Narrative

The frequency statistics attempt to assess how often any part of a publisher's data is substantively updated.

For the purposes of these statistics an update is assumed to have taken place on any given day when the most recently recorded transaction date (across a publisher's entire portfolio) is observed to have changed to a more recent date. This approach has been adopted as transactions are the most numerous and most frequently updated elements in the reporting of activities. 

The table of statistics records the number of days in each of the last twelve calendar months (the current month is also displayed for informational purposes, but is not used in the assessment) on which the most recently recorded transaction date was observed to have changed. The dashboard maintains a statistical snapshot of each day, and has been doing so for over a year, which allows for this data to be recalculated on a nightly basis using historical recordings.

Assessment

To assess these statistics one also has to take into account how long a publisher has been publishing to IATI. This is calculated based on when a publisher first appears in the statistical snapshot.

These statistics are then assessed as follows:

 

For publishers of 1 year or more

(EDIT 19/02/2015: 6 months -> 6 month periods)

Updates reported in ...	Assessment
9 of the last 12 months	Monthly
3 of the last 4 quarters	Quarterly
2 of the last 6 month periods	Six-monthly
1 of the last 12 months	Annual
None of the last 12 months	Less than annual

For publishers of six months or more

Updates reported in ...	Assessment
4 of the last 6 months	Monthly
2 of the last 2 quarters	Quarterly
1 of the last 12 months	Annual

For publishers of three months or more

Updates reported in ...	Assessment
3 of the last 3 months	Monthly
1 of the last 6 months	Annual

For publishers of less than 3 months

Updates reported in ...	Assessment
1 of the last 3 months	Annual

 

Exceptions

• Future transaction dates are not allowed in IATI and disrupt these statistics. For example a publisher might today have a transaction date reported for each month going forward for a year and never refresh their data. The statistics would, over the next year, as each of these future dates move into the past, assess the publisher to have a frequency assessment of monthly. For this reason not only are all future dates excluded, but the activities in which they are reported are excluded from future assessment until such time that a publisher's entire portfolio no longer contains any future dates.

Comparison with Global Partnership Indicator methodology

This methodology is substantially different.

In the Indicator methodology the IATI Registry log dates were analysed to assess when updates had been made. This approach was flawed for two reasons:

• Firstly, following an upgrade to the underlying software platform on which the IATI Registry is hosted it was no longer possible for a machine to reliably calculate when data were actually updated.
• Secondly, the Registry would log any change, no matter how trivial. A spelling correction, for example, would count as an update. Similarly if a publisher's file was inaccessible one night, its reappearance the following night would count as an update.