The United States Central Intelligence Agency maintains a public site called the World Factbook at www.cia.gov/ library/publications/the-worldfactbook/. There you’ll find a wealth of variables about all the countries of the world. Let’s examine some of them as possible participants in a regression model. (Note: The following odd-numbered exercises will lead you through an analysis of these data. At each step you are asked to comment. There are no strict right or wrong answers to these questions, but there certainly are things you should notice and be able to comment on. So, taken together, these are an exercise in thinking about and building a multiple regression model.) Let’s consider the following selected variables: Life expectancy (years) Population Birthrate (births/1,000 persons/year at midyear) Obesity% (% of a country’s population with a BMI Ú 30.0) Education (expenditure as %GDP) GDP/capita Imports/capita (US$, exchange rate basis) Exports/capita Airports/capita (total number of airports or airfields recognizable from the air) Cell phones/100 inhabitants Here are histograms of the variables:
a) Comment on these variables. Which are suitable for use in a regression model? Which would benefit from re-expression (and, if so, why)? Note any other issues. Here are histograms in which some of the variables have been re-expressed: b) Comment: Are these variables now suitable for use in a regression model?
