[with thanks to a former university Head of Examinations for input and discussion]

Recent years, and weeks, have seen a move away from the traditional examination context, where candidates gather in large halls to write on paper, to candidates being assessed using computers, in small groups or individual work spaces. In this change, the role of the invigilator (also known as "proctor") remains important; there may be new opportunities to use technology to assist in the invigilation process. However some of the technological systems being proposed appear to significantly expand, or even to contradict, the role of the invigilator. Such changes require particularly careful scrutiny, as they are likely to change both the purpose and the effect of examinations.

Invigilator's Role

The role of the invigilator is to help all candidates complete their exams to the best of their abilities, in as supportive an environment as possible. Significant departures from appropriate conduct, either of examinations in general or of the specific paper being assessed, must be recorded and should be corrected where possible. These records will then be used by academic staff to determine how, and in some cases whether, the candidate's assessment should be marked. An invigilator must only prevent a candidate completing their assessment if this is the only way to prevent disruption to others.

Management of the Examination

The invigilator has three main roles in managing the conduct of the examination:

• Time-keeping: the invigilator will generally provide interval warnings, for example when there is one hour, or ten minutes to go. Each candidate should be able to choose how much more precisely to keep track of time, to minimise their own stress level;
• Faults in papers: if candidates have questions about the examination paper – for example because it appears to have an error or be unclear – the invigilator(s) should be able to refer these promptly to a member of academic staff and then to disseminate their ruling to all candidates in all locations;
• Managing/recording unexpected situations: in a physical exam hall the invigilator is responsible for recording incidents such as fire alarms or water spills and maintaining, so far as is safely possible, compliance with the exam rubric. The impact of such disturbances on individual candidates' performance is assessed later, by academic staff. In an online context incidents could include, for example, loss of internet connectivity.

Improper conduct that should be detected

The invigilator is expected to detect three main types of improper conduct (note that details of these may vary depending on the exam, for example whether specified external resources are allowed, and whether annotations are permitted):

• Wrong candidate taking exam: the identity of candidates should be checked to confirm that the individual who sits the exam is who they claim to be. If a candidate cannot show a matching identity document they should normally be permitted to sit the exam (in case they have good reason for lack of ID) but the circumstances must be recorded. Similarly, if a candidate who is not on the list wishes to sit a paper, they should normally be allowed to do so, as this may be the result of simple mistakes by student or institution. Both situations can be resolved later.
• Candidate using unauthorised materials: the exam rubric should state what additional resources (if any) the candidate is allowed to consult; use of resources not on the list should be prevented or detected.
• Candidates colluding (either within or outside room): unless the exam rubric permits collaboration, candidates' attempts to communicate with others should be prevented or detected.

Improper conduct that is not expected to be detected

It is important to remember that there are some types of improper conduct that invigilators are not expected to detect. These include those that would require extreme preparation by candidates, or where detection would require unacceptable levels of intrusion and stress for them. In particular, invigilators are not expected to detect:

• Anything that cannot be detected during the exam session (e.g. that requires a recording of the whole examination process);
• Anything that requires long-term preparation (e.g. candidate substitution after obtaining false ID documents);
• Anything that requires intimate inspection/monitoring/restriction of candidates.

Methods (face-to-face)

During a traditional examination, invigilators use a number of techniques to detect improper conduct:

• At the start of the exam, or at least early in the period, identity documents are checked against the candidate's appearance (see above for if the candidate does not present an ID); invigilators are not expected to be experts in detecting forgeries;
• Where candidates are allowed to bring materials into the examination, these are checked to ensure they meet requirements (for example on annotations);
• If there is a grace period for latecomers to enter the exam, candidates are prohibited from leaving their desks until after that period;
• If candidates are permitted to leave their desks and return they are supervised, but not intimately, while away;
• Invigilators continually survey the examination hall from a distance; if they see something that, based on their experience, raises concerns then this may trigger…
• Invigilators periodically inspect the desk area: the candidate has warning of their approach.

Thus invigilation consists of a combination of continuous, coarse-grained observation from a distance and occasional close-focus inspection. To reduce stress, the candidate is aware when the latter is taking place: a candidate who believes they may be under continuous close-focus human surveillance or recording is likely to experience considerable stress and will not perform their best.

Methods (remote/digital)

Technology should be used to contribute to the invigilation process, not merely to allow it to be conducted remotely. Simply providing a continuous video-conferencing link that effectively seats the invigilator on the candidate's desk fails the invigilator, the candidate and the technology: the invigilator has to do at least as much work, the candidate is placed in a more stressful environment, and the technology is badly under-utilised.

Rather than merely a communications medium that removes any distance (even the length of the exam hall) between the invigilator and the candidate, technology should aim to be part of the invigilator's alerting and record-keeping process: detecting abnormal situations, recording these for later consideration, and drawing the attention of a human invigilator in cases where either correction or further investigation is required. Where possible, technological innovation should support the invigilator's full role: allowing it to be performed with less stress (for both candidate and invigilator) and more effectively than a physical human presence. For example:

• An e-proctoring system might take a snapshot of the candidate's work at the point of any alert, break, or interruption. This extension of the face-to-face practice of ruling a line on the script after any fire alarm allows the work to be checked for sudden bursts of "creativity" or correction that might suggest a candidate had been consulting unauthorised materials during the break;
• An e-proctoring system might detect unexpected sounds, such as the turning of pages in a book, record and alert the human invigilator to these;
• An e-proctoring system might detect, record and alert on changes in typing cadence, suggesting that the individual entering the assessment is no longer the intended candidate;
• An e-proctoring system might detect, record and alert on unusual patterns of system or network activity that may suggest the candidate is engaged in unauthorised activity;
• An e-proctoring system might detect, record and alert on environmental changes, such as loss of network connectivity;
• Etc. etc….

[UPDATE: colleagues in Norway have pointed out that many of these 'in-machine' mechanisms already exist in systems designed for sitting computerised exams in traditional exam hall environments]

Since digital systems cannot understand the full human context of what may take place during the examination, their most likely role is in the continuous, "distance" monitoring aspects of invigilation. If an e-proctoring system detects behaviour that it does not understand, or believes may be suspicious, it should raise an alert. The circumstances around the alert can then be checked, either by a human invigilator during the exam or, if this is not possible, by a member of academic staff, using the system's records and their knowledge of the individual candidate, afterwards.

 "E-proctoring" systems that do no more than reproduce, badly, the face-to-face invigilation process should be regarded with suspicion.

Other Factors

Finally, remember that the invigilation process is not the only thing that prevents candidates cheating in examinations. Invigilation/proctoring, whether in person or remote, must be part of an assessment system that is designed to be resistant to cheating. Other means to this end include:

• Before assessment: considering alternative ways of assessing or verifying performance; designing assessment so that it is hard for candidates to benefit from collusion.
• During assessment: considering open book exams (so the risk of "unauthorised sources" is designed out) or, conversely, conducting on-line exams using kiosk systems that only allow candidates to view permitted resources.
• After assessment: using a moderation process to identify candidates whose examination performance differs markedly from what was expected from their work in class; also to review invigilators' records for anything that might significantly affect the performance (either negatively or positively) of an individual or group.