Core Facility Metrics

Introduction

As part of the recent public call one of the topics which was raised was how feasible it was to quantitatively assess the output of a core facility. One of the actions from this call was that we would try to put together a list of metrics which could be used to measure how well a core was performing and that this might be useful for both helping to assess your own core, or to compare your core with others to see how you compare.

Metrics

Background Information

This information doesn't measure the core facility, but would help other people seeing the later metrics to put them in some kind of context.

1. Number of members of staff
   1. Number of general bioinformaticians
   2. Number of programmers
   3. Number of database administrators
   4. Number of statisticians
   5. Number of computer technicians
   6. Number of admin/management staff
2. Method of support (chargeback / grants / core funded etc)
3. Number of staff to support
   1. Ratio of biologists : bioinformaticians

Core Metrics

These metrics could be calculated over any arbitrary period of time. They could be applied to a facility as a whole or to specific individuals or groups of individuals.

1. Proportion of work hours attributable to:
   1. Billable or Assignable work for biologists
   2. Maintainance of existing facilities or tools
   3. Development of new facilities or tools
   4. Developing, Presenting or Receiving training
   5. Other enrichment activities (reading / writing papers, attending conferences etc)
   6. Administration
   7. Other activities
   8. Unaccounted hours
2. Number of jobs completed per day per person
3. Average number of active jobs per person
4. Average number of pending jobs per person
5. Average elapsed time on active jobs
6. Average length of time to start working on a job
7. Average length of time to complete a job (elapsed time)
8. Average length of work on a job (billable or active time)
9. Ratio of active time to elapsed time

Working out how best to summarise job lengths might be tricky. In our core I know that there are certain jobs which will drag on for months whilst most jobs are completed within a day, so a simple mean would not be appropriate. Perhaps expressing the results as box-whisker plots (maybe on a log time scale) would give a realistic impression).