Module Components

• Overview

Topics

• Partial Correlation
• Modeling Independent Variables in Regression Models
• Multiple Linear Regression
• Multiple Logistic Regression

NRSG 795: BIOSTATISTICS FOR EVIDENCE-BASED PRACTICE

Module 9: Model Building-Multiple Independent Variables

Overview

In the previous few modules we examined simple relationships via regressions and epidemiological measures.  However, considering the complex phenomena that are studied in health care, two-variable analyses cannot adequately capture the multiple influences on these phenomena. Moreover, many studies are not experimental and analysis must control for the lack of random assignment and other confounding variables (what is confounding?). Multivariate statistical approaches have become increasingly common – and complex – as computers have made powerful analytical capabilities accessible and affordable.

After an overview of partial correlation, we cover some general concepts important in model building and issues to look out for when including multiple covariates (independent variables). We then expand on what we know from simple regression to perform a multiple linear regression, where the dependent variable is interval/ratio level and the predictors are either interval/ratio or dummy variables. We also introduce you to multiple logistic modeling, where the dependent variable is dichotomous and the predictors are either interval/ratio or dummy variables. Other variations of regression are used when the dependent variables are not normally distributed such as when the dependent is a count or a rate (e.g., Poisson) but we do not describe them in this course.

 

Objectives

At the conclusion of this module, the learner will be able to:

• Understand ramifications of introducing more than one predictor variable in a model
• Recognize variations in linear and logistic regression, including uses and advantages
• Interpret regression analyses as presented in manuscripts

Directions

There are four topics. The first introduces you to the idea of working with multiple variables using partial correlation.  Then there is a module that covers some important aspects and definitions needed for building models with more than just one predictor variable. The last two topics introduce you to multiple linear regression and multiple logistic regression.

The required readings and videos are assigned within the subtopics. Learning activities within the subtopics are designed to help you apply what you learned.