Sampling

Sampling is the process of selecting a smaller group (a sample) to represent a larger population, so that the firm can draw conclusions about the whole population without surveying everyone.

One reason is cost. Surveying the entire population — which could be millions of customers — would be extremely expensive. By using a sample, the business gathers useful information at a much lower cost, freeing money for other activities. Provided the sample is representative, its results can still be treated as a reliable guide to the views of the whole population.

The first limitation is sampling bias. If the sample is not chosen properly — for example by surveying only people who happen to be passing (convenience sampling) — some groups are over- or under-represented. This skews the results, so the firm could draw the wrong conclusions and target the wrong customers.

The second limitation is sample size. If the sample is too small, the results may be due to chance rather than a real pattern in the population, making them unreliable. Increasing the size improves reliability, but larger samples cost more money and take more time.

Together, these mean a firm cannot fully trust sample results: they are a useful guide, but the way the sample is chosen and its size must be considered before relying on the conclusions.

The reliability of market research depends heavily on how the sample is chosen, because the firm uses the sample's results to make conclusions about the whole population.

A well-chosen sample improves reliability. Random or stratified sampling gives every group a fair chance of being included, so the sample is representative of the population — its make-up reflects the make-up of all customers. Findings from such a sample can be trusted as a guide to how the whole market would respond, leading to better decisions.

A poorly-chosen sample reduces reliability. Convenience sampling, for example, surveys only whoever is easy to reach, so certain groups are over-represented and others left out. This causes sampling bias: the results reflect a skewed slice of the market, not the whole, so the firm may design the wrong product or target the wrong customers. Sample size also matters — a too-small sample may give results driven by chance rather than a real pattern.

Therefore the way a sample is chosen is central to reliability. The most reliable research uses a representative method (random or stratified) with an adequate sample size — but these cost more, so firms face a trade-off between reliability and cost. A firm should choose a more representative method when the decision is important enough to justify the expense.

There are reasons the firm can place some trust in the results. Surveying 100 customers is better than no research at all: it gives real, first-hand feedback from actual customers at low cost and quickly, which can highlight obvious issues and broad preferences. If the shop's customers are typical of the firm's wider market, the results may give a reasonable guide to what customers think.

However, there are strong reasons for caution. The sample is a convenience sample — only customers leaving one shop were asked — so it is likely to be biased: it excludes people who shop online, at other branches, or who chose not to enter the shop at all, including potential customers the firm has failed to attract. The sample of 100 may also be too small to represent a large customer base reliably, meaning the results could be down to chance. As a result, the data may not reflect the views of the whole market and could lead to a wrong decision.

Judgement. How far the firm should trust the results depends on the importance of the decision and how typical the shop's customers are. For a minor, low-cost decision, the survey is a reasonable, affordable guide and can be trusted to a moderate degree. For a major, costly decision, the firm should not rely on it alone, because the convenience sampling and small size make it too unreliable; it should gather a larger, more representative sample (e.g. across several locations and online) before committing. On balance, the results are a useful starting point but should be treated as a rough guide, not a firm basis for an important decision — the firm should weight its trust according to what is at stake.

A sample is a smaller group chosen to represent a whole population, used because surveying everyone is too costly and slow. Businesses must rely on samples to make decisions, but how far they can trust the results depends on how the sample is chosen, its size, and the importance of the decision.

The case that a business can rely on a sample. A well-chosen sample is a practical and reliable substitute for surveying the whole population. Using random or stratified sampling, the sample's make-up reflects the population, so its findings can be treated as a sound guide to the whole market. This allows the firm to gather reliable information at a fraction of the cost and time of a full census, and to make timely decisions. For most everyday decisions, a representative sample of adequate size gives results a firm can act on with confidence.

The case that reliance should be limited. First, a biased sample (e.g. convenience sampling) over- or under-represents groups, so the results mislead — and bias cannot be cured simply by making the sample bigger. Second, a too-small sample may give results due to chance, not a real pattern. Third, even a representative sample becomes out of date if customer tastes change between the research and the decision. Fourth, respondents may say one thing but do another, so stated views overstate real behaviour. For these reasons, sample results are a guide, never a guarantee, and over-reliance can lead to costly mistakes.

Weighing it up (criterion). How far a firm can rely on a sample depends on the quality of the sample (method and size) and the stakes of the decision. A representative, adequately-sized, up-to-date sample can be relied on heavily, especially for routine decisions. A biased, small or dated sample should be relied on little, particularly for major, irreversible decisions where the cost of being wrong is high.

Judgement. A business can rely on a sample's results to a significant extent, but only when the sample is well-chosen and the decision is not too high-stakes. The most defensible conclusion is that reliance should be proportionate: trust a representative, adequately-sized sample as a strong guide for ordinary decisions, but for major decisions either invest in a larger, more representative sample or treat the results cautiously and combine them with other evidence. So a sample is a powerful but imperfect tool — reliable enough to base most decisions on, provided the firm checks how it was chosen and never treats its findings as a certainty.