SUMMER TERM – NEA (INDEPENDENT INVESTIGATION) AND SKILLS

The NEA is 20% of the A-level and is marked out of 70. It is a written report of 3,000–4,000 words, internally assessed and externally moderated.

You define your own question, linked to any part of the specification. You must use fieldwork data you collected yourself (individually or in a group), alongside your own research and/or secondary data. You must use both quantitative and qualitative data where appropriate.

Edexcel NEA mark area	Marks
1. Purpose of the independent investigation	12
2. Field methodologies and data collection	10
3. Data representation, analysis, interpretation and evaluation of techniques	24
4. Conclusions and critical evaluation of the overall investigation	24
Total	70

Topic 22: Purpose – identifying a suitable question and hypothesis

Key ideas

• A good title:
  • Is focused and manageable: a small area, a few weeks of data collection.
  • Is linked to theory (e.g. longshore drift, sediment cells, the psammosere, place perception or regeneration).
  • Can be answered with data you can collect yourself.
  • Is geographically specific: name the place.
• Research questions and hypotheses:
  • A hypothesis is a testable statement, e.g. "Mean pebble size decreases with distance eastwards along Chesil Beach."
  • The null hypothesis (H₀) states that there is no significant difference or relationship. It is used for statistical testing.
  • You can also use sub-questions, e.g. 2–4 smaller questions feeding the main aim.
• Literature review / context:
  • Explain the theory, e.g. LSD and sediment sorting, or ICZM.
  • Summarise previous studies and secondary data, e.g. Environment Agency LiDAR, SMP documents, census data, the Index of Multiple Deprivation (IMD).
  • Locate your study area using maps (and GIS).
• Coasts-linked NEA ideas (they link to your October fieldwork):
  • Does pebble size and roundness change along a beach (LSD)?
  • How effective are the groynes at a named resort?
  • How do vegetation and soil change along a sand dune transect (succession)?
  • How do residents' perceptions of coastal management vary with distance from the defences?

Self-test

Q1: What is a null hypothesis? A: A statement that there is no significant relationship or difference between variables (e.g. "There is no significant difference in pebble size between site A and site B"). The statistical test assesses whether it can be rejected.

Q2: Give three features of a good NEA title. A: Focused and manageable in scale; clearly linked to geographical theory and the specification; answerable with primary data you can collect (and place-specific).

---

Topic 23: Fieldwork methods, research and selecting equipment

Key ideas

• Sampling:
  • Random: every point has an equal chance of selection (e.g. random number tables or GIS-generated points). It avoids bias, but may miss small areas.
  • Systematic: regular intervals (e.g. every 10 m along a transect). It gives even coverage and is easy, but risks missing patterns or coinciding with periodic features.
  • Stratified: samples proportional to sub-groups (e.g. by age group for questionnaires, or by dune zone). It is representative of known sub-groups.
  • Opportunistic / convenience: whoever or whatever is available. It is quick but biased.
  • Sample size: larger samples are more reliable. Aim for at least 30 for questionnaires where possible, and at least 10 pairs for Spearman's rank.
• Typical physical (coastal) methods and equipment:

Method	Equipment	Measures
Beach profile	Ranging poles, clinometer, tape measure (or dumpy level)	Beach gradient and shape, e.g. either side of groynes
Sediment analysis	Callipers/ruler; Powers' roundness index chart	Long axis (mm); roundness (angularity)
Longshore drift	Float/dye, stopwatch, tape, compass	Rate and direction of drift
Wave frequency	Stopwatch	Waves per minute (constructive <8; destructive >10)
Dune transect	Quadrat (e.g. 1 m²), percentage-cover estimate, soil pH meter, anemometer, infiltration ring	Vegetation succession and environmental gradients
Cliff/defence survey	Field sketch, photos, bipolar assessment	Condition of defences, mass-movement features

• Typical human methods:
  • Questionnaires (closed and open questions).
  • Interviews.
  • Environmental quality surveys (bipolar scores).
  • Land-use mapping.
  • Pedestrian and traffic counts.
  • Photo and media analysis.
• Risk assessment:
  • Tides: check tide tables and work on a falling tide.
  • Cliffs: rockfalls.
  • Slippery rocks, weather, and lone working.
• Ethics:
  • Informed consent and anonymity; avoid sensitive questions; respect for property.
  • Minimise environmental damage (e.g. trampling dunes).
  • GDPR: store personal data securely.

Glossary

Term	Definition
Primary data	Data you collect yourself in the field.
Secondary data	Data collected by others (e.g. census, Environment Agency, Met Office).
Quantitative data	Numerical data (e.g. pebble length, pedestrian counts).
Qualitative data	Non-numerical data (e.g. interviews, photos, opinions).
Accuracy	How close a measurement is to the true value.
Precision	How consistent repeated measurements are.
Reliability	Whether the method gives consistent results if repeated.
Validity	Whether the method actually measures what it claims to.
Pilot study	A small trial run to test methods before full data collection.

Self-test

Q1: Why might you use stratified sampling in a questionnaire? A: To make sure sub-groups (e.g. age groups or residents versus tourists) are represented in proportion to their share of the population. This reduces bias.

Q2: Name the equipment for measuring a beach profile. A: Two ranging poles, a clinometer and a tape measure (alternatively a dumpy level, or a GPS or LiDAR comparison).

---

Topic 24: Information collection, data representation and analysis (statistics)

Data presentation – choose the right technique

Data type	Good techniques
Change along a line	Kite diagrams (dune vegetation), cross-section and profile graphs
Spatial patterns	Choropleth maps, dot maps, proportional symbols, GIS layers, isolines
Flows	Flow lines, desire lines
Relationship between two variables	Scatter graph with line of best fit
Distribution / spread	Dispersion diagrams, box plots
Large ranges	Logarithmic scales
Composition	Pie charts, divided bars, triangular graphs
Inequality	Lorenz curves (and the Gini coefficient)
Images	Annotated photographs, field sketches, satellite/aerial images (oblique and vertical)

Descriptive statistics

• Measures of central tendency:
  • Mean: the sum divided by n. It is affected by outliers.
  • Median: the middle value. It is resistant to outliers.
  • Mode: the most frequent value.
• Measures of dispersion:
  • Range: the simplest measure.
  • Interquartile range (IQR): upper quartile minus lower quartile. It is resistant to outliers.
  • Standard deviation (SD): the spread of values around the mean.
• Worked example (central tendency and dispersion):
  • Data: 12, 15, 11, 18, 14, 20, 16.
  • Mean = 106 ÷ 7 = 15.1; median = 15; range = 20 − 11 = 9.
  • Sample SD ≈ 3.2. Most values lie within about ±3.2 of the mean.

Inferential statistics

General method for every test:

1. State the null hypothesis (H₀).
2. Calculate the statistic.
3. Find the degrees of freedom (where relevant).
4. Compare the result with the critical value at the 0.05 (95%) and 0.01 (99%) significance levels.
5. Accept or reject H₀.
6. Interpret geographically.

Spearman's rank correlation coefficient (rs)

• Use: to test the strength and direction of the relationship between two variables. You need at least about 10 pairs.
• Formula: rs = 1 − (6Σd²) / (n³ − n).
• Worked example: distance along a beach (in the direction of LSD) against mean pebble long axis.

Site	Distance (m)	Mean pebble size (mm)	Rank distance	Rank size	d	d²
1	0	62	10	1	9	81
2	50	58	9	3	6	36
3	100	60	8	2	6	36
4	150	51	7	4	3	9
5	200	47	6	6	0	0
6	250	49	5	5	0	0
7	300	40	4	7	−3	9
8	350	38	3	8	−5	25
9	400	35	2	9	−7	49
10	450	31	1	10	−9	81
					Σd²	326

• Calculation: rs = 1 − (6 × 326) / (1000 − 10) = 1 − 1956/990 = −0.98 (to 2 d.p.).
• Critical values for n = 10: 0.648 at the 0.05 level and 0.794 at the 0.01 level.
• Result: |−0.98| > 0.794, so there is a strong negative correlation, significant at the 99% level. Reject H₀.
• Interpretation: pebbles get smaller with distance along the beach, which is consistent with attrition and sorting during longshore drift.
• Correlation does not prove causation. Consider other factors, such as cliff sources along the way.

Chi-squared (χ²)

• Use: to test whether observed frequencies differ significantly from expected frequencies. It needs counts (not percentages), and every expected value should be at least 5.
• Formula: χ² = Σ (O − E)² / E.
• Worked example: the number of quadrats containing marram grass in four dune zones (60 quadrats in total). H₀: there is no difference between zones, so E = 15 per zone.

Zone	O	E	(O−E)²/E
Embryo	28	15	11.27
Yellow	17	15	0.27
Grey	9	15	2.40
Slack/mature	6	15	5.40
		χ²	19.33

• Degrees of freedom = categories − 1 = 3. Critical values are 7.81 (0.05) and 11.34 (0.01).
• Result: 19.33 > 11.34, so the difference is significant at the 99% level. Reject H₀.
• Interpretation: marram distribution differs between zones, which is consistent with succession.

Student's t-test

• Use: to test whether the means of two samples differ significantly (e.g. pebble size at two sites, or beach width either side of a groyne).
• Formula: t = (x̄₁ − x̄₂) / √(s₁²/n₁ + s₂²/n₂). Degrees of freedom = n₁ + n₂ − 2.
• Worked example:
  • Site A (updrift of the groyne) pebble lengths: mean 51.7 mm, SD 4.85, n = 10.
  • Site B (downdrift): mean 40.5 mm, SD 3.03, n = 10.
  • t = 11.2 / √(2.357 + 0.917) = 11.2 / 1.809 ≈ 6.19.
  • df = 18; critical values are 2.10 (0.05) and 2.88 (0.01).
• Result: 6.19 > 2.88, so the difference is significant at the 99% level.
• Interpretation: the groyne appears to trap coarser material on its updrift side.

Gini coefficient and Lorenz curve

• Lorenz curve: plot the cumulative % of the population (x-axis) against the cumulative % of income or resource (y-axis). The diagonal represents perfect equality.
• Gini coefficient = area between the diagonal and the Lorenz curve ÷ total area under the diagonal. It ranges from 0 (perfect equality) to 1 (perfect inequality).
• Worked example: the five quintiles of a population receive 5%, 10%, 15%, 25% and 45% of income.
  • Cumulative shares: 5, 15, 30, 55, 100.
  • Area under the Lorenz curve (trapezium rule, width 0.2): 0.2 × (0.025 + 0.10 + 0.225 + 0.425 + 0.775) = 0.31.
  • Gini = (0.5 − 0.31) / 0.5 = 0.38. This is moderate inequality, similar to the UK's income Gini, which is roughly 0.33–0.36 depending on the measure.

Measurement error and sampling

• Measurement error comes from:
  • Instrument precision (e.g. ±1° on a clinometer).
  • Human error (e.g. reading a scale, subjective percentage cover).
  • Sampling error: the sample does not represent the population.
• Reduce errors by:
  • Repeating readings and taking the mean.
  • Calibrating equipment.
  • Using a pilot study.
  • Using a larger sample.
  • Keeping the same operator.

Self-test

Q1: When should you use chi-squared rather than Spearman's rank? A: Chi-squared compares observed and expected frequencies (counts) in categories. Spearman's tests the correlation between two sets of paired, ranked continuous data.

Q2: An rs value of 0.55 with n = 10 – is it significant at 0.05? A: No. The critical value at 0.05 for n = 10 is 0.648, and 0.55 < 0.648, so you accept H₀ (no significant correlation).

Q3: What does a Gini coefficient of 0 mean? A: Perfect equality: every person or group has an equal share.

Q4: Give two ways to improve the reliability of field data. A: Increase the sample size and repeat measurements, taking the mean. You could also run a pilot study or calibrate the equipment.

---