How do YOU explain double counting?

Hi everyone

Every couple of months I host a session about “Using IATI data” with staff from the European Commission. It’s very interesting to meet people who need data, and we get through a lot in the three hours. It also includes a “d-portal quiz” (thanks to @amys for that idea - I’ll detail that in another thread)!

During the session, people get familiar with the concepts of IATI, and start to use the data. d-portal is very useful, but we do also check other sites such as devtracker and open SIDA, plus pull spreadsheets out of the datastore, and even talk about IATI feeds into AIMS and FTS.

At some point in every session a participant will stop, reflect and then say something like:

  • hold on, if IATI is about different flows, and some of those are related, how do we avoid double counting?

I’ll usually then reply with something like:

  • yes, good point! Well, the idea is that as activities are linked together by different publishers, data users will know not to add them together

Sometimes I explain this in a hopeful way. At other times I find I’m more assertive. At no times am I 100% confident!

Sometimes people accept this. Other times, there’s more questioning. We often start to explore the model of publishing IATI, and that it doesn’t help us to consider it as a single database we sum together. In all cases, it’s a rewarding conversation, albeit conceptually challenging (for me!)

With this in mind, I wanted to ask the brilliant people out there in our community: how do you describe double counting & IATI?

Have you a way of explaining it that works? Do you have slides? Have you used props? All ideas and experiences are gratefully accepted!

Lastly - please please please: let’s not defeat ourselves in this thread by trying to build or discuss a technical solution for double counting! It’s a thing, and will be with us for a while. What I’m specifically interested in is how we can explain (in words, pictures and hand gestures)it to others, as I think many of us have had to do this…

2 Likes

Eagerly waiting to see the brilliant ideas (I,m sure) people have found to explain this! It comes up every time I describe IATI or present the D-Portal as well.

Errrr, how about…“IATI double counting is like the news, something happened once, and you can read about it from different sources (different newspapers), and some of them will have slightly different information about it (in the same way your local paper might mention the street something happened, whereas the national paper might link it to a new law designed to stop it. These different perspectives can help you to build the whole picture (but that is a lot of work at the moment)”.

Great question @stevieflow! I usually say something like: of course if you add up all organisations’ data then you will get double counting; this is the same in IATI data as it is in data collected through other mechanisms. The easiest place to start is to select only certain organisations (or types of organisations). For the purposes of my work, this generally means including only donor organisations (bilaterals and multilaterals). Within those, there will be some double-counting, but it is usually easy enough to work out where the double counting is. So in this way, you get the widest coverage of spending at fairly low cost.

Then only add data from other types of organisations (e.g. NGOs, private sector) in if you can be sure you are avoiding double-counting - i.e.

  • if they are spending money independently raised, and they are clear about this;
  • if they are spending money received from organisations you’ve already included, then only include them if they have included traceability elements in IATI.

I have also ended up discussing/explaining ‘non-financial double counting’ e.g. when locations or sectors are reported differently by different organisations - which can make it look like a project covers more things than it really does. On the other hand, with many fields (e.g. the descriptions) these can happily be ‘double counted’ to see if it adds extra info, or extra languages etc.

@bill_anderson, Couldn’t the difference in the figures be explained not necessarily because of double counting but that what gets reported to OECD DAC isn’t the same group of activities as may be reported in IATI? IOM aren’t OECD reporters so I’m not certain but if it is limited to development activities only then all the humanitarian stuff may be in IATI and not in OECD. Perhaps you already factored that in when you pull your comparative data.

There are so many reasons for double counting already stated that I don’t think we will ever avoid it when we are talking about data compared between two reporting standards. I’ve been reading the old/revived thread on traceability and it seems to me that we all still have to nail that one down within IATI to make it obvious where there may be double count.

Multilaterals only report core funding to CRS (as earmarked funding from bilaterals is reported by them). What CRS therefore loses (and IATI gains) is the account of what is happening on the ground.

(@stevieflow I know you don’t want complicated technical solutions in this thread, but …
If all multilaterals clearly reported incoming funds on all earmarked activities [quite a few now do] there is a relatively simple methodology that can be put in place to handle the issue)

1 Like

Thanks

I’d like to know how this helps in supporting someone to understand IATI. What we would be saying is that “the theory is fine, but practice doesn’t support it (yet)”. How do we say that when someone is asking us about double counting?

Same as you would explain a UK train timetable to a commuter.

I had to explain this on Tuesday and so tested out several of these options:

  1. Explain that the the tools the IATI has to deal with this aren’t (yet) used by publishers in a way that can solve this @bill_anderson
  2. Explain that you shouldn’t use all the data at the same time @markbrough
  3. Explain that you need to add local knowledge to work out what you can add together (my current approach)

As we discussed, we came to the conclusion that all of these suggest that the problem is the expectation that IATI data, whilst machine readable, should not be considered ‘machine > machine’, but rather ‘machine > informed user > machine’ as it currently stands - with the followup conclusions that we lacked good tools for users to use to manipulate (case-by-case user input) how data from multiple publishers in IATI was combined and that this was a current barrier.

1 Like

I think this is very useful - thanks @matmaxgeds. I guess the follow-on from this is “how do we inform users?”…

Could we get to a point where we replace “informed user” with “select a model”?

Simplistically:

  • Model 1 - top down. For use by those interested in funding. (Ignore expenditure at the bottom of a delivery chain)
  • Model 2 - bottom up. For those interested in operations and expenditure. (focus on expenditure at the bottom of the chain and cancel out donor disbursements that fund this expenditure)
  • Model 3 - back to “informed user”
1 Like

I like that innovation - it sounds worth pursuing to me as a way of breaking us out of this impasse between the different demands on IATI (e.g. traceability vs ‘not an accounting system’) I imagine it fitting well with the user stories and helping to make sure the relevant parts of the standard meet their needs (by recognising that the different users use the standard differently, and use different parts of the standard).

PS - another double counting discussion in Somalia today

@bill_anderson In case it matters, not all multilaterals currently report to CRS.