250 PART 5 Looking for Relationships with Correlation and Regression

Using Logistic Regression

Following are typical uses of logistic regression analysis:»

» ^{To test whether one or more predictors and an outcome are statistically}

significantly associated. For example, to test whether age and/or obesity

status are associated with increased likelihood to be diagnosed with

Type II diabetes.»

» ^{To overcome the limitations of the 2x2 cross-tab method (described in}

Chapter 12), which can analyze only one predictor at a time (and the predictor

has to be binary). With logistic regression, you can analyze multiple predictor

variables at a time. Each predictor can be a numeric variable or a categorical

variable having two or more levels.»

» ^{To quantify the extent or magnitude of an association between a particular}

predictor and an outcome that have been established to have an association.

In other words, you are seeking to quantify the amount by which a specific

predictor influences the chance of getting the outcome. As an example, you

could quantify the amount obesity plays a role in the likelihood of a person

being diagnosed with Type II diabetes.»

» ^{To develop a formula to predict the probability of getting an outcome based}

on the values of the predictor variables. For example, you may want to predict

the probability that a person will be diagnosed with Type II diabetes based on

the person’s age, gender, obesity status, exercise status, and medical history.»

» ^{To make yes or no predictions about the outcome that take into account the}

consequences of false-positive and false-negative predictions. For example,

you can generate a tentative cancer diagnosis from a set of observations and

lab results using a formula that balances the different consequences of a

false-positive versus a false-negative diagnosis.»

» ^{To see how one predictor influences the outcome after adjusting for the}

influence of other variables. One example is to see how the number of

minutes of exercise per day influences the chance of having a heart attack

after controlling for the for the effects of age, gender, lipid levels, and other

patient characteristics that could influence the outcome.»

» ^{To determine the value of a predictor that produces a certain probability of}

getting the outcome. For example, you could determine the dose of a drug

that produces a favorable clinical response in 80 percent of the patients

treated with it, which is called the ED80, or 80 percent effective dose.