If you’re looking at a list of sampling methods and trying to spot the one that would not produce a representative sample, the answer is almost always the method that lacks random selection. Convenience sampling, voluntary response sampling, and judgmental (purposive) sampling all fail to produce representative samples because they let something other than chance determine who gets included. Methods built on random selection, such as simple random sampling, stratified random sampling, and systematic random sampling, are the ones designed to reflect a population accurately.
Understanding why certain methods fail comes down to one principle: every member of the target population needs a known, nonzero chance of being selected. When that requirement is violated, the sample drifts away from the population it’s supposed to represent.
What Makes a Sample Representative
A sample is representative when the results you get from studying it match, within a reasonable margin of error, what you’d find if you could study the entire population. That match depends on whether the key characteristics of the sample, things like age, income, health status, or opinions, are distributed the same way they are in the broader group. Random selection achieves this because it gives every person an equal (or at least calculable) probability of being chosen, which prevents any single characteristic from being systematically over- or underrepresented.
Probability sampling methods accomplish this in different ways. Simple random sampling is the most straightforward: every individual has the same chance of selection, like drawing names from a hat. Stratified sampling divides the population into subgroups first (by age bracket or region, for instance), then randomly samples within each subgroup. Cluster sampling randomly selects entire groups, such as schools or neighborhoods, and then studies everyone within those selected clusters. All of these qualify as representative because randomness is built into the selection process.
Methods That Fail to Be Representative
Non-probability sampling methods skip random selection entirely, and that’s where representativeness breaks down. Here are the most common ones you’ll encounter on a test or in practice:
- Convenience sampling: You pick whoever is easiest to reach. A researcher surveying people in a shopping mall on a Tuesday afternoon will miss anyone who works during the day, lives far from the mall, or avoids malls altogether. The findings can only be generalized to the narrow slice of people who happened to be accessible, not the broader population.
- Voluntary response sampling: You put out a survey and wait for people to respond on their own. The problem is that people with strong opinions, especially negative ones, are far more likely to participate. Volunteers also tend to differ from non-volunteers in predictable ways: they’re generally better educated, more sociable, and more approval-seeking. For studies involving stress or risk, volunteers skew toward higher sensation-seeking and risk-taking traits. The result is a sample that overrepresents extreme views and underrepresents the moderate middle.
- Judgmental (purposive) sampling: The researcher hand-picks participants or locations they believe are “typical.” For example, choosing one town thought to represent the whole country. This injects the researcher’s assumptions directly into the sample, and there’s no way to verify whether those assumptions are correct. The sample reflects the researcher’s judgment, not the population.
- Quota sampling: The researcher sets targets for certain subgroups (50 men, 50 women, for instance) but fills those quotas using non-random methods. The proportions might look right on the surface, but the individuals within each quota weren’t randomly chosen, so hidden biases persist.
- Snowball sampling: One participant recruits the next, who recruits the next. This creates a sample clustered around a particular social network, missing anyone outside that chain.
Why Randomness Is the Dividing Line
The mathematical reason is straightforward. When every person in a population has a calculable chance of being selected, you can also calculate the margin of error, the range within which your sample results likely reflect the true population values. Without random selection, that calculation is impossible. You have no way to quantify how far off your results might be, which means you can’t claim your findings apply beyond the specific people you happened to study.
This isn’t just a technicality. In probability sampling, the distributions of key characteristics in your sample will, on average, mirror the population. In non-probability sampling, certain types of people are systematically excluded. That exclusion is called undercoverage bias, and it distorts results in proportion to how different the excluded group is from the included one. If you survey college students about national voting preferences, for example, you’ve excluded everyone who isn’t a college student, and those groups may vote very differently.
How Bias Creeps Into Real Surveys
Even well-intentioned sampling can go wrong in two main ways beyond the initial selection method. Undercoverage bias happens when part of the population never makes it into the pool you’re sampling from. If your sampling frame is a phone directory, you’ve automatically excluded anyone without a listed number. Nonresponse bias happens when people who were selected simply don’t participate, and the people who skip the survey differ in meaningful ways from those who complete it. Both problems shrink the representativeness of your sample even if the original design was random.
One of the earliest and most famous failures happened in 1934, when a large purposive survey produced results so unreliable that statistician Jerzy Neyman used it as proof that future surveys needed to rely on random sampling. That failure helped establish the standard that persists today: if you want results you can generalize, randomness isn’t optional.
How to Spot the Right Answer
When a multiple-choice question asks “which of these would not produce a representative sample,” look for the method that removes randomness from the selection process. If the option describes selecting people based on who’s available (convenience), who chooses to respond (voluntary), or who the researcher thinks is typical (judgmental), that’s your answer. If the option involves random selection of any kind, it’s designed to be representative.
A quick way to test each option: ask yourself, “Does every person in the target population have a chance of being picked?” If the answer is no, the method won’t produce a representative sample. Convenience sampling pulls from whoever is nearby. Voluntary response sampling pulls from whoever feels motivated enough to participate. Both systematically exclude large portions of the population, making them the most common correct answers to this type of question.