Statistics Canada
www.statcan.gc.ca
Common menu bar links
Summary
If you decide that a graph is the best way to present your information, then no matter what type of graph you use, you need to keep in mind the following 10 tips:
10 tips to make your graphs great!
- convey an important message
- decide on a clear purpose
- draw attention to the message, not the source
- experiment with various options and graph styles
- use simple design for complex data
- make the data 'speak'
- adapt graph presentation to suit the target audience
- ensure that the visual perception process is easy and accurate
- avoid distortion and ambiguity
- optimize design and integrate style with text and tables
| Graph type | Description |
|---|---|
| age pyramid | Represents age structure of a population. |
| vertical bar graph | Compares important data values. Displays data better than horizontal bar graphs, and is preferred when possible. |
| dot graph | Displays a comparatively large number of categories when category order is unimportant. Best used when portraying category values in descending order. |
| histogram | Shows discrete or continuous variable data in a similar way to column graphs, but without the gap between the columns. |
| histograph (frequency polygon) | Depicts continuous variable data. Smoothes abrupt changes which may appear in a histogram |
| horizontal bar graph | Compares important data. Useful when category names are too long to fit at the foot of a column. |
| line graph | Depicts data over time. |
| pictograph | Favoured by professional graphic artists, although students can create simple pictorial presentations as well. Comparisons must be accurately depicted and respect the scale. |
| pie chart | Compares a small number of categories. Values should be markedly different, or differences may not be easy to decipher. Labelling pie segments with their actual values overcomes this problem. When data points are similar, the pie chart's message may be misunderstood. A bar graph may be better in this case. |
| scatterplot | Measures two or more variables thought to be related. |
