Application
Now that you have completed all of the activities in this chapter, use the concepts and techniques that you've learned to respond to these questions.
Scenario: Return to the NHANES SRS data table. Exclude and hide respondents under age 18.
Perform a regression analysis for adult respondents with BMI as Y, and use waist circumference and gender as the independent variables. Discuss the ways in which this model is or is not an improvement over a model that uses only waist circumference as a predictor.
Perform a regression analysis extending the previous model, and this time add an interaction term between waistline and gender.
Scenario: High blood pressure continues to be a leading health problem in the United States. In this problem, continue to use the NHANES SRS table. For this analysis, we'll focus on just the following variables, and focus on respondents between the ages of 12 and 19.
Perform a regression analysis with systolic BP as the response and gender, age, weight as the factors. Report on your findings.
Perform a regression analysis with systolic BP as Y and gender, age, weight, and diastolic blood pressure as the independent variables. Explain fully what you have found.
Scenario: We'll continue to examine the World Development Indicators data in BirthRate 2005. Throughout the exercise, we'll use BirthRate at the Y variable. We'll broaden our analysis to work with other variables in that file:
Earlier we estimated a quadratic model with maternal mortality as the independent variable. To that model, add the categorical variable MatLeave90+, which indicates whether there is a national policy to provide 90 days or more maternity leave. Evaluate your model and report on your findings.
Investigate whether there is any interaction between the maternity leave dummy and the rate of maternal mortality. Report on your findings, comparing these results to those we got using the model without the categorical variable.
Scenario: The United Nations and other international organizations monitor many forms of technological development around the world. Earlier in the chapter we examined the growth of mobile phone subscriptions in Thailand. Let's repeat the same analysis for two other countries using the Cell Subscribers data table.
Develop a log-linear model for the growth of cell subscriptions in Denmark. Compute the annual growth rate.
Develop a log-linear model for the growth of cell subscriptions in Malaysia. Compute the annual growth rate.
Develop a log-linear model for the growth of cell subscriptions in the United States. Compute the annual growth rate.
We've now estimated four growth rates in four countries. Compare them and comment on what you have found.
Scenario: Earlier in the chapter we estimated a logistic model using the Parkinson's disease (PD) data. The researchers reported on the development of a new composite measure of phonation, which they called PPE.
Run a logistic regression using PPE as the only regressor, and Status as the dependent variable. Report the results.
Compare the results of this regression to the one illustrated earlier in the chapter. Which model seems to fit the data better? Did the researchers succeed in their search for an improved remote indicator for PD?
Scenario: From time to time a historian discovers an unsigned manuscript, musical composition, or work of art. Scholars in these fields have various methods to infer the creator of such an unattributed work, and sometimes turn to statistical methods for assistance. Let's consider a hypothetical example based on the Sonatas data.
Run a logistic regression using Parta and Partb as independent variables and Composer as the dependent variable. Report the results.
(Challenge) Suppose that we find a composition never before discovered. Parta is 72 measures long and Partb is 112 measures long. According to our model, which composer is more likely to have written it? Why?
Scenario: The United Nations and other international organizations monitor many forms of technological development around the world. Earlier in the chapter we examined the growth of mobile phone subscriptions in Thailand. Let's look at a single year, and examine the relationships between adoption levels for several key communication technologies. Open the data table called MDG Technology 2005.
Develop linear and quadratic models for the number of cellular subscribers per 100 population, using telephone lines per 100 population as the independent variable. Which model fits the data better?
Develop linear and quadratic models for the number of cellular subscribers per 100 population, using personal computers per 100 population as the independent variable. Which model fits the data better?