This type of visualisation depicts items stacked one on top (column) of the other or side-by-side (bar), differentiated by coloured bars or strips. A stacked graph is useful for looking at changes in, for example, expenditures added up over time, across several products or services. The graph integrates different data sets to create a richer picture of (the sum of) changes.
Items are "stacked" in this type of graph allowing the user to add up the underlying data points. Stacked graphs should be used when the sum of the values is as important as the individual items. Stacked graphs are commonly used on bars, to show multiple values for individual categories, or lines, to show multiple values over time. Thus, stacked graphs must always work with positive values.
Stacked bar graphs are often used in evaluation to show the full scale of survey responses, from Strongly Disagree to Strongly Agree, for each survey question. There are two options with the stacked graph – one that shows individual elements (raw data) and one that shows results as a percentage. For the latter, the stacked values should add to 100%.
Stacked line graphs often show how quantities have changed over time, such as school racial composition, where each racial category would correspond to a strip in the graph.
While stacked graphs are helpful for conveying multiple levels of meaning simultaneously, they also have some drawbacks. Though it is fairly easy to interpret the values for the first bar or first strip in the graph, it can be difficult to judge the exact widths of any subsequent strips, or to compare the widths of two strips. If accuracy or comparisons are of primary importance, a line graph might be the better option.
Stacked bar chart teaching example
Source: David Shellard & Mark Vogelgesang
2D stacked column chart
Made in Microsoft Excel
Advice for CHOOSING this option
Stacked graphs display cumulative values. Thus, consider whether it makes sense to add up your values; for example, it is again better to use a line graph rather than adding the price of one company to the price of another, or adding the prices of products over time.
Advice for USING this option
Because stacked graphs can be hard to interpret, it may be best to use them in analysis and first decide whether the graph will be meaningful during reporting. It can be most meaningful if there is an obvious pattern in the graph or if the stacked bars are ordered from greatest to least according to one of the quantities in the stack. Consider whether a stacked bar or column is appropriate; in cultures where people read from left to right, a stacked bar chart (rather than a stacked column) may be more easily interpretable. There is no right answer, try both to see which one works best for your data.
As for whether to show percentages or raw data, that is also a decision to make based on the data and desired message. A good test is whether your data points are part of a whole (thus you would choose percentage) or comparison points that wouldn’t naturally be summed as a percentage (e.g., total widgets produced by shift).
Other ways to visualise changes over time
Displaying information as a series of data points connected by straight line segments, on two axes.
Comparing change between two points in time with a line.
Split Axis Bar graph
Regraphing comparison between two points in time by simply graphing the change that has occurred in that time frame.
IBM, Many Eyes: Stack Graph, Stack Graph Guide. Retrieved From http://www-958.ibm.com/software/data/cognos/manyeyes/page/Stack_Graph.html
Stacked area chart vs. line chart – the Great Debate by Andy Kriebel http://vizwiz.blogspot.com/2012/10/stacked-area-chart-vs-line-chart-great.html
Jon Peltier discusses alternatives to a stacked bar chart http://peltiertech.com/WordPress/stacked-bar-chart-alternatives/
Quantitative Displays for Combining Time-Series and Part-to-Whole Relationships by Stephen Few http://www.perceptualedge.com/articles/visual_business_intelligence/displays_for_combining_time-series_and_part-to-whole.pdf