Sampling Methods: The Six Named Types
Four slides in the quantitative research week carry the whole of this subject's sampling taxonomy. This page gives you every word of it, the two distinctions worth keeping, and the exact line at which naming a method stops being the same thing as knowing how to draw one.
Four slides, six names, and where they sit in the week
Sampling enters the quantitative research week on a slide titled "Selecting the sample" and is finished four slides later. Three of those slides are titled "Types of sampling (1)", "(2)" and "(3)", and between them they name six types and one residual category. Everything the supplied teaching source says about choosing the units you collect data from sits in those four slides: the population and sample definitions come earlier, and the week turns to data custody straight afterwards.
That shape matters. This is not a chapter on sampling with worked procedures; it is a vocabulary list delivered as one-line glosses, and the difference is the difference between naming what you did and being able to do it.
Two questions are posed there and neither is answered on that slide or on any slide after it. The word "First" promises a sequence whose second step never arrives, and "consider the purpose for collecting the data" is neither a procedure for representativeness nor a size rule. The size question is taken up on its own at how large should a sample be, where the finding is that the material poses it and stops.
The example is the deck's only worked illustration of any sampling type, and it does not work: it states a decision to select customers and then executes the opposite one by choosing managers of supplier companies. The ten in it is a scenario figure — an illustrative value inside an example, not a recommended size, a minimum or a threshold. Nothing in the source says ten is enough, appropriate or typical.
The six definitions, exactly as the source gives them
Every gloss in the table below is quoted from the slides. Where a type gets three lines, three lines is the entirety of its treatment in the subject.
THE SIX NAMED TYPES AND THE RESIDUAL CATEGORY, AS THE SUPPLIED TEACHING SOURCE DEFINES THEM
| Type | Definition as given | What that actually supplies |
|---|---|---|
| Purposive | "selecting particular respondents because of certain attributes or based on their particular ability to contribute to your purpose" | A selection criterion — attributes, or ability to contribute |
| Random | "chosen as the result of some random or chance process"; "generally all the objects in the population have an equal chance of being chosen"; "avoids biases that enter when trying to select a representative sample" | A mechanism, a property and a claimed benefit |
| Simple random | "a random operation based on a chance process of some kind" | A mechanism — and nothing that distinguishes it from "Random" |
| Stratified | "useful technique when you know something about the population of objects you are sampling from"; "helps to divide the population into groups or strata of like-objects you can study separately and re-assemble back into an estimate for the population as a whole" | A precondition, and a two-step idea: divide, then re-assemble |
| Systematic | "taking an observation at regular intervals e.g. every 5th person. This is best if the list from which you are sampling is in random order but can cause problems when there is some regularity or pattern in the list" | A mechanism, a precondition and a failure mode |
| Cluster | "used when your population falls into clusters or groups"; "different from stratified sampling because the clusters are naturally occurring and may not necessarily be made up of like individuals" | A precondition, and an explicit contrast with stratified |
| Non-random (residual) | "includes quota, judgement sampling, self-selected sampling"; "includes methods that obtain representative population samples" | Three sub-labels — quota, judgement, self-selected — plus a claim discussed below |
Six types and one residual category are counted from the slides; the source states no count. "Every 5th person" is the source's own example of an interval, not a rule for choosing one.
The two distinctions worth keeping
Two things in this taxonomy are better than the rest of the week and should be used rather than apologised for.
Stratified — groups you construct
- Precondition stated: you already know something about the population you are sampling from.
- Sequence sketched in one sentence: divide the population into strata of like objects, study them separately, re-assemble into an estimate for the population as a whole.
- The defining property is likeness — a stratum is a group of like objects.
- The only justification the slide gives is "identifying and studying different needs", which is an analytic goal rather than a sampling rationale.
Cluster — groups that are already there
- Precondition stated: your population falls into clusters or groups.
- The discriminator is given explicitly: clusters are naturally occurring and "may not necessarily be made up of like individuals".
- Nothing is said about how many clusters to take, or how to select within one.
- No example of either a stratum or a cluster appears anywhere in the week.
That pair is the best-constructed distinction in the week. Each type gets a precondition and the discriminator sits in a single sentence, which is more than the subject's other method catalogues manage — compare the method catalogue, where ten techniques are named and none is defined. If you can say which of the two you used and why, you are working at the top of what this material supports.
Systematic sampling is the second one worth keeping, because it is the only entry that supplies a mechanism, a precondition and a failure mode together: take an observation at regular intervals; this works best when the list is already in random order; it causes problems when the list has a pattern in it. That failure mode is real and it is the kind of thing that catches a project researcher sampling from a scheduling roster, a risk register or an issue log, where the order of the list is rarely accidental.
Three places the definitions do not hold together
These are defects in the source. They are reported here rather than tidied, because tidying them would mean deciding what the source meant — and on at least one of the three, the repair would change the claim being made.
The three internal contradictions
Random and simple random are not distinguished
Random is "chosen as the result of some random or chance process". Simple random is "a random operation based on a chance process of some kind". The second definition uses the word being defined and adds nothing to the first. On the source's own wording the two entries are the same thing presented as two, and a reader is left with a taxonomy in which one item is a synonym of another.
Purposive is filed opposite random and belongs under non-random
The first types slide pairs purposive with random as its two entries. The third then creates a non-random category and populates it with quota, judgement and self-selected sampling. Selecting respondents for their attributes or their ability to contribute is by construction not a chance process, so the deck's own three-slide sequence puts the same method on both sides of the random and non-random line without noticing.
The non-random line asserts the opposite of the slides either side of it
As given: "Non-random … includes methods that obtain representative population samples." The slide before it says random selection avoids the biases that enter when trying to select a representative sample. The slide after it makes self-selection — one of the three non-random methods just named — the mechanism of the week's only named bias.
The residual category and its three bare labels
Quota, judgement and self-selected sampling are named on the third types slide and that is the whole of their treatment. No definition, no example, no procedure, no condition of use. A reader who wants to know what quota sampling is must get it elsewhere, and the source does not say where.
Two further non-probability methods that most methods texts carry — snowball and convenience sampling — appear nowhere in the supplied material at all. The residual category names three methods and does not say that its list is partial: a set of labels presented as though it were a treatment.
What this taxonomy does not give you
Sampling was one of the two largest gaps on this library's record, and this week closes part of it. Being exact about the boundary matters here, because what remains open is the operational half.
THE SAMPLING GAP: WHAT CLOSED AND WHAT DID NOT
| What a reader needs | Position in the supplied material | Closed? |
|---|---|---|
| A vocabulary of methods | Six types named and glossed in one to three lines each, plus a residual category holding three more labels | Substantially |
| A sampling frame | Never named, in this week or anywhere else in the supplied material. The one slide that depends on it calls it "the list from which you are sampling" | No |
| A sample size rule | No rule in the teaching material's own voice. The phrase "sample size" does not occur in the week's slides at all | No |
| A procedure for any of the six | None. No randomisation mechanism, no rule for defining strata or for allocating a sample across them, no rule for choosing clusters or how many, no rule for the systematic interval or its start point | No |
| A criterion for choosing between the types | None. Three of the six carry a precondition and three carry none; no two types are ever compared on cost, precision, bias, effort or data quality | No |
The counts in this table are counts of the supplied material, derived by audit. None is a research finding and none is a standard.
Working with a taxonomy you cannot execute
The position is an unusual one: names that are correct as far as they go, attached to procedures you will have to source elsewhere. The checklist below is this library's synthesis rather than the source's instruction — the source gives no method-selection guidance at all — but every item is anchored to something the slides do say.
Before you write the sampling paragraph of your report
- Name the type you actually used, in the source's vocabulary, and say which of the six it is.
- State the precondition you relied on: what you already knew about the population (stratified), what natural groupings existed (cluster), or what list you drew from (systematic).
- Say where your procedure came from. If the randomisation method, the strata allocation or the interval rule came out of a methods text, cite that text — not this material, which supplies none of them.
- Describe the list you sampled from, since the source gives you no term for it: what it contained, who was on it, who could not be, and how current it was.
- If your selection was purposive, say so and give the attribute or the contribution criterion you selected on. Never describe purposive selection as random.
- Record what you cannot claim. Without a frame and a size justification, generalising from your sample to a wider population is an argument you have to make, not one your method makes for you.
Where this sits against the week's stated objectives
Sampling is named in none of the three learning objectives this week states — those are about interpreting concepts and measures, perceiving reliability and validity, and evaluating external and internal sources. It is named in the later data collection week's only stated objective, and it was taught there in a single sentence that recommended random selection and immediately withdrew it as usually impossible. The week that actually teaches sampling does so under objectives that do not mention it.
Across the six weeks of teaching material now on record the subject states fourteen learning objectives. None is fully delivered, four are partially delivered and ten are not delivered at all. That reconciled audit is at the six-week objective audit, and the consolidated account of what is missing is at what this material does not teach. The practical consequence is on this page: use the names, source the procedures, and say which is which.
What to carry forward
- Six types are named — purposive, random, simple random, stratified, systematic, cluster — plus a non-random residual holding quota, judgement and self-selected sampling as bare labels.
- Strata are constructed groups of like objects; clusters are naturally occurring and their members need not be alike. That single sentence is the most useful thing in the taxonomy.
- Systematic sampling turns on list order, and the source never names the list — sampling frame is absent from the entire dataset.
- Report the three internal defects rather than repairing them, especially the non-random line that appears to have lost a negation.
- The material lets you name a method and defend a distinction. It does not let you execute one, size one, or justify choosing it over another — take those from a text you cite yourself.
Frequently asked questions
What is the difference between random and simple random sampling in this material?
There is none that the source states. Random is defined as "chosen as the result of some random or chance process" and simple random as "a random operation based on a chance process of some kind", which uses the word being defined and adds nothing. The two are presented as separate entries in the taxonomy and are not distinguished by their definitions, so report them as the source leaves them rather than inventing a difference.
Does the source tell me how to draw a stratified sample?
No. It states a precondition — that you know something about the population — and a two-step idea of dividing into strata of like objects and re-assembling into a population estimate. It gives no rule for defining the strata, no rule for allocating your sample across them, and no worked example. The procedure has to come from a methods text you cite yourself.
Is purposive sampling random or non-random here?
The source puts it on both sides. It is paired with random on one slide, while the non-random category two slides later holds quota, judgement and self-selected sampling. Selecting respondents for their attributes or their ability to contribute is not a chance process, so the filing is internally inconsistent; this page reports the inconsistency rather than resolving it.
The non-random slide says non-random methods obtain representative samples. Is that right?
It contradicts the slide before it and the slide after it. The preceding slide credits random selection with avoiding the biases that arise when trying to select a representative sample, and the following slide makes self-selection the mechanism of the week's only named bias. The sentence reads as though a negation has been lost, but this library quotes it as it stands rather than repairing it.
How do I choose between the six types for my project?
The material gives you no criterion. Three of the six carry a precondition you can test — do you know something about the population, does it fall into natural clusters, do you have a list — and no two types are ever compared on cost, precision, bias, effort or data quality. Use the preconditions to rule types in or out, then take the choice rule itself from a methods text and cite it there.
What is a sampling frame, and where does the source define it?
It does not. The term appears nowhere in the supplied material, in any week. The systematic sampling entry depends on the idea and calls it "the list from which you are sampling" without naming it, defining it, or saying where such a list comes from and how you check that it is complete.
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 three slides titled "Types of sampling", which carry the whole of the subject's sampling taxonomy; and the same week's study notes, whose sampling vocabulary differs from the deck's.
- 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' sampling passages. Cited here only to mark the register of those passages; its figures belong to its own discipline.
- Bryman, A. 2016, Social Research Methods, 5th ed., Oxford University Press, Oxford. Standing reference for the sampling terms this material names without defining.
- O'Leary, Z. 2017, The Essential Guide to Doing Your Research Project, 3rd ed., Sage Publications, London.
- Trochim, W. M. K. 2006, Research Methods Knowledge Base, http://www.socialresearchmethods.net/kb/probform.php — cited in the supplied source's own reference list.
- Veal, A. J. 2005, Business Research Methods: A Managerial Approach, Longman.
Suggested questions for Ask KEVOS
- Show me the six sampling types with the source's exact definitions side by side.
- Which sampling distinctions in this material are safe to rely on, and which are defective?
- What would I have to add to this material before I could actually draw a stratified sample?
- How should I describe a purposive sample in a research report without overstating it?
- Where does the supplied material discuss the list you sample from, and what does it call it?
- Summarise every gap in this subject's sampling coverage in one list.