CS/SE Instructors Can Improve Student Writing without Reducing Class Time Devoted to Technical Content: Experimental Results – Paul V.Anderson

The problem, in a nutshell: Employers report graduates have strong technical skills, but they lack communication skills. Paul did a 4 year study with 14 different institutions, both with computer scientists and science communication researchers. The fundamental outcome is:

We need to integrate technical and communication work, because the two are integrated in industry as well. 

Interestingly enough, projects and labs for courses are often realistic, they perform small exercises like a code review. But, a part is missing, the communication of the results to the owner of the code. Paul wants to address this with Reader-oriented Integrated
Technical and Communication Instruction, or RITCI.

What matters here is that there are two things: scenarios (what needs to be communicated) and genres (different documents that are often produced in the same way, like a requirements document)

Experiment

Hypotheses: Using RITCI without taking time away from course contents will help both communication and technical abilities.

The experiment was ran un CSC216 at NC State. It is a course on Programming concepts, following introductory programming and has three projects consisting of 1) writing a test plan and 2) implementing the tests. So the genre here is the ‘test plan’.

The experiment compared two groups:

2015-05-21 17.14.09

For the treatment group, a number of changes were made to the curriculum:

2015-05-21 17.15.08

Changes were made in the following manners:

  • For the treatment group, the assignments stressed that the test plan should have the reader in mind, by adding a few sentences where it was explained the test plan should enable the reader to execute the test cases. More specifically they explained the background of the reader (everyone who can use Eclipse should be able to do it)
  • The control group got a template that mentioned “you need to make a test plan” whereas the treatment group was provided a  with a model test plan that explained in a lot of detail how to execute given tests. The model test plan also included actual test cases with IDs, descriptions and expected results. They stressed often that anyone should be able to use it
  • The rubric was changed for the treatment group as well, by adding reader-oriented fields. For example:

2015-05-21 17.23.35

I think the most important thing to understand here is that the assignment moved away from: you have to write this assignment for you or for the lecturer, but you are writing for a concrete, realistic reader. A small difference, with, as it will show, a huge impact.

Results

Paul had a number of computer science and software engineering lecturers graded all test plans and it turned out that students in the treatment group scored better both on the technical skills and the communication skills!
2015-05-21 17.26.57