Skip to main content
eScholarship
Open Access Publications from the University of California

Dr. Armor's Regressions

  • Author(s): de Leeuw, Jan
  • et al.
Abstract

Definition. Gap analysis is defined by Dr. Armor as the following two-step procedure. It is not a standard or particularly well-known data analysis procedure, but there are many variations of it in the social and educational research literature. As we shall argue below, all these variations share the same fundamental flaw. There is a binary variable BLACK in the regression analysis indicating if a student is black or not. We first do a regression analysis using only the BLACK variable (and a constant). This gives a regression coefficient for BLACK, say β0, which is simply the difference in mean test score of blacks and non-blacks. This is called the raw gap (for this particular test and this particular group of students).

Main Content
Current View