Good points from both of you. I suggest the usual system (when
such things are published) is to provide one map that gives the number of
surveys done for all sites, which is a constant for all species and other maps
separately for each species. The existence of incidental records obscures any
such analysis but really there is no way around this. On face value, I agree
that the colours and sizes of the 'dots' should surely be
reversed - the larger and more colourful, the better the sighting/reporting, and
the fainter and paler, the least. Although maybe it requires being seen
in practice to assess if that looks good. But I also think that Steve's map was
not addressing the issue of the reliability of the reporter or accuracy of the
identification (these are separate issues and better approached in other
ways - as in whether the record is on the system at all). I expect his use of
reliability is just about the usefulness of the sample size (based on a
start point that all records are accurate identifications). I think that on a
quick look whilst the map is worth
having, it is visually confusing, noting that it is full of symbols but
only The cells with the squares are ones where Brown Treecreeper has been
reported, so that the ones without squares are not, and so are minimally
relevant to the species. I suspect that the idea that Steve has put forward with
his map would be much easier to assess for usefulness by picking a far more
common and widespread species than a Brown Treecreeper. That way I am guessing
the information would be easier to see, by virtue of fewer
gaps.
Philip
Steve - I am only Michael Lenz' wife and not a
'bird-watcher' though also trained in scientific disciplines and have been
'around' bird-watching for most of my life, so please accept comments in
the light of this.
I really like you analyses of the various species which
you have been putting on the chat line - thank you.
Certainly seeking to overcome problems of reporting
rates is a good aim - however, I think the complex map is not very
helpful.
On a minor point, the colours and sizes of the 'dots'
should surely be reversed - the larger and more colourful, the better the
sighting/reporting, and the fainter and paler, the least ? But more
importantly, the spread indicates the ease and prevalence of reporters, rather
than the reporting itself. Quite obviously, there will be more reporting
at the easily accessible areas. Next, and much harder to quantify, is the
reliability of the reporter. For example, if Michael is the only person to
record the sighting (eg. north eastern tip of Lake George which is very
inaccessible, but which he visits) then that could be taken as a 'sure' record,
given his life-time experience and dedication, however, if I went to the lower
slopes of Mt Ainslie on a stroll, I could 'record' but might be mistaken.
Unfortunately, this is an awful minefield to get into, for while I readily admit
my lack of knowledge, many others do not - nor should they, perhaps, as it sets
up a hierarchy which we would not want in COG and people would become
discouraged.
For these, as well as the reasons you set out (nil
reporting etc. does not mean that there were no birds there) I would think that
it is not particularly useful to follow up with 'reliability indicators'.
Most people will 'know' that a reporting rate map has its flaws, but that these
are perhaps insurmountable in terms of absolute 'scientific' proof.
Hopefully this input will increase
discussion.
Janette
Subject: [canberrabirds] Brown Treecreeper reporting rate and the
overloaded map
Attached is a map showing the reporting rate for
Brown Treecreeper from 1981 to 2011.
It illustrates one way to overcome one of the
fundamental problems with reporting rate maps: they do not show how many sheets
contributed to the reporting rate. The basic problem is that the number of
sheets submitted for a gridcell is an important factor in determining the
reliability of the reporting rate. Any species reported on the
only sheet submitted for a gridcell gets a 100% reporting rate for that
gridcell. A gridcell with a reporting rate of zero means little if only one
sheet has been submitted and means nothing if no sheets have been submitted for
the gridcell.
The map indicates the quality of the reporting rate
by the circles in the gridcells. The smaller the circle and the less
obvious the colour, the better the reliability of the reporting rate. Cells with
red circles indicate a very low reliability, yellow is better but the data
should still be used with caution and very small green or clear circles indicate
the best data reliability. The cells with the squares are ones where Brown
Treecreeper has been reported. For instance, the map indicates that around
Gunning is a good place to see Brown Treecreeper (high reporting rate with a
good number of sheets submitted) but B08 may not be as good (it has a high
reporting rate but only one sheet has been submitted).
Gridcells without any sheets have nothing in them
on this map but this needs to be improved to distinguish them more easily from
the 50+ sheets cells. Also the number of categories for the number of sheets
could be changed (eg 0,1-2, 3-10, 10-50, 50+) or reduced.
While I recognise that map is currently overloaded
and the graphic design could be greatly improved, does including a
reliability indicator improve the ability to get useful information off the
reporting rate map?
Steve
|