As I come off grading the reports turned in for my Applied Bayesian Time Series course and as I head into the next quarter where I will teach Statistical Consulting, I have been thinking about how to describe good writing to students. Now, I really claim to be no expert on the topic, but I think that I can generally avoid many of the obvious pitfalls. These pitfalls are usually not obvious until somebody points them out at which point you cannot believe you ever fell into them in the first place. It was a happy coincidence that a blog I normally read, environmental economics, posted some advice for `economical writing.' In particular, I liked this point
"5. Tables are writing." This is a big one for me. One of my committee members told me that your data and regression tables should be standalone. Don't include your mnemonic in the list of variables; i.e., use "Smoke less" instead of "SMKLSS."
This is relevant for figures as well. The reader should not need to read the text to understand the figure. This means that the axis labels, legends, and caption should tell the story. Some of the points I just don't understand
"15. Watch out for bad words." For example: via, intra, and/or, respectively, thus, overall, basic, factor, etc. We had a nice discussion of "respectively."
Huh? Perhaps this is particular to how these words are used, but I don't think they are inherently bad in their own right. I am a fan of Strunk & White for their brevity. They boil down English grammar into a few specific rules and principles. Of course, there are always times that rules need to be broken.

blog comments powered by Disqus


25 March 2010