Sampling Strategies — The Core Skill

🎯 Sampling Strategies — The Core Skill

It is almost never possible to measure everything in a study area. A beach might contain millions of pebbles. A high street might receive 10,000 visitors on a Saturday. A river channel might be 5 km long. Sampling is the strategy you use to select a manageable number of data points that are still representative of the whole.

Choosing the right sampling strategy is a heavily tested exam skill because it directly affects how reliable and valid your conclusions can be. A biased sample produces biased conclusions, no matter how carefully you measure.

Strategy	How It Works	Best Used When…	Geography Example	Advantage	Weakness
Random	Sample locations or individuals are selected using a random number generator or random number table — no human judgment in selection	The study area is broadly uniform and you want to remove all researcher bias from the selection process	Measuring pebble sizes on a beach: assign grid references to the beach, use random numbers to select 30 grid squares, measure the pebble nearest the grid centre	Eliminates researcher bias; statistically valid; every point has an equal chance of selection	May miss spatial patterns by chance; some zones may be over- or under-represented; impractical in unsafe or inaccessible terrain
Systematic	Data is collected at fixed, regular intervals — every X metres along a transect, every Nth person, or at evenly-spaced time intervals	You expect a gradient or progressive change and want data that reveals the pattern across the whole study area	Measuring river velocity at sites every 100 m downstream for 1 km; recording environmental quality every 200 m along an urban transect from city centre to suburb	Simple to execute; guarantees full spatial coverage; good for detecting trends over distance; repeatable by other investigators	If the interval coincidentally aligns with a repeating feature (e.g., every road junction), it may over-represent that feature; misses features that happen between measurement points
Stratified	The population or study area is divided into sub-groups (strata), and samples are taken from each sub-group in proportion to its size or importance	The population or area has distinct sub-groups that must all be represented — e.g., different land use zones, or different age groups of people	Questionnaire on a high street: if 50% of visitors are shoppers, 30% are workers, and 20% are residents, sample those proportions from each group. River study: if the channel is 40% riffles and 60% pools, measure velocity in proportion to that ratio	Ensures all sub-groups are proportionally represented; reduces the risk of one group dominating the data; more representative overall	Requires prior knowledge of sub-group proportions, which may not be available; more complex and time-consuming to organise in the field
Opportunistic (convenience)	Sample whoever or whatever is conveniently available — the people who approach you, or the data points easiest to access	Quick reconnaissance; early-stage qualitative research; when other methods are genuinely impossible	Stopping people on the street who seem willing to answer; measuring only the pebbles you can reach easily at the water's edge	Quick and easy; useful for pilot studies; avoids safety risks from accessing difficult terrain	Highly biased — the sample represents whoever/whatever is most accessible, not the whole population. Not acceptable as the primary sampling method for a formal investigation

Why Sampling Strategy Affects Your Conclusions

Consider two groups studying pedestrian flows on the same high street. Group A uses systematic sampling — they count pedestrians at every 100 m interval from one end to the other. Group B uses opportunistic sampling — they stand outside a coffee shop because it was convenient. Group B's data shows high footfall throughout the morning. Group A's data reveals that footfall drops sharply beyond 400 m from the main bus stop. Only Group A's sampling strategy was capable of detecting the pattern the hypothesis predicted.