Assignment 2a
Matrix Algebra - Part 1 (OLS)
Robert S. Erikson (1972) "Malapportionment, Gerrymandering and Party Fortunes in Congressional Elections" American Political Science Review, 66: 1234-1245.
Erikson's analysis examines the relationship between popular vote and the percentage of seats in the House of Representatives won by the Democratic Party during the 1950's and 1960's. A corrolary issue is whether the Supreme Court's decisions in the early 1960's (Baker v. Carr, Reynolds v. Simms) "upset the applecart", that is, changed the translation of votes into seats.
Here you're asked to consider the following questions. What does the bivariate relationship between votes and seats look like? What happens when we consider the effect of the Court's intervention? (In other words, are the elections after 1962 any different than the earlier elections?)
The Erikson data is available on "sobek":
ASCII: eriksn72.dat
SPSS: eriksn72.por
STATA: eriksn72.dta
Part 1
- Compute OLS regressions "by hand" (via spreadsheet)
- Compute OLS regressions via STATA reg subroutine
- Compute univariate and bivariate statistics via matrix algebra
- Compute OLS regressions via matrix algebra
To compute OLS estimates using matrix algebra you must use the matrix language capabilities of either STATA or SPSS. Introductions to each can be found by following these links: