Planning Fieldwork Data Collection

Plan evidence that answers the enquiry

Fieldwork planning connects a geographical question to evidence that can answer it. Before choosing a site or preparing a recording sheet, decide what you need to compare or measure. For example, pedestrian counts could help investigate whether a regenerated shopping area attracts more people. Counting people somewhere convenient, without considering whether it represents that area, would be much less useful.

A workable plan specifies what will be recorded, where, when and how much. These choices affect whether the evidence supports a convincing conclusion.

Choose suitable locations and times

Select specific sites that provide relevant information and capture the contrasts your enquiry needs. An investigation of spatial change needs locations across the area, rather than several observations in one small corner. An investigation comparing different environments needs sites from each environment.

Sites must also be safe and accessible. Check permission, access routes and conditions before collecting data. A beach may be inaccessible at high tide; a river may be too deep to approach safely; a tourist survey outside the tourist season may find very few respondents. These are not just practical inconveniences: they can prevent the planned evidence from being collected.

Precautions should match the hazard. For pedestrian counts, choose a position away from traffic that does not obstruct the pavement. For physical fieldwork, follow the agreed safe boundaries and do not enter unsafe water or approach unstable ground to reach a planned sample point. If access changes, record any change to the plan rather than silently replacing a site with a more convenient one.

Timing matters because geographical conditions change. A city-centre count at lunchtime may differ from one early in the morning. Choose times that suit the question, and use comparable times when comparing places. If the enquiry concerns change through the day, sample at several times rather than treating one count as typical of the whole day.

Build a useful data collection sheet

A data collection sheet should make recording quick and consistent while preserving the information needed for later comparisons. Use a clear title, labelled columns, units in headings and enough space for every planned observation. Include site identifiers, dates and times so results can be matched to their locations and conditions.

Suppose you are comparing pedestrian flows at several shopping streets. A recording sheet could use this structure:

Define the counting rule before starting: for example, count each passage across an agreed line in either direction. This measures passages, not necessarily different individuals, because someone may return. All groups should use the same rule and duration so their totals can be compared.

For a measurement sheet, give each variable its own column and state its unit, such as width in metres. Provide separate spaces for repeated measurements rather than replacing the first reading with a later one. For categories or scores, define the categories or scale so different observers interpret them consistently.

Try the sheet before fieldwork. A short trial can reveal missing columns, unclear instructions or insufficient space. Recording is complete when every planned observation has an identifiable entry, with any missing result or departure from the method explained. A blank should not be mistaken for a measured zero.

Establish a fair comparison using a control group

A control group provides a baseline against which another group or set of sites can be compared. In geographical fieldwork, this may be a comparison area rather than a laboratory group.

Suppose you want to investigate the effect of regeneration on pedestrian flow. You could compare regenerated streets with similar streets that have not been regenerated. The non-regenerated streets act as the control. Choose streets with comparable functions and collect counts using the same rules, durations and times.

The factor being compared is regeneration status. Other conditions, such as count duration, are control variables kept consistent. The control group does not make the streets identical: differences in transport access or shop types could still influence pedestrian flow. A good plan therefore identifies remaining differences and avoids assuming that regeneration alone caused any contrast.

Choose a sample that reflects the population

The population is everything your investigation aims to draw conclusions about: perhaps all pedestrians using an area, all pebbles on a beach or all possible measurement locations in a field. Usually there is not enough time to investigate the whole population, so you select a sample.

A representative sample reflects important variations within that population. Its quality depends on both the number of observations and how they are selected. A large sample from one convenient location can still overlook most of the area.

Random sampling: selection by chance

Random sampling gives each member of the population an equal chance of selection. Number the eligible people, objects or locations and use random numbers to select them. For an area, randomly generated coordinate pairs can identify sampling positions within its boundaries.

This reduces selection bias because the investigator does not choose locations that look especially interesting or easy to measure. However, chance can produce clusters or leave some parts poorly covered, particularly with a small sample. Reaching scattered points may also take time. Random sampling is useful when you want to avoid choosing observations according to personal preference, but it does not guarantee a representative result on every occasion.

Systematic sampling: selection at regular intervals

Systematic sampling uses a fixed interval. Examples include taking a river-depth reading every 50 centimetres across a channel or selecting every tenth person passing a survey point. Regular spacing is straightforward to organise and helps reveal change along a route or across an area.

However, the interval can miss important features. Consider 100 possible positions numbered across a grid. Selecting positions 10, 20, 30 and so on samples every tenth position, but it also selects only the last column of this particular grid.

A small patch at positions 65–67 would be missed completely. When judging systematic sampling, examine both the interval and the spatial arrangement of the selected points. A regular numerical sequence does not necessarily provide good geographical coverage.

Systematic sampling can also operate through time. Measuring wind speed every 10 minutes in a sand-dune ecosystem is systematic. So is counting city-centre pedestrians for five minutes every hour. Choose an interval short enough to capture the variation relevant to the enquiry, but manageable within the available time.

Opportunistic sampling: selection when available

Opportunistic sampling uses people or observations that are readily available. For example, an interviewer might approach available pedestrians who are willing to answer questions, rather than selecting them using random numbers or a fixed interval.

It is quick and practical when time or access is limited. If a planned sampling site cannot be accessed safely, an accessible site nearby may allow data collection to continue. Record the replacement and the reason for it, because changing the site can change what the sample represents.

However, opportunistic sampling can over-represent convenient places or people who are easy to approach, while omitting those who are busy, absent or unwilling to participate. Its results should not automatically be treated as representative of all users of the area. Justify it by explaining the practical constraint, while recognising how more planned coverage of sites, groups or times could improve the sample.

Stratified sampling: include relevant groups

Stratified sampling divides the population into relevant groups and samples each one. These groups might be age categories in a questionnaire or different land-use areas in a town.

This ensures that important groups are included. If the aim is to represent the population as a whole, sampling in proportion to each group's share can be appropriate. If the aim is to compare groups, equal numbers from each group can make the comparison more balanced, but those numbers may not reflect their proportions in the whole population.

Selection within each group can be random or systematic. The main limitations are that you need information about the groups beforehand and must avoid leaving out relevant groups through poorly chosen categories.

Justify the sample size and the final plan

There is no single sample size suitable for every investigation. More observations generally provide a better opportunity to capture variation and reduce the influence of an unusual individual result. However, collecting more data takes time, and rushed recording can introduce errors.

Base your choice on the variation expected, the area or groups to be covered, the time available and the effort needed for each observation. A varied town may need more sampling locations than a small, relatively uniform area. Repeated observations at one site help assess consistency, but do not replace observations from other relevant sites or times.

A strong justification explains the trade-off. For example, systematic counts across several streets provide spatial coverage and can be completed within the fieldwork period, but repeating them at other times would be needed before making a claim about typical daily pedestrian flow. The best plan is not simply the easiest or the largest: it is one that collects relevant, reasonably representative evidence safely and consistently.

Planning essentials

• Collect evidence that directly answers the enquiry.
• Choose specific sites for relevance, coverage, safety and access.
• Check timing: tides, seasons and daily changes can affect what is available.
• Recording sheets need site identifiers, dates/times, labelled variables, units, repeats and notes on problems.
• Use consistent recording rules; distinguish a missing result from zero.

Sampling choices