At the bottom of every bill on Richmond Sunlight is a comment form, with a textarea for typing a message and a few text input fields for name, URL, and e-mail address. On January 30, I significantly increased the size of the textarea, making it more than twice as large in area. My theory was that doing so would increase the average length of comments on the site.
Yesterday I graphed the average number of characters in each comment by week. Week 4 is when I changed to a larger textarea:
While hardly a scientific study, I’m convinced.
I doubt there’s enough here for statistical significance…
You should do a study: configure the system randomly to give everyone one of five sizes of box, and see if there’s a correlation to comment length.
Unfortunately, the session is over, so my source of data is kaput. :) But yes, Roger, you’re right that a proper analysis would require just such a test. Some day I’ll do it.