I find reading scientific articles hard. It's not a particularly efficient way for me to learn. I prefer conference presentations, with gesticulation and animations and lots of colour. I know people who have the opposite preferences, and that's OK.
But I’m a PhD student, and reading papers is unavoidable.
I've been reading a paper this last couple of days and realised that in my years as a PhD student I've learnt of a few tools/techniques that make my reading far more time-efficient. I thought I'd share some of them here.
I've been reading a paper this last couple of days and realised that in my years as a PhD student I've learnt of a few tools/techniques that make my reading far more time-efficient. I thought I'd share some of them here.
Here's the paper I'm reading: doi.org/10.1016/0042-6… for demonstration. Since I’m presenting this as a live(ish) demo, there are going to be some dead ends (things I try that don’t work), but that’s fine because that’s life.
(I like using DOIs rather than URLs because they're stable, they avoid subconscious bias based on journal prestige, they are unequivocal (e.g. for similarly named articles), and they make it easy to use @Sci_Hub (which obviously I don't use or endorse))
So, here's the process I go through when reading the paper:
Someone recommends I read the paper. They provide the details "Werner and Walraven 1982". No title, no journal. Cool.
I've set up Firefox with a "smart keyword" so that when I type "sch werner walraven 1982" into the omnibox it takes me directly to the Google Scholar results page for "werner walraven 1982". So speedy.
I hit the @zotero firefox connector button and the metadata and pdf is downloaded to my PC - because who's got time to actually find the pdf download button on a publisher's website right?
This works automagically when I'm on campus (because @UCL pays squillions to access almost everything: edchamberlain.wordpress.com/2011/12/09/a-m…) but if I were offsite, and the article wasn’t gold open access, I would fall back on the @unpaywall firefox add on.
(If I was more piratey I might consider the zotero-scihub add-on: github.com/ethanwillis/zo…)
Now I actually get to the part where I read the article, digitally highlighting with wild abandon and making notes of the bajillion other things I need to read in able to be able to understand everything in this paper.
I take regular breaks, and note that my ability to concentrate is worse than normal today.
Finished.
Before I invest too heavily in this paper (second reading, skimming the references I’m not familiar with etc.) I'd like to understand the context of this paper.
Did it cause a revolution? Was it panned? Have the ideas in it been developed/usurped since?
Before I invest too heavily in this paper (second reading, skimming the references I’m not familiar with etc.) I'd like to understand the context of this paper.
Did it cause a revolution? Was it panned? Have the ideas in it been developed/usurped since?
My first port of call is the @scite plugin. This sits unobtrusively on the side of my browser and gives me a quick indication of how supporting the following (citing) literature was/is.
For this paper it says that there's 1 supporting, 23 mentioning and 0 contradicting (this is roughly average in my experience, for this field and this age of paper).
For a newer paper I might glance at the @altmetric details to see if anyone has said anything interesting about it anywhere, or if it’s resulted in patents or Wikipedia articles etc. (I’d use the Altmetric Bookmarklet: altmetric.com/products/free-…)
Also for newer papers I might check out @PubPeer to see if there has been any post-publication peer reviewing.
Neither great for old papers though.
So, clicking through the @scite link (takes me here: scite.ai/reports/10.101…) shows me snippets of text from the papers which cite this paper.
So, clicking through the @scite link (takes me here: scite.ai/reports/10.101…) shows me snippets of text from the papers which cite this paper.
This saves me manually finding the papers and working out where they cite this one specific paper, nice. I read through and see that most of them just mention it in passing as an early example of this experimental technique (achromatic setting). Fine.
The one of special interest is the one marked as 'supporting' (doi.org/10.1016/0042-6…). It seems like @scite use a black box to make these distinctions, but it’s enough to make me actually skim the paper and see what they’re saying.
(Summary: they use the assumption laid down by W&W that adaptations occur along straight lines, despite others having found minor variations from this. Actually not all that supporting.)
This is fun but @scite has only "Over 400m classified citation statements". Good for detail (means I don't have to trawl through all the other papers looking for 'how' the paper is mentioned) but poor for breadth (I think it only has access to OA pubs).
And so it’s back to the big boy. On Google Scholar, I click one of my favourite buttons: 'Cited by n' (in this case n = 116). Citation ≠ endorsement, but together they do give an indication of where these ideas led to.
I recognise a review paper that I know well and trust, pull it up in Zotero and CTRL+F the first author. Only mentioned in passing as an example of the method. Cool. Fine. Whatever. Next.
I realise that what might be really useful is a thesis which summarises this line of study (and also confirms whether I have actually understood the meat of the paper).
It seems a common idea that ‘nobody reads theses’. I do. I think they’re great. I get a plain-speaking, textbook style, ground-up, non-space-constrained summary where the explicit goal is to distil. Fiddly to get hold of though.
I fiddle around with Google Scholar Advanced search trying to make it give me ‘theses which cite X’. Seems like this is impossible (please correct me), and so I turn to a database I’m aware of: ProQuest Dissertations & Theses Global™. UCL has a subscription, thankfully. Sorry.
(I wish openthesis.org and oatd.org were better populated, but they’re not. The world sucks.)
On ProQuest I do an advanced search asking for every thesis that includes ‘werner walraven’ in the document text. 375 results, the top 7 of which look relevant. I hit the Zotero button and quickly get copies (incl pdfs) automatically downloaded.
Flicked through the abstracts for all of them. One (the top result) has a neat rundown of the state of thinking around 2000 on the interplay between chromatic adaptation and colour constancy (v. relevant for understanding the W&W paper). Neato.
(Color constancy from chromatic adaptation. Landsberger, David Mark. Brown University. 2001. search.proquest.com/pqdtglobal/doc…)
This is where I took a break to go for a walk, think things through, and make my next plans.
I’m going to go and read some supporting papers on some subjects that I really should know more about, and would help me to understand this paper better (Stiles’ pi mechanisms). TTFN! (en.wikipedia.org/wiki/TTFN)
OMG you read this far. Hope you’ve found this interesting/useful.
HT to @dsquintana for showing that threads about academic things are cool and can be useful.
Please reply with ways I can do this better.
HT to @dsquintana for showing that threads about academic things are cool and can be useful.
Please reply with ways I can do this better.
