Regional picture: is it representative?
Regions with small numbers of survey returns are less likely to be representative of all households from the region.
Larger numbers, greater confidence?
Since its inception in 2007, some regions have very low numbers of survey returns to the Garden Bird Survey. For example the number of surveys from Gisborne is typically between 10 and 20.
The results from regions with very low numbers of participants are less likely to be representative of the region than results from regions with more participants.
Also, for regions with low numbers of participants, we have less confidence that the results from those regions are truly representative of the entire region.
Location of participants?
All other things being equal, we would be more confident in regional averages from Wellington (which has more than 700 surveys per year) than from Gisborne.
Of course it is still possible that the respondents from Gisborne are spread randomly across the region, and those from Wellington are clustered in one small area, in which case the Gisborne results would be more representative of their region.
How to deal with this issue?
One simple way to account for this is to join some regions together. This is a valid approach when regions are adjacent to each other and there is not large variation in bird abundances between those adjacent regions. Previous analyses from the GBS have combined data from Gisborne with Hawke’s Bay, Bay of Plenty with Waikato, Taranaki with Manawatu-Wanganui, and Nelson City with Tasman (see link below).