Conclusions and Evaluation of the Enquiry

You must establish a Causal Link by explaining why the data supports the conclusion. This requires linking your results to established geographical theory, such as the Bradshaw Model for physical geography or the Burgess/Hoyt models for human geography.

The Conclusion Logic Formula:

1. Verdict: "My hypothesis that [Hypothesis] was [Supported/Rejected]."
2. Evidence: "The data shows a [Trend]. For example, at Site 1..."
3. Theory: "This matches the [Model Name], which suggests [Process]."
4. Counter-Evidence: "However, Site X was an anomaly..."

Spotting and Explaining Anomalies

Sometimes the most valuable piece of data is the one that completely breaks your expected pattern.

An Anomaly is a result that does not fit the overall trend or expected geographical theory. For example, finding a large 15 cm boulder in the lower course of a river where the trend line suggests 2 cm pebbles.

On a scattergraph plotting Distance Downstream ($x$) against Pebble Size ($y$), anomalies are easily identified visually as data points sitting far away from the line of best fit. Statistically, they can be flagged using the Interquartile Range (IQR), as anomalies fall significantly outside the upper or lower quartiles.

The impact of an anomaly depends heavily on your sample size. In small samples, anomalies are highly influential and can skew your mean, leading to tentative conclusions. Large samples of 50 or more help "smooth out" the impact of irregular results.

When dealing with familiar fieldwork, you must explain why the anomaly occurred. This could be due to a localised bank collapse, a sudden gust of wind affecting equipment, or simple human miscounting.

Evaluating Your Enquiry: Accuracy, Reliability, and Validity

If another student repeated your exact fieldwork tomorrow, would they get the exact same results?

When evaluating your enquiry, you must provide a balanced assessment of its strengths and weaknesses. AQA distinguishes between Accuracy, which is how close a measurement is to the true value, and limitations of scale, which refer to sample size.

Reliability refers to the consistency of your results and whether they are repeatable. It is often compromised by small sample sizes or unpredictable random errors.

Validity questions whether your data actually measures what was intended. For example, a pedestrian count on a Monday morning has low validity for measuring how "busy" a town is overall, because it does not represent typical behaviour.

To correctly evaluate your enquiry, you must provide a definitive judgement that weighs strengths against weaknesses. Here are worked examples modelling this balanced approach:

• Evaluating Validity: "Our conclusion is largely valid because we surveyed multiple sites; however, its overall validity is limited as we only measured velocity in July during a drought, representing 'low flow' rather than typical annual behaviour."
• Evaluating Reliability: "The conclusions are partially reliable. While the large sample size of 100 helped reduce random error, the use of an orange as a float instead of a flow meter introduced inconsistency due to wind interference."

Most GCSE fieldwork suffers from the "snapshot in time" limitation. Taking measurements on a single dry day limits the validity of your conclusions regarding long-term, annual river discharge processes.

Data that relies on personal opinion, like Environmental Quality Surveys, introduces Subjectivity. This lowers overall validity due to Operator error/bias.

Identifying Data Collection Errors

Even with the best planning, fieldwork is messy and measurements are rarely perfect.

Errors generally fall into two categories. A Random Error is an unpredictable mistake made by chance, such as misreading a stopwatch. A Systematic Error is a consistent mistake in every measurement, like using a tape measure stretched by 2 cm or a flow meter that wasn't zeroed.

You must also identify Sampling Bias, which occurs when a sample is non-representative (e.g., pragmatic sampling by only surveying friendly-looking people). This bias can be fixed using Stratified Sampling, which divides a population into subgroups to ensure all demographics or geographical zones are represented.

Equipment limitations also frequently cause errors. Reading a ruler at an angle causes a Parallax Error. Furthermore, using simple floats (like oranges) to measure river velocity only captures surface speed and is highly affected by wind, limiting reliability.

When asked to evaluate data collection errors in the exam, use the Three-Step Appraisal framework to ensure you structure your evaluation to cover the error, its effect, and the solution:

1. Identify: State exactly what the error or limitation was (e.g., "Our traffic count was only 5 minutes long").
2. Impact: Link the error to its effect on the data or conclusion (e.g., "This is a limitation of scale; we may have missed a 'pulse' of traffic from nearby traffic lights, reducing reliability").
3. Improvement: Provide a specific, actionable fix (e.g., "Conduct counts for 15 minutes and repeat at different times of day, such as 8 AM and 1 PM").

Suggesting Improvements

Understanding what went wrong in your fieldwork is only half the battle; examiners want to know how you would fix it next time.

Quantitative improvements involve increasing your sample size (e.g., from 10 to 50) to improve reliability and reduce the influence of random errors and anomalies.

Methodological improvements focus on increasing accuracy by upgrading equipment. For example, swapping a standard ruler for digital callipers, or a float for a digital flow meter.

Temporal improvements boost validity by repeating the study at different times. Conducting surveys across different seasons, or on both weekdays and weekends, gives a much more representative picture.

Finally, primary "snapshot" data can be evaluated against secondary data sources. Checking your fieldwork results against Environment Agency flood maps or Census data helps verify long-term trends and improves the overall reliability of your conclusions.