Version 4.11.3.0 brings a new "Distribute Work" functionality to EPPI-Reviewer Web and a range of improvements to the Microsoft Academic Graph (MAG) powered features (they still are "on invitation", more information is here). This release also includes a number of bugfixes and enhancements, for both Version 4 and EPPI-Reviewer Web.

Microsoft Academic

We're in a phase or rapid development as these features are being road tested and their current weak spots are being identified. Since these features are still experimental and accessible only on invitation, listing them all here is inappropriate. Within the EPPI-Centre, we are starting to incorporate MAG-powered features into our workflow, a process that will allow us to better understand what features are both useful and suitable for the wider community of EPPI-Reviewer users.

New Feature: Distribute Work wizard

In the collaborate tab of EPPI-Reviewer Web, there is a new button in the main toolbar, called "Distribute Work". Clicking it provides access to a new "distribute work" wizard, which "joins together" the functionalities used to assign (and track) work. Currently, in most cases, users need to first "Create reference groups" and then use these to "Create coding assignments". The new wizard allows users to perform and "join together" these, via a series of guided steps. The effect is that it is now possible to quickly produce rather complex assignments, including the much requested feature of assigning a precise number of items to specific people. The wizard will also perform a number of "sanity checks" to ensure that the users' instructions are consistent and will not be counter-productive.

It's important to note that this wizard does not replace the existing functionalities, but complements them. Common scenarios where this wizard will be useful are:

• When starting a new screening/coding round. This is true especially for new EPPI-Reviewer users, who would otherwise need understand how both "reference groups" and "coding assignments" are expected to be used in conjunction.
• When the aim is to distribute a fair amount of work to numerous collaborators, especially if there is a need to distribute the work unevenly.

In other situations, such as redistributing already assigned work, it is possible that this new wizard will not be particularly helpful, which is why the pre existing set of functionalities might still be useful.
Please feel free to provide feedback on this new feature; it represents a significant new development in our efforts to make working in EPPI-Reviewer simpler and less demanding. Should this feature prove to be popular, we plan to provide a similar one, specifically aimed at organising screening rounds.
Please also note that this feature is not currently documented, we will include it in the online help soon and will also release a "how to" video dedicated to it.

Bugfixes and enhancements

In the Items Details page of EPPI-Reviewer Web, clicking on the "info" button for any given code would make the page suddenly scroll to the bottom, which resulted in considerable inconvenience: to resume work, people would have to (frequently) scroll back to where they started. This problem is now solved.

When deleting a code from a coding tool, EPPI-Reviewer (both versions) would first run a check. This check reported how many items have been assigned to the code in question. We have added a second check, to see if the code in question is used to drive a coding assignments. From now on, EPPI-Reviewer will refuse to delete a code that is currently used to drive a coding assignment and will advise to delete the assignment first. Before, users could attempt to delete the code, which would generate an error and result in a partial deletion (would remove the code from the tree, but would not actually delete it).

In very rare occurrences, fetching the details about personal coding assignments ("My work" in EPPI-Reviewer Web, and "My work allocations" in the "My info" of EPPI-Reviewer 4) and the "review stats" (EPPI-Reviewer 4) / "Coding progress" (EPPI-Reviewer Web), could fail with a timeout. We tweaked the underlying queries in a way that should prevent such problems in the foreseeable future.

Random Allocations could sometime fail to complete, especially when allocating large numbers of items (starting at around 25,000). We've done some work on the data-routines which could sometimes time-out (creating this problem), the result is now much faster (up to 7 times), with the bigger improvements being obtained for large allocations (the more items and groups are used, the bigger the improvement). [This has no impact on the new Assign Work wizard, which uses its own dedicated logic on the server side.]

Future plans

While we will keep working on the MAG-powered features, the next major improvements in our pipeline are:

• A new deduplication/grouping algorithm, which is finally in the last rounds of testing/refining and will surpass all the major shortcomings of the current system.
• We also plan to port the "setup priority screening" features in EPPI-Reviewer Web, which we hope will be ready for the next release.