In Problem 19 in Chapter 5 and Problem 14 in Chapter 8, data from the 1990 Census for 26 randomly selected Metropolitan Statistical Areas (MSAs) were discussed. Of interest were factors potentially associated with the rate of owner occupancy of housing units. The following three variables were included in the data set:
OWNEROCC: Proportion of housing units that are owner-occupied (as opposed to renter occupied)
OWNCOST: Median selected monthly ownership costs, in dollars
URBAN: Proportion of population living in urban areas
It is also of interest to see whether the average owner occupancy rate differs between metropolitan areas where 75% or more of the population lives in urban areas and metropolitan areas where less than 75% of the population lives in urban areas, while controlling for ownership costs. For this purpose, the following additional variable is defined:

a. State an ANACOVA regression model that can be used to compare the two MSA types, controlling for average monthly ownership costs.
b. State the model that should be used to check whether the ANACOVA model in part (a) is appropriate. Using the computer output given next, carry out the appropriate test.
c. Using ANACOVA methods and the accompanying output, determine adjusted mean owner occupancy rates for the two types of MSAs, and test whether they significantly differ from one another. (The unadjusted average owner occupancy rate for MSAs where at least 75% of the population lives in urban areas = 64.5%; the unadjusted average owner occupancy rate for MSAs where less than 75% of the population lives in urban areas = 69.25%.)
Question continues to the next page…