Line Graph

History

The idea of specifying the position of a point using two axes, each reflecting a different attribute, was introduced by René Descartes in 1637 (what are now called Cartesian coordinates). During the following century, graphs were used to display the relationship between two variables, but they were all hypothetical pictures and were not based on empirical data. William Playfair, who has been described as an “engineer, political economist, and scoundrel,” is credited with inventing the line graph, pie chart, and bar chart. He first used line graphs in a book titled The Commercial and Political Atlas, which was published in 1786. In it, he drew 44 line and bar charts to describe financial statistics, such as England's balance of trade with other countries, its debt, and expenditures on the military.

Types

Perhaps the most widely used version of the line graph has time along the horizontal (x) axis and the value of some variable on the vertical (y) axis. For example, it is used on weather channels to display changes in temperature over a 12- or 24-hour span, and by climatologists to show changes in average temperature over a span of centuries. The time variable can be calendar or clock time, as in [Page 719]these examples, or relative time, based on a person's age, as in Figure 1. Graphs of this latter type are used to show the expected weight of infants and children at various ages to help a pediatrician determine whether a child is growing at a normal rate. The power of this type of graph was exemplified in one displaying the prevalence of Hodgkin's lymphoma as a function of age, which showed an unusual pattern, in that there are two peaks: one between the ages of 15 to 45 and another in the mid-50s. Subsequent research, based on this observation, revealed that there are actually two subtypes of this disorder, each with a different age of onset.

Figure 1 Prevalence of Lifetime Mood Disorder by Age and Sex

Source: From “The Epidemiology of Psychological Problems in the Elderly,” by D. L. Streiner, J. Cairney, & S. Veldhuizen, 2006, Canadian Journal of Psychiatry, 51, pp. 185–191. Copyright 1991 by the Canadian Journal of Psychiatry. Adapted and used with permission.

Also widely used are line graphs with a continuous variable, such as weight, displayed on the abscissa, and another continuous variable (e.g., serum cholesterol) on the ordinate. As with time graphs, these allow an immediate grasp of the relationship between the variables, for example, whether they are positively or negatively correlated, whether the relationship is linear or follows some other pattern, whether it is the same or different for various groups, and so on. This type of display is extremely useful for determining whether the variables meet some of the assumptions of statistical tests, which might require, for example, a linear association between the variables.

...