A **Bar Graph** is a method of displaying and comparing discrete data sets. Bar graphs display categories as rectangles whose height (if vertically displayed) or length (if horizontally displayed) are proportional to each other by the proportionality of the data they are displaying. The graph shown displays the number of dog owners per breed in a sample with a rectangular bar displaying the discrete value for each breed. Breeds with a larger quantitative value of owners have a taller rectangle while breeds with fewer owners have a smaller rectangle. This is a particularly useful method of displaying data to make relative comparisons. As opposed to a histogram, there are spaces between each bar indicating that “breed” is not a continuous measurement but rather a discrete individual grouping; a categorical variable, not a quantitative variable.

The wikipedia article on bar graphs has a useful introductory discussion and provides a brief examination of bar graph mechanics. But for a more detailed and exhaustive work on bar graphs the Syracuse University math department has a useful review page. The Catalyst Movement put out an interesting infomercial of the construction of a three-dimensional freestanding bar graph filled with used soda cans. While the information displayed in the bar graph, the categorical variable rate of obesity in the nation in a given year, may be correct, there is no other data. The Catalyst Movement states that soda consumption has doubled and implies a cause-effect relationship. To have confidence of a correlation or even association would require more data be displayed which is not appropriate for a bar graph while the causality implicated in the video requires data from a more scientific controlled experimentation.

[…] clustered bar graph, also called a grouped bar graph, is visually very similar to a “regular” bar graph, with one important difference in the data being presented: a second categorical variable is used […]