How Large Should a Sample Be?
A finding rather than a topic. This page marks exactly where the subject's guidance on how much data is enough runs out, separates the teaching material from the two external documents bound in behind it, and then shows you how to arrive at a number and write it up with nothing from the material at your back.
A question on its own slide, and nothing after it
The slide that opens the sampling sequence is titled "Selecting the sample". It poses two questions, one after the other, before anything has been explained. The second is the one this page is about.
Both questions are abandoned. The representativeness question is picked up glancingly on later slides and contradicted there — the taxonomy page, sampling methods, records where. The size question is not picked up at all.
Those are counts of the supplied material, made by audit; the source states none of them. They are worth having in front of you because the absence is easy to misread as your own oversight. It is not. A reader who works through the whole week looking for a sizing rule will not find one, and the week never says that it is not going to supply one.
What the material does supply that bears on size
Three statements in the teaching material touch the question without answering it. Each is worth knowing, because together they are what you can legitimately attribute to the subject when you write your method up.
EVERY STATEMENT IN THE TEACHING MATERIAL THAT BEARS ON SAMPLE SIZE
| Statement, as given | Where it appears | What it settles |
|---|---|---|
| "Populations are usually too big or items too inaccessible for the whole population to be examined. One may have to be satisfied with examining only a part, or sample, of the total population." | The slide introducing sampling | Why you sample at all. Not how many — the whole of its sizing language is "usually too big", "too inaccessible" and "only a part" |
| "you have many objects that you could collect data on but you only have the resources to collect data on a few of them" | The same slide | That resources bound the number. "A few" is not quantified, and resources are never decomposed into time, money, access or skill |
| "Higher costs to examine the whole population than a small sample and could easily exceed the value of the survey results." | The slide headed "Sampling selection limits", under a sub-heading "Cost/size" | That cost, not precision, is the operative constraint. The sub-heading names size and the body never returns to it — see whole population or sample |
| "Census: The process of collecting information about every member of a population" | The week's key terms slide | The name of the alternative to sampling. The word is defined once and never used again, including on the slide that argues against doing one |
None of these four statements carries a number. The subject's entire treatment of how much data is enough is qualitative.
Where the only numbers in the dataset come from
There are concrete sample-size figures in the supplied material. They are not in the teaching material. Understanding that distinction is the whole of this section, because everything that goes wrong with sample size in this subject goes wrong at exactly that seam.
The week's study notes run to five numbered sections. Not one sentence of them is written in the subject's own voice: the first two sections are lifted from a national research-data service guide, and the last three from a published web-journal article that the week sets as required reading. That article is a paper on quantitative research design in sport and exercise science. Its fifth section, headed "5. Sample size", is reproduced in the notes word for word.
THREE KINDS OF SOURCE, AND WHAT EACH CAN BE USED FOR
| Source | Register | What it can support |
|---|---|---|
| The week's slide deck | The subject's own voice — teaching material | Definitions of population and sample, six sampling type names, one named bias, a cost argument. No size rule of any kind |
| The week's study notes | Two external documents reproduced verbatim, with no running attribution | What the external documents say. Never what the subject teaches — the notes contain no sentence of the subject's own |
| The set reading | One author's published article, from 2000, in sport and exercise science | Its own discipline's design conventions, attributed and bounded. It is set as reading; it is not a standard anybody must meet |
The article's own examples are athletes, physiological measures and controlled trials, and its stated subjects are tissues, cells, animals or humans.
There is one more wrinkle for anyone going looking. The slide that sets the reading names it by the opposite of its title — the deck calls it a qualitative research design article, and the article at the address given is titled "Quantitative Research Design". A student following the words rather than the link would search for the wrong paper.
What the reproduction leaves out
The borrowed sample-size section was not reproduced whole. Five passages were deleted, and the deletions are more consequential than the reproduction, because they remove the scaffolding that the retained text stands on.
- The section opens by promising three approaches to sizing a sample — via statistical significance, via confidence intervals, and "on the fly". The subsection on confidence intervals is deleted. A list of three is followed by two headings.
- That deleted subsection is the only place in anything supplied to this library where a confidence interval is defined. The subsection retained immediately after it uses the term repeatedly and relies on it completely.
- A subsection on how the validity and reliability of your measures change the sample size you need is deleted in full — the one place the article connects measurement quality to sample size.
- The subsection on pilot studies is deleted in full. It is the one passage in the article addressed to a student researcher without resources, and the nearest thing in the dataset to the pilot stage that the questionnaire material names and never describes.
- The retained text states a significance threshold and a detection convention, both inside quoted material, and defines neither. The statistical term for the second of those conventions never appears, so a reader cannot look it up from what is given.
Defending a number the source will not give you
You still have to put a number in your method chapter, and your examiner will still ask how you arrived at it. What follows is this library's synthesis rather than the source's instruction — the source prescribes no procedure — but it is built so that the gap is stated rather than concealed, which is the position you want to be defending.
Arriving at a defensible number
Decide what the number has to do
A sample assembled to describe a situation, to compare two groups, or to support an inference about a wider population carries different obligations. The subject never makes this distinction, and it is the one that governs everything after it.
Take the rule from a text you have actually read
Whatever sizing approach you use, it comes from a methods text, a supervisor's instruction or a disciplinary convention. Name that source in your write-up. Do not attribute it to this teaching material, which supplies none.
Record the constraint the source does name
Resources bound the number, and the material says so. If your size was set by access, budget or the time available, say so plainly. That is an honest answer and it is the one constraint the subject actually names.
Report the achieved number separately from the planned one
How many you approached and how many responded are two different figures, and the difference is where non-response bias lives. Report both.
State what the size does not license
Without a sampling frame and a stated precision, a claim about the wider population is an argument you are making, not a property of your method. Write the limitation in yourself; nothing in the source will write it for you.
If you are asked what the subject says
Answering accurately under questioning
The boundary, and where it sits against the objectives
This week closed a large part of the library's sampling gap: six named types, three usable distinctions, one named bias with a sound causal chain and one genuine cost argument. Sample size is one of the four things it did not close, alongside a procedure for executing any sampling type, any mention of a sampling frame, and any criterion for choosing between the types.
Sampling is named in the later data collection week's only stated learning objective, where it was taught in a single sentence recommending random selection and immediately withdrawing it as usually impossible. Across the six weeks of teaching material now on record, fourteen learning objectives are stated: none fully delivered, four partially delivered, ten not delivered at all. The reconciled audit is at the six-week objective audit. Treat the absence of a sizing rule as a documented feature of this material rather than as something you have failed to find.
What to carry forward
- The question is posed in the subject's own words and answered nowhere in the subject's own voice.
- The material's only size-related content is qualitative: populations are too big, items too inaccessible, resources too few, and a census usually costs more than it is worth.
- Every concrete figure in the dataset belongs to a sport and exercise science article reproduced in the study notes, and none of them is a project management standard.
- The borrowed answer arrives with its own foundations deleted, including the only definition of a confidence interval anywhere in the supplied material.
- Source your sizing rule elsewhere, cite it where you got it, report planned and achieved numbers separately, and state the limitation yourself.
Frequently asked questions
Does the supplied teaching material state a minimum sample size?
No. It states no minimum, no proportion, no rule of thumb and no worked calculation. The question is asked on the slide that opens sampling and is not answered there or anywhere later, and the phrase "sample size" does not appear in the week's slides at all.
There are numbers in the week's study notes. Can I use them?
You can cite them as what they are — figures from a published article on quantitative design in sport and exercise science, reproduced verbatim in the notes. You cannot present them as the subject's guidance or as a standard for a project management study. They are set out with their attribution and their limits on the sample size in quantitative studies page.
So how do I decide how many people to survey or interview?
Take the rule from a methods text you have read and cite it there, or state the resource constraint that set the number, which is the one constraint the material itself names. Then report how many you approached and how many responded as separate figures, and state what your size does not allow you to claim.
Is ten a reasonable sample, given the example on the slide?
The slide's mention of ten supplier companies is a scenario figure inside an illustration of purposive selection, not a size recommendation. Nothing in the source says ten is enough, appropriate or typical, and it happens to sit on the same slide as the size question purely by layout.
Why does this page not just print the figures from the set reading?
Because putting them on a page about what the subject says would make one discipline's conventions read as the subject's answer, and the subject has none. The figures live on the page dedicated to that article, where the attribution, the discipline and the non-transferability are attached to them.
What should I write in my limitations section about sample size?
State how the number was arrived at, name the source of any rule you applied, and say what the size prevents you from claiming — particularly any generalisation to a wider population, since the material supplies no sampling frame and no precision statement. Writing the limitation yourself is the only option the source leaves you.
References and source attribution
- The supplied teaching source: the quantitative research week's 23-slide deck, principally the slide titled "Selecting the sample" and the slide titled "Sampling selection limits"; and the same week's five-section study notes, no sentence of which is in the subject's own voice.
- Hopkins 2000, 'Quantitative Research Design', a sport and exercise science web journal, vol. 4, no. 1 — the article set as the week's required reading and the source of the study notes' sample-size section. Cited to identify the register of that section; its figures are its own discipline's.
- Bryman, A. 2016, Social Research Methods, 5th ed., Oxford University Press, Oxford. Standing reference for the sampling and sizing vocabulary this material uses without defining.
- O'Leary, Z. 2017, The Essential Guide to Doing Your Research Project, 3rd ed., Sage Publications, London.
- Veal, A. J. 2005, Business Research Methods: A Managerial Approach, Longman.
Suggested questions for Ask KEVOS
- What exactly does the supplied material say about how many participants I need?
- Show me the difference between the teaching material's voice and the reproduced article in the study notes.
- How should I word a sample-size justification when my source material gives me no rule?
- What was deleted from the sample-size section reproduced in the study notes, and why does it matter?
- Which numbers in this dataset are safe to quote, and which belong to another discipline?