Fiona Bradley

Workflow, routines, and reference managers

It’s funny how fast routines can be established. In the month following my move to Sydney and starting formal PhD work I quickly settled into a daily working pattern at one of the city’s major university libraries before starting my new job. I arrived in the morning and stayed all day, sitting in the same seat most days. Many of the student faces around me were the same everyday. Tweets about #PhDlife #WritethatPhD emphasize writing everyday, setting small goals. Alongside this are frequent conversations about workflow, tools, and routines. No wonder so much of the discussion in research from libraries, publishers, and startups alike is about workflow. Once a tool becomes a trusted, reliable part of your workflow it makes it hard to switch.

One of the longest standing tools in my workflow is Zotero. Zotero has been a trusted friend for well over a decade now, I’ve used it to write numerous reports, articles, and conference presentations. Unfortunately, given the slightly different demands of PhD writing, I’ve hit a snag.

I do all my notetaking, drafting, and writing in Scrivener. It’s an incredibly flexible piece of software (more below). However getting referencing software to interact with Scrivener properly is a major headache.

Instead of native cite-as-you-write integration that you’d find in in LibreOffice or Word, to use a referencing tool with Scrivener you need to copy an unformatted version of your reference into Scrivener and translate it later. For Zotero, that looks something like: { | Ashwill, & Norton, 2015 | | |zu:13542:P8UMW7ID}.

Need to get text out of Scrivener to share with collaborators or supervisors? To convert Zotero references, you need to:
paste references as scannable cite -> Compile -> RDF/ODF scan -> convert to citations or makers -> open in LibreOffice -> change preferences -> save as bookmarks -> save as docx -> open in Word -> share.

This works more or less ok if you only need to do it once at the end of a project. The real pain begins if you want to incorporate comments or changes back into Scrivener. Scrivener copies track changes and comments from Word perfectly, but Zotero cannot convert references, once formatted, back into a scannable cite, the version needed to keep on writing in Scrivener. Even at this early writing stage I can see this becoming a major roadblock later on as it would require me to manually copy and paste reference details every single time the text leaves Scrivener. This could potentially introduce all sorts of errors.

So, reluctantly, this means adding another tool into my workflow. I will continue to use Zotero as my main tool for creating references, tagging and annotation for all my projects. Plugins like ZotFile help keep my PDFs organised, and extracting comments from PDFs is not perfect but functional.

Next to Zotero, I’ll add Endnote for citations only. Endnote does not have a problem with formatting and unformatting references in Word, which means I can quickly switch referencing formats when I need to move text back into Scrivener. Sorted. But there are other challenges that arise from switching or using multiple reference managers:

  • Zotero can export references as RIS, for import into Endnote. However, this does not preserve Zotero’s folder structure or any links to full text.
  • You can manually edit the RIS files to point to the correct file location, but if you use Endnote Online to sync your references, the sync process will immediately delete the links. I have 1500 references in my library. This would mean manually exporting every folder/collection from Zotero separately, and reattaching every PDF reference by reference. No thanks.
  • Zotero and Endnote deal with organisational author names in different ways, and these too all need to be manually edited (eg OECD is imported as Oecd by Endnote. Just. Why).
  • Endnote also lacks the ability to extract annotations from PDFs.

I’d rather be writing than spending my evenings fixing this stuff. One benefit of using both though is that Endnote should be slightly zippier without managing PDFs and notes etc as well. And, I’ll still have access to Zotero after I graduate, with all my notes and tags and folders, unlike Endnote.

Another task I’ve been working on is cleaning up metadata imported from Google Scholar, usually by going back to the publisher’s website now that I have institutional access. Like many students, I only had access to open access content while I was preparing my application. Ponder that when you consider the challenges of getting into graduate school in the first place – not having access to all the works you need to prepare a strong proposal. Thankfully, I got most of what I needed via open access, but a lot of the metadata left much to be desired. Records can and do disappear from Google Scholar more often than you’d think. However, sometimes the metadata on publishers’ websites is more detailed than reference managers expect. They can’t parse this sort of thing and often leaves a blank since they can’t figure out what year to include:

Issue Online
21 February 2017
Version of Record online:
03 February 2016

The reason I’m willing to persist with switching to a hybrid Zotero-Endnote approach is that for better or worse, my discipline is still oriented around Word/Endnote, and because Scrivener is now central to my workflow. I keep all of the summaries of articles I read in Scrivener, copies of emails exchanged with my supervisor, and my drafts. I can create new versions in the same file with a click, roll back if needed, and move around bits of text and storyboard whenever needed. I can add keywords and references to other documents to each file, meaning that if I want to write up a chapter around say, compliance or China, I need only to search and everything I need will be right there. If I need to go back to the source, ZotFile creates hyperlinks when it extracts annotations from PDFs, and clicking on those takes me directly to the annotation on the file. I was dragging PDFs into Scrivener for a while, but now just type notes directly into a floating scratchpad. I do still need to export out of Scrivener to share work with journals or supervisors, but that’s an easy step to compile the work right at the end.

For those working with Scrivener in more detail and curious about setup, here’s how my binder is organised:

  • Chapter outline
    • Intro
    • Literature Review
    • etc – remainder of chapters
  • Chapter 1 CURRENT DRAFT (this is what I’m currently working on)
    • Intro
    • Research Questions
    • Scope
    • Literature review
    • etc
  • Summaries
    • Nye (2012)
    • Wendt (1992)
    • Easterley (2006)
    • etc – as noted above every article/book gets a separate page. I then add the Endnote reference at the top so I’m sure which work I’m referring to.
  • Ideas
    • Lots of pages with points to pursue, data sources and so on
  • Quotes
    • Compelling finds along the way
  • Feedback and comments
    • copy and paste of correspondence with supervisors so I can track progress and how I’ve responded
  • Admin
    • University guidelines, my bio, forms etc