Fieldwork methods & questionnaires

Fieldwork methods & questionnaires

1. Enquiry and examination scope

Fieldwork investigates a geographical question using evidence collected in a place. It links classroom processes to observations and tests whether an explanation fits the evidence. The enquiry cycle has five stages: planning and preparation → data collection → processing, presentation and analysis → interpretation and conclusion → evaluation. Evaluation can also improve decisions during collection.

For the upcoming \(2027\) HKDSE, the prescribed fieldwork modules are Managing Coastal Environment and Building a Sustainable City, confirmed by HKEAA’s final circular of \(19\) June \(2025\). Fieldwork is assessed through compulsory written Paper \(1\) Section B, worth \(15\%\) of the overall subject assessment; it is not SBA. These chapter practices are original learning materials, not official examination papers.

Cohort comparison: \(2026\), historical — Climate Change, Changing Industrial Location, Managing Coastal Environment. \(2027\) — Managing Coastal Environment, Building a Sustainable City. \(2028\) — Building a Sustainable City, Combating Famine. These are fieldwork pools, not replacements for the full compulsory syllabus. Industrial location remains in the wider course. The \(2028\) label does not claim that this chapter supplies a complete famine case.

2. Stage 1: turn an idea into a testable plan

Start with a manageable question: “Is the beach steeper where surface sediment is coarser?” is testable; “Is this coast good?” is not precise. A hypothesis is an expected relationship, not a fact to prove at all costs. Define variables, units, measurement procedures, study area and time before visiting.

Coastal plan: compare gradients and median sediment diameter on several transects, using the same beach zone and similar tidal conditions. City plan: compare pedestrian comfort on shaded and unshaded streets while recording temperature, land use, traffic and time. State a plausible process: coarser, more permeable beach material can reduce backwash, while shade can reduce incoming solar energy at a street surface.

Primary data are collected for this enquiry, such as measured slope or a questionnaire response. Secondary data were collected for another purpose, such as a published tide table, earlier map or census. Use secondary data to choose safe times and add context, but check date, scale and definitions before combining it with your observations.

Plan equipment, access, recording sheets, roles and a pilot. Review tides, weather and escape routes on coasts; avoid wave impact and unstable cliffs. For city work, use safe pavements and crossing points. Follow school supervision procedures; cancel or relocate if conditions are unsafe. Participation in interviews is voluntary: explain the purpose, avoid identifying details and accept refusal.

3. Sampling and questionnaires

A sample should represent the population relevant to the question. Simple random sampling gives each unit in a defined sampling frame an equal chance. Systematic sampling uses a fixed interval after a random start, such as every fifth pedestrian; periodic patterns can bias it. Stratified sampling divides the population into relevant groups, then samples within each, often in proportion to group size.

Purposive sampling selects sites or people for a stated reason, such as a coastal engineer; it can supply useful expertise but is not statistically representative. Convenience sampling uses whoever or wherever is easiest. It saves time but may miss commuters, residents at work or inaccessible parts of the beach. “More samples” helps only if coverage and selection also improve.

For \(100\) interviews in a population that is \(60\%\) residents and \(40\%\) visitors, a proportional design allocates \(60\) and \(40\). Define how each person is selected within each stratum, spread interviews across places and times, and record refusals without replacing them only with especially willing friends.

Ask one neutral question at a time. Replace “Don’t you agree the new promenade is beautiful and safe?” with two questions: “How would you rate the promenade’s appearance?” and “How safe do you feel using it?” Give balanced responses from very poor/unsafe to very good/safe, including a middle response and “not used/do not know” where relevant.

Specify a recall period: “What was your main travel mode for your most recent journey here?” with one-choice categories walking, cycling, bus/minibus, rail, private car/taxi, other. Include an open question asking why. Closed questions are easy to compare; open responses reveal reasons but take longer to code. Pilot the wording, avoid duplicate categories and do not infer actual behaviour solely from stated preferences.

4. Stage 2: measure consistently

For a beach profile, set a transect approximately perpendicular to the shoreline, record its endpoints and take readings at fixed intervals or breaks of slope. Use equal-height marks on two vertical ranging poles, a level or clinometer, and a tape. Sight between equal-height marks; record the slope angle and the distance convention. If \(L\) is distance along the slope, \(\text{vertical change}=L\sin\theta\) and \(\text{horizontal distance}=L\cos\theta\); do not treat sloping distance as horizontal without justification.

Measure sediment at preselected points rather than picking the largest visible pebbles. For pebbles, use a ruler or callipers to measure the intermediate axis with the same rule at every site; for sand, use suitable grain-size comparison or sieving methods. Record units and sample size. Repeated transects represent spatial variation better than many measurements at just one point.

For urban environmental quality, define a scoring rubric before recording. For example, litter: \(1\) \(=\) many items throughout the fixed observation area, \(3\) \(=\) a few scattered items, \(5\) \(=\) none seen. Score each site over the same area and duration. Have observers practise and compare scores; subjective ratings are not direct pollutant measurements.

For temperature, use a calibrated instrument at a consistent height, shielded from direct sun, after allowing it to stabilise. Compare sites simultaneously or rotate the order to reduce time-of-day bias. For pedestrian counts, fix the counting line and interval, define whether both directions count, and prevent double counting. Record wind, cloud, unusual events, absences and instrument problems.

5. Stages 3 and 4: analyse and conclude

Check units, transcription and missing values before analysis. Tabulate raw data and calculated statistics separately. Use a profile for beach relief, a scatter plot for paired grain size and gradient, a bar chart for travel modes, and a map for site patterns. Quote the number of observations and display variability rather than reporting only a mean.

Worked coastal example, synthetic: transects A/B/C have median pebble diameters \(12/24/36\,\mathrm{mm}\) and gradients \(1:20\), \(1:10\), \(1:5\). Convert to percentages: \(5\%\), \(10\%\), \(20\%\). Coarser material is associated with steeper profiles in these three transects. Greater infiltration may weaken backwash, but exposure and recent storms could also differ. The evidence supports the local association, not a universal law or a proved cause.

Worked city example, synthetic: shaded streets S1/S2/S3 have temperatures \(29/30/31{}^{\circ}\,\mathrm{C}\); unshaded streets U1/U2/U3 have \(32/33/34{}^{\circ}\,\mathrm{C}\). Means are \(30{}^{\circ}\,\mathrm{C}\) and \(33{}^{\circ}\,\mathrm{C}\), a \(3{}^{\circ}\,\mathrm{C}\) difference. Shade is a plausible explanation, but material, street orientation and observation time should be matched before attributing the difference to trees.

Conclude against the original question: supported, partly supported or not supported, with figures and a geographical explanation. Consider contrary evidence. A questionnaire may show perceived comfort, while a thermometer records air temperature; compare the two without treating them as identical. Recommend an action only as far as the evidence allows, considering cost, access and different users.

6. Stage 5: evaluate and improve the enquiry

Distinguish accuracy (closeness to the true value), reliability (consistency when repeated), representativeness (how well the sample covers the target population) and validity (whether the method measures what the question asks). A thermometer can give consistent but wrongly calibrated readings: reliable results can still be inaccurate.

Link every limitation to its effect and a remedy. Only surveying a promenade at noon misses evening users → repeat at varied times. One transect beside a groyne may not represent the whole beach → stratify by distance from structures and sample several transects. Different observers interpreting “clean” differently reduce comparability → use anchored criteria and shared training. A larger biased sample does not remove selection bias.

More repeated readings reduce random fluctuation but cannot fix a systematic calibration error. Match controls, calibrate instruments and change design where necessary. Explain a practical trade-off: more sites improve spatial coverage but leave less time per site; a longer questionnaire may yield detail but increase refusal.

For a written fieldwork answer, specify what is done, where, how often, with which instrument, and why it answers the enquiry. “Use a questionnaire” earns less than a neutral question with a sampling method and a reason. For evaluation, name the actual limitation of the supplied study, not a memorised list. Refer to figures in conclusions and distinguish a practical improvement from an entirely different investigation.

Self-check: interviewing friends at one café is convenience sampling, not random sampling. Equal counts in two unequal population groups are not proportional sampling. A failed hypothesis is a valid result if the enquiry was defensible. Safety and consent are part of planning, not marks gained by taking risks.

7. Build a usable recording sheet and pilot

A coastal recording sheet should include site and transect IDs, date/time, tide stage, weather, start/end locations, distance convention, slope readings, pebble measurements, observer and problems. A city sheet should include site ID, shade category, surface type, measurement time, instrument height, repeated temperatures, respondent category and anonymous response code. These fields allow later checks; memory is not a substitute for a record.

Synthetic repeat example: a shaded-site thermometer reads \(29.8\), \(30.0\) and \(30.2{}^{\circ}\,\mathrm{C}\) after stabilising. \(\text{Mean}=30.0{}^{\circ}\,\mathrm{C}\) and \(\text{range}=0.4{}^{\circ}\,\mathrm{C}\). The small range suggests consistency over that short interval, but does not show that the instrument is correctly calibrated or that every summer day is similar. Keep the three readings, not only the mean.

In a pilot, two observers give litter scores of \(2\) and \(4\) at the same site. First compare what each counted and the area observed; agree visible score anchors and repeat the pilot. Simply averaging the scores to \(3\) hides the disagreement. If the enquiry asks about air quality, a litter score alone is not a valid air-quality measurement: choose a relevant measurement or clearly redefine the question.

For questionnaire coding, give each closed category a code without treating unordered codes as measured quantities. \(\text{Walking}=1\), \(\text{bus}=2\) and \(\text{rail}=3\) do not imply rail is three times walking. Summarise counts and percentages instead of averaging these category codes. Group open responses into clearly defined themes and retain a short anonymous evidence record for checking the coding.

An effective final recommendation links evidence to a realistic action and a check: “Trial shade at two exposed waiting areas, then repeat matched temperature and comfort measurements at the same times.” Identify who benefits and who may lose space or access. Do not recommend a whole-city programme from one convenience sample without stating that the evidence is local and preliminary.