3.2 Statistical graphics

The main elements of a statistical graph are largely the same as those in a statistical table: the title or caption, the explanatory components, the graph itself and its footnotes. Their quality requirements are quite similar, too. However, the graph is often looked at without reading the appended text. This is why the graph has to be understandable as such.

We should again begin our examination of the figure from its caption, which should tell us the subject covered, the variables described and the time period. The unit of measurement is often illustrated in the scale of the figure, as is the classification of variables. The legend indicates the meaning of the symbols used (such as colours).

There are five basic types of statistical graph: bar graph, line graph, scatter plot, pie chart and cartogram.

The most commonly used among these statistical displays is the bar graph. In a bar graph, the length of the bar represents the quantity described. Since the length of the bar is crucial to understanding the message, it is important that it is based on a real scale with equal intervals. If the scale is truncated (i.e. if some values have been omitted from the scale), the figure will be misleading. In most cases it is best to draw the bars horizontally. Vertical bars are used for describing changes over time; in this case the horizontal axis describes the passage of time. It is also good to organise the data by size of the described subject so the reader gets a clearer picture of the distribution.

Production of fossil and renewable energy in Finland 1995, 2000 and 2005 TJ

Source: Statistics Finland, Energy 2007

Household dwelling-units and form of tenure in Finland 2006

Source: Statistics Finland, Housing 2007

Another option instead of using a bar is to use image symbols or dots corresponding to a certain number of units.

Source. Statistical Yearbook of Finland 2007

One type of bar graph is the population pyramid, where different age groups are placed on the vertical axis by age, with women on the right hand side and men on the left hand side of that axis.

Finnish population by age and gender at 31 December 2006

Source: Statistics Finland, Population 2007

The age structure of the population can be illustrated with the population pyramid.

Line graphsare often used to illustrate variation and trends of development in different phenomena. The most typical use of the line graph is in the representation of time series (changes taking place over time).

Marriages and divorces in Finland 1987-

Source: OSF: Changes in marital status, 2008

The key factor with regard to interpreting the line graph is the curvature of the line. This is why it is possible to truncate the scale without this affecting the message. On the other hand, the shape of the figure has a major impact on the curvature of the line. By stretching the figure it is possible to make even rapid changes seem much slower, and by narrowing the figure even minor changes seem quite dramatic.

In a scatter plot, observations or clusters of observations are depicted using dots or image symbols placed in a co-ordinate system.

Life expectancy (2007) and GDP/capita (2006) in European countries

Source: World in numbers, (29 Oct. 2008)

In this kind of graph it is possible to examine the size and location of different groups in relation to two variables, which sheds useful light on the interdependence of these variables. The example above shows that wealth correlates positively with a higher life expectancy.

Pie chartsare also used a lot. They are not, however, well suited for purposes of comparisons. It is difficult to compare two pie charts, indeed even comparisons between the slices within one chart can be difficult.

Household consumption 2006

Source: OSF: Household consumption 2006 http://stat.fi/til/ktutk/index.html (29 Oct. 2008)

Pie charts are often presented in 3D, which further complicates the task of comparing the size of the individual slices.

A cartogramis a thematic map showing geographical statistical information in the shape of dots, symbols, bar graphs or pie charts. Another option is to colour regions in the map according to the relative intensity of the characteristic measured.

Example of a qualitative regional classification map.