Data Visualization: Charts, Histograms, and Scatter Plots
Data visualization in geography is the use of charts and graphs to show patterns, trends and relationships in data; the main types include bar charts (for categories), histograms (for the distribution of a continuous variable), line graphs (for change over time) and scatter plots (for the relationship between two variables). In brief: choose the graph that matches the data and the question. A bar chart compares separate categories with gaps between the bars, a histogram has touching bars that show how many values fall into equal ranges, a line graph joins values in time order and a scatter plot with a line of best fit shows whether two measurements are related. Every graph needs a title, labeled axes with units and a suitable scale.
A clear graph makes patterns visible and supports honest conclusions, while a poor one can hide or distort them. This guide explains how to choose and construct the main charts, how to read scatter plots, how to avoid misleading graphs, compares the chart types in a table, works through a histogram and a correlation example, outlines the history and corrects common mistakes.
How do you choose a graph?
First, identify the type of data. Categorical (discrete) data, such as land use types or transport modes, are shown with bar charts or pie charts. Continuous data, such as heights, temperatures or pebble lengths, can be grouped into classes and shown on a histogram. Time series are shown on a line graph. Two numerical variables measured for the same cases are shown on a scatter plot. Next, think about the question: comparison (bar chart), distribution (histogram), trend (line graph), relationship (scatter plot) or proportion (pie chart or divided bar). Good graphs follow common rules: a descriptive title, axis labels with units, an even scale starting at zero for bars, a clear key if more than one series is shown, consistent colors and a source line for secondary data.
How do histograms and bar charts differ?
A bar chart compares separate categories. The bars have gaps between them, and their order can be changed (for example, sorted by size). A histogram shows the distribution of a continuous variable: the data are divided into class intervals of equal width (such as 0–2 cm, 2–4 cm and so on), and the height of each bar shows the frequency (the number of values in the class). The bars touch because the scale is continuous, and the order cannot be changed. The shape shows the pattern: a symmetrical bell shape (normal distribution), skewed (a long tail to one side) or having two peaks. In geography, histograms can show the sizes of pebbles, the heights of waves or the ages of residents, and the shape can reveal, for example, that most pebbles are small but a few are large. Choose between 5 and 10 classes, and avoid classes that are too wide (they hide detail) or too narrow (they give a noisy pattern). The mean can be estimated from grouped data by using the midpoint of each class.
How do scatter plots show relationships?
On a scatter plot, each case is a point, with the independent variable (the one you think causes change, such as distance from the source) on the horizontal axis and the dependent variable (the response, such as pebble size) on the vertical axis. A line of best fit is drawn through the middle of the points, with about as many points above as below. The pattern shows the type of correlation: a positive correlation (both increase), a negative correlation (one increases as the other decreases) or no correlation (no pattern). The closer the points are to the line, the stronger the relationship. Outliers, points far from the others, should be examined: they may be errors, or interesting exceptions. A statistical measure of the strength is the correlation coefficient, such as Spearman’s rank coefficient. An important rule is that correlation does not prove causation: two variables may be related because of a third factor or by chance.
What other charts are used in geography?
- Line graphs: for time series, such as temperature or river discharge over time. Storm hydrographs combine a line for discharge and bars for rainfall.
- Climate graphs: bars for monthly rainfall and a line for temperature, with two vertical axes.
- Pie charts and divided bars: for the proportions of a whole, best with a limited number of categories.
- Population pyramids: back-to-back bar charts of age groups for males and females.
- Kite diagrams: for the distribution of species along a transect, such as on sand dunes.
- Radar (star) graphs: to compare several variables for a place.
- Triangular graphs: for three components adding up to 100%, such as sand, silt and clay in soil, or employment in three sectors.
- Cumulative frequency graphs and box plots: to show the median and spread of data.
Chart types compared
| Chart | Best for | Data type | Geographic example |
|---|---|---|---|
| Bar chart | Comparing categories | Categorical | Land use in five wards |
| Histogram | Showing a distribution | Continuous, grouped | Pebble length at one site |
| Line graph | Showing change over time | Time series | River discharge during a storm |
| Scatter plot | Showing a relationship | Two numerical variables | Pebble size and distance downstream |
| Pie chart | Showing proportions | Parts of a whole | Share of employment by sector |
Worked example: histogram and scatter plot
The data are hypothetical.
| Class (cm) | Frequency | Midpoint (cm) |
|---|---|---|
| 0 to under 2 | 3 | 1 |
| 2 to under 4 | 8 | 3 |
| 4 to under 6 | 12 | 5 |
| 6 to under 8 | 5 | 7 |
| 8 to under 10 | 2 | 9 |
- Total: 3 + 8 + 12 + 5 + 2 = 30 pebbles, which is the sample size.
- Modal class: the class with the highest frequency is 4 to under 6 cm (12 pebbles).
- Relative frequency: the modal class has 12 ÷ 30 = 40% of the pebbles, and the two lowest classes together have (3 + 8) ÷ 30 = 36.7%.
- Estimated mean: using midpoints, the sum is 1 × 3 + 3 × 8 + 5 × 12 + 7 × 5 + 9 × 2 = 3 + 24 + 60 + 35 + 18 = 140. The mean is 140 ÷ 30 = 4.67 cm. The distribution is slightly skewed to the right, since the tail of larger pebbles is longer than that of smaller ones.
- Scatter plot: pebble length (cm) measured at eight sites that are 2, 5, 8, 11, 14, 17, 20 and 23 km from the source are 14.2, 12.5, 10.8, 9.9, 7.5, 7.9, 5.2 and 4.1. Plotting distance on the x axis and length on the y axis shows points that fall from top left to bottom right, a strong negative correlation, with a small exception at 14 km and 17 km, where the sizes are 7.5 and 7.9.
- Gradient of the line of best fit: a line from (2, 14.0) to (23, 4.0) has a slope of (4.0 − 14.0) ÷ (23 − 2) = −10 ÷ 21 = −0.48 cm per km, so the pebble length decreases on average by about 0.5 cm for every kilometer downstream.
How did graphical methods develop?
William Playfair invented the line graph, bar chart and pie chart in the late 18th and early 19th centuries. Florence Nightingale used her “rose” diagrams in the 1850s to show the causes of deaths in the Crimean War, and Charles Joseph Minard’s map of Napoleon’s march to Moscow (1869) combined geography, time and numbers in one image. Francis Galton introduced the scatter plot and regression in the 1880s, and John Tukey’s exploratory data analysis (1977) popularized box plots. Today, software such as spreadsheets, R, Python and GIS can produce graphs quickly, and interactive dashboards are widely used, but principles of clarity and honesty stay the same, as emphasized by Edward Tufte.
Which mistakes are common?
Mistake 1: Using a bar chart with gaps for continuous data
Continuous grouped data need a histogram, with touching bars and equal class widths on an axis with a continuous scale.
Mistake 2: Starting the axis of a bar chart above zero
A truncated axis exaggerates differences between bars. If a different scale is needed, state it clearly or use another chart type.
Mistake 3: Joining scatter plot points with a zigzag line
Scatter plots show a relationship with a single line of best fit. Joining each point suggests a time sequence that does not exist.
Mistake 4: Concluding that correlation proves cause
A relationship in a graph shows association. Cause needs further evidence, such as a mechanism and the exclusion of other factors.