Data Stewardship and Consultation

Executive Summary

To move for­ward and improve how data stew­ard­ship is imple­ment­ed, the impacts of cur­rent and past approach­es need to be well under­stood. Qual­i­ta­tive inter­views were car­ried out with data stew­ards at TU Delft to get a bet­ter under­stand­ing of con­sul­ta­tion and the impact of data stew­ard­ship. The results of this work can also be used as an infor­ma­tive ref­er­ence point for the cur­rent state of data stew­ard­ship and in par­tic­u­lar the cur­rent role of data stew­ards.

The ques­tions asked main­ly sur­round­ed three areas, the scope of data stew­ards’ role, com­mon con­sul­ta­tion top­ics and the dif­fi­cul­ties encoun­tered. Regard­ing the scope, some large roles were report­ed­ly miss­ing. The com­mon con­sul­ta­tion top­ics han­dled by data stew­ards were iden­ti­fied as ful­fill­ing the orig­i­nal needs found in the 2018 report¹ and ques­tion­ing dif­fi­cul­ties yield­ed a num­ber of sug­ges­tions from data stew­ards for help­ing improve the cur­rent approach.

Introduction

Data stew­ard­ship has been a grow­ing part of research for many years now. Con­cerns over integri­ty and repro­ducibil­i­ty have dri­ven this along with an increas­ing need for large-scale data han­dling². The FAIR guid­ing prin­ci­ples for research data stew­ard­ship were first put for­ward in 2014 and serve as the core con­cepts behind data stew­ard­ship. FAIR stands for find­abil­i­ty, acces­si­bil­i­ty, inter­op­er­abil­i­ty and reusabil­i­ty and it is a cor­ner­stone of research pol­i­cy that all work car­ried out should be fol­low the FAIR prin­ci­ples. Mak­ing research FAIR how­ev­er requires changes in research cul­ture, pol­i­cy and train­ing schemes.

One approach to improv­ing data stew­ard­ship is through pro­vid­ing data stew­ards as dis­ci­pli­nary data man­age­ment con­sul­tants for researchers. Data stew­ards were first appoint­ed at TU Delft fac­ul­ties in 2017 and were imple­ment­ed to address data man­age­ment needs across the cam­pus. Each data stew­ard has sub­ject spe­cif­ic exper­tise to pro­vide rel­e­vant and effec­tive sup­port to spe­cif­ic data man­age­ment needs of each field. The orig­i­nal scope set out for data stew­ards was to pro­vide the fol­low­ing:

• Tai­lored con­sul­ta­tions on data man­age­ment prac­tice
• Advice on the use of data man­age­ment tools
• Help with data man­age­ment plans and fun­ders poli­cies
• Help with meet­ing jour­nals’ require­ments for data avail­abil­i­ty
• Infor­ma­tion about data repos­i­to­ries
• Advice on increas­ing impact and data shar­ing
• Ded­i­cat­ed work­shops and infor­ma­tion ses­sions on research data man­age­ment
• Infor­ma­tion about work­ing with con­fi­den­tial research data

Under­stand­ing how the scope changes over time can pro­vide an indi­rect mea­sure of and insights into the impact of data stew­ards on research.

To under­stand the impacts of data stew­ard­ship, the Data Stew­ard­ship Coor­di­na­tor set out to qual­i­ta­tive­ly analyse the pol­i­cy imple­men­ta­tion, train­ing and con­sul­ta­tion across fac­ul­ties and gen­er­ate an infor­ma­tive report to be used as ref­er­ence mov­ing for­ward. Here we dis­cuss the inter­views with data stew­ards regard­ing con­sul­ta­tion with some reflec­tions and sug­ges­tions from the data stew­ards.

Methodology

A set of semi-struc­tured qual­i­ta­tive focus group inter­views were car­ried out with data stew­ards to ask ques­tions about their work, specif­i­cal­ly, fre­quent­ly dealt with top­ics, how accu­rate the scope is, what has changed over time and what dif­fi­cul­ties they encounter. The inter­vie­wees were select­ed based on their role as data stew­ards who reg­u­lar­ly car­ry out con­sul­ta­tions and were informed that the find­ings would be used to cre­ate a report on the impact of data stew­ard­ship. Inter­views last­ed on aver­age 40–60 min­utes and were not audio/video record­ed but respons­es and key notes were tak­en dur­ing the inter­views.

Areas of Data stewardship Consultation

Data stew­ards car­ry out an array of roles and con­sult on many top­ics depend­ing on the needs of their fac­ul­ty. The answers col­lect­ed can be loose­ly grouped into three areas: The scope of data stew­ards, dif­fi­cul­ties encoun­tered and con­sul­ta­tion top­ics.

Scope

Data stew­ards were asked about the accu­ra­cy of the scope and whether they could iden­ti­fy any impor­tant respon­si­bil­i­ties miss­ing as well as any­thing that may be out of date or no longer rel­e­vant. All data stew­ards agreed the scope was most­ly com­plete.

As part of data con­sul­ta­tion, data stew­ards advise researchers on pub­lish­ing open sci­ence and man­ag­ing research code. These were report­ed as impor­tant parts of data con­sul­ta­tion that are not acknowl­edged in the scope. Beyond con­sul­ta­tions, data stew­ards act as com­mu­ni­ca­tors link­ing researchers to the resources and staff they need; they also assist in devel­op­ment of research data man­age­ment pol­i­cy by pro­vid­ing feed­back and input. Both activ­i­ties were report­ed as impor­tant but miss­ing from the scope.

When asked about whether any parts of the scope were no longer rel­e­vant the data stew­ards were most­ly in agree­ment that every­thing present was accu­rate and a rel­e­vant part of the tasks they car­ry out. How­ev­er, a few points were brought up regard­ing ele­ments of the scope that were rarely dealt with. Com­pli­ance with jour­nal poli­cies was sug­gest­ed to be the least dealt with issue out of those present with­in the scope.

Difficulties in Data stewardship

To under­stand bet­ter the role data stew­ards are play­ing in aid­ing researchers, we asked them ques­tions regard­ing what takes up the most time, the most dif­fi­cult top­ics to aid researchers with and whether there are suf­fi­cient resources avail­able to effec­tive­ly help peo­ple in con­sul­ta­tion. It’s impor­tant to note that there’s a large vari­ety in the needs of researchers in dif­fer­ent fac­ul­ties and there­fore the answers may reflect that.

Most data stew­ards agreed that review­ing data man­age­ment plans took up the most time due to the fre­quen­cy of the requests, it was also iden­ti­fied that tai­lored con­sul­ta­tions were time inten­sive also, with advice on data stor­age and per­son­al data being com­mon­ly brought up. As for what issues were dif­fi­cult to resolve, the response depend­ed on the fac­ul­ty. Large col­lab­o­ra­tions and com­plex cus­tom arrange­ments were report­ed as some of the most dif­fi­cult issues to deal with along with confidential/personal data and legal issues.

Common RDM issues for DSs

To get a more com­plete under­stand­ing of the sup­port data stew­ards pro­vide, we formed a list of cat­e­gories of the top­ics data stew­ards con­sult on and asked what the main issue they dealt with was for each:

• Data man­age­ment plan­ning – Per­son­al data-relat­ed ques­tions are most fre­quent. Licens­ing issues, ques­tions about IP were also cit­ed as being a fre­quent issue to devel­op in the DMP process.
• Data stor­age, back­up, and archiv­ing – A large issue was researchers using per­son­al com­put­ers for data stor­age rather than secure stor­age and being unaware/unsure about options for stor­ing data, par­tic­u­lar­ly with large/public data sets.
• Data impact and shar­ing – Researchers strug­gling with what to and not to share, along with ques­tions being asked about the rules regard­ing con­fi­den­tial and per­son­al data shar­ing.
• Licens­ing – A wide range of ques­tions on the top­ic includ­ing what license to choose, cus­tom licens­es and soft­ware licens­es.
• Data man­age­ment tools – The biggest prob­lem was the use of non-licensed, non-approved and poten­tial­ly inse­cure tools.

Discussion

When attempt­ing to under­stand impacts of data stew­ard­ship, it’s impor­tant to look at the goals set and orig­i­nal intent for the cur­rent sys­tem. Both the pol­i­cy frame­work doc­u­ments and reports from 2018 can be looked at for this¹. In the 2018 report, researchers were con­sult­ed, and the pri­ma­ry needs were iden­ti­fied. Con­sid­er­ing the answers giv­en by data stew­ards and the orig­i­nal scope we can see how the orig­i­nal needs are being met by cur­rent data stew­ard activ­i­ties (Fig­ure 1).

Fig­ure 1: Table show­ing the orig­i­nal needs iden­ti­fied in the 2018 report¹ along­side the cur­rent activ­i­ties of data stew­ards based on answers pro­vid­ed regard­ing com­mon top­ics they con­sult on and the accu­ra­cy of the scope.

The answers regard­ing the scope and com­mon con­sul­ta­tion top­ics match up well with the orig­i­nal goals and serve as an indi­ca­tor of how data stew­ards are impact­ing research through ful­fill­ing the demands for advice and con­sul­ta­tion.

Suggestions

From the answers col­lect­ed, sev­er­al issues were brought up along with some sug­ges­tions of how they could be reme­died. Regard­ing updat­ing the scope, the opin­ion of data stew­ards was that gen­er­al­ly it was still applic­a­ble and valid how­ev­er, there were aspects where they sug­gest­ed improve­ments could be made. Two large roles were miss­ing, the liai­son role and their activ­i­ties in fos­ter­ing pol­i­cy devel­op­ment which could be used to update the scope.

When asked about poten­tial resource lim­i­ta­tions, the data stew­ards agreed on the pri­ma­ry lim­it being in exper­tise. The data stew­ards sug­gest­ed that more sup­port with legal and privacy/personal data issues would be help­ful, espe­cial­ly giv­en how they report­ed legal issues being so time cost­ly.

References

1.         Mar­ta Teperek. Views on Data Stew­ard­ship — report of pre­lim­i­nary find­ings at the Fac­ul­ty of Tech­nol­o­gy, Pol­i­cy and Man­age­ment (TPM) at TU Delft. (2018).

2.       Boeck­hout, M., Ziel­huis, G. A. & Bre­de­no­ord, A. L. The FAIR guid­ing prin­ci­ples for data stew­ard­ship: fair enough? Eur. J. Hum. Genet. 26, 931–936 (2018).