# Australian Bureau of Statistics

 ABS Home #search{vertical-align:-3.5px; }
 Understanding statistics

 Module 2: Describing, Clarifying and Presenting Data 3. Data Displays 3.4. Selecting the appropriate graphic display As a ‘rule of thumb’ you might like to use the following recommendations to represent information graphically. You will find them useful when you are analysing the way data are presented in journals, newspapers, magazines and on the internet. 1. Bar charts In bar charts, the bar height is a frequency or count. Bar charts are used to represent qualitative data, i.e. data in categories measured on nominal or ordinal scales; qualitative discrete data, eg. number of visits per month to a doctor. 2. Histograms Histograms are a special case of bar charts. They are used for continuous data (the ‘bars’ touch each other, whereas in a bar chart there are gaps between the bars – though you should also note that many books and journals do not obey this distinction); to represent relative frequencies of quantitative measurements of one variable and the relative frequencies add to 1. 3. Line graphs Line graphs are used to indicate a relationship between 2 variables. Scatterplots can also be used for this purpose. Line graphs indicate trends such as those that might happen over time. A line between successive points should only be drawn to show a trend over time. 4. Pie charts In these charts, the circle represents 100% so the segments of the circle can be used to show percentages or proportions. A pie chart should usually be used only with a small (say 4-8) number of categories. To make the relationships easier to see the segments should be ordered from large to small with the largest segment starting at 12 o'clock (by convention). It is useful for the darkness of the segments to be ordered in the same way as the magnitude of the segments. It is generally more visually satisfying to make the smallest segment the darkest. Some authors require the sectors to be ranked in the clockwise direction, but it is ranking of the sectors which is most important. Research[12] has shown that people read proportions more accurately in a pie chart than they do in a stacked bar chart. However, the best simple graphical display for comparing proportions is the unstacked bar chart showing both numbers and proportions on the vertical scale[13,14]. Clearly, pie charts have some limitations with respect to presenting data, but they tend to be a commonly used graphical display so you should be aware of their strengths and weaknesses. 5. Pictorial graphs or pictographs Used to indicate simple numeric differences between groups. However, the use of objects can distort relationships and misleading information is more likely to be presented. For pictographs to be effective each symbol must have a stated value. If you increase the size of the object then you are distorting the data and going against the dimensions rule for graphic displays. For example, if you double the height and the width of an object, you increase its area four times.