A recorded presentation is a short prepared talk that each student records and hands in. As an assessment it keeps what a talk tests, choosing what matters, ordering it and saying it to an audience, and moves the listening out of lesson time.
EdProcto, which publishes this guide, carries the shape as Presentation: the student records the talk as sound in the browser, the slides arrive beside it as a file, and you mark both against your rubric. A large class answering unseen questions is another job, covered in oral exams at scale.
Most of the research is on live talks at universities. Every source was read on 10 October 2026: Brookhart (2018) in full, the other studies as an abstract or an ERIC record.
What does research say about assessing oral presentations?
It says to mark what is said apart from how it is said, to treat one marker's score as moderately reliable, and to use students' own ratings for learning before marks.
What should a presentation rubric separate?
Content from delivery, at the least. Carlson and Smith-Howell (1995) tried several speech evaluation forms and found that any of them scored speeches reliably and validly, provided it covered those two constructs.
Brookhart (2018) reviewed 46 studies of rubrics in higher education. Taking the criteria one at a time suits feedback, she writes, and one overall decision is quicker for grading. Of her 29 analytic, descriptive rubrics, 14 had four levels.
Structure, visuals and answering questions are our additions to the marking criteria, each on its own line so that polished slides cannot stand in for an argument.
How reliable is one marker?
Moderately, on one hearing. Magin and Helmore (2001) analysed four years of engineering presentations, each rated by teams of staff and of students. Staff agreed far more than students did, yet one teacher's rating had an intra-class correlation of only 0.40 to 0.53, and it took the average of two to four peers to match it. For one four-level presentation rubric in Brookhart's (2018) table of reliability evidence, scores agreed exactly 66% of the time and within one level 98%.
Carlson and Smith-Howell (1995) are more hopeful: a lack of extensive rater training did the scores no major harm.
Can students assess themselves and each other?
For feedback, yes; for marks, with care. In Langan et al. (2008), peer marks tracked tutors' more closely than self-marks did and ran higher, while female students rated themselves below their tutors' grades.
Self-assessment pays off on the next talk. Ritchie (2016) video-recorded two presentations from each student; the section required to rate its first recording on a rubric scored higher on the second.
What does recording change for anxiety, rehearsal and fairness?
Little for nerves, in the one comparison we could read, and a good deal for rehearsal. Braun (2017) compared science students presenting in teams, in class or online with tools such as screencasts. Presentation anxiety was similar in both, and academic performance did not differ.
Menzel and Carrell (1994) found better speeches from students who had prepared longer and rehearsed aloud more often. A recording turns rehearsal into takes, and students do not have equal evenings, so we let a clock cap the takes.
Does accent sway the mark?
It can, through the listener. Kang and Rubin (2009) describe reverse linguistic stereotyping: what a listener believes about a speaker's group distorts the judgement of the speech. A recording cannot hide a voice. So the rubric below has no line for accent, and delivery asks only whether the words can be followed. Our guide to spoken answers has the studies of intelligibility.
How do you write the task for a recorded presentation?
Put one question, a time limit, a named audience and what must be shown into the prompt itself.
- Ask one question that needs a position. "Should the café charge for cups?" produces an argument; "Talk about packaging" produces a list.
- Set the time. Three to five minutes suits a class task.
- Name the audience. "The owner of the café" tells a student what to leave out.
- Say what must be shown. A recommendation in the first half minute, one calculation, one risk.
- Settle the slides and the takes. The recording is sound, so slides come as a file: three at most, as one PDF. Allow takes, and let the clock cap them.
- Write the question that follows. Nobody is in the room to ask one, so set it in advance, seen only after the talk: "What is the weakest point in your case, and how would you answer it?"
- Build it. In EdProcto: a file upload question for the slides, the Presentation question, a spoken answer for the question that follows, and a short answer switched to "Not marked: a reflection, not a question" for the student's own rating.
What do students need to know about the recording?
Who will hear it, what it is for and when it will be deleted. A student's recorded voice is personal data. The UK's Information Commissioner's Office (n.d.) says personal data may be kept only as long as it is needed, with retention periods set in a policy and the data erased or anonymised afterwards, and that the law itself fixes no periods. Your school's or institution's own policy decides, and nothing here is legal advice.
Where speaking is the obstacle and not the skill, the spoken answers guide says what to offer.
What does a four-level rubric for a presentation look like?
Five lines, four described levels, and nothing that cannot be heard or seen in what was handed in.
| Line | 4 | 3 | 2 | 1 |
|---|---|---|---|---|
| Content | Answers the question; each claim has a correct figure or source | One claim unsupported or one figure wrong | Several claims unsupported or wrong | Does not answer the question |
| Structure | Position in the first half minute; parts announced; ends inside the limit | Position clear by the middle; easy to follow | Position only at the end, or hard to follow | No position, or cut off at the limit |
| Delivery | Followed in one hearing; pace and stress mark the key points | Followed throughout; even pace, little stress | Parts need a replay: rushed, mumbled or read flat | Mostly cannot be followed |
| Visuals | Each slide makes one point the talk uses; readable | Slides support the talk; one crowded or unused | Slides repeat the script | No slides, or unrelated ones |
| Questions | Direct answer, with a reason from the talk | Direct answer, general reason | Answers another question | No answer |
Each line scores 4 to 1, twenty points in all; nothing handed in scores 0. The levels describe: none says "excellent" or "average", the evaluative wording Brookhart (2018) advises against. With sound alone there is no line for eye contact or gesture. Stop listening at the limit, so a talk that runs over loses its ending.
How long does marking take, and how do two markers moderate?
About nine minutes a student for a four-minute talk, and about seven hours for a class of thirty with a second marker. The minutes are our assumptions; put your own in.
| Step | Sum | Minutes |
|---|---|---|
| Calibration | 2 markers × 4 talks × 9, and 20 each to compare | 112 |
| First marking | 26 talks × 9, in EdProcto beside the rubric | 234 |
| Second listening | 9 talks × 7 | 63 |
| Total | 409 |
Nine minutes is five of listening, one on the slides, two and a half to score and write a note, and half to open and save; a second hearing, with no note, takes seven. The nine talks heard twice are six drawn at random before marking and an assumed three within a point of the pass mark.
Recording returns lessons, not marking time: thirty talks heard live, at seven minutes each with the changeover, fill 210 minutes.
Moderation between two markers:
- Calibrate on four talks. Both score them line by line on paper before either speaks. Two levels apart means the line's wording needs work.
- Mark the rest, keeping the sample's marks on paper, unsaved.
- Hear the sample again. The second marker scores the nine without seeing the first marks.
- Compare, save, release. Within one level on every line, the first mark stands; two levels apart, listen together and agree.
Marking alone, score five talks again a day later.
What feedback on a presentation can a student use?
A score for each line, one moment in the talk, and one change for next time. Van Ginkel et al. (2017) compared feedback sources for 144 first-year students: the teacher's did most for how students presented, and self-assessment by itself did least. Murillo-Zamorano and Montanero (2018) found that peer feedback with a rubric lifted presentations by 10%, against 5% for a teacher's comments straight after the talk, and that the gain had gone by the follow-up. One round is not enough.
- Line scores first. "Content 4, structure 3, delivery 2."
- One moment, by the clock. "From 3:05 you speed up."
- One change. "Cut one example and finish by 3:45."
- Their rating beside yours. The student scores each line before seeing the mark; the gap is the conversation.
- A second talk within a few weeks, so the feedback gets used.
A worked example: one task, two students, one rubric
The task and both students are invented for this guide.
Audience: the owner of a café that sells 300 hot drinks a day, 60% of them to take away. A disposable cup and lid costs the café 9p. Should the café charge 20p for a disposable cup? Four minutes, three slides at most. Give your recommendation in the first half minute, show one calculation from these figures, and name one risk with your remedy.
A full answer holds this sum: 60% of 300 is 180 cups a day. If a third of those customers bring a cup, the café saves 60 cups at 9p (£5.40) and takes 20p on the other 120 (£24): £29.40 a day better off, before any lost custom.
| Line | Student A | Student B |
|---|---|---|
| Content | 4: £29.40 a day, assumption stated | 2: charges all 300 drinks; a risk with no remedy |
| Structure | 3: recommendation arrives at 1:10 | 4: recommendation first, parts announced |
| Delivery | 2: from 3:05 the risk needs a replay | 4: followed in one hearing, figures stressed |
| Visuals | 3: clear slides, one never used | 2: the slides are the script |
| Questions | 3: direct, general reason | 2: answers another question |
| Total | 15 of 20 | 14 of 20 |
B gave the talk a listener enjoys more, and one impression mark could not show why A finishes ahead. Five lines can.
The note for A: "Content 4, structure 3, delivery 2. Your £29.40 is right and you said what it assumes. From 3:05 you speed up: cut one example and finish by 3:45." B's note points to the 180 take-away cups.
How does EdProcto build and mark a recorded presentation?
EdProcto is the best tool for a recorded presentation when lessons are too short to hear every student, because the slides, the talk, the question that follows and the student's own rating arrive as one timed attempt, each beside your rubric with its own points: the separation the studies above ask for.
- Building. Presentation is in the builder now. Type the task into "Prompt". "Model answer (what a full answer contains)" and "Rubric (what earns the points)" appear beside each recording when you mark. Set "Points" to 12 for content, structure and delivery, and to 4 each on the slides and the question that follows.
- Settings. Switch off "Shuffle the questions for each student" and "Allow going back to change an answer", so the question that follows stays unseen until the talk is recorded. Put "Minutes overall" at 30, room for about three takes, and "Show the score" at "When released". Leave "Full screen" and "Leaving the tab" off: the student has the slides open.
- Recording. The student presses "Start recording" and the browser asks for the microphone and nothing else: the recording is sound. "Stop recording" sends the take to EdProcto's server, kept with the attempt, and a player lets the student hear it; a take still running when the clock runs out is stopped and kept by itself. "Record again" makes a new take, and the newest one sent is the answer. A take can be up to 25 MB, so record a full-length one on a phone yourself first; we have not measured the minutes in each browser. The slides question takes "Up to five files, 25 MB each."
- Marking. At hand-in each recording and file shows "to grade" and counts as zero until you mark it. The grading page shows one submission: the files as links, a player for each recording, "Points" in half points, "Note for the student" and "Save my grade". A recording is marked by the person who hears it; "Ask Procto AI" sits on typed answers, from the Tutor plan. Then press "Release the scores" on the results page: each student sees the score and your notes by opening the quiz's link again on the device they used.
Free has ten quizzes a month, thirty students a quiz and the core templates. Tutor opens all 111 templates, the recorded ones among them, for 100 students a quiz; Educator takes 500 and adds the full analysis. Pricing has the plans. A recording is evidence of a prepared talk in a student's voice, not proof of who wrote the script.
Quick questions
How long should a recorded presentation be?
Three to five minutes for a class task: a working rule, not a research finding. Thirty talks of four minutes are two hours of listening.
Should students be allowed to re-record a presentation?
Yes, for a prepared talk: Menzel and Carrell (1994) tie rehearsal to better speeches. Cap the takes with the quiz's overall time limit.
Can peers mark recorded presentations?
For feedback on a rehearsal, yes. For the mark, average several: Magin and Helmore (2001) needed two to four peer ratings to match one teacher's.
Does EdProcto record video or sound for a presentation?
Sound. In EdProcto's Presentation shape the browser asks for the microphone, and the slides are handed in as a file in the same quiz.
References
- Braun, M. (2017). Comparative evaluation of online and in-class student team presentations. Journal of University Teaching and Learning Practice, 14(3), Article 3. https://doi.org/10.53761/1.14.3.3
- Brookhart, S. M. (2018). Appropriate criteria: Key to effective rubrics. Frontiers in Education, 3, Article 22. https://doi.org/10.3389/feduc.2018.00022
- Carlson, R. E., & Smith-Howell, D. (1995). Classroom public speaking assessment: Reliability and validity of selected evaluation instruments. Communication Education, 44(2), 87–97. https://doi.org/10.1080/03634529509379001
- Information Commissioner's Office. (n.d.). Principle (e): Storage limitation. Retrieved October 10, 2026, from https://ico.org.uk/for-organisations/uk-gdpr-guidance-and-resources/data-protection-principles/a-guide-to-the-data-protection-principles/storage-limitation/
- Kang, O., & Rubin, D. L. (2009). Reverse linguistic stereotyping: Measuring the effect of listener expectations on speech evaluation. Journal of Language and Social Psychology, 28(4), 441–456. https://doi.org/10.1177/0261927X09341950
- Langan, A. M., Shuker, D. M., Cullen, W. R., Penney, D., Preziosi, R. F., & Wheater, C. P. (2008). Relationships between student characteristics and self-, peer and tutor evaluations of oral presentations. Assessment & Evaluation in Higher Education, 33(2), 179–190. https://doi.org/10.1080/02602930701292498
- Magin, D., & Helmore, P. (2001). Peer and teacher assessments of oral presentation skills: How reliable are they? Studies in Higher Education, 26(3), 287–298. https://doi.org/10.1080/03075070120076264
- Menzel, K. E., & Carrell, L. J. (1994). The relationship between preparation and performance in public speaking. Communication Education, 43(1), 17–26. https://doi.org/10.1080/03634529409378958
- Murillo-Zamorano, L. R., & Montanero, M. (2018). Oral presentations in higher education: A comparison of the impact of peer and teacher feedback. Assessment & Evaluation in Higher Education, 43(1), 138–150. https://doi.org/10.1080/02602938.2017.1303032
- Ritchie, S. M. (2016). Self-assessment of video-recorded presentations: Does it improve skills? Active Learning in Higher Education, 17(3), 207–221. https://doi.org/10.1177/1469787416654807
- van Ginkel, S., Gulikers, J., Biemans, H., & Mulder, M. (2017). The impact of the feedback source on developing oral presentation competence. Studies in Higher Education, 42(9), 1671–1685. https://doi.org/10.1080/03075079.2015.1117064