Evaluating Data Visualizations
As an information consumer, you need to be critical of data visualizations like any other information source.
Although the information is presented in an eye-catching way, it is possible for the data to be misinterpreted, over-simplified or over-complicated.
Below are some tips to help guide you through the evaluation process.
Tips to evaluate Data Visualizations
1. Truncated axis
The baseline for any graph should begin a "0" for both the X and Y axis. When a graph begins at a number other than zero, the data can be skewed and misinform the reader.
2. Manipulating the scale
The scale for the both the X and Y axis should increase proportionately in order for the reader to accurately interpret any changes in the data. Compressing or expanding the scale can make the changes in the data seem more or less significant than they actually are.
3. Cherry picking data
Sometimes designers only include certain data points to reinforce the intent of the graph. Any missing data should be indicated in the rather than completely excluded.
4. The wrong graph for the data
It is important to choose the right graph for the data. If you use the wrong graph, the data can be skewed which can confuse the reader.
5. Failed calculations
Failed calculations are often represented in pie charts. Slices should always add up to 100 and show parts of a whole.
6. Correlation implying causation
Just because two sets of numbers follow a similar path doesn’t mean there’s a correlation.
7. Violating standards of conventions
Over time, data visualizations have established standards which readers expect when they read any visualization. For example, on a map visualization, the darker the shade indicates a higher level of density. Graphs that reverse this convention can confuse the reader and cause them to misinterpret the information.
8. Too hard to understand
Visualizations should make data easier to understand and not confuse the reader. If the visualization includes too much data, colors, etc. the reader will have to work too hard to understand the message of the graph.
9. Bias
Purposeful Bias is a deliberate attempt to influence data, most likely to take the form of data omissions or adjustments.
Selective Bias is slightly more discreet where the sample population is not representative of the population at large or the population of interest. One example is: Surveying college students about legal drinking age.
Content for this list was compiled from: 5 Ways Writers Use Misleading Graphs To Manipulate You and https://guides.zsr.wfu.edu/interpretdataviz
Examples of Poorly Designed Visualizations
Inverted axis
Pie Chart Adds up to more that 100
Bar heights should represent value
Line Chart should follow data over time
