Policy Significance - how to express "not screened" (blank)? Clarification help needed

Via some work with a publishing organisation I took at look at the Policy Marker & Significance code lists:

The source for these is the OECD DAC codelist spreadsheet.

Where I am slightly “confused” is that the OECD DAC list has a specific value of (blank):

(blank) not screened

Whereas a value of 0 is different:

0 not targeted

It is my understanding that a (blank) value in a CRS return indicates a specific choice in the data return, it is not purely a missing value. This indicates something different to a 0, conceptually.

However, I am unclear how this (blank) can be expressed via IATI XML, because:

  • The IATI PolicySignificance codelist does not contain a (blank) value

  • It does not seem to be useful to publish the @significance attribute with a blank value, or missing entirely - eg:

policy-marker vocabulary="1" code="3" significance="" 


policy-marker vocabulary="1" code="3" 

I’m interested to know if others have discovered this, and how they have worked with/around it. Or - if it is not an issue.

I wonder if @Herman or @OJ_ might be able to help here…

Hi Steven,
Your interpretation is correct i.m.o. Blank in CRS terms just means: ‘we do not know, because we didn’t screen it’. I agree that the most simple way to express this in IATI is not publishing the policy marker.

Thanks @Herman

I agree that the most simple way to express this in IATI is not publishing the policy marker.

It would be interesting to check with publishers this is so, and that use of 0 has not represented this use case.

I am quite sure that some Publishers are using ‘0’ as their default value - this disserves no respect, and should not be encouraged nor set precedence. We must make sure that empty fields are understood correctly. Even if that should lead us to do strange things like adding ‘-1’ to the values - meaning ‘blank’, if we find reason to doubt the communicative power of absent fields.

Thanks all for your previous comments and just to add that as a general principle I would concur with @Herman in that if a publisher is unable to provide a valid codelist (or other) value for any relevant attribute then that attribute should not be included in the published data at all (rather than being included and left blank).

I’d like to flag that this slight complexity around Policy Significance attribute may work against ideas around utilising this element for a general “tag” on an activity (as discussed at #TAG2017 Standards Day)

–> @YohannaLoucheur @rory_scott