Project 34
Project description

In Project 33, you pretended that the only data available was for the bird with ID number 3 (which we will call Bird 3). However, given the actual available data set, explore the following two alternative approaches:

  1. Choose a single day (say Day 33 to be concrete) and consider all of the birds on that day. Use this to develop a model to predict wing length from culmen length, and compare this approach (and its validity and scope of inference) to that in Project 33.
  2. Now use a simple linear regression model based on the data from all days and all birds to develop a method of predicting wing length from culmen length. Conduct a full regression analysis with this model, and then state why this approach is not at all valid (and might even be considered to be quite deceptive).

Background on the data set

This data set consists of biometric measurements of snow petrel nestlings in Hop Island, East Antarctica. The data collection is described as follows in the article Growth and development in antarctic fulmarine petrels by Peter J. Hodum, Karen L. Gerhart, and Wesley W. Weathers:

The primary field site is Hop Island (68 degrees 53 minutes S 77 degrees 50 minutes E) in the Rauer Island group, approximately 40 kilometers south-southwest of Australia's Davis Station. We arrived on station on 22 October 1995 and began the field season on the island on 29 October. We departed the island on 31 March 1996. During the 1995-1996 season, we studied diet composition, chick energetics and growth, adult energetics, and breeding success in populations of snow petrel (Pagodroma nivea), cape petrel (Daption capense), antarctic petrel (Thalassoica antarctica), and antarctic fulmar (Fulmarus glacialoides). The extended nature of the field season allowed us to follow all four species through the entire breeding season, from pre-breeding attendance through to fledging.

Known-age chicks were measured every 3 days throughout the nestling period from the day of hatching. These measurements continued until the chicks fledged.

This article also contains further information about and analysis of the data.

We at the University of Puget Sound Data Hoard are grateful to Peter Hodum for making this data available to us.

Variables in the data set
The variables in the data set are as follows:
id(number)id number of bird
daydaysday number of measurement
massgramsmass of bird
wingmillimeterswing length of bird
tarsusmillimeterstarsus length (less technically, leg length)
culmenmillimetersculmen length (less technically, beak length)
Link to the data set
The full data set in csv format is at: