Whenever I read the newspaper, magazines, or online articles, I am pretty interested in data visualization. I like to see the story that is told by the chosen type of chart or graph, the formatting, the word choice, etc.
Recently, I saw a couple of graphs in periodicals I regularly read that had a similar violation. The first was in my college’s quarterly alumni magazine. It is a graph of the percentage of students in cohorts who qualify for Pell Grants in a short piece on class mobility. Below is my rough re-creation of the graph:
The second was in a weekly newspaper we receive for news about our county. It is a graph of the recent budgets of the county’s public schools in an article about budget approval. Below is my rough re-creation of that graph:
The shared violation for the above graphs is that neither y-axis begins at zero.
Part of the power of graphs is the ability of our visual system to automatically detect patterns. This ability is exploited when graphs are presented with a non-zero baseline. Patterns and differences are magnified, resulting in visual misrepresentations.
Below are new versions of the above graphs with a zero baseline:
Do you notice a visual difference between the originals and the “new” charts with non-zero baselines? Do you think it is ever okay to use charts with non-zero baselines?