Implementer

[andylolz]

Description

The implementer of the activity is the organisation which is principally responsible for delivering this activity.

Note: This test has not been modified since the 2016 Aid Transparency Index

Proposed test

IATI 2.0x

For each current activity,
if `activity-status/@code` is one of 2, 3, 4
then participating-org[@role="4"] should be present

IATI 1.0x

For each current activity,
if `activity-status/@code` is one of 2, 3, 4
then participating-org[@role="Implementing"] should be present

[YohannaLoucheur]

Recent work based on IATI data (some of which was presented to the TAG) show that the quality of the information published about the implementing organization is very important to make the data usable. Any way by which the quality can be assessed would thus increase the value of this test, and of the Index. Just having something in this field should no longer be enough.

Based on our work on Tanzania data, I would suggest to (strive to) test that meaningful information on the implementer is provided, especially the following:

• Is the name of the organizations provided, not a generic term like "CSO"? Spot checks could be done on a sample of projects - in our experience, publishers use generic terms everywhere or not at all, so a fairly small sample should suffice. Or perhaps comparing the contents (which a machine can do): if the same terms appear more than eg 25 times in this field, it could be flagged for a human to check whether there are indeed more than 25 projects with the same org, or whether it's a generic term.

• Is the org ID number provided? I would expect scores to be fairly low initially, but we can see the progress over time. Perhaps the Implementer element in the Index could be a composite score, and the ID would be given a low weigth (but still count). At a minimum we should expect ID numbers for organizations based in donor countries, since they do have registration agencies.

• Is the type of organization provided, and is it accurate? Accuracy could be done with a spot check, perhaps completed with business rules (e.g. for big organizations that have their own channel codes; cross-checking the type of org with the type of aid)

Finally, note that the above would also be relevant for other org roles, especially funding.

[andylolz]

Many thanks for raising this, @YohannaLoucheur. We’re interested in exploring this, and have been discussing similar ideas (on both org names and org IDs) with @rolfkleef and @markbrough.

[andylolz]

Thanks for comments here. This test will remain as proposed.

There is certainly some useful data use work to be done in this area, but it probably isn’t for the index. It would be great to score publishers on the use of org IDs, but unfortunately the infrastructure isn’t there yet to make this possible.