Title: Creating More Effective Graphs
Author: Naomi B. Robbins
Publisher: Wiley (2005)
Any book that quotes an author who says tables are preferred over “dumb pie chart[s]. The only worse design than a pie chart is several of them. . . given their low data-density and failure to order numbers along a visual dimension, pie charts should never be used” is a book after my own heart.
I’ve been fascinated by the idea of what makes data displays meaningful and effective, given the diversity of learning styles and approaches of educators who use data. After wandering through Amazon.com for a few minutes, I stumbled over name of the Tufte who has spent considerable time research and writing about visual data displays. Tufte lead to Robbins and her book “Creating More Effective Graphs” as it serves to bring together many of the concepts presented by Tufte and his contemporaries.
Robbins takes real examples, explains why they are useful or not, and how to fix errors. As a highlighter-notes-in-the-margin reader, I highlighted several passages and immediately applied several of her recommendations in future data displays. Although Victoria Bernhardt devotes a chapter in the 2nd edition of her book Data Analysis for Continuous School Improvement to the issue of presenting data, Robbins book takes data displays and the construction of data and sets out the reasoning behind decisions. I agree with Robbins statement that Excel’s default settings do more harm than good . . . rumor has it though that Excel 2007 gets rid of many of quirks that get in the way of good data displays. I’m keeping it on my shelf next to my Excel books and pulling it out for a quick consultation when I’m preparing a new graph. I’m not happy with the trellis charts I’ve built but they are definitely better than what I was creating before!
For what it’s worth, Bernhardt breaks one of the hard rules of data – when your count of students (or n) gets below 5, suppress the student ID. For example, in her perception graph on page 202 of the book mentioned above, she disaggregates students’ responses by ethnicity (Total, Black, American Indian, Asian, White, and Other). In her summary notes, she says “Asian students are in highest agreement with all items.” There are two Asian students. Two students - who if this was an actual school could be identified by anyone looking at the data that had a passing familiarity with the student body. Imagine if these students were siblings. It could be interesting experience for their parents to be sitting in a meeting and hear the results of the school survey. Especially if the Asian students had the lowest level of agreement. . . Nothing like a little public disclosure to make a student feel comfortable about responding to a survey. Makes me wonder why she didn’t include the Asian responses with the “Other” group. This leads back to the issue of respectful data displays. I’d be curious how often we convey unintended messages with our data displays. . .