Dear Kate,

I suspect this is not a bug, but the direct consequence of (mis)using a code-set "multiple user data entry" mode. Code-sets should be placed in "multiple user" mode only when you wish to conduct a double/multiple blind coding/screening round. "Single user" mode is safe for all other tasks. We are aware that the mode-names are somewhat misleading and are about to change them (we hope to include this change in the next update, hopefully in mid June).
In detail: the "multiple user data entry" mode is designed to allow double/multiple blind coding, so that two or more reviewers may be able to assign codes to the same items without seeing each other's decisions.

When a code-set is in "multiple user" mode, different things may happen as the result of applying a code to an item:
1) if the item has no codes from the set, or if the item has "uncompleted" codes, the new code will be assigned as "uncompleted" and will be visible to the person that added it only. Reporting/searching features of EPPI-Reviewer will not detect this coding.
2) if the item already has "completed" codes in the set, the new code will be added to the "completed" collection, and will be visible to all users (and appear in reports, frequencies tables, searches, etc).

I've given a look at your review, and some items have "completed" codings from the screening set, some other have uncompleted codes, mostly applied by Katherine. What I think happened is that you had some items already completed, "bulk assigning" selected items to a given code would work straight away from them. The same operation (select some items, right click on a code and "assign selected items to this code") to items that don't have any completed coding in that set would add the new codes as "uncomplete" (or private) and the other members of the review will not see them.

To solve this issue, you can bulk-complete the coding that was already added (if my "diagnosis" is correct!), this is done on the "Review Statistics" tab (low right on the main screen), you can expand the lines of uncompleted codes so to see to who they belong, and then click on the green "v" little button to complete them all in one go (on a per-person basis). This should make all your "disappeared" codes visible again. Once you have completed all the coding that you wish, you may want to switch the code set to "single user" once more, so to avoid having the same issue again.

I hope this helps,

Sergio

Oh,

I'm sorry Kate, you are right: I completely misunderstood your issue.

The checkbox on the main "Documents" list can seem to behave erratically, especially because most lists are very long and sometimes the selections are (justifiably, see below) lost while they are not visible due to scrolling. The end result is that proceeding in the way you are is possible but a little dangerous, as you may indeed assign less items than the intended to a given code. There is little we can do to improve this situation as the table follows the standard conventions, and does so very reliably (departing from the standard may be even more confusing).

You may know the following, but I will spell it out explicitly as it may be useful to everyone; the "standard" convention works as follows:

1. Clicking anywhere on a given row will select (or unselect) the checkbox on the given row (you don't need to click on the checkbox). Any other selected checkbox will be unchecked. [I believe this is why pre-existing selections sometimes disappear without notice]
2. The most intuitive (and dangerous) way to add the "check" to more rows is to click on the checkbox of other rows, this will leave the already added checks unaltered, but is dangerous because if you happen to click anywhere else, the already present checks will disappear (as per 1.), if you have scrolled away and the previous selections are not visible, this may happen without noticing.
2. Fortunately, the behaviour above can be modified by pressing some keyboard keys while clicking on an row (not the checkbox).
3. The CTRL key is the most useful: tells the table to only act on the clicked row, leaving all other pre-existing selections unaltered.
4. The Shift key is powerful but can be easily mis-used: it tells the table to select (or unselect) all the items that are between the currently active row (usually the last one that was clicked) and the row that is being clicked. It is sometimes useful in combination with custom sorting (for example if you are sorting by Year and want to select all references from 2010).

As a result, when I need to go through a long list and want to select some of the items, I keep the CTRL key pressed at all times, scroll up and down as needed and click on the rows (never aiming to the checkbox at all) that I wish to select or unselect. This is faster that aiming at the checkboxes (requires less precision) and is much safer as well. Overall, I hope it will help you out!

Sergio