Statistical Applications for Environmental Analysis and Risk Assessment

Chapter 5Graphics for Data Analysis and Presentation

5.1 Introduction and Overview

Plotting the data is always a recommended first step in performing a statistical analysis, as the graphical review provides immediate insights into the nature of the data, identifies unusual data values or patterns, and suggests the appropriate types of follow-up quantitative analysis. Graphical exploratory tools are available for various data types, including single-variable (i.e., univariate) data contained in a single data sample, single-variable data contained in multiple data samples, and bivariate (i.e., involving two variables) or multivariate (involving multiple variables) data contained in single or multiple data samples.

For univariate single data samples, commonly used graphical analysis methods include box and whiskers plots (or simply, box plots), probability plots, histograms, and quantile plots. Two univariate data samples can be compared using quantile–quantile plots or double quantile plots, while side-by-side box plots provide an effective method for evaluating two or more univariate data samples. A single bivariate data sample is most easily visualized using the familiar two-dimensional scatterplot, which is simply a plot of the matching data values from the two variables that may be connected by lines if desired to form a line graph or time series plot (if one of the variables is time). Multiple bivariate data samples can be represented as individual scatterplots on the same ...