Developing a Workforce to Support Research Reliant on Data and Compute The University of Melbourne , Australia 0000 0003

We describe the construction, operation and evaluation of the Melbourne Data Analytics Platform; a group of academics whose mission is to support research requiring non-trivial data analysis or compute at the University of Melbourne. Like many organisations, the University of Melbourne recognised many years ago that digital methods for generating, analysing and storing data would be fundamental to the business, teaching and research activities of the university. Our comprehensive university is composed of ten faculties that include Law and Fine Arts through to Science and Medicine, and it was envisaged that each faculty would require some support, possibly at diferent levels of complexity, for research methods that relied on computing and data. Such recognition led to the formal establishment of the Petascale Campus Initiative in 2018 within the university. The Initiative had the triple aim of increasing afordable access to computer hardware for research; building infrastructure, policies, processes and procedures for supporting research data management; and developing a sustainable workforce to assist researchers. The aim of this paper is to describe the process that we used for the formation of the workforce, which has successfully operated for about three years under the name Melbourne Data Analytics Platform.

 Data analysis Research support Evaluation Introduction

While the idea of “Big Data Enabled Transformations” may well be considered little more than a myth propagated by large tech companies to increase profits [ 7 ], the promise of transformation has certainly captured the imagination of many organisations. The reader can no doubt bring to mind several examples, and has probably experienced some rfist hand. While our university is no diferent in some regards1, our eforts to establish a support workforce for research in the areas of data and compute have been considered. 1 For example, establishing a “business intelligence” unit to harvest insights from student enrolment, finance, and other data

Wang and colleagues [ 11 ] identify key links between the transformative power of big data capabilities, practices and benefits concluding that “a focus on strategic view has great potential to help balance the number of studies of big data from technological and managerial perspectives (p. 74)”. Such potential is echoed other studies of large organisations [ 3, 9 ] as scholars suggest that too much focus on either techno-optimism (seeing big data and AI as a panacea) or policy pessimism (disregarding big data and AI in light of entrenched practices) can undermine balanced research [ 10 ]. In practice, therefore, when developing a research support workforce we must be wary of these two extremes and remained focused on the intended outcomes the workforce should achieve.

The focus of modern academic research endeavours is to have “impact”. In an Australian context, this is defined by the Australian Research Council [ 2 ] as “the contribution that research makes to the economy, society, environment or culture, beyond the contribution to academic research.” While impact evaluation has a long history (see Gertler et. al [ 5 ] and references therein), academic institutions are still coming to terms with how to best evaluate the “impact” of academic research. As we are looking to establish and sustain a workforce that supports research rather than lead independent research, we can finesse this issue by being aware that the researchers we support will be seeking impact, but it is up to them to define and evaluate the success of their own research. Success of our workforce, therefore, is reliant on the researchers we support succeeding by whatever metrics they choose. 3

A Basis for Hiring

As the workforce we were building at Melbourne was new, we needed a deal of flexibility and agility in recruiting, setting operational structures in place, and evaluating outcomes. It is tempting in such an environment to operate as a “disruptor” and claim special status within the organisation, relying on governance to be wilfully ignorant of the new “tech startup”. But Whitchurch and Gordon [ 13 ] warn that fluid projects relying on professional networks and relationships run the risk of ignoring formal structures that enforce transparency and equity. Similarly, Klein [ 6 ] advocates for central oversight and management, but not “monolithic control”. Rather than deliberately subvert university processes we opted to make use of a “work focus” category in the HR policies of the university.

Existing at the university are two formal categories of employees: Academic and Professional staf. Academics are responsible for undertaking teaching and research, while Professional staf provide supporting infrastructure such as finance, HR, legal compliance, student record keeping and so on. There is a subcategory (or “work focus”) of Academic called the Academic Specialist which allows for an Academic to focus on one particular activity; for example, teaching. With this point in mind, we adopted the option to create “Research Data Specialists” to make a team of academics who were focused on supporting research using digital methods.

In our initial analysis, the category fit nicely with Whitchurch’s denfiition of a “third space” of employee that sat between the Academic and Professional categorisations [ 12 ]. The term third space, originally borrowed from cultural studies, embraces the notion that our workforce will require both in-depth academic training and collegial support. Neither a traditional academic nor solely a professional member of staf, the “third spacer” can only be fully recognised through an emerging set of new concepts. Eventually, we came to understand the work of Whitchurch as primarily concerned with Professional staf transitioning to work that demanded a deep understanding of academic research. By choosing to make use of the Academic Specialists focus at our university, we were more concerned with academics who would be asked to take on some duties more akin to Professional staf.

We came to adopt a more powerful narrative that is grounded in the concept of interdisciplinarity [ 6 ]; not only between academic disciplines, but between the two major workforce cultures that exist within the institution. Repko and Szostak [ 1 ] define interdisciplinary work as being in a contested space between the cultures associated with particular disciplines. Of particular note, they argue the term “interdisciplinary” is preferred for university and related research settings because it evokes a sense of critical thinking that is often needed to evoke key questions that eventually lead to the production of new knowledge.

Table 1 (adapted from Repko [ 1 ]) lists traits of an interdisciplinarian that we saw as being key to people we employed into the workforce. While it is unlikely that we can ever find individuals that have all of these traits, the list does provide a framework for hiring decisions and ongoing development of our workforce. Of note, our experiences pointed to a love of learning, tolerance of ambiguity and an appreciation of diversity as the most important traits. Following soon after those three, openness and adequacy are also key to successful team building.

These key traits also fit with the observations by Palmer and Neumann [ 8 ] who note that humanities scholars who undertake interdisciplinary research can be grouped into two behaviours: “exploration” or “translation”. In our case, the “exploration” of new fields is dictated by the researchers that we support, and so we need staf who are “translators”. In particular, the processes and activities that these staf will undertake are learning, contextualising and converting information [ 8 ]. 4

Establishing the Workforce

With a framework in mind for the types of employees we wanted, and the style of operation we wanted to encourage, we chose to hire in two phases. The first was an internal recruitment round, figuring that existing employees of the university would have their own networks of researchers that they could easily support as a ifrst round of activities and that they would have existing knowledge and social support for navigating university systems and processes. Once this group was established, we would augment them with an hiring round open to both internal and external applicants. Traits Enterprise Love of learning Reflection Willingness to assume risk to achieve outcomes Enthusiasm for learning in new situations in ways that are novel, adaptable and pertinent to the problems ahead Evaluation of conflicting lines of information, controversial stances that leads to an ability justify important decisions Tolerance for ambiguity and Acceptance of understanding as a constant process paradox amid complexity that may never be complete, and an attitude to remain open to new information and processes Receptivity of other disci- Openness to a range of disciplinary perspectives, plines and their perspectives and a willingness to work with those embedded in disciplinary ways of thinking Willingness to achieve ‘ad- Appreciating the diference between achieving an equacy’ in multiple disci- adequate understanding at the expense of mastery plines of a discipline area Appreciation of diversity

Respect for people holding difering views, and an awareness of own biases, as problems and solutions emerge Willingness to work with Collaborative approaches and thinking that manothers ifest through intellectual, interpersonal and group communication skills Humility

A learned state of mind that leads to further learning and greater respect of experts and others

The initial group was aforded a large degree of autonomy to build their own internal structures processes and operational principles. The motivation for this was twofold. Firstly to reinforce that the employees were Academic Specialists, and thus had freedom within the confines of university policy and strategy to choose what they worked on and how they worked. This was supported by the university’s academic promotion processes which call for an argued case for each individual, and does not set arbitrary numeric targets and goals that must be met. Secondly, in order to attract top software engineers and data scientists from industry the university cannot compete on salary, but can compete by ofering academic freedom and flexibility in the workplace. We want the team to be an energetic, supportive and intellectually stimulating environment to attract top talent.

In keeping with the idea of an argued case for academic promotion, we set boundaries for the workforce in the style of “claims” for which they could argue as being met. These claims are outlined in Table 2. Every six months, leadership of the group gathers evidence and assesses if each claim is weak, moderate or strong using a mix of qualitative and quantitative methods. As Freeth and Vilsmaier [ 4 ] argue, studies that detail the “lived experience of navigating” interdisiciplinary research projects are rare. Despite the promise of working across and within disciplines, practitioners must balance the oftconflicting diferences of such work that teeter between observation versus participation, curiosity and care, and impartiality versus investment. Further to these tensions that require balancing, the ongoing operation of our platform2 2 “the network of interactions and synergies becomes the platform, not simply the hardware structures and software strategies that facilitate them.” [6, p.67] has its own characteristics that require balancing as outlined in Table 3. An important part of our workforce development is learning to acknowledge and accept of these tensions as an inevitable part of a complex working environment [ 4 ]. This agility in working practice and goal setting is also essential in the leadership and management of the workforce [ 6 ]. Transactional services such as data The Platform must ofer an acawrangling and cleaning must be demic style of working in order to performed for researchers to lift attract excellent staf on a univertheir ability to make use of more sity pay scale.

complex methods.

Training There are many researchers who Delivery of training (as opposed to would benefit from introductory education) is not an academic accourses on basic tools, concepts and tivity. Steering researchers towards methods. well constructed self-learning resources is preferred.

The Platform is a support activ- Platform staf are highly motivated ity, therefore should not have a high academics who want recognition profile and should not overshadow and a vibrant community of practhe researchers it is trying to sup- tice. Further, internal funding is port. It should not be a “brand”. reliant on the Platform having a strong reputation within the University.

In summary, meeting the demands of creating a competent workforce that can meet the growing demands of data intensive research depends on creating a strong framework. We found success in constructing our perspective on an argument-based approach that was grounded in the widely used promotion and performance criteria as we sought to identify and enhance the traits of an interdisciplinary team. Perhaps counter-intuitively, the team was given latitude to then create their own methods of working together rather than, for example, senior academics imposing their own disciplinary views of “how things should be done”. We ourselves were fortunate that the University allowed us the freedom to craft a response to a complex environment that now avoids the pitfalls of having tightly coupled SMART goals to defined projects that can no longer be defended in contemporary institutional setting characterised by fast-paced, interdisciplinary research.

Importantly, however, it is worth noting that our experiences remain uncommon; that is, we see that the University as a whole will require a long period of adjustment to create new frames on how it seeks to manage, stimulate and retain a workforce capable of working across disciplines in data intensive research. Eventually, senior leadership will need to consider new approaches that, indeed, must foster success in the competitive global environment.

 1. Allen F. Repko , R.S.: Interdisciplinary Research: Process and Theory . SAGE Publications Inc, United States ( 2008 ) 2. Australian Research Council: Research impact principles and framework (March 27 , 2019 [Online]), http://arc.gov.au/policies-strategies/strategy/research-impactprinciples-framework 3. Desouza , K.C. , Jacob , B. : Big data in the public sector: Lessons for practitioners and scholars . Administration & Society 49 ( 7 ), 1043 - 1064 ( 2017 ). https://doi.org/10.1177/0095399714555751, https://doi.org/10.1177/0095399714555751 4. Freeth , R. , Vilsmaier , U. : Researching collaborative interdisciplinary teams: Practices and principles for navigating researcher positionality . Science & Technology Studies 33 ( 3 ), 57 - 72 ( 2020 ) 5. Gertler , P.J. , Martinez , S. , Premand , P. , Rawlings , L.B. , Vermeersch , C.M.J. : Impact Evaluation in Practice, Second Edition . World Bank ( 2016 ), https://openknowledge.worldbank.org/handle/10986/2550 6. Klein , J.T. , Schneider , C.G. : Creating Interdisciplinary Campus Cultures : A Model for Strength and Sustainability . John Wiley & Sons, Incorporated ( 2010 ) 7. Ossewaarde , M.: Digital transformation and the renewal of social theory: Unpacking the new fraudulent myths and misplaced metaphors . Technological Forecasting and Social Change 146 , 24 - 30 ( 2019 ). https://doi.org/https://doi.org/10.1016/j.techfore. 2019 . 05 .007, https://www.sciencedirect.com/science/article/pii/S0040162518315804 8. Palmer , C.L. , Neumann , L.J.: The information work of interdisciplinary humanities scholars: Exploration and translation . The Library Quarterly: Information, Community, Policy 72 ( 1 ), 85 - 117 ( 2002 ). https://doi.org/https://www.jstor.org/stable/4309582 9. Pencheva , I. , Esteve , M. , Mikhaylov , S.J. : Big data and ai - a transformational shift for government: So, what next for research? Public Policy and Administration 35 ( 1 ), 24 - 44 ( 2020 ). https://doi.org/10.1177/0952076718780537, https://doi.org/10.1177/0952076718780537 10. Vydra , S. , Klievink , B. : Techno-optimism and policy-pessimism in the public sector big data debate . Government Information Quarterly 36 ( 4 ), 101383 ( 2019 ). https://doi.org/https://doi.org/10.1016/j.giq. 2019 . 05 .010, https://www.sciencedirect.com/science/article/pii/S0740624X18302326 11. Wang , Y. , Kung , L. , Wang , W.Y.C. , Cegielski , C.G. : An integrated big data analytics-enabled transformation model: Application to health care . Information & Management 55 ( 1 ), 64 - 79 ( 2018 ). https://doi.org/https://doi.org/10.1016/j.im. 2017 . 04 .001, https://www.sciencedirect.com/science/article/pii/S0378720617303129 12. Whitchurch , C. : Reconstructing identities in higher education: The rise of 'third space' professsionals . Routledge, London and New York ( 2013 ) 13. Whitchurch , C. , Gordon , G. : The role of middle managers. In: In Reconstructing relationships in higher education: Challenging agendas , pp. 84 - 102 . Routledge, London ( 2017 )