Configure iteration order of records in Screening/Tagging/Extraction#251
Just curious what determines the study order in the “normal” workflow (e.g. clicking on Screening/Tagging/Extraction instead of using Study Inspector). I’ve had 2 cases in the first 10 studies where they referenced another paper that I have in my nest that I haven’t reviewed yet, so I’ve gone back to Study Inspector to read “out of order” in order to have an accurate time sequence and read the old paper first. I can’t seem to figure out what the sorting logic is currently, but it might be cool to have a way to set the order preference (just like you can do in Study Inspector). And it may be more of a “nest-level” setting rather than yet another button on the interface to toggle all of the time. For my nest, I don’t see why I would want to change mid-study.
Anywya, not a big deal, but could be helpful/interesting. You guys probably already thought about this as well, but wanted to share my thought.
Good question! We don’t provide any specific guarantees on order; however, iteration order is approximately:
- Number of times the study was skipped by the current user (i.e. skipping a reference sends it to the back of the queue)
- For screening only, predicted inclusion probability (highest first)
- If no model has been trained, this doesn’t influence order.
- Time of import (earliest first)
The reason I say approximately is that:
- Collaboration screws this order up. If someone else is concurrently reviewing the “next” study in the queue you move to the next study in the queue
- We consider time of import sorting mostly an implementation detail, and it may be changed in the future.
This is an interesting idea, what types of sorts on records would you like?
Gotcha - I was wondering if the import/search order was in there somewhere. :) I’m the only one in my nest, so the searching is probably driving.
Off the top of my head, the only order I could think of that would be useful/interesting is oldest first (to get the historical picture) or newest first (skip right to the latest and greatest). Once there are tags applied, I could see how sorting by tags could be interesting, but that gets more complicated pretty quick and may be too much mental work to set up and may be too late to be useful. Also, I’m realizing that, at least for my study, it’s really only useful in the Tagging portion because that’s where I’m reading the study and pulling out useful information and that’s where the study order can provide additional context as far as the time-based aspect of it. My study is pretty much tagging-only and I’m looking for comments/discussion points in additional to objective details and that’s why it matter. Once I’m past the Tagging point, the order of studies in Extraction doesn’t really matter since it’s just data.
Apologies if I’m going too deep into the weeds here; but hopefully the relatively unedited “stream-of-consciousness” approach is useful.
Makes sense! Publication date does seem like the most obvious candidate. The other one I was considering was # of matching user keywords in TIAB, but that’s a can of worms and not terribly useful in many scenarios.
Agreed
@Jade Thurnham Made pretty much exactly this request in https://nested-knowledge.nolt.io/257
@Karl Holub My apologies I didn’t see this previous request sooner! Thanks for directing me over here.