Problem 1

For this problem we will be working with the Ericksen data set for describing the percentage of the population not counted in the US Census from 1980. In this data set we have different sociological variables and a factor saying whether the data point is a major city, state, or remaining part of the state. You should use the help file to help you understand all the variables. Using the regression techniques we have learned in this course use the rest of the covariates to describe the variable undercount. In 3 pages or less (including plots) describe the process you used, decisions you made to result in your final model. You should also give some insight to what your model says about describing the undercount of the 1980 Census. Describe the advantage of the model you fit in terms of the problem (does it predict well, is the inference there, or both). The largest model that can be understood in this data is the second order model. Finally compute a confidence interval for Minnesota and a prediction interval for West Virginia using your model. In an appendix attach the R summary of your final model and diagnostic plots.

library(car)

data(Ericksen)