Chapter 13

Causality

Show Causality, Mechanism, Structure, Explanation. If one variable causes another, then show both on the same graph.

Content Counts Most of All

Analytical presentations ultimately stand or fall depending on the quality, relevance, and integrity of their content. The only reason to have data is to help further an explanation. Any data that does not help clarify the relevant issues should be omitted. If your data is no good, then get better data

Bar Graph

Bar graphs are used to compare discrete categories on a common measure. In other words, the measure is quantitative and the categories are qualitative. For example, compare sales figures (quantitative) in different geographic regions (qualitative).

Integration of Evidence

Completely integrate words, numbers, images, diagrams. Annotate key data points on the graph. As a courtesy to the reader, include explanations right on the graph rather than burying them in the text.

Line Graph

Line graphs compare continuous categories on a common measure. They are often seen in business, especially to show trends over time. The time appears on the X axis, and the quantitative data to be measured appears on the Y axis. Examples are sales, stock market trends, or mortgage rates over time. A text box is often placed near these graphs to explain the reasoning behind changes in trends.

Pie Chart

Much like a bar graph, a pie chart compares discrete categories on a common measure. Unlike a bar graph, a pie chart can only show one series of data. Although pie charts are commonly used in the media, a bar graph is a better way to convey the same information in a way that allows for much easier comparisons between categories. The relative height of bars is easier to compare than pie slices that must be mentally rotated and aligned for comparison. Bar graphs also allow for multiple data series to be compared on the same graph, but pie charts are limited to one data series. Overall, pie charts should be avoided.

Scatterplots

Scatterplots do the best job of adhering to Tufte's principles of showing multivariate data and causal relationships. In a scatterplot, both sets of data are quantitative. The cause (independent variable) appears on the X axis, and the effect (dependent variable) appears on the Y axis. For example, in economics, scatterplots can be used to show trends in price versus demand. Greater demand for a product or service (independent variable) leads to higher price (dependent variable). In spite of their explanatory power, scatterplots are rarely found in business.

Comparisons

Show comparisons, contrasts, differences. Comparison is the fundamental act in statistics. Always ask the question, "Compared to what?" The comparison often is in time, such as a trend over the last five years. A comparison of multiple trends would be even better.

Multivariate Analysis

Show more than 1 or 2 variables. It is a complex world, the more variables you show, the better chance that you have of offering an explanation.

Documentation

Thoroughly describe the evidence. Provide a detailed title, indicate the authors and sponsors, document the data sources, show complete measurement scales, point out relevant issues. Documentation establishes the credibility of the evidence. Are these numbers something that we should believe?

Evaluate a Graphic

• To make sense of large amount of data, it is often displayed in a chart or graph • When created carefully and specifically, graphics transfer a huge amount of data to readers enabling them to learn and retain much more than if the data was all written out as text • Follow Tufte's Principles to avid bad data