Geographic Methodology and Fieldwork: Design, Sampling, Measurement and Data
The Geographic Methodology and Fieldwork domain tests how geographical knowledge is produced: hypothesis design, sampling, measurement, data handling and honest reporting. Questions pose a small field or data scenario and ask which method fits, or what is wrong with the design.
Research design
- Begin with a question, then a testable hypothesis; define variables and identify the independent and dependent ones.
- Distinguish primary data (measured by you) from secondary data (census, national statistics, satellite products, climate normals).
- Quantitative data suits measurement and statistics; qualitative data suits perception, interviews and observation.
Sampling
- Random: every unit has an equal chance; avoids bias but may miss rare cases.
- Systematic: every nth unit along a line or transect; easy in the field but can align with a periodic pattern.
- Stratified: split the area into strata (altitude belts, land-use types) and sample within each; best where the area is visibly varied.
- Quadrats for vegetation cover; transects for change along a gradient such as a beach or slope profile.
Instruments and accuracy
- GPS for position (state accuracy, typically a few metres), clinometer for slope angle, anemometer for wind, soil auger and pH meter, flow meter for discharge, rain gauge and thermometer.
- Repeat measurements and report the mean; distinguish precision (repeatability) from accuracy (closeness to the true value).
- Record date, time, weather, location and observer, otherwise the data cannot be checked or repeated.
Worked example
You want to compare soil erosion on north- and south-facing slopes above a village. Reasoning: erosion varies with slope angle, vegetation cover, soil and aspect. Use stratified sampling, with the two aspects as strata and several sites in each. At every site measure slope angle with a clinometer and record vegetation cover from quadrats, keep soil type and altitude band constant, and take three replicates. If grazing pressure also differs between the slopes, that is a confounding variable and must be reported rather than ignored.
Traps that cost marks
- Claiming causation from a correlation between two mapped variables.
- Sampling only along the road because it is convenient - a biased, unrepresentative sample.
- Mixing units (metres with centimetres, per mille with per cent) inside one calculation.
- Omitting sample size and site location, making results unverifiable.