Module Components

• Overview

Topics

• Describe Associations: Cross Tabulation/
  Contingency/Pivot Tables
• Statistically Testing Association: Chi-Square
• Measuring Disease
• Simple Logistic Regression

NRSG 795: BIOSTATISTICS FOR EVIDENCE-BASED PRACTICE

Module 8: Associations Between Nominal or Ordinal Variables

Overview

In health care, many of our outcomes of interest are in terms of yes/no.  You may hear sound bites on the radio that says drinking 1 glass of red wine per day decreases the likelihood of getting colon cancer. Remember in addition to the type of comparison, the level of measurement of the variables and basic assumptions are fundamental to determining what analytical procedure to use. In this module we continue to review various ways tests to measure the association (or relationship) turning our attention to nominal or ordinal variables.

In linear regression, dependent variables are interval level variables (e.g., satisfaction score) and an ordinary least squares (OLS) is used to estimate the “best” equation to predict the dependent variable. When the dependent variable is a dichotomous variable, (e.g., yes infection/no infection), we cannot use linear regression where assumptions of normality cannot be met. Instead, we use logistic regression.  While the approach is very similar to linear regression in terms of model building, there are no beta coefficients to interpret. Instead, odds ratios are produced which are more easy to interpret in terms of how they influence the likelihood of having the outcome (e.g., infection).

When examining relationships among nominal or ordinal level variables, the first step is often preparing a table that reflect how the variables relate to each other. These tables are known under various names including cross tabulation and contingency tables. Summary tables developed from individual observations in Excel or other software are called pivot tables. Information such as percentages and statistical tests (e.g. chi square, odds ratio, relative risks) can be derived from cross tabulations, particularly when they are in the form of a 2x2 table. While chi square testing provides a means to test if an association is significant, odds ratios are important because they provide information on the magnitude or strength of the association. Odds ratios can be calculated using a simple formula or via logistic regression models.

Objectives

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

• Develop tables (e.g., cross tabulation, contingency or pivot tables) that summarize the relationship between variables.
• Calculate odds ratios and relative risk given frequency data.
• Identify the correct statistical procedures to use when exploring associations (or relationships) with a dependent variable

• Describe the findings of an association found in a table or figure.

Directions

There are four topics. The first introduces you to creating tables to describe the associations.  Chi square and odds ratios are the most common statistical tests to measure associations between two categorical variables. The fourth topic introduces you to logistic regression in its simplist form where we only have one predicting variable. 

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