Your first problem is due to overlapping groups of duplicates. This is explained in detail on the "Overlapping duplicate groups" chapter in the Manual (from page 29), it happens when you import some references, run duplicate checking, import some more, run duplicate checking again and so on.

You can handle the situation manually, or use the  "reset" functions in the "Manual/advanced" tab, in case you have already marked many items as duplicates, you will want to to use the first "reset" option, the one that retains the work already done (it's explained on-screen). You will then re-run duplicate checking and get a clean list of groups, with no overlaps.

Re going through groups manually: you can use the Advanced Mark Automatically feature in "Manual/Advanced" to lower the similarity thresholds, and/or to allow to mark as duplicates items that have already been coded. Normally, this second limitation is enforced to prevent loosing work done elsewhere.

I hope this helps, please let us know if you need further guidance.

Sergio

Both reset options will delete all existing Duplicates Groups. The first option however preserves the "marked as duplicate" flag, as well as the master information, meaning that after the reset, the items that are already marked as duplicates will remain in their current state, and the system will know of what item they are a duplicate (the master). This information remains visible through the "item details" window: if an item is the master of some duplicates, it will show there, the same applies if an item is marked as the duplicate of something else.

After clicking "Get new duplicates" once more, new groups will be created, but items that are already marked as duplicates will not be included in any of the new groups. This will allow you to deal with the items that still need to be looked at, without having to re-do what has been done already.

In short, the first option allows you to avoid having to manually untangle overlapping groups while ensuring you won't need to re-evaluate items that you've already marked as duplicates. The drawback of this system is that in case you made a mistake, and marked as duplicate some items that are in fact unique, it will be difficult to spot the mistake (because the old groups have been deleted). Since your review is big I'm guessing that you will never want to manually look at all groups, so in your case this isn't a problem at all.

The second "reset" route is simpler: it deletes all groups, and re-instates all duplicates to their original "not a duplicate" state. This option is recommended if and only if you think you've made some big mistake and wish to start from square one.

I hope this helps,

Sergio

No, if you click the "reset" button, a window will open. With a long description of the two options that are available there.

Don't worry, the system is designed in such a way that it is virtually impossible to delete your work without noticing.

Hi,
I'm not sure about what you did exactly, can you state the following?
1) I assume you: used the first "reset" option, clicked "get new duplicates", went through the whole lot, then clicked "get new duplicates" again. This time new groups appeared. Is that the case?
2) while going through the steps above, have you by any chance undeleted a source? This would explain the new groups.

If a study is marked "not a duplicate", you reset the groups (both options), and "get new", the study will most likely re-appear and should be marked as "not a duplicate" again. Sorry for not mentioning this before.

Sergio

Hello Deny,

The screening process in a systematic review is a method of determining which of your studies are relevant to your review question. If you have a complex screening tool then looking at each study would probably be necessary. If you are just trying to eliminate items based on basic criteria such as date you could try running a search. You can search on the date field in the search tab. By identifying items to exclude through searching you can then bulk exclude a number of items. If you are trying to exclude or include items on study type or the region you might also be able to eliminate a number of items in one go as well through searches. Searching on particular terms might be a bit risky as those terms could be referring to different concepts than what you might expect. Using this method could reduce the number of items that you need to examine individually.

A second method you might wish to consider is using the text mining functions in EPPI-Reviewer to help identify the items that are most likely to be includes. This would allow you to put your resources where they would be most effective. You will find details on the text mining functions in the user manual under 'Text mining' (page 77). This might not eliminate the need to look at each item but it will allow you to look at the most likely candidates first.

Best regards,

Jeff