From a reply to a colleague's email...
You're correct that significant changes should be based on sufficient data. But what constitutes “sufficient data” may differ in every case. The primary safeguard of the integrity of our programs and our assessment procedures is that evidence-based decisions about changes are to be made by the program’s faculty.
Some programs, for some of their learning outcomes, have several semesters’, even several years’, data. But if a program only has data on a given learning outcome from a single semester, the faculty may very well decide that the appropriate “action” is to continue collecting data until there’s enough to justify doing something. In such a case, the making of that decision and the meeting and discussion from which it arose are, themselves, assessment-related actions that should be reported. More on that point below.
The Assessment Report (as distinct from the Assessment Plan) calls for reporting the assessment-related actions that were taken during a given academic year. There are several types of assessment-related actions, but they fall into two main categories and assessment results may yield evidence bearing on both types of actions:
- Changes in the instructional program for the purpose of improving students’ learning. Guided by assessment results, there’s an infinity of things that a faculty might do to improve their students’ learning and achievement of the program’s learning outcomes. These possibilities include modifying the program’s learning outcomes, themselves. Just a couple of other examples are developing a curriculum map in order to better specify the courses in which the various outcomes are addressed, and deciding on a core set of course-level learning outcomes for particular courses.
- Changes in assessment procedures. Guided by assessment results, the faculty may decide to add, delete, or modify the means by which student achievement of learning outcomes are measured. For instance they may improve a rubric, develop a plan for portfolios of student work, or figure out ways in which the measurement of some program-level learning outcomes can be embedded into certain courses.

