Last year, Dan Butler and David Broockman published an article showing that African American constituents who wrote to state legislators were less likely to receive a response than white constituents were. Jayme Neiman, a graduate student at the University of Nebraska, has applied this same framework to a new idea -- the quality of written correspondence. She details the results in her MPSA paper "Does Quality Matter? State Legislative Response to Constituent Communication."
Basically, she sent out e-mails to a random selection of state legislators across the country. Half received a well-written request for information on registering to vote, and the other half received a poorly-worded, poorly-spelled piece of drivel on the same topic. Neiman reports than 62 percent of the well-written e-mails received a response, while only 45% of the poorly-written ones did. Legislators were also more likely to respond themselves (rather than refer the letter to a staffer for response) to the well-written ones.
From my own experiences answering mail for politicians, poorly-written letters are less of a concern than crazy ones -- aliens, fluoride conspiracies, etc. But I admit that's a lot harder to operationalize in a study.