Sunday, 21 March 2010

Some thoughts about Citizen Science

There has recently been an upsurge in interest in "Citizen Science" which I take to mean using 'ordinary people' to record scientific observations rather than science professionals. 

From a (slightly cynical) view of the scientific community (and a realistic view of public administration in the post-Reagan era) this is a pretty good deal since instead of:
  • paying researchers and field assistants you get members of the public to perform similar functions at no cost; and
  • having expensive laboratories etc the work is done largely from people's homes.
It is also a good deal for the citizens since it allows them to contribute to informed decision making and provides a record of their activities for posterity.  Both important 'feel-good' factors.

Since I am involved in the process in various ways I thought I would set down a few thoughts about the philosophy involved and the (surprisingly large number of ) examples of citizen science with which I am either aware or involved.

Some Definitions
Often a member of the public who may not have any scientific training whatsoever.  Or they may be the former head of a science organisation mainting an interest in their field of work once they have retired.  Or an active scientist whose work is also their hobby.  Most likely any Citizen Science group will have a mixture of all of the above.

They will generally have some interest in the topic being studied, acquired through social links or the media but possibly not the obsession with detail required for professional science.

Almost by definition they will be operating as unpaid volunteers, although in some cases the coordinating body may provide facilities and materials for the projects being undertaken.

Some branches of science endeavour (quantum mechanics; DNA sequencing spring to mind) are not amenable to Citizen Science because of the facilities involved.  However I see the basic tenets of 'the scientific method' - adherence to facts and repeatability of procedures as being a crucial aim of Citizen Science.

In the scientifically ideal world observations are undertaken in a tightly controlled environment in which the only variable factors are those being investigated.  At the simplest this is test-tube science: the environment is within a small glass tube and the only variables are the chemicals added thereto.  Of course even this is subject to external variations:
  • adding chemicals in a different order may give different results; and/or
  • the temperature of the lab may affect the speed or strength of the reaction; and/or
  • the tube may be shaken, not stirred; and/or
  • the observer may perceive resuls differently.
However by comparison to an exercise such as recording "the number of birds seen in an area over a period of time" such varibility is trivial.  Just about every word in that phrase is open to debate:
  • The number - cardinal (1,2,3 ... etc) or binary (1 or 0 for presence/absence; are there rules for counting indidual birds - ie what is the minimum number seen? );
  • Birds (very few people want to include bats or reptiles ...; generally the birds will need to be classifed to species);
  • seen (include heard as well?);
  • in an area (this a real mess: what area; how big, include flying over or just those that land  - and several variations on this theme);
  • period of time (does this have to be a standard period? is it the period in the area or the period actually spent looking at birds?)
A scientific data collection will include definitions of all those items (and possibly more).  By way of contrast the lay 'citizen' might want to report something like "a list of species I see while walking the dog'.  In some cases analyst will assume that the rules are rigorouly followed while the observers tend to regard the rigid "rules" as more like plastic "guidelines".

To some extent this can be overcome by seeking metadata (ie information about the reported data) which will reveal the approach adopted by the observer.  Of course the problem becomes one of how much metadata is collected.  If there is to much people will not report the metadata or will abandon the project.  Also if there is too much data the analyst may havedifficulty in working out how to include it in their analysis (or the variable:observation/ratio may preclude some types of analysis - for example regression won't work if there are more variables than observations).

In practise I belive that one of the defining attributes of citizen science is that the data will include a lot more "noise" than a formal science project.  In some cases this can be overcome by the "law of large numbers" in which the extra obersations make it possible to come up with a believable measure of the average despite the vagaires of the individual observers.  For example, a formal study may find it possible to monitor the number of birds seen at 10 sites for 3 months while a citizen science project could have 100 observers active across a whole year.  However the likely presence of extra noise does mean that the role of the "coordinator" is crucial in such surveys.  They should be active in
  • deleting the most egregiousness departures from the rules (eg accumlating observations over a week rather than reporting for an 'instant'); and
  • advising analysts where the rules have been followed but unusual results obtained (eg a site includes a large pond or section of lake) thus atracting many waterbirds.
Examples of Citizen Science
I am not including in this a range of bird banding projects of which I am aware.  I regard them as formal science, even though the operators of the projects are not necessarily employed as professional scientists.

Canberra Bird Atlas: a good example of allowing a wide range of observations and collecting metadata.  I have some doubt about the extent to which metadata is reported (and the extent to which it is used) but the model is very sound.  Results are available for use by analysts and summarised in the Annual Bird Report.
Canberra Garden Bird Survey: This tends to be based on firm rules and requires a reasonably careful look at reports to ensure they are followed (for some species in general).  However the length of the time series (and the duration of some observers/sites  - 20+ years involvment) is a great source of strength.  Results are available for use by analysts and summarised in the Annual Bird Report.  They are also summarised in Birds of Canberra Gardens of which a second, hard-copy edition has been published in early 2010.
Australian Bird Atlas: Similar to Canberra Bird Atlas with added rigour for some observations where a standard 2Ha site is monitored over a long period. Results are available for use by analysts and summarised in the birdata pages on the internet.
Australian Native Plants Society (ANPS):  The ACT chapter undertakes a series of Wednesday Walks in which obeservers visit areas and record the plants growing there.  The metadata is not too well documented but the rigour of species identification is very strong (false negatives are far mor likely than false positives).  The results are published as plant lists for the areas. (Other elements of ANPS do other things including plant propogation.)
Fungimap: This is a project run through the Royal Botanic Gardens in Melbourne through which obsevers around Australia report on the location of a range of Target Species of fungi.   The project has published a book "Fungi Down Under" and will also offer advice on the identification of fungi if a photograph is submitted (as evidenced by my earlier post). 
Frogwatch:  This project is run by a catchment management group based in Canberra.  They offer a report each year on the frogs identified (through recording calls during the breeding season).  They also provide a good array of kit (recordings of calls etc) to assist observers. 
Waterwatch:  This is run by another Catchment Group covering the Molonglo River.  They have paid employees and undertake a range of activities, including training for citizen observers in various useful pursuits (use of GPS; water quality monitoring etc).  They put out a number of reports on issues affecting the catchment of the Molonglo River.
Greening Australia:  The principal function of this group is getting vegetation back into the scenery (which for some reason doesn't fit my definition of scince).  However the ACT Chapter have undertake a project assessnig the use of revegetation projects by birds.  The results of this are published as a very useful book 'bringing birds back'.  The material on that link clearly shows the scientific thrust of the project.

There are also a range of natural history groups in the ACT area (Field Naturalists; Hepetological Society; Friends of Grassland, Orchid Society) which have laudable environmental protection aims and a broad membership.  However as far as I am aware - not being a member of any of these groups - they don't do organised data collection and thus I don't count them as Citizen Science.  That being said if someone wished to persuade to organise their data they'd be a very useful source of data about wildlife and its interactions.

Another point to consider is that there is considerable cross-membership between these groups with many people being members of two and several being involved in 3 or more groups.

No comments: