Supporting Tables for
As in chapter 6, our initial analysis of the elderly is based on the entire sample of unified school districts, and our measure of public opinion is not included. In addition to our usual controls, the regression model controls for the overall migration rate for the school district's county, so we do not attribute to senior migrants what is a more general impact or correlate of high in-migration rates. Because the percentage of elderly migrants is highly skewed (skewness = +3.8), we use the natural log of elderly migrants; although the substantive interpretations are basically the same, this improves the specification and fit of the model substantially; see table A7.1.
In tables A7.2 and A7.3, we add interactions between property tax reliance and our two measures of elderly population size as well as circuit breaker use (coded 1 for states that used circuit breakers).
Table A7.1. Effect of Long-Standing and Migrant Elderly on Per-Pupil Instructional Ex-
penditures (N = 9,174)