Judgement Sample

A judgement sample (also called a purposive sample) is a non-probability sample in which the researcher selects which units to include based on their own judgment about who is relevant or representative, rather than selecting at random. Instead of giving every member of the population a known chance of inclusion (as probability sampling does), the researcher deliberately picks the sample — choosing the customers, respondents, or cases they judge most appropriate, informative, or representative for the study. Judgement sampling is one of several non-probability methods (alongside convenience and quota sampling). It can be efficient and useful when the researcher's expertise genuinely guides good selection, but because the selection depends on judgment rather than chance, it carries the biases of that judgment and does not support the statistical generalization that random sampling allows.

Judgement samples are used because they are practical and can be purposeful: when a researcher needs specific, hard-to-reach, or expert respondents, when probability sampling is infeasible, or when expert judgment about who is relevant adds value, judgement sampling lets the researcher target exactly the cases they want. For exploratory research, expert interviews, or studies of particular segments, a judgement sample can be more useful than a random one. But the method's defining feature — selection by judgment — is also its defining risk: the sample reflects the researcher's judgment, assumptions, and blind spots about who is relevant or representative, which may not match reality. So judgement samples trade the statistical defensibility of random sampling for the targeting and convenience of purposive selection, and that trade must be made knowingly.

The key contrast is with probability sampling and with representativeness. Probability (random) sampling gives every population member a known chance of inclusion, which supports representativeness and lets findings be generalized to the population with quantifiable confidence. A judgement sample does neither: because the researcher picks the sample by judgment, there is no random mechanism ensuring the sample mirrors the population, and no statistical basis for generalizing from it with known confidence. So a judgement sample is typically less representative than a well-drawn probability sample, and its findings cannot be generalized to the population in the rigorous, quantifiable way random samples allow. The representativeness of a judgement sample rests entirely on whether the researcher's judgment about who to include was actually right.

This is where the bias risk lives. Because a judgement sample is chosen by the researcher, it is prone to the selector's biases — conscious or unconscious assumptions about who is typical or relevant that skew the sample. The researcher may pick cases that confirm expectations, over-represent familiar or accessible groups, or misjudge who is representative, producing a sample that reflects judgment rather than reality. And unlike random sampling, there is no mechanism to correct for this — the bias is built into the selection. So a judgement sample relates to representativeness as a method that may or may not achieve it, with no guarantee: it can be representative if the judgment is excellent, but it is structurally prone to bias because it depends on judgment. This distinguishes it sharply from random sampling, which builds representativeness into the method rather than relying on the researcher to achieve it.

Using a judgement sample appropriately means recognizing it as a non-probability method that is useful for specific, purposeful, or exploratory research but not for generalizing to a population with statistical confidence. It means using judgement sampling where its targeting is genuinely valuable (reaching specific or expert respondents, exploratory work, cases probability sampling cannot reach), applying the best possible judgment to selection while staying alert to one's own biases, and — crucially — being honest about the method's limits: not presenting judgement-sample findings as if they were representative, generalizable population estimates. Good practice uses judgement samples for what they are good for and reaches for probability sampling when representativeness and generalization matter.

The failures are using a judgement sample where representativeness is needed (then generalizing biased, non-representative findings to a population as if they were rigorous estimates), letting the researcher's biases skew selection unchecked, and presenting purposive-sample results with unwarranted statistical confidence. A team that hand-picks 'typical' customers to interview and then reports their views as the voice of the whole market has misused a judgement sample. The discipline is to use judgement sampling deliberately and honestly — for the targeted, exploratory, or expert purposes it serves, with awareness of its bias risk, and without overclaiming representativeness or generalizability — recognizing it as a convenient, purposeful, but bias-prone alternative to the random sampling that representativeness and statistical generalization actually require.

A judgement sample — chosen by the researcher's judgment rather than at random — is a convenient, purposive but bias-prone method, useful for targeted or exploratory work but no substitute for random sampling.

Etymology: source.

Search interest for this term over the last five years:

View interest-over-time on Google Trends →

Areas of marketing where judgement sample is a core concern: