Making a difference: Impact report 2017

Appendices

Appendices

Appendix 4 - Notes on the report

General

  1. Many AMRC charity funders choose not to use researchfish® and may gather their information on outcomes from research through different methods. The report contains data from 29% of AMRC membership, and so this report does not completely represent the whole medical research sector as all charities differ vastly in the way and types of research that they fund.

  2. Of the funders which are included, often not all their awards which they have been awarded are on the system and so will not be accounted for in this report. For example, many charities will ask their PIs to report on the outputs from their Centre awards in a different format, not using researchfish®. Furthermore, some funders have awards which are on the system but may not be included in the report as the funder has chosen to exclude them from sharing with AMRC. This is usually because they contain confidential information which the PI is not happy to share.

Quantitative analysis

General

  1. Due to rounding of numbers and percentages, figures may not always equal the totals or 100%.

  2. If further funding amounts were given in different currencies then researchfish® automatically applied the exchange rates based on the rates at the point at which the output has been entered into researchfish®. The equivalent amount in Pounds Sterling is given in the download field [Funding Organisation Further Funding Value].

  3. When calculating the time since award start date figures, the year was used. This means that even if the award started 31 December 2013 then it will be counted as starting in 2013, and outputs collected at May 2016 will be 3 years since award start date.

  4. Collaborations can be made up of multiple partners. For this report, each separate partner is counted as a ‘collaborator’.

Categorisation

  1. When funders upload awards to researchfish®, it is not mandatory to assign grant type classifications (e.g. project, PhD studentship) to them. Where possible AMRC have added these classifications to them based on data which has previously been submitted to AMRC by funders. However if there was no classification available then awards were coded as unknown. Of 5287 awards, 1008 had a grant type of unknown (19%)

  2. The coding of awards by HRCS codes was taken from various different data sources. Where possible, coding was taken from the 2014 UKCRC HRCS report dataset (http://www.hrcsonline.net/pages/data) as these codes have been double-coded to ensure quality control. Some funders coded the awards themselves but if not then AMRC coded the remaining awards.

  3. Some awards were either outside of the Health Research Classification System or were unable to be coded as the awards had insufficient information. These awards were coded as unknown. 1% (60) of the awards were coded as unknown for both research activity and health category codes. Of 5287 awards 5052 (96%) were categorised to research activity codes and 5208 (99%) were categorised with health category codes.

Data cleaning

  1. Often an output was attributed to more than one award from the same funder or from different funders. Where possible, when calculating the high level analyses (for example total number of publications produced by all funders) then these were de-duplicated. Duplicate outputs from the same principal investigator were excluded by removing any outputs which had identical output IDs (assigned to the output by researchfish®). However it is far harder to identify identical outputs which were created by multiple principal investigators as these will have different output IDs.

  2. For the breakdowns of the outputs by different categories (e.g. health category), outputs were not de-duplicated. This means that often the sum of the number of outputs in categories will be higher than the figure for the total number of outputs.

  3. AMRC does not have the in-depth knowledge of the awards which the individual charities have. Therefore cleaning of the dataset was at the very top-level and so the data used is based purely on what the principal investigators inputted into researchfish®. Where there were any seemingly obvious outliers (e.g. a principal investigator reported a multi-million award as further funding) these were further explored by AMRC by searching through online databases or contacting the charitable funder.

  4. It is likely that some awards may not have been through a submission period yet. To limit the number of these included in the report, any awards which did not have an Outcome Owner UID were removed as these have not been assigned a PI yet.

  5. Publications were removed if they were reported to have occurred prior to the start date of the award it is attributed to or if they were not yet published.

  6. At one time, the ‘spin outs’ section was used to collect information on collaborations with large companies as well as information on spin outs from the PIs own lab. In this report, where possible, these collaborations which were reported as spin outs were removed.

  7. If there was no month given with the year of an output, then January was used as the default. This is consistent with other funders who have analysed researchfish® data (i.e. http://www.aihealthsolutions.ca/media/Annual-Impact-Report-2014-2015.pdf).

  8. All the data is correct as of May 2016. However as the data in the system is live then researchers may change their outputs in the future or the data could change due to further cleaning or mapping.

  9. There were 15 awards where the start or end date of the award were unknown, and 2 awards where the end date preceded the start date. These dates were estimated based on database and website searches.

A4.1 List of Funders