Project FeederWatch home page
Canadian FeederWatch page
Project FeederWatch
Overview Instructions and Updates Data Entry  link Data Retrieval News About Birds and Bird Feeding

How are FeederWatch data analyzed?

Return to Data Retrieval Index Page

 

 

 

 

 

 

 

 

 

 

 

 

When all Data Booklets and online data submissions are received, FeederWatch scientists begin their fun. They merge all the data from each participant, run the data through a series of editing programs, then, for each feeder species in each region, begin summarizing the data into two variables: "percentage of feeders visited" and "mean group size." Both variables provide an index of species abundance.

"Percentage of feeders visited" is calculated as the number of FeederWatch sites that reported a species at least once over the winter divided by the total number of FeederWatch sites in the area of interest (see example, below). This value can be from 0 to 100% and is often calculated for specific states or FeederWatch regions.

Example: How to calculate percentage of feeders visited and mean group size

In this example, only 8 participants are signed up for Project FeederWatch, and they have submitted data about Species A from the following 8 locations.

Location
FeederWatch count
(highest number of Species A
seen at one time)
Fairbanks, Alaska
Eugene, Oregon
Salt Lake City, Utah
Edmonton, Alberta
Ames, Iowa
Halifax, Nova Scotia
Bangor, Maine
Winter Park, Florida  

3
5
2
5
7
Not reported
Not reported
Not reported

Percentage of feeders visited by Species A
Percentage = 5 Count Site reports/8 Total Count Sites = 63%

Mean group size of Species A
Group size = (3 + 5 + 2 + 5 + 7 ) / 5 = 4.4 birds/site

"Mean group size" is calculated by computing the geometric mean* (or average) of all the FeederWatch counts for a species. More accurately, this is the average maximum group size, as FeederWatch participants report the largest flock of each species seen during a count period. The averaging does not include "zero" counts, that is, counts from FeederWatchers who did not report the species at their site (see example).

The mean group size can vary quite a bit depending on the species; those that tend to roam in large flocks (e.g., Pine Siskin) have mean group sizes from 1 to over 100, whereas species that tend to feed alone or in pairs in winter (e.g., Downy Woodpecker) are usually seen in much lower numbers.

*In calculating "mean group size" FeederWatch scientists use what is called a geometric mean, which is a method of calculation that de-emphasizes rare high counts to provide a mean that more closely matches what a typical FeederWatcher is likely to see.

Sometimes, FeederWatch scientists blend these two measures together and factor in how many weeks the bird was reported within a season. The resultant number allows them to "see" several things with one number--how widely the bird is distributed, how numerous it is at feeders, and how likely it is to visit feeders within a season.  This number is known as the "FeederWatch Abundance Index."

What can you analyze at your own feeder?

At your own feeder, you can calculate the mean group size of each species you see. You can also calculate the "average number of birds per count period," which is similar to the FeederWatch Abundance Index but it can be calculated at a single feeder as opposed to over a large geographical area. To calculate the average number of birds per count period for a species, add up all your counts for that bird and divide by the total number of count periods (including those weeks that you didn't see this bird, but not including weeks that you didn't count at all). 

© 2002-2004 Express written permission required for use of images or text on these pages.
FeederWatch is a joint research and education project of:
Cornell Lab of Ornithology Home Page
Bird Studies Canada