Calculating the national average
Regional differences in the abundance of a species coupled with the changing proportion of survey returns from each region can make it appear as though the species has changed over time, even though it is only the proportion of survey returns that is changing!
Here we illustrate why it is important to develop a robust process for calculating the national average number of tūī for the Garden Bird Survey.
The simplest approach: providing misleading results?
![fig 1 Figure 1: The average number of tūī per garden for New Zealand (Data source: Garden Bird Survey).](https://oldwww.landcareresearch.co.nz/__data/assets/image/0004/83785/varieties/thumb.jpg)
When we examine the time series that results from a simple average of bird counts, it would appear as though the mean number of tūī per garden has increased from 2007 to 2013. But is this really the case?
Digging a bit deeper: understanding survey participation rates
![fig 1 Figure 1: The percentage of total survey returns by region (Data source: Garden Bird Survey).](https://oldwww.landcareresearch.co.nz/__data/assets/image/0004/83812/varieties/thumb.jpg)
The apparent increase change in tūī per garden appears to be mostly an artefact of the changing proportion of respondents, especially from Canterbury.
Taking into account garden numbers helps
![fig 1 Figure 1: The average number of tūī per garden for New Zealand, both unweighted and weighted by region, from the GBS for the period 2007 to 2013.](https://oldwww.landcareresearch.co.nz/__data/assets/image/0007/83788/varieties/thumb.jpg)
When calculating a national average, the results must be adjusted to reflect the proportion of gardens from each region.
Other factors are also important
![fig 1 Figure 1: The average number of tūī per garden for New Zealand, unweighted, weighted by region, and weighted by region and garden type, from the GBS for the period 2007 to 2013.](https://oldwww.landcareresearch.co.nz/__data/assets/image/0008/83789/varieties/thumb.jpg)
When calculating a national average, the results must be adjusted to reflect other factors such as the type of garden.
Regional picture: is it representative?
![fig 3 Figure 3: The average number of tūī per garden for New Zealand and Canterbury only. (Data source: Garden Bird Survey).](https://oldwww.landcareresearch.co.nz/__data/assets/image/0006/83787/varieties/thumb.jpg)
Regions with small numbers of survey returns are less likely to be representative of all households from the region.