|
Activity Number:
|
138
|
|
Type:
|
Contributed
|
|
Date/Time:
|
Monday, August 4, 2008 : 10:30 AM to 12:20 PM
|
|
Sponsor:
|
Section on Statistical Education
|
| Abstract - #301905 |
|
Title:
|
Graphical Exploration and Suggestions for the Selection of a Regression Model Using Real Data
|
|
Author(s):
|
Irene Schiattino+ and Claudio Silva and Carmen Acuna*+ and Rosa Montaño and Verónica Iglesias
|
|
Companies:
|
Universidad de Chile and Universidad de Chile and Bucknell University and Universidad de Chile and Universidad de Chile
|
|
Address:
|
Escuela de Salud Pública. Facultad de Medicina. Avenida Independencia Nº 939, Santiago, , Chile Moore Avenue, Lewisburg, PA, 17837,
|
|
Keywords:
|
Least squares estimation ; weighted least squares ; maximum likelihood
|
|
Abstract:
|
The wide availability of statistical software might lead us to rush into the model fitting process and to bypass the use of efficient graphical representations of the data without taking the time to analyze the coherence and the logic of the strategies employed. Both, the modeling process and the graphical representation should consider the context induced by the statistical theory and the area of application. In this work we discuss, from an educational perspective, several approaches for fitting a regression model when heteroscedasticity is present. The work emphasizes the use of regression graphics (scatterplots, residual plots, scatterplot matrix, three-dimensional plots, smooth curves, etc.), and compares different model fitting methods. Data from two Chilean research projects are used.
|
- The address information is for the authors that have a + after their name.
- Authors who are presenting talks have a * after their name.
Back to the full JSM 2008 program |
