How Readers Discover Content in Scholarly Journals Summary Edition Comparing the changing user behaviour between 2005 and 2012 and its impact on publisher web site design and function. By Tracy Gardner and Simon Inger Published by Renew Training Copyright © Renew Training, Tracy Gardner and Simon Inger ISBN 978-0-9573920-3-8 Renew Training Fernhill, Church Lane, Drayton Abingdon, United Kingdom The full version of this report (123 pages and 90 figures) is available in Kindle (for Kindle, iPad, PC and Mac) and PDF formats, the entire survey results data set upon which this report is based, and the analytical framework are available for purchase at Summary Edition Page 1 ACKNOWLEDGEMENTS This work would have not been possible without the support of the following organisations, who between them executed hundreds of thousands of invitations to journal readers asking them to participate in this survey. Our heartfelt thanks go out to them all. BMJ Group CABI Cambridge University Press IOP Publishing Nature Publishing Group Palgrave Macmillan Publishing Technology RSC Publishing SAGE Summary Edition Page 2 TABLE OF CONTENTS Acknowledgements 1 Table of Contents 2 Introduction 3 Methodology 4 Demographics 5 Discovery Resource Preference 8 Discovery Resource Conclusions 16 The Role of the Library 18 Search Engine Preference 20 Device Preference 21 App Use 22 Publisher Web Site Features 23 Conclusions 24 About the Authors 25 Summary Edition Page 3 INTRODUCTION This summary report is the output of a large scale survey of journal readers (n=19064) about journal content discovery conducted during May, June and July of 2102. While statistics and analytics can tell us some of this information, there are many gaps in the knowledge that these can provide which we have endeavoured to fill by asking readers what how they discover journal content. Summary Edition Page 4 METHODOLOGY This research carries on from, and expands upon, previous research undertaken in 2005 and 2008 (also by Simon Inger and Tracy Gardner) and attempts to follow the trends in behaviour over that period of time. Naturally, each time the survey is repeated, the authors have sought to keep the questions as consistent as possible with the questions in earlier surveys whilst keeping terminology current and tracking new developments. For this reason the three key questions on reader behaviour were modified a little, some options being reclassified and additional options created. However, since those questions don’t limit how many starting points the reader acknowledges as being important, this approach should have minimal impact on the results for any option present in the survey all the way from 2005 to 2012. Other questions were dropped completely, since the conclusions from these in 2008 are now so widely accepted as fact (and easily checked with analytics) that these were not tested. These included asking readers where links from discovery products would take them in publisher web sites, the answer being predominantly at the article level. The full methodology is available in the full report available at: Summary Edition Page 5 DEMOGRAPHICS Figure 1 As shown in Figure 1, the respondents to the survey come from an excellent regional spread and this allows for significant regional (and in some cases country) breakdowns, see below. 13308 people told us their country (and hence region). Figure 2 9406 people came from the top 15 countries shown above, i.e. 71% of those who indicated a country. Summary Edition Page 6 Figure 3 The sector-breakdown is also very good, with sufficient numbers in all but Charity/NGO to allow for further breakdowns by subject, region, income and so on. 18958 people told us which sector they worked in. Figure 4 The numbers of respondents by job role allows for meaningful comparisons to be made for all roles save for, perhaps, journalists and marketing/PR/sales roles. There are sufficient responses within academic researcher, lecturer and student categories to allow for considerable further demographic analysis within these groups, including by subject, region and income. 17403 people told us their job role. Summary Edition Page 7 Figure 5 When embarking on this project, our aim was to get around 1000 responses in each subject area, so that detailed demographic analysis would be possible within subject with minimal error-bars. This was achieved for six of the subject areas, and three more subject areas got around 600 responses, which still allows for some useful further sub-division. Computer Science, Earth Science, Environmental Science and Mathematics are the least wellrepresented in the data, although in absolute terms there are enough individuals in these areas to allow for a useful subject-based analysis, even if not when combined with a further demographic.