The danger of a single story

  Sometime last semester, I went for a data presentation where the book - The Visual Display of Quantitative Information by Tufte-Ian - was discussed. The presenter spoke about what is good data visualization and what is not using interesting examples from popular sources. It was an interesting seminar which made me realize how many plots I had made without bearing the message nor my audience in mind. I realized how I had been making plots just for the sake of adding visualizations. Of course, an engineering report would look bland if it has no plot to show. The presenter talked about pie-charts as an example of a needless plot because it does not effectively show the relationship between variables of little difference. This made sense to me. He said he hated pie-charts and I realized I didn't like pie-charts either. I remember how I feel reluctant whenever I had to interpret it in an aptitude test. I left the seminar thinking about the effectiveness of pie-charts compared to bar-charts and similar charts then concluded that pie-chart was the least effective. I even felt pie-charts had no use whatsoever, why do people still use the plot?

  However, I enrolled in an applied statistics course the following semester and heard another narrative. While the instructor introduced the course, he listed some descriptive tools and their uses. He said pie-charts are used to visualize parts relative to a whole. That was the most important sentence I heard in the class. Just like what I heard the previous semester, it made perfect sense to me because I could not think of a better plot which gives a clearer insight like pie-charts about samples as part of a whole. What did I learn from all these?

  There is danger in a single story. That a narrative makes sense does not mean that it is absolute and binding in every case.