This lab shows how to create geo-spatial data at the Census public-use micro area level. These data are then integrated into the person-level data. You are asked to compare different fixed and random effects models for the integrated data.
The lab uses pumsbxak.sas7bdat (AK 1% PUMS from Census 2000).
1.
Create average income and house size (rooms) within PUMA5 area.
2. Link those data to the person-level file.
3. Build a fixed-effects model of the relation between log wage and salary
income and sex, age, educational attainment, average income in the area, and
average house size in the area.
4. In the same model replace the area variables with fixed generic geographic
heterogeneity controls.
5. In the same model replace the fixed generic geographic heterogeneity controls
with random ones.
6. Interpret the results.