Evaluating Your Investigation — The Highest-Mark Stage

⚖️ Evaluating Your Investigation — The Highest-Mark Stage

Evaluation is not listing everything that went wrong. It is a structured, critical assessment of the quality of the entire investigation — one that makes a reasoned judgment about how much confidence you can place in your conclusions. A strong evaluation addresses three distinct dimensions.

The Three Evaluation Dimensions

Dimension	What It Asks	How to Improve It
Reliability	Would you get the same results if you repeated the investigation using the same method? Unreliable data varies due to random error, inconsistent technique, or changing conditions between measurements.	Take repeat measurements at each site and calculate the mean; use a larger sample size; standardise the method carefully (e.g., always measure velocity at 0.6 × channel depth); use the same equipment throughout; note and control for changing conditions
Validity	Does your data actually measure what your enquiry question requires? Data can be reliable (consistent) but invalid (measuring the wrong thing). For example, the float method measures surface velocity — but mean channel velocity at 0.6 × depth would be a more valid measure of overall flow.	Use multiple complementary methods (triangulation); cross-check primary data against secondary data; choose equipment specifically suited to the variable being measured; control confounding variables
Representativeness	Is your sample representative of the whole population or study area? A small sample, or one biased towards accessible or convenient locations, may not reflect the broader pattern.	Increase the sample size; use stratified sampling to ensure all sub-groups are represented; collect data at different times of day and different seasons; compare your sample characteristics against known population data

Common Fieldwork Weaknesses and Specific Improvements

Weakness	Why It Matters	Specific Improvement
Data collected on one day only	Weather, time of year, and unusual events can all affect results; results may not be representative of typical conditions	Collect data on three different days across different seasons; compare results and note any variation
Small sample size (e.g., 8 questionnaire respondents or 5 river sites)	Small samples are easily distorted by one or two unusual values; statistical tests become unreliable with fewer than 10 data pairs	Increase questionnaire sample to at least 30 per zone; increase river sites to 10+ to allow valid use of Spearman's rank
Only one data collection method used	A single method can only measure one dimension of the phenomenon; it cannot identify whether the pattern holds from multiple angles	Triangulate using a second method — e.g., supplement pedestrian counts with an environmental quality survey at the same sites
Personal subjectivity in bi-polar surveys or land use classification	Different observers may score the same location differently, reducing reliability	Use a structured recording sheet with clear criteria for each score; have two observers assess the same site independently and compare scores
Questionnaire carried out at one time of day (e.g., Tuesday morning only)	The sample will over-represent people available at that time — retired people, those not in work — and under-represent commuters and students	Repeat the questionnaire at different times (morning, lunchtime, evening) and on both weekdays and a weekend; stratify the sample by time slot
Float method for river velocity	Surface velocity is typically 20–25% faster than mean channel velocity; results overestimate true discharge	Use a flow meter at 0.6 × channel depth to measure mean velocity directly; or apply the standard correction factor (multiply surface velocity by 0.8)