Anchor Links Example

1. Assessment values and guidelines

Assessment values help to make consistent decisions on assessment. This chapter explains what the current TU Delft vision on assessment is, for each of the following four aspects: 

  1. What the value of assessment is at the TU Delft?
  2. What the TU Delft assessment values and resulting quality requirements for assessment are?
  3. How the TU Delft defines and maintains assessment quality? 
  4. How the combination of all TU Delft assessment tools support this?

The scope of the TU Delft vision on assessment is good quality assessment. The vision on assessment follows the vision on education3. Since the vision on education differs per faculty or even per programme, a directive central vision on assessment would diminish the freedom for the faculties’ (and programmes’) visions on education and assessment. That is why the TU Delft vision on assessment focusses on good quality assessment, and that faculties and programmes will have different visions on education and as a result, different visions on assessment.

1.1 The value of assessment

1.2 Assessment values and quality requirements

The assessments at TU Delft should be fair, meaningful, and feasible. The TU Delft strives towards these three assessment values and they form the basis of decisions on assessments. They are the basis for the following 9 quality requirements for assessment (see Figure 3) that match the TU Delft vision on education3, the TU Delft vision on teaching and learning9, and the assessment framework for the higher education accreditation system of the Netherlands10. Since it is impossible to obtain all requirements simultaneously, the TU Delft strives for a balance between the nine assessment requirements.

The value ‘fair’ consists of 1) transparency, 2) validity, and 3) reliability. The value ‘meaningful’ consists of meaningfulness 4) for learning and 5) for life. The value ‘feasible’ consists for students of 6) studyability and 7) inclusivity, and for teaching staff of 8) time and resources and 9) competence.
Figure 3: The three TU Delft assessment values and nine quality requirements for assessment.

The value ‘fair’ consists of 1) transparency, 2) validity, and 3) reliability. The value ‘meaningful’ consists of meaningfulness 4) for learning and 5) for life. The value ‘feasible’ consists for students of 6) studyability and 7) inclusivity, and for teaching staff of 8) time and resources and 9) competence.

1.3 Assessment quality

The TU Delft achieves good quality assessment because all aspects of assessments (building blocks) are consistent with each other and because both their consistency as well as their individual quality is monitored and improved in the Plan Do Check Act (PDCA) cycle13 of the Quality Assurance System of the TU Delft. This section discusses what these building blocks are and how this improvement takes place.

See running text. The red, surrounding blocks (6, 7, and 8) support and are the fundament of the central blocks (1-5).
Figure 4. Assessment building blocks at the TU Delft that determine the assessment quality.

See running text. The red, surrounding blocks (6, 7, and 8) support and are the fundament of the central blocks (1-5).

Assessment at the TU Delft can be divided into eight aspects that we will call building blocks of assessmentj (adapted from the assessment web4, see Figure 4):

  1. TU Delft Assessment framework
  2. Faculty assessment policy
  3. Programme assessment plan
  4. Course assessment plan
  5. Assessment
  6. Assessment competence (of everybody with an assessment task)
  7. Assessment organisation (all levels)
  8. Legal framework (all levels)

These building blocks range from national level to assessment level. The TU Delft vision on education and the national legal framework are the basis for the TU Delft assessment framework, which includes the assessment vision. This framework, and the faculties’ educational vision and policy form the basis of the faculties’ assessment policies. The latter forms the basis for the assessment programs of BSc and MSc programmes, which are influenced by the programmes’ intended learning outcomes. From the programme assessment plans, the course assessment plans sprout, based on their learning objectives, with the actual tests at the bottom of this chain. The assessment organisation and assessment competence support all previously mentioned building blocks.

A faculty’s assessment policy can deviate from the assessment framework, if this is motivated by the faculty’s vision on education and if this does not negatively influence assessment quality.

At least every six years, the assessment quality of each of these building blocks is systematically assessed and improved by checking its requirements. Assessment is of good quality if each building block supports the assessment values (fair, meaningful and feasible), adheres to the previously described nine quality requirements for assessment, and meets the block’s conditions as described in the following chapters. The conditions are summarized per building block in the ‘Introduction and summary’ and summarised chronologically in Appendix A.1.

1.4 Digital assessment tools: goal and landscape


b Students will of course learn from formative assessment. Learning will positively influence their course result (grade or pass/fail).

c Feedback that is structured per applicable LO or assessment criterion. The term is used as the opposite of ‘unstructured feedback’, in which not all LOs or criteria are systematically assessed.

d This is important if there are two parallel courses. If one of them has fully voluntary formative assessments while the other course has small assessments that counts for the course grade, students tend to focus on the second course.

e or exit qualifications

f In assessment literature, assessor bias is considered to mainly influence reliability (instead of validity), probably because most assessment effects are caused by the order in which the student work is assessed (sequence effect, norm shift, contamination effect, etc.11) and are therefore in general randomly distributed over the students.

g In exam settings, fraud is related to the conditions in which the exam was administered, and not related to the validity of the assessment itself. Therefore, it is listed under reliability. In assessments without invigilation (e.g. take-home assignments and projects), fraud like plagiarism and free-riding are explicitly forbidden, like it is in professional situations. Therefore, fraud is not assumed to influence the validity of the assessment, but the reliability. I.e., we assume that the reproducibility of the assessment is jeopardized.
However, in case of widely available tools (like AI tools) that graduates will use in their professional life, lecturers must assume that students will use this tool during non-invigilated assessments and adjust the learning objectives and assessment criteria accordingly. If the AI tool is not taken into account, the validity of the assessment is jeopardized.

h If this is appropriate and feasible for the course.

i See model TER2 art. 25.1: “Students with the [sic] support need means students who are held back due to a functional limitation, disability, chronic illness, psychological problems, pregnancy, young parenthood, gender transition, or special family circumstances, for example in relation to informal care”. Accessibility can be increased by for example using dyslexia-friendly fonts, colourblind-friendly colours, and clear instructions, and by offering TU wide facilities like the extension of exam time.

j ‘Building blocks of assessment’ are called ‘entities’4, ‘pyramid layers’5, and ‘building blocks’6 in literature. Since we are a technical university, ‘building blocks’ seems fitting.

k In an assessment matrix on exam level, learning objectives (LOs) are rows, and levels of the taxonomy (e.g. Bloom’s) are columns. Cells contain subquestion numbers and corresponding numbers of points/weights. Per learning objective (row), the cell(s) that correspond to its level of the taxonomy is/are highlighted.

l In a consistency check table, learning objectives (LOs) are rows, and deliverables/processes are columns. Cells contain the applicable assessment criteria and their weight (or points). Columns with formative assessments contain criteria with 0% weight.

m A scoring guide contains the number of (+) points per correct step, and (if applicable) a list of mistake/omission and corresponding subtraction points (-). For essay questions, a rubric can be used.

n R&G: Every faculty has their own abbreviation. See E.5

o ‘Structured feedback’ is feedback that is structured along the learning objectives or assessment criteria.

p Examples: the availability of user-friendly test result analysis will facilitate and stimulate lecturers to use them to adjust the answer model and increase the reliability of the grade; increasing the ease of giving feedback will stimulate lecturers to give good quality feedback; facilitating partial grading of open-ended questions will increase grade reliability compared to single answer numerical questions.

q Möbius is replaced by Ans as standard exam tool as of 2023.

r Currently, on-campus exams use a TU Delft based safe exam environment.

s Online proctoring exams can only be administered as a last resort, and after approval of the Board of Examiners19.

Anchor Links Example