Values and quality requirements

Students and lecturers should experience assessment as fair, and as stimulating learning. These are our assessment values to which we strive while creating assessments. In order for that to happen, there are quality requirements for assessment that need to be met. However, some of the quality requirements counteract each other, because if you make a perfect assessment, it will costs a lot of time. Therefore, you will balance these quality requirements.  

The quality requirements for assessment are (for a more detailed description, see the TU Delft Assessment manual): 

A valid assessment measures what it is intended to measure: 

• It measures the performance of students of all learning objectives, at the correct level (e.g. Bloom level). And nothing more. The assessment should cover the learning objectives, and nothing more. 

• The assessment method matches the learning objectives and learning activities 

• The amount of questions during an assessment corresponds to the importance of the learning objectives and the time spent on this learning objective during the course. 

In reliable assessments, the grade accurately reflects how well individual students master the LOs. 

Reliability can be split in three parts: 

1. Specific:

   Only students who achieved the learning objectives at the minimum acceptable level will pass. This implies the following: 

   1. Check for fraud & plagiarism. 

   2. Check for contribution to group work. 

   3. Grading should have enough precision. This implies that large questions that are automatically graded as ‘correct’ (all points) or ‘incorrect’ (no points) should be prevented or mitigated, for example by having students explain their calculation method and granting partial points for partially correct answers. 

2. Test-retest reliability:  
   Students would get the same grade if they would redo the assessment.  
   This implies: 

   1.  that assessment is very clear as in what it expects of students: A question is not susceptible to different interpretations. e.g. Students should know the required level of detail of the answer in order to get a full score. 

   2. That the influence of chance on the grade is minimized 

3. Objectivity: 
   There should be no rater bias, i.e. who is doing the grading and at what time the grading takes place should not influence the grade of the student, and neither should the identity of the student influence the grade.  

   1. Assessors should receive clear grading instructions that should ensure objective assessment, minimize bias and align the grade with the learning objectives and assignment instructions. Other grading-related measures include anonymous grading and grading per question (instead of per student). 

   2. If possible, grading should take place anonymously. 

Transparency means that it is clear to students on what grounds and how they are tested. Information about the assessment and grading is accessible before, during, and after the assessment: 

• The learning objectives that are assessed and the criteria they are assessed on 

• How their grades will be established 

• The test itself: about points per question, admissible aids, cut-off score, date, and available time 

The assessment should be feasible for both your students and you. 

• Students: It should be feasible for students to finish the assessment within the allotted time. In addition, study delay should be prevented: Check for conflicts with deadlines and assessment slots in other courses. 

• Lecturers: The assessment should be feasible for the lecturers, examiners and assessors within the allotted time. Take into account assessment construction, the actual assessment, giving feedback, and marking. 

Numbers of students and assessment methods affect the feasibility. 

The assessment should make the students learn: 

• Students learn while making the assessment 

• Students should receive structured feedback before they hand in something for a grade 

• The assessment plan(ning) within a course should stimulate regular studying