Designing Fieldwork and Sampling Strategies

Planning evidence that answers the enquiry

A fieldwork enquiry needs more than a collection of measurements. The evidence must relate to the question, cover suitable locations and be recorded consistently. Planning these decisions before leaving the classroom makes it easier to compare results and judge how confidently they answer the enquiry.

For example, investigating how a river changes downstream requires measurements at several points along its course. Measuring repeatedly at one convenient site would describe that site, but would not show downstream change. The locations, sampling method and recording sheet therefore need to work together.

Designing a useful data collection sheet

A data collection sheet is a planned form for recording measurements, observations and contextual information. Start with the evidence needed to answer the question, then give each item a clearly labelled place on the sheet.

For a river investigation, the sheet could be organised as follows:

Put units in the column headings rather than relying on the recorder to remember them. Plan a consistent level of precision appropriate to the equipment. Extra decimal places do not make a measurement more accurate.

A sheet also needs enough space for the planned observations. Recording several river depths without their distances from the bank loses information about the channel's shape. Recording width, depth and measurement positions allows a cross-section to be drawn later.

The River Tillingbourne investigation illustrates these design decisions. Its recording sheet included units, space for qualitative site notes and repeated flow measurements that could be averaged. Improvements included consistent precision for depth readings, recording the intervals between them, collecting enough measurements to show the channel shape, and using photographs to help explain the results.

Test the sheet before the main visit. A short trial can reveal unclear headings, missing units or insufficient space. Check it again at each site before moving on, while omissions can still be corrected.

Selecting specific fieldwork locations

Choose locations because they help answer the enquiry, not simply because they are easy to reach. Consider the range of conditions that needs to be represented and whether sites can be compared using the same method.

A transect is a line across an area along which observations are made. It is useful when investigating change across space. In Sheffield, a litter investigation used systematic sampling along a transect from the train station into the city centre. This arrangement linked each survey site to its position along the route.

For a similar enquiry, sites at regular distances would provide evidence of how litter varies along the route. Surveying an equivalent area at each site would make the counts more comparable. However, one route would not necessarily represent the whole city: other streets might have different uses or conditions.

Safety and access can restrict site choice. Plan permitted routes, avoid unsafe river conditions and keep survey positions away from traffic. Follow the teacher's risk assessment and work within the agreed area. If a planned site cannot be used safely, record the change and its reason; do not silently substitute a more convenient location.

Sampling a larger population

The population is the whole set being investigated—for example, all residents of a town or all pebbles on a beach. A sample is the smaller set actually studied. Sampling saves time, but the aim is to capture the important variation in the population rather than an unusually narrow part of it.

Random sampling uses chance-based selection. For a questionnaire, number the members of a suitable list and use randomly generated numbers to select respondents. For sites, numbered locations can be selected in the same way. This reduces the investigator's personal influence on selection. Nevertheless, a small random sample may miss important groups or areas, and some selected locations may be inaccessible.

Systematic sampling selects at regular intervals—for example, every 10 metres along a transect or every fifth passing person. It is straightforward to organise and gives regular coverage, making it useful for investigating spatial change. Its limitation is that fixed intervals may miss small features or coincide with a repeating pattern. The interval therefore needs to suit the scale of the enquiry.

Opportunistic sampling selects whoever or whatever is available or convenient. A student might question people willing to stop in an accessible pedestrian area. This can be quick, safer and more practical, and approachable respondents may be less likely to refuse. However, it is likely to be less representative. Choosing people who look friendly, similar in age or unhurried can introduce unconscious bias. An apparently casual choice is not the same as random selection.

Stratified sampling first divides the population into relevant groups, then samples from each in proportion to its share of the population. For example, a residents' survey could include age groups in the same proportions as the town's population, with random selection within each group. This helps ensure that important groups are represented. It requires information about the population beforehand and takes more planning than simply interviewing available people.

The best choice depends on the enquiry. Systematic sampling suits change along a route; stratified sampling suits a population with important differences between groups. Opportunistic sampling may be necessary under practical constraints, but its limitations must remain part of the interpretation.

Justifying sample size and coverage

Sample size means the number of people, objects or sites studied. Decide it alongside the sampling method, considering the population's size and variety, the time available and how long each observation takes.

Five questionnaires from a supermarket workforce of 1,200 would provide a weak basis for describing the workforce as a whole. Likewise, measuring ten pebbles on a beach two kilometres long would give very limited coverage. More observations are useful only if they also cover the variation relevant to the question: collecting many pebbles from one small patch still leaves most of the beach unrepresented.

For a questionnaire, consider coverage of different groups, locations and times. A survey conducted only during working hours might exclude many people at work. If the enquiry instead concerns particular decision-makers or groups affected by a change, fewer, longer interviews may provide useful depth—but should not be presented as representing every resident.

A good justification explains the trade-off: the sample should be broad enough to support the conclusion while remaining manageable within the available time and safe working conditions.

Establishing a control group

A control group provides a comparison not exposed to the change being investigated. It helps distinguish the possible effect of that change from differences that might have occurred anyway.

Suppose an enquiry investigates whether installing new litter bins reduces street litter. The intervention sites would receive new bins; the control sites would not. Choose control streets that are otherwise as similar as possible—for example, in land use, pedestrian activity and existing cleaning arrangements. Record litter in both sets of streets before and after installation, using the same survey area, procedure and comparable times.

If litter falls in both sets of streets, another influence, such as changed cleaning arrangements, might explain part of the reduction. A greater fall at the intervention sites would support the suggestion that the bins helped, but would not prove it: unnoticed differences between streets could still affect the results.

A control group is different from controlled variables, which are factors kept consistent during measurement. Both strengthen a comparison, but neither guarantees that every other influence has been removed.

Recording and site selection

• Tailor the sheet to the enquiry: clear headings, units, site identifiers, measurement positions, repeats and contextual notes.
• Use consistent precision appropriate to the equipment.
• Select sites that capture relevant variation and allow comparable measurements; record safety or access changes.
• A transect helps investigate spatial change, but one route need not represent the whole area.

Choosing a sample