Statistical Graphics of Pearson Residuals in Survey Logistic Regression Diagnosis

• Activity Number: 142
• Type: Contributed
• Date/Time: Monday, August 2, 2010 : 8:30 AM to 10:20 AM
• Sponsor: Section on Survey Research Methods
• Author(s): Stanley S. Weng*+
• Companies: National Agricultural Statistics Service
• Keywords: Pearson residuals normality graph ; Statistical graphical modeling

Abstract

For survey data logistic regression, the model fitness has been assessed through a set of diagnostic statistics, borrowed from the logistic regression in generalized linear models. However, for survey data, these statistics may need to be reexamined. In practice, we have observed their irregular behaviors, which make their established statistical criterion suspect. This presentation reports our use of Pearson residuals normality graphs as graphical diagnostic statistics, to assess survey data logistic modeling, as in a recent NASS study. Statistical graphics summary may provide broader scope and more elaborated information than analytical summary. The statistical graphs of Pearson residuals showed their diagnostic ability, and their careful reading may reveal delicate diagnostic information on modeling effects. We illustrate the statistical graphical modeling process with our analysis.