Hello,

 I have an issue with the reconciliation fonction, when i want to use it with my data extraction tool; After selected "View" and then click on reconcile,  i have this kind of error message:

"[Arg_Null ReferenceException]

Arguments : Les chaînes de ressource de débogage ne sont pas disponibles.

http://go.microsoft.com/fwlink/?linkid=106663&Version=5.0.20125.00&File=mscorlib.dll&Key=Arg_NullReferenceException"

Could you help me to solve this problem please?

Hi again,
this appears to be a genuine bug, and something that was only half corrected already. Specifically, it is happening when a comparison is created in quite peculiar circumstances:

The error shows up if and only if the second reviewer in the comparison has applied to the items in the comparison some codes that belong to a branch that has been deleted (after applying the codes).

I can solve this once and for all, and make sure the error will never show up again, but I will need to publish a new version of EPPI-Reviewer to make the fix public. We don't have enough updates to justify a new release at the moment, so, in order to solve your immediate problem, I might proceed in two different ways, and would like you to pick the preferred option. Both options require to delete data, so I will not proceed without your explicit instructions.

Option 1: this is the less destructive. I could remove the codings applied to codes that belong to deleted branches in the "Data extraction" set. It's important to note that because the branches are deleted, no information about this coding is available or visible in EPPI-Reviewer, however, the information is still there as a precaution: if deleting a branch was accidental, it allows us to recover the lost information (branch structure and associated coding) very quickly. This option removes the data needed to recover the "associated coding", but retains what's needed to reconstruct the deleted branches.

Option 2: this deletes more data. I could simply remove completely all the information associated with the deleted branches (both coding and branch structure). As before, this information is already sealed off and unreachable from within EPPI-Reviewer. Thus the only reason why we might not want to eliminate is if you think one or more branches in the "data extraction" sets might have been deleted by mistake.

Thus the criteria is the following:

A: if you think that one or more branches were deleted by mistake, I should not delete anything and restore the missing branches instead.
B: if you are confident that the codeset has the correct shape (no codes-branches are missing) then I can proceed with option 2.
C: I don't really see a scenario for choosing option 1. If you can't decide between A and B we should discuss further and agree on the next step once we've clarified the details. I can easily send you (via email) the list of codes that belong to deleted branches, in case that would help.

I apologise once more for the inconvenience,

Sergio

Hi Matthias,

I've applied Option 2: all stale codes that belonged to deleted branches in your "data extraction" set have been deleted. With them, also all the "item-code" associations (what we normally call "coding") have been removed, so, unless you'll delete new branches, the error will not show up again. I've checked your comparisons and I can confirm that this worked as expected.

If you'll delete another branch, the error may reappear in the short term. If this happens, please get in touch and we'll revive this discussion (apply option 2 again, if feasible). However, this is happening because of a bug: as soon as a new version of EPPI-Reviewer will be published (I don't have a set date, it's likely to be January, but may be later), the error will disappear once and for all and no one will need to take extra precautions to avoid it.

In the mean time, in case you prefer not to ask for our help, the easiest way to avoid getting the error again (if and only if you need to delete a branch which contains codes from a codeset where reconciliation needs to be done) would be:

- Do not delete a whole branch (avoid deleting the branch root).
- Instead, delete each single code, starting with those that have no children (and/or delete those that have "to be reconciled" codes applied, and when done, delete the whole remaining branch in one go).

Let me stress that we don't wish EPPI-Reviewer users to go through this process, it's an unnecessary waste of time, so please remember that after the next update no additional precaution will be need.

More in general, I'd like to offer some unrequested remarks:
Reconciling the differences of a full data-extraction exercise is always tricky. The reconciliation window doesn't really help too much, it's designed to aid the reconciliation of screening exercises (where it visually helps spotting the differences). In case of data extractions, the differences are normally too many to reconcile in this way.

However, the reconcile window helps in 3 ways:
1. You can pick which version to complete. By contrast, completing a coding from the "Document Details" window only allows to complete the coding that belongs to the currently logged-on user.
2. You can view and save the data shown there in tabular format. This then allows to discuss the differences and decide what version to complete and what amendments to apply to it (see below for a popular alternative).
3. You can click on the link showing the item ID and thus open up the item details (this could be useful both before and after completing the coding, depending on how you are planning to proceed).

Note that from the "Document Details" window, in the "Coding Record" tab you can produce a different comparison report, which will show the codes applied in a tree-like form (instead of tabular). I think this latter form is slightly better, but I guess it depends on the specific circumstances. To get this report, select the two lines for the "Data Extraction" coding and click "Run Comparison" (opens a popup, so watch out for the popup blocker!).

Please do not hesitate to get back in touch whenever you'll experience any difficulty: we are here to help.
Thanks again,
Sergio
 

Hello Matthias,

What we will need to know to see if the missing codes exist is the review and the codeset it was deleted from. It will also help to know where in the codeset the code that was deleted was situated.

Best regards,

Jeff

Hello Matthias,

We are able to look for orphaned child codes. If a parent code is deleted it is gone for ever but its child codes become orphans that we can bring back (by creating a new parent).

I had a look at this review and I was a bit confused. In the 'Quantitative data' branch of the 'Data extraction tool' the deleted parent codes you have listed are still there. I can see codes called 'Withdrawal rate due to AE,  n (%)', 'Proportion Female,  n (%)', etc. in the coding tool. Am I misunderstanding what you are saying?

Best regards,

Jeff