© Ugur Akinci
Line graph or line chart is one of the most frequently used types of graphic displays in technical communication.
Here is an example:
Consumption of Fiber by U.S. Apparel Industry, 1994-2001
(Millions of pounds)
(Source: U.S. Department of Commerce, Bureau of Industry and Security)
The most important aspect of a line graph is its use of “categories” for the X-axis (HORIZONTAL). This is usually a some sort of time variable like days, weeks, months, years, seasons, etc.
To be more technical about it, the X-axis represents an “interval level” data: that is, (i) the consecutive units of this type of data are separated from one another by the same discreet quantity and (ii) they are lined up on a linear scale.
For example, not only the year 1995 comes right after 1994, but there is exactly the same amount of time (12 months) between the years 1994 and 1995 as between 1999 and 2000, etc.
The Y-axis (VERTICAL) is similarly represented by interval level data, but it never represents time. It represents any quantity that can be measured and expressed by INTEGER numbers.
For example, a logarithmic scale (either for Y or X axis) cannot be used for a linear chart.
A linear chart implies a TIME TREND. By connecting the dots we are claiming that there is some sort of trend and the data is not completely random.
However, a linear chart stops short of two things:
- It does not prove the implied trend, and
- It does not tell us anything about the possible cause of the implied trend.
This is why, even though very widely used and popular, a line graph really is not very useful in terms of understanding the PROVEN REASONS why trends emerge, continue, and disappear. It’s an introductory level visual aid that SUMMARIZES time-related trends, if any.
For example, in the above graph, “we can tell by one look” that the fiber consumption went up from 1994n to 1995 and then it fell steadily until 2001.
However we have no idea about two things:
(1) We have no idea why the fiber consumption first went up and then fell down since the only “independent variable” (X axis) we have is “years” which does not suggest any causality relationship to fiber consumption.
(2) We also are not sure whether the increases and decreases were as sooth and continuous as the “connected dot pairs” suggest.
For example, what if every June the fiber consumption skyrocketed by 200% and then fell back to its average value? Would such spikes be represented by this chart? Of course not, yet it would be something very significant to investigate if such spikes indeed did exist.
So continue to use the line graph with the above mentioned reservations in mind. Use it for practical presentation purposes but be smart and critical about it as well.