Anchor Links Example

3. TU Delft assessment agreements

This chapter describes TU Delft agreements on assessment. Some of these agreements are specific for the TU Delft. Therefore, this chapter is relevant for new teaching staff and other employees, especially in case they have been working in other universities. Most agreements originate from the models for the Teaching and Education Regulations (model TER, Onderwijs en Examenregeling in Dutch) and from the model for the Rules and Guidelines of the Board of Examiners (model R&G, Regels en Richtlijnen van de Examencommissie in Dutch). In case the operationalisations differ per faculty or if there are exceptions, this is indicated in the text.

3.1 TU Delft assessment terminology

In this assessment framework, we use the terminology that lecturers and programme management use, to optimise the readability of this document. However, some terminology is different in laws, regulations, and tools. These are discussed below, together with an overview of assessment characteristics. In addition, Appendix F contains the following tools:

  1. Appendix F.1: Glossary (assessment terminology with definitions / descriptions)
  2. Appendix F.2: Translations English-Dutch (assessment terminology)
  3. Appendix F.3: Acronyms (often used abbreviations)

3.2 Planning of assessments

To ensure studyability for students (Quality requirement 6) and feasibility for teaching staff in terms of time and resources (Quality requirement 8), the model TER contains the following agreements on the planning of assessment:

3.3 Written exam specific information: registration

To enable the scheduling of resources that are required for written exams (Quality requirement 8), the following agreements exists in the model TER:

3.4 Fraud prevention and detection

Fraud is defined as ‘any act or omission by a student that makes it fully or partially impossible to properly assess the knowledge, insight and skill of that student or another student’.22

Fraud prevention and detection helps to ensure that an assessment result correctly reflects how well an individual student masters the learning objectives (Quality requirement 3, leading to fairness). In order to create fair assessment, the TU Delft takes fraud prevention and fraud detection measures23 that facilitates catching students who commit fraud. This has a preventative effect.24

3.5 Scoring and grading

For transparent (Quality requirement 1) and reliable (Quality requirement 3) grades/results, as well as a feasible process for examiners (Quality requirement 8) the following agreements are in the model TER:

3.6 Test result analysis to adjust grading guide & grades

The faculty’s assessment policy (see 4.1) describes how lecturers are advised/required to do a test result analysis to check for the need to adjust the answer model or grading guide/assessment sheet (see 1.3). The policy also describes in what case what methods examiners should use to adapt the grade calculation and/or the cut-off scorell. This is to ensure the reliability (Quality requirement 3) of grades/results.

3.7 Communicating grades & feedback, student review & discussion of assessed work, appeals and the validity of results

The model TER includes processes that stimulate learning from assessments (Quality requirement 4) by giving students the right to review. In addition, it includes processes to ensure fairness of grading by giving students insight in their scored and graded work (Quality requirement 1) and by the appeal process in case students do not agree with their result.

3.8 Ownership and archiving of assessments and student work

3.9 Guidelines remote assessment & fraud prevention

In the case remote assessment is needed, the TU Delft delivers good quality assessment (the quality requirements in 1.2) and specifically a healthy balance between 1) quality assurance (fraud prevention measures, Quality requirement 3), 2) privacy concerns, 3) enabling students in demonstrating how they master the learning objectives (Quality requirement 2 & Quality requirement 7), and limit stress for students (Quality requirement 6). This has resulted in limiting the use of online proctoring in remote assessment as laid down in the online proctoring regulations19. Online proctoring can only be used in exceptional cases where other remote fraud prevention measures are insufficient and where remote assessment is the only option, and only after approval from the concerned board of examiners. This can be the case in specific individual cases, like students who cannot come to the exam hall due to chronic health issues (Quality requirement 7).

For remote assessments, the following guidelines for remote assessment28 apply:

  1. The assessment assesses all learning objectives in a reliable way (Quality requirement 1 & Quality requirement 2).
  2. Fraud prevention measures do not hinder student performance (Quality requirement 3 & Quality requirement 6), i.e., aims to limit stress for students in these assessments.
  3. Helpdesk: the examiner is available for students during the assessment.
  4. Practice exam: The examiner provides a practice exam to enable students practice with the setting, questions & tools (Quality requirement 1 & Quality requirement 3).
  5. Feasible: The assessment is feasible for both students and lecturers (Quality requirement 6, Quality requirement 8 & Quality requirement 9).
  6. Extra time: Students with disabilities receive the required extra time (Quality requirement 7).
  7. Privacy: The assessment complies with the privacy regulations.
  8. Transparency: The examiner communicates assessment details to the students via the LMS (Brightspace) and email (Quality requirement 1).

3.10 Guideline for use of (AI) tools in assessment

Below, the initial 8 guidelines for teaching staff (June 2023) on the use of (AI) tools by students in non-invigilated assessments are listed. The guidelines are in development. On this page, a more extensive and recent version can be found.
These guidelines are only relevant outside exam-like environments, in which students will likely use available (AI) tools.

  1. Discover the possibilities and limitations of (AI) tools and discuss them with the students.
  2. Promote safe use of AI tools and plugins and do not reveal personal, internal or confidential information.
  3. Be transparent and explain choices. Discuss with students how they can follow the Code of Conduct29 in the context of AI tools. Communicate changed expectations to students.
  4. Attribute correctly: Inform students on how they should correctly attribute the use of AI-tools.
  5. Reduce the need of students to rely on AI tools by making them feel confident: Have sufficient feedback moments and regularly check the progress of individual students.
  6. Focus on the students’ process if the course is heavily influence by AI tool use: Shift assessment criteria towards the process, track progress using version control.
  7. Take fraud detection measures & report suspicions of fraud to the board of examiners: Consider doing oral authenticity checks to check if it is likely that the student produced the deliverable by themselves.
  8. Rethink the course, including learning objectives and course assessment planpp.

3.11 Assessment adaptations for students with a support need

Students who encounter obstacles during assessments due to e.g. functional limitation, disability, chronic illness, psychological problems, young parenthood, gender transition, or special family circumstances may request adjustments of assessment (TER art. 25, Quality requirement 7), after consultation of Horizon (desk for studying with a disability or extra support question) for standard support facilities, or their academic counsellor for customised adjustments.

Standard assessment support facilities include 10 minutes per hour extra exam time for students with e.g. dyslexia. Customised assessment adjustments depend on the individual situation of the student and may include changes in assessment type, timing, permitted aids (e.g. dictionaries) and location (TER art. 25.4). This includes exemptions from attendance requirements.

For customised assessment adjustments, the board of examiners (or the mandated academic counsellors in some faculties) will evaluate the student requests on the following criteria30:

  1. If possible, the adjustment must still allow assessment of the learning objectives of the course at the required level. If this is not possible, individual degree programmes of students should still cover and assess all final attainment levels of their degree programme (TER art. 25.1).
  2. The adjustment must be efficacious for the student: it should be suitable and necessary (Wgbh/cz31 art. 2.1).
  3. The adjustment should not place a disproportionate burden on the faculty / TU Delft (Wgbh/cz art. 2.1), in terms of time and money32.

3.12 Composition of assessment committees for graduation projects

The board of examiners establishes rules on the composition of the assessment committee for the graduation project in order to secure assessment competence (see 6.3 for examples & guidelines, model R&G art. 25).

3.13 Graduating with honours or cum laude


t Each faculty will have different values.

u Remote and hybrid exams are normally not allowed by boards of examiners, except for in lockdown situations or in case of students with specific support needs.

v Multiple select: a multiple choice type where multiple options can be selected. Should only be used in specific situations. See here.

w ‘plagiarism scan’ can be digital in case of digital work, but also manual. Assessors always need to be vigilant about fraud by detecting suspiciously similar work or mistakes.

x Example: If a programme prepares students for writing a thesis in different courses in which students write a report that is assessed on writing skills, they can define ‘report’ as a separate assessment method that is explicitly mentioned in the assessment programme. This can help to make learning lines more explicit.

y Mandatory deadlines are considered summative assessments (or ‘examinations’ in terms of the TER).

z Teaching weeks are numbered p.w where p is the period number (1-4 are the regular periods, 5 is the summer period that is only used for resits), and w is the week number (1-10). September 1st typically falls in week 1.1. See https://www.tudelft.nl/en/student/education/academic-calendar

aa There are some exceptions. Examples: EE has two resit weeks in week 5.2 and 5.3; ME use eight ‘octals’ instead of 4 ‘quarters’.

bb Because not all first year bachelor students can be expected to find a room before the start of their first year.

cc Students who registered for the exam can enter the exam until 30 minutes after the start of the exam.

dd Examples: During lockdowns, 1) some lecturers administered digital exams in which students could answer one question at a time, without the possibility to access previous questions; 2) online proctored digital exams during lockdowns, stress was caused because students e.g. feared that roommates or family members would walk into the room, or that there would be technical error.

ee At IDE, a ‘NI’ (niet ingeleverd, not delivered) is chosen if a deliverable was not delivered (in time).

ff Example: if students participate in a project/computer lab, but do not hand in the summative assignments (in time).

gg This has at least been the case since 2006. In other Dutch universities, a (rounded) 5.5 is considered a pass grade.

hh In de model TER and model R&G, the term ‘interim examination’ is used. The term ‘interim’ is very appropriate for midterm exams that test the first half of course as apposed to the final exam that tests the entire course. However, for courses that consist of an exam and a practical, the term ‘interim’ is not appropriate. Therefore, we use the more neutral term ‘partial’ here.

ii In some programmes, the regular assessment of a course consists of a midterm and final exam, while the resit consists of one large exam that covers both regular exams. This is typical for BSc year 1 courses.

jj This has at least been the case since 2006. Other Dutch universities keep the last grade.

kk During the pandemic, the results of some online exams were changed into pass/fail instead of grades because of the lower reliability. This arose the question of whether a ‘pass’ was considered higher than a 6.0 or not.

ll The Cohen-Schotanus procedure is advised for score-grade calculation adjustments, but not in resits (because the student population is not representative). ME uses Angoff cut-off score calculation beforehand combined with Hofstee cut-off score adjustments afterwards. See TU Delft assessment manual.16

mm In case students appeal against course results to their BoEx, the BoEx is legally obliged to forward the appeal to the EAB. After receiving the appeal, the EAB will request the BoEx to mediate between the student and the examiner.

nn Student work could be altered after returning it to the student, which hinders e.g. a second opinion.

oo Examples per assessment category:
      o exam: exam, answer model (including grading guide)
      o oral exam: used questions, cases, scenarios, etc.; grading guide/rubric/assessment form
      o project/assignment: manual, mandatory template, rubric, assessment form

pp For group work / projects, consider e.g. checking the transfer of skills & knowledge by adding an individual exam on project related cases. However, consider study load as well.