10 Common Examples of Bad Graphs and How to Avoid Them

Data visualization is a powerful tool for communicating complex information, but it can also be the source of misinformation or confusion. This happens when graphs are not designed effectively, leading to 'bad' or misleading graphs. Let's explore some examples of these to understand what to avoid when creating graphs.

By examining these examples, we can learn about common pitfalls in data visualization and how to prevent them. Let's dive right in.

Axis Manipulation

Manipulating the axis scales or ranges to emphasize or hide data can lead to misleading or incorrect representations. This is often done to make small changes seem dramatic, or to conceal large changes.

A notorious example is the 'Hockey Stick' graph prevalent during the climate change debate, where the y-axis was truncated at particular values to exaggerate post-1990 temperature spikes. This makes the trends appear more dramatic than they actually are.

Starting Axis from Zero

Starting your axis at zero is not always necessary and can sometimes lead to disproportionate scale. For instance, if you have a bar chart with values from 10 to 20, starting the axis at zero will make the 10 look minimal compared to 20, although a 10-point difference is significant.

A better approach would be to start the axis at a reasonable value, such as the minimum value, to highlight changes accurately. Comparison with a scale, rather than the origin, can help viewers understand the true context better.

Contrasting Value Scales

Using different scales for contrasting datasets can make comparison challenging. While it might seem intuitive to do so with very different magnitudes, it can lead to viewers falsely interpreting the data.

For example, if you plot 'Number of Customers' (in thousands) and 'Revenue' (in millions) on the same graph, the revenue bar would look negligible despite having a much higher value, thus misrepresenting the dataset.

Unsuitable Chart Type

Choosing the wrong chart type can bias data representation and interpretations. Each type of chart has a specific purpose, and deviating from these can lead to erroneous conclusions.

For instance, line charts are typically used to display changes over time. Using a line chart to compare discrete values (like bar chart would be more suitable for) can lead to incorrect interpretations, such as assuming a linear progression between data points.

Pie Charts vs Bar Charts

Bar charts are typically better at comparing discrete values between different categories. Pie charts, on the other hand, are useful for showing proportional data, not quantity.

For example, if you're showing sales figures for different regions, a bar chart would be more appropriate. Using a pie chart could lead to viewers overestimating the proportion of sales compared to the actual figures.

3D Charts

3D charts are often unnecessarily complex and can disguise actual data values. They are also more difficult to create and interpret accurately. This can lead to incorrect interpretations, such as viewing depth as a measure of value rather than perspective.

A 2D chart would present the same data more clearly and without adornment, allowing viewers to focus purely on the data.

Labeling and Legend Issues

Incomplete or inaccurate labeling, legends, or titles can make data either confusing or impossible to interpret. Conversely, too much information can clutter the graph and cause visual noise.

For instance, a line chart without a corresponding legend or title might be presented as a demonstration of different trends, but without context, the meaning is unclear. Similarly, overlabeling can distract the eye, making it hard to focus on the most important data.

Missing Legends

Legends are crucial for multi-series charts, where different lines or shapes represent different datasets. Without a legend, viewers may be unsure what each line represents, leading to incorrect interpretations.

A clear, concise legend helps viewers understand at a glance what different data series mean.

Misleading Titles

An inaccurate or misleading title can completely alter the viewer's interpretation of the graph. A title that twists or exaggerates the data can effectively turn the graph into a lie.

For example, a graph titled 'Sales Decline' for a stable or slightly increasing sales data can mislead viewers into thinking sales are plummeting—despite the data actually indicating otherwise.

Knowing what to avoid in creating graphs can significantly improve the clarity of your data visualizations. By ensuring your graphs are accurate, easy to interpret, and allow viewers to draw the correct conclusions, you can use graphs to effectively communicate your data.

Remember, the goal of data visualization is presentation, not decoration. Keep your graphs clean, simple, and focused on the story you want to tell. With this in mind, you're well on your way to creating effective and useful graphs that truly inform and enlighten.