on Human Rights (2003), at
tp: //ww w.e chr .co e.i nt/ Lib rar yDo cs/ DG2 /HRH AND /DG 2-E N-H RHA ND-
01(2003).pdf>; D. Korff, `The Standard Approach Under Articles 8±11 ECHR and
Art icl e 2 ECH R' (2 009 ), a t p:/ /ec .eu rop a.e u/ jus tic e/n ews /e ven ts/
conference_dp_2009/presentations_speeches/KORFF_Douwe_a.pdf >.
108 The harms arising out of a failure to share data have been explored in social policy
literature and in context of acute failures in social services resulting in great harm
and impact to individuals and communities: see The Victoria Climbie
ÂInquiry
(2 00 3) , at h tt ps :/ /w ww . go v. uk /g o ve rn me nt /u p lo ad s/ sy s te m/ up lo ad s /
attachmen t_data/fi le/273183 /5730.pdf >; `Paedoph ile Jailed fo r Raping Girl '
(2006), at .bbc.co.uk/1/hi/engla nd/southern_counties /4926482.stm>;
C. Bellamy et al., `Information-Sharing and Confidentiality in Social Policy:
Regulating Multi-Agency Working' (2008) 86 Public Administration 737, at
dx.doi.org/10.1111/j.1467-9299.2008.00723.x>; ```Mistakes Were Made.'' HMIC's
Review into Allegations and Intelligence Material Concerning Jimmy Savile
between 1964 and 2012' (2013), at
medi a/r evi ew- int o-al leg ati ons -an d-in tel lig enc e-m ater ial -co nce rnin g-j imm y-
savile.pdf>; A. Jay, `Independent Inquiry into Child Sexual Exploitation in
Ro t h e r h a m 1 9 9 7 ± 2 0 1 3 ' ( 2 0 1 4 ) , a t
independent_inquiry_cse_in_ rotherham.pdf>; Press Association, ```Catastrophic
Failure'' Allowed Convicted Killer to Murder on Day Release' (2015), at
www .th eg uar dia n. com /uk -n ews /20 15 /ma r/2 3/c at ast rop hi c-f ail ur e-a ll owe d-
convicted-killer-to-on-day-release>; `Risks Associated with Sharing/Not Sharing
Information' (2015), at
07/Risks-associated-with-sharing-information.pdf>.
ß2016 The Author. Journal of Law and Society ß2016 Cardiff University Law School
Second, there is uncertainty as to the necessity of introducing new laws,
as argued above. In light of Brexit, one of many unresolved questions is how
the United Kingdom will proceed in updating its own data-protection
regime. Whether or not it remains part of the European Economic Area (and
thus bound by the forthcoming GDPR), it is likely that an essentially
equivalent legislation will be adopted to withstand determinations of
adequacy for data transfers between the United Kingdom and the EU.
Regardless of the precise terms and conditions of Brexit, the United
Kingdom is unlikely to diverge greatly from EU law in respect of data
protection. As for introducing new data-sharing laws, and as the Law
Commission has noted, `. . . misunderstanding and confusion about what the
law requires can also point to a need to simplify or codify the law to address
its complexity and make it more accessible to practitioners.'
109
The
multitude of existing gateways for sharing and using data could be
simplified into more discrete and overarching data-sharing legislation,
currently being considered in an open policy process. Even so, such a legal
exercise alone will not be enough to deliver the necessary culture change.
For this to be realized, the role of principled and discretionary decision
making within a governance framework must be recognized. If a data
custodian has to determine whether a proposed use of their administrative data
is `in the public interest', this will require exercises of judgement that are not
currently supported by the problematic culture surrounding data. In this sense,
sector-specific guidance that promotes public interest sharing would ideally be
provided by a body with overarching authority ± such as the United
Kingdom's Information Commissioner's Office
110
± or other relevant, sector-
specific professional bodies; this could ensure the guidance had the requisite
authority and legitimacy to instil confidence. However, the challenge of cross-
sector uncertainties and disparities remains. Full harmonization across sectors
is highly aspirational, and probably unattainable. Rather, there needs to be
mutual recognition of diverse approaches under the common objective of
sharing public authorities' data in the public interest. Yet again, this requires
more evidence of best practice, especially inter-sectoral best practices, and the
capture and dissemination of experiences of multi-agency data initiatives that
have demonstrated responsible and mutual sharing.
111
Third, the authoritative research in this area ± the Law Commission's
2014 report ± only reflects the data-sharing practices and laws of England
and Wales ± similar scoping research must be undertaken in both Scotland
and Northern Ireland, where devolved powers are likely to mean divergence
in practice. This makes it even more imperative to determine the true nature
391
109 Law Commission, op. cit., n. 4, p. 25.
110 The United Kingdom's Information Commissioner's Office regularly publishes
sector-specific guidance which often become best practice (for example, its
`Anonymisation Code of Practice'): see .
111 As recently considered in context of local authorities: Copeland, op. cit., n. 70.
ß2016 The Author. Journal of Law and Society ß2016 Cardiff University Law School
of law's role in responsible sharing. The same is even truer if international
sharing is contemplated.
112
The interoperability of governance regimes is
key to success here. Initiatives such as the ADRN and Farr Institute are
dedicated to this end. Further scoping work is necessary on the cross-sectoral
and country-specific differences in cultural attitudes and resulting practices
surrounding different types of data, data custodians, and data users in the
United Kingdom and beyond.
While this article focuses on the broader components to achieving good
governance across the inherently varied context of administrative data,
efforts must simultaneously consider the benefits of designing frameworks
built for interoperability but that are sensitive to and provide for the varying
cultures, needs, and resources specific to particular public authorities arcoss
the public sector.
CONCLUSION
This article has addressed the broad range of legal and ethical concerns
arising out of the reuse of administrative data for research, using the United
Kingdom and experience in working with the ADRN as an exemplar. It puts
forward a novel decision-making template to help public authorities navigate
the complexities discussed herein, and to distinguish between purely legal,
ethical, and cultural factors that impact upon this process. This template
crucially exposes the problems associated with legal myths and perceived
controversies and how they contribute to the creation of a problematic
culture of caution impacting on data use by public authorities. Our original
contribution lies in linking two key elements to address cultures of caution:
the deployment of the decision-making template to reveal what is at stake as
a measure to assess preparedness for data reuse, and the adoption of a public
interest mandate to set and deliver priorities and to promote necessary
cultural shifts in attitudes and practice. As has been demonstrated, while
simplifying the law in this area would certainly diminish legal uncertainty,
the problematic culture that has arisen in context of administrative data
requires separate and dedicated attention.
392
112 Not least because of the recent European Court of Justice decision rejecting the
adequacy of the EU-United States Safe Harbor agreement: Schrems v. Facebook
(2015) C. 362/14.
ß2016 The Author. Journal of Law and Society ß2016 Cardiff University Law School

To continue reading

Request your trial

VLEX uses login cookies to provide you with a better browsing experience. If you click on 'Accept' or continue browsing this site we consider that you accept our cookie policy. ACCEPT