Charting and Graphing Tips

Before You Develop a Chart or Graph

  • Develop questions
  • What relationships or comparisons are you interested in making?
  • What hypothesis would you like to test?

Consider the Following:

  • Begin by looking at the data sets and decide which one(s) you would like to investigate.
  • Review the general characteristics of the data.
  • Decide which type of graph would best illustrate the data, such as line, bar or scatter.

Choose Which Type of Graph to Use

Don’t worry if you’re not positive that you have chosen the best type of graph for the data. If it does not work out as you expected, you can easily choose a different type of graph, using the same data. Below are descriptions of various graph and chart types and how they are typically used. If possible, use software (e.g., Microsoft Excel or Adobe Illustrator) to dynamically demonstrate how to create and edit charts and graphs.

  • Scatter (XY) plots – Scatter plots are used for examining relationships between two types of data. If you know that one of the types of data depends on the other (for example, waves depend on wind), the dependent variable (effect) goes on the Y-axis (vertical left) and the independent variable (cause) goes on the X-axis (bottom). The dots are not normally connected in a scatter plot, though a ‘best-fit’ or regression line may be drawn through the points to illustrate the relationship. The dots are connected only if you expect that a measurement taken between the two X values would yield a Y value midway between their associated two Y values.
  • Line Graphs – These graphs are a hybrid of the scatter plot and bar graph. They are used for looking at the relationship between two continuous types of data. Line graphs are appropriately used when the independent variable is evenly spaced (e.g., whole numbers) or non-numeric (e.g., station codes) and the information is expected to be continuous. For example, you might reasonably expect that if you look at a station between these two geographically you would find an intermediate Y value. The points are connected in a line graph.
  • Pie Charts – These charts are used to depict data indicating fractions (percentages) of a whole. Pie charts are useful with fewer data points.
  • Bar Graphs (Columns)  Bar graphs are used for making comparisons between discrete cases or to look for trends, such as over space or time. Bar graphs are appropriately used when one of the axes — usually the X — is discontinuous, evenly spaced (e.g., dates) or non-numeric (e.g., station codes).

Other Types of Bar Graphs

  • Clustered column. A clustered bar graph is used to simultaneously depict several series on one graph. This is the same as putting multiple series on a scatter plot or multiple lines on a line plot.
  • Stacked. This is a combination of two or more bar graphs into a single image. This should be used only where the sum of the series has meaning. For example, graphing particulate carbon for one series and dissolved carbon for a second series, which add to total carbon, is appropriate. Although, adding temperature, plus oxygen, would not be appropriate because the total bar height would have no independent meaning.
  • Broken stick (100% stacked bar). When comparing two sets of percentage data, typically two pie charts side-by-side are used to illustrate the comparison. However, in comparing multiple sets, a broken stick graph is more typical. Like the individual pie charts, each column of the broken stick chart represents 100% subdivided into pieces.
  • Three-dimensional. This is used to simultaneously compare three data sets, which are believed to be simultaneously related. The X-axis (horizontal) is used for the independent variable, the Z-axis (vertical) is used for the dependent variable; the variable graphed along the Y-axis (depth) may be a second independent variable or also dependent on X. For example, if you suspect that wave height is dependent upon both wind speed and water depth, you might graph wind speed (independent) on the X-axis, water depth (also independent) on the y-axis, and wave height on the Z-axis.
  • Surface plot. This is similar to a 3-D column but should only be used for continuous data.
  • Bar graphs with multiple, independent variables. Bar graphs are used for making comparisons between discrete cases or to look for trends (usually over space or time). Bar graphs are appropriately used when one of the axes (usually the X) is discontinuous – evenly spaced (e.g., dates) or non-numeric (e.g., station codes).

Developed by Eastern Michigan University. Edited by Michigan Sea Grant.