Identifying who is being funded

Back in 2018 we focused some of our labs work on organisational identifiers – the important numbers that help us to differentiate between organisations. We wrote a bit about identifiers and why they are important here and here. Our work last year focussed on how many of the grantmakers publishing to the 360Giving Standard are using identifiers and how we can help more to do so.

In this blog, we are sharing some highlights from our internal report, written by charity data expert David Kane. The assessment is based on grants downloaded from GrantNav in June 2018 and January 2019.

Because of the large proportion of grants that are accounted for by The National Lottery Community Fund (formerly the Big Lottery Fund), they are separately considered below.

How many grants have an identifier?

The good news is that half of 360Giving publishers are using external unique identifiers with a coverage of between 75%-100%.

Table 1 Number of funders by the proportion of their grants that have a valid identifier scheme

However, when looking at grants as a whole, we need to do better as only half of the grants include useful identifiers.

The data in table 2 below is based on a check that the organisation identifier used looks like it has a valid scheme in the field Charity Number, Company Number and Recipient Identifier. It does not check whether the actual number is valid.

Table 2. Proportion (%) of grants with a useful organisation identifier

There are several reasons why a funder wouldn’t use an identifier:

1. The group they are funding in England and Wales has a turnover of less than £10k, which means they don’t have to register with the Charity Commission. See more here.
2. Local sports clubs don’t have to register with the Charity Commission but have their own list (with no identifiers).
3. Some religious institutions with an income of under £100k don’t have to register.
4. Some funders just don’t collect identifiers in their systems.

Which identifiers are grantmakers using?

As of January 2019, the leading identifier in the 360Giving dataset is the charity number issued by the Charity Commission for England and Wales. 88% of funders use this GB-CHC scheme to identify recipients. This is followed by UK Companies House (COH), Office of the Scottish Charity Regulator (SC), Education institutions number (EDU), the NHS, and the Northern Ireland Charity Commission (NIC). A full table is provided in the report.

Table 3 Grants and identifiers per identifier scheme

How many identifiers are valid and why?

We have found that in 360Giving data CHC numbers are usually valid with a 6% error rate. COH numbers, however, tend to be less valid with a 35% error rate. Scottish charity number has a 20% mistake rate, and NI charity numbers are almost 100% valid.

The top three causes of invalidity are:

1. Zeros: A very important number, but people and machines tend to ignore leading zeros or trailing zeros. This is because if treated as numbers rather than text, most software will delete or ignore them. In the case of Scotland, 0 is also mistaken sometimes and is added instead of the letter “O”.
2. Use of numbers from other regulators: People tend to mix between Companies House numbers and Charity Commission numbers or between charity regulators’ numbers.
3. Use of historic numbers: Companies and charities are sometimes closed down and opened under a different name. Grantmakers don’t always use the most up-to-date numbers.

Finding duplicates and identifying organisations

We also tried to add identifiers to organisations without an external identifier. This process was carried out for a sample of 35,000 grants (excluding grants from The National Lottery Community Fund).

This process found a charity or company number for around 10,000 of grant recipients, of which around 7,100 were universities funded by the Wellcome Trust (matched to the Royal Charter Company number for the university).

Universities are currently a challenge in our identifiers quality quest and we are looking at the best way to tackle this.

We also dedicated some of our research to finding duplicates. We define duplicates in this case as an organisation that had two different organisational identifiers. For this, we used DeDupe, a tool that helps in finding similar rows in datasets. More on this process is outlined in our report.

We checked the whole dataset for duplicates and found 13,000 organisations that were duplicates of another in the dataset and don’t have a usable external identifier.

What does the future hold?

In 2019, we are continuing to work with grantmakers on improving the quality of their identifier data. We are exploring what tools we can build to help with adding organisational identifiers and to help address the Recipient Org:Organisation Type and fit the Standard (see a discussion about this on our forum).

What can you do to help us in the journey for better quality data?

• Read the full report.
• Join our discussions on the forum.
• If you are a 360Giving publisher and want to improve your identifiers – get in touch as we want to help!