1 Digital repositories and versions of academic papers Frances Shipsey and Louise Allsop, VERSIONS Project Library, London School of Economics and Political.

1

Digital repositories and versions of academic papers

Frances Shipsey and Louise Allsop, VERSIONS Project

Library, London School of Economics and Political Science

ALISS Christmas Special: Libraries and Open Access Scholarship

British Library Conference Centre, 11 December 2006

The VERSIONS Project (www.lse.ac.uk/versions)

• VERSIONS: Versions of Eprints – user Requirements Study and Investigation of the Need for Standards

• Funded by the Joint Information Systems Committee (JISC) under the Digital Repositories Programme

• London School of Economics and Political Science (LSE) lead partner ; Nereus – consortium of European research libraries specialising in economics – associate partners

• Eprints – Economics – European

• July 2005 to February 2007

Nereus – a network of European economics research libraries www.nereus4economics.info

Economists Online – a pilot search service - http://nereus.uvt.nl/eo

Economists Online – institutional pages

Economists Online – an author page

Economists Online – content is stored in Nereus partners’ institional repositories

LSE institutional repository cover sheet

Notes indicate that differences between this version and published version remain

Focus on economics

• Established preprint culture – working papers and use of RePEc archive – discipline is already open access?

• Sue Sparks report on disciplinary differences:

‘What is the single most essential resource you use, the one that you would be lost without?’ Economists responded:

• 18.2% preprints• 9.1% postprints• 54.5% journal articles• 18.2% datasets

Sue Sparks. JISC Disciplinary Differences Report. Rightscom Ltd, August 2005. Appendix C, Table 43. http://www.jisc.ac.uk/uploaded_documents/Disciplinary%20Differences%20and%20Needs.doc

Issues regarding versions

• Do authors have the ‘usable’ version to deposit in IRs?• Can they produce it (easily) on request?• Level of awareness about publisher permissions• What are their attitudes towards making these versions

publicly available?• Any differences between UK and other European

countries regarding population of repositories• Experience of researchers – is it a problem sorting

through multiple versions?• Citing other authors’ work - issues

VERSIONS Project – User requirements study 2006

• Online survey ‘Versions of academic papers online – the experience of authors and readers’, conducted May-July 2006

• 464 responses from academic researchers• 76% of researcher respondents from economics and

econometrics• A variety of roles, from PhD student through to

professor• Good geographic spread• In addition, 133 responses from stakeholders - separate

survey, not covered in as much detail here

VERSIONS Survey of academic researchersCreation, storage and dissemination of versions

• Research active (50% wrote 4 or more papers in the past 2 years)• Interviews with researchers showed very large numbers of

revisions being produced and kept (as many as 60 or 70 in some cases)

• More difficult for researchers to retrieve older papers• May be left in a previous institution• Different servers, PCs and other storage media – dispersal• Problems with older software packages• Some much older material not available in digital form• Drafts and revisions not clearly labelled, so researcher cannot now identify

wanted version

How many versions do researchers produce?

• Researchers regularly produce numerous outputs from a single research project

• 59% typically produce 4 or more different types of research output per project, 33% produce 5 or more (Question 4)

57

408363

273

153

6619 6 1

050

100150200250300350400450

Which types of output do researchers produce?

• Journal articles are the most common output, with a wide range of others preceding, accompanying or following them (Question 4) 0

50

100

150

200

250

300

350

400

450

500

Conference paper

Presentation

Working paper-no quality control

Working paper-quality control

Mem

ber series-NB

ER

,IZA

Journal article-refereed

Journal article-unrefereed

Report for funding body

Book chapter

Book

Dataset

Thesis

Other

Which versions do researchers keep?

The majority of respondents personally keep / plan to keep major, but not all, revisions of their research papers stored in electronic form (e.g. computer or network drive) at the end of the process:

36%

54%

8%2%

0% Keep all revisions

Keep majorrevisions but not all

Keep the latestrevision that Iworked on only

Other

Do not keep

Permanent storage by authors of multiple versions of their journal articles

VERSIONS survey of researchers Q7: ‘Which of the following versions of a paper, that you have written for publication in a refereed journal, would you personally keep (eg on your own computer or network drive)?’

Revision stage Percentage of respondents who keep this stage permanently

Number of respondents who keep this stage permanently

Early draft version(s) before circulating to anyone, other than co-authors

39.9% 185

Draft version circulated to colleagues or peers for feedback before submission

53.9% 250

Version submitted to a journal for peer review 78.9% 366

Final author version produced by yourself/co-authors – agreed with the journal following referee comments

90.7% 421

Proof copy (publisher-produced version) 62.5% 290

Final published version (publisher-produced PDF) 91.8% 426

Satisfaction with management of personal versions

Survey respondents were split between satisfaction and dissatisfaction with the organisation of their own revisions and versions, on their own computers or storage mediums (Question 9):

49%48%

3%

0%

Yes

No, not completely

Don't know

Don't produce researchoutputs

Personal version management - examples

• Systems detailed include dating in file name, version control by number, by software system, and retaining the latest version only.

‘Every filename includes a date e.g. hello111206. That way it is easy to find the latest version among co-authors and for myself’.

‘I use version numbers e.g. “paper 2.1.doc”, changing the second number with each edit of any significance and the first number if there is a milestone in the process – team review / change of direction etc. I keep the milestone versions in a backup folder within the main folder that the document is developed in.’

‘Version control system – CVS or Subversion.’

‘I usually throw away everything as soon as I have a new version. Unless, for example, the new version is in another language or has some substantial changes in it, so that I may need the first version for some other purpose.’

Examples - continued• Problems cited include changing between computers, hoarding too many

versions, management of co-authored papers, maintaining an awareness of differences between versions, insufficient naming systems and accidental editing of the wrong versions ‘I am using two different computers for my work. As a result, sometimes I have my work at different stages on the two computers. I would like to find an easy way to get both systems up to date at any point in time.’

‘I think that I keep too many old versions. I like to keep several while working on the paper in case files get corrupted etc. but I seldom go back afterwards and delete all unimportant versions.’

‘The problem is that co-authors sometimes do revisions on the wrong version. We don’t agree which is the latest version.’

‘What I am not so efficient at is distinguishing between versions with limited differences, and those where substantial changes have been made.’

‘I do not have a consistent renaming system. This causes major problems in finding the correct version after a long period.’

Responsibility for secure storage of different versions (Questions 10-13)

0

50

100

150

200

250

300

350

400

450

500

Early

dra

ft

vers

ion

(s)

Dra

ft to

co

lleag

ues/p

eers

Su

bm

itted

to

jou

rnal-p

eer

revie

w

Ag

reed

with

jou

rnal p

ost-

refe

reein

g

Pu

blis

her p

roo

f

Fin

al p

ub

lish

ed

vers

ion

(ofte

n

PD

F)

No

ne o

f these

Do

n't k

no

w

Authors/Co-authors Authors' institutions (inc. libraries) Publishers Subject repositories

Dissemination of different versions

• In addition to refereed academic journals and university/institutional collections, respondents disseminate their full text research findings through a range of other channels (Question 17):

- Personal website / homepage [301 respondents]

- University website for working paper / discussion paper series [256 respondents]

- REPEC (IDEAS, EconPapers) [209 respondents]

- SSRN [181 respondents]

- Other [60 respondents]

Which versions?

• Respondents were asked which versions of their academic papers they were interested in making openly accessible, if permitted (Question 19):

116191

274

100

385

3 30

50100150200250300350400450

Draft to

colleag

ues/p

eers

Version

for p

eerreview

Fin

al version

,p

ost-refereein

g

Pu

blish

er pro

of

Fin

al pu

blish

edversio

n(o

ftenP

DF

)

Do

n't kn

ow

Do

n't p

rod

uce

Copyright awareness

• The survey revealed significant uncertainty relating to copyright issues among its respondents:

53.7% of respondents reported limited or no understanding of which version(s) of academic papers, intended for publication in refereed academic journals, they are allowed to disseminate in full text, in which locations, and at which times

46

136

182

40

10

0

50

100

150

200

Full understanding Someunderstanding

Limitedunderstanding

No understanding Don't know

Num

ber o

f res

pond

ents

Copyright awareness in relation to repositories

• 68.3% of researchers stated that they were unsure whether publisher copyright agreements permit them to place final author versions into institutional repositories:

Question 16a, ix) To what extent do you agree with the following: ‘I am unsure whether the publisher copyright agreement permits me to provide this version [final author version for use in an institutional repository]’?

138

179

36 36

72

30

50

100

150

200

Stronly agree Slightly agree Slightlydisagree

Stronglydisagree

Don't know Don't producepapers

Num

ber o

f res

pond

ents

Citing versions in the face of change

• Respondents were asked how they prefer to cite earlier versions of papers that have subsequently been published in journals (Question 24):

12

22

58

33

339

0 50 100 150 200 250 300 350 400

Don't know

Do not cite any version of the paper if I have notread the published version

Cite earlier author version that I have read

Cite published version and earlier author version thatI have read

Cite published version only

Responses indicate that many researchers spend time reading both versions to ensure no major differences

Multiple versions – experience of readers

• 93% of respondents reported finding more than one full text version / copy of a paper online (Question 22)

• When asked whether it is generally quick and easy to establish which version(s) they wish to read, respondents answered as follows (Question 23):

54%41%

5%

Yes

No

Do not find multiple versions

Recent projects and initiatives on versionsOngoing standards development work:• NISO/ALPSP Working Group on Versions of Journal Articles

- Publisher-led group, with larger review group made up of publishers, librarians and other stakeholders- Draft documents including Terms and Definitions for versions (March 2006): Author’s Original, Accepted Manuscript, Proof, Version of Record, Updated Version of Recordhttp://www.niso.org/committees/Journal_versioning/JournalVer_comm.html

Two JISC activities during 2006:• RIVER – Scoping Study on Repository Version Identification

- Led by Rightscom Ltd, with partners London School of Economics and Political Science Library, University of Oxford Computing Services- Defined two broad classes of requirement for version identification (Collocation and Disambiguation), and defined a tentative typology of ‘versions’ (March 2006)http://www.jisc.ac.uk/uploaded_documents/RIVER%20Final%20Report.pdf

• JISC Eprints Application Profile Working Group- Approach based on Functional Requirements for Bibliographic Records (FRBR) and DCMI Abstract Model, more detail and structure than Dublin Core (June – August 2006) – work going forward through DCMI Task Group- Deals with versions very wellhttp://www.ukoln.ac.uk/repositories/digirep/index/Eprints_Application_Profile

How can social science librarians advise academic authors?• Provide information to researchers about permitted use of different

versions of journal articles through the SHERPA/RoMEO database: www.sherpa.ac.uk/romeo.php

• Explain the term ‘postprint’ in researchers’ own language: eg ‘accepted version’ or accepted manuscript’

• Strongly encourage researchers’ to keep these ‘accepted manuscript’ versions (in Word as well as in PDF) and to obtain them from co-authors if they don’t have the latest

• Encourage use of date in authors’ manuscript versions (date manuscript completed)

• Provide general information about depositing papers • BOAI Self-Archiving FAQ - http://www.eprints.org/openaccess/self-faq/ • Peter Suber’s Open Access Overview -

http://www.earlham.edu/~peters/fos/overview.htm

• If appropriate, provide advice about ongoing management of versions – consider whether this can be incorporated into training for research students

How can librarians improve version identification in their IRs?• Store metadata in richer format than simple Dublin Core, for

example Eprints Application Profile• Add version identification metadata• Consider use of cover sheets (some search engines take users

directly to the full text document, bypassing metadata)• When evaluating repository software, include version identification

in criteria• Look at RIVER report recommendations to universities (IR

managers)• Monitor JISC work on version identification and future guidelines• Encourage your library to develop a policy on version identification

in the IR

Reproduced by permission of the National Library of Ireland and X Communications

Thank you!

Frances Shipsey [email protected]

Louise Allsop [email protected]