I really enjoy crafting words. Beyond this blog and Facebook status updates (believe it or not, I sometimes agonize over them), academic papers and my dissertation are my primary writing outlets.  To me, the tedious aspects of writing are the elements tangential to the act itself, like structuring the paper and formatting the document.  One of my favorite writing-related activities is to rewrite a passage, seeing how much more concise I can make it.  Not only does this channel my competitiveness in a productive way, it also allows me the freedom to write initial drafts thinking more about the ideas than how to express them.

Here are a couple of examples from a methodological paper about work I did at the Consulting Center with the Pennsylvania Fish & Boat Commission.  Disregard the symbols; they are a byproduct of writing in LaTeX.  I make no claim that this is great writing, but I think you’ll have to admit that the rewrites are improvements over the originals.

Sample 1 Original (90 words):

What may seem as an arbitrary level of specificity for the computation of these elements (for instance, why calculate $g_{m}$ only for each mode, but $b_{jkl}$ for each combination of geostrata, month, and day type but not mode) is a consequence of our method of calculation which is intuitive but requires a lot of data.  In some cases, there was simply not enough data to estimate every combination and somewhat subjective decisions had to be made about which factors to consider.  A more satisfactory method would involve a model-based regression approach.

Sample 1 Rewrite (79 words):

What may seem an arbitrary level of specificity for the computation of these elements (for instance, why calculate $g_{m}$ only for each mode, but $b_{jkl}$ for each combination of geostrata, month, and day type but not mode) is a consequence of our estimation method which is intuitive but data-intensive.  In some cases, we lacked sufficient data and somewhat subjective decisions had to be made about which factors to consider.  A more satisfactory method would utilize a model-based regression approach.

Sample 2 Original (116 words):

As an illustrative example, suppose only 5\% of agent wait time throughout the day is spent during the 800 hour, while 10\% is spent during the 1000 hour.  In this case, if there was an equal amount of effort being expended during the two hours, twice as much would be captured in the 1000 hour because of the extra time interview agents were spending during that hour.  Thus, to standardize these interview effort quantities we would divide by 0.05 any effort obtained based on an interview during the 800 hour and by 0.10 any effort obtained during the 1000 hour.  Note that these new values have no units and are used simply to standardize the wait times to to be used to produce the distribution.

Sample 2 Rewrite (95 words):

For example, suppose only 5\% of agent wait time is spent during the 800 hour, while 10\% occurs from 1000-1100.  If an equal amount of angler effort was expended in each hour, twice as much would be captured in the 1000 hour because of the extra time agents spent conducting interviews during that hour.  Thus, to standardize these interview-elicited effort quantities we would divide by $0.05$ any interview-elicited effort obtained in an interview during the 800 hour and by $0.10$ any effort obtained during the 1000 hour.  Note that these standardized values are unitless and are used only to produce the distribution.