Geographical Techniques and Fieldwork: Data and Methods
Geographical Techniques and Fieldwork tests whether you can collect defensible data and choose the right statistical or graphical method.
Core rules
- Sampling: random, systematic (every nth), stratified (by subgroup) and opportunistic. State the sampling frame and acknowledge bias.
- Fieldwork: pilot the questionnaire, use closed and open questions, avoid leading questions, and record date, time and location. Repeat readings to check reliability.
- Descriptive statistics: the mean is affected by outliers; the median is robust for skewed data; the mode is useful for categories. Range and interquartile range measure dispersion.
- Graphs: bar charts compare categories; line graphs show time series; scatter graphs test association; choropleth maps show density by area; isoline maps show continuous surfaces. Correlation does not prove causation.
- Spearman rank correlation: rs = 1 − (6Σd²) ÷ (n(n² − 1)); values range from −1 to +1.
Worked example
Ten sites are ranked on distance from a road and on soil erosion. Σd² = 44 and n = 10. rs = 1 − (6 × 44) ÷ (10 × 99) = 1 − 264/990 = 1 − 0.267 = 0.733. This is a strong positive association: erosion tends to increase with distance from the road. It does not establish that distance causes erosion; slope and land use may be confounders.
Traps
Do not use the mean for heavily skewed income data without saying so. Do not claim a sample is random when it was chosen for convenience. Do not infer causation from correlation. Do not forget n in the Spearman formula.
Key points
- Sampling methods include random, systematic, stratified and opportunistic.
- The median is robust to outliers; the mean is not.
- Scatter graphs show association; correlation does not prove causation.
- Spearman rs ranges from −1 to +1 and uses Σd² and n.
- Pilot studies and repeated readings improve reliability.