Revise Planning & Analysing Whole Investigations for Biology 6 (IAL) WBI16 (A2 Level) — revision notes and instant AI marking. Free to start.
📖 Revision notes · preview
Edexcel IAL Biology • Unit 6: Practical Skills in Biology II
Planning & Analysing Whole Investigations
🗺️ Big idea: A whole investigation is planned backwards. Decide what graph and what statistical test will answer the question, and that tells you what data you need — how many groups, how many readings, paired or unpaired. Plan forwards and you often finish with data no test can handle.
Summary — What This Topic Covers
Planning from the analysis backwards
Choosing the test before collecting the data
Deciding how many replicates, and why sample size matters
Pilot studies and what they are for
Writing a conclusion the data can actually support
1. Plan Backwards from the Analysis
ASK → WHAT TEST → WHAT DATA YOU MUST COLLECT
─────────────────────────────────────────────────────────────────────
Do two groups t-test two sets of measurements, continuous,
differ? roughly normal, ideally n ≥ 10 each
Do observed chi-squared COUNTS in categories, not means or
counts match percentages; expected values from a
expected? stated hypothesis
Are two Spearman's rank pairs of measurements from the same
variables individuals or sites, at least 7 pairs
related?
The rule that saves whole investigations
Chi-squared needs raw counts. If you record percentages or means, the test cannot be done and the data cannot be rescued afterwards. Decide the test first.
The commonest planning failure
Collecting one measurement per condition. Any test needs a spread to work with — a single value has no variability, so no test can tell signal from noise.
2. Sample Size and Replicates
More replicates narrow the spread of the mean, making a real difference easier to detect
Around 10 per group is a reasonable target for a t-test at this level; three is usually too few
In the field, more quadrats reduce the effect of patchiness
Sample size must be balanced against time — say so, and justify the number you chose
Why it matters statistically
A small sample gives a large standard error, so the confidence interval is wide and even a real difference may not reach significance. Increasing n is the most reliable way to strengthen a conclusion.
3. Pilot Studies
Key Term
A pilot study is a small preliminary run used to refine the method before the real investigation begins.
Establishes a sensible range for the independent variable
Checks the timing — whether readings can be taken fast enough, or the reaction is over too quickly
Confirms the concentration or dilution gives a measurable response
Reveals practical problems while they are still cheap to fix
Easy marks
Almost any choice in a plan can be justified with "a pilot study showed…". It is one sentence and it converts an arbitrary decision into a reasoned one.
4. Randomisation and Bias
Key Term
Bias is any systematic tendency in how samples are selected or measured that makes them unrepresentative.
In the field, place quadrats using random coordinates, not by eye — "choosing a representative spot" is exactly how bias enters
Assign organisms to groups at random, so that any pre-existing difference is spread evenly
Where possible, measure blind to the treatment, so expectation cannot influence a judged endpoint
Use a transect rather than random placement only when studying a gradient — that is systematic sampling, and it is the right choice there
5. A Conclusion the Data Supports
✗ "This proves that light intensity controls growth."
— one species, one site, one season; correlation, not proof
✓ "There was a significant positive correlation between light intensity
and shoot length (rs = 0.81, n = 12, p < 0.05), suggesting that light
availability is associated with growth in this species at this site.
A correlation does not establish cause: soil depth also increased
along the transect and may contribute."
Quote the test statistic, n and the significance level
Say correlation, not cause, unless you manipulated the variable
Limit the claim to the conditions and organisms tested
Name a confounding variable if one is plausible — it shows judgement
Practice Questions
Practice Question 1
A student plans to test whether two populations of snails differ in shell height. Explain what data they must collect and why the analysis should be decided first.
Practice Question 2
Explain why a chi-squared test cannot be carried out on data recorded as percentages.
Practice Question 3
State two things a pilot study might establish before a full investigation into enzyme activity.
Practice Question 4
A student places quadrats "where the vegetation looks typical". Explain the problem and describe a better method.
What to Memorise
Plan backwards from the testChi-squared needs raw countst-test needs individual measurementsSpearman needs paired data, n ≥ 7One reading per condition = no test possiblen ≈ 10 per group is a sensible targetPilot study justifies your choicesRandom coordinates, not by eyeTransect for a gradientCorrelation ≠ cause
Concepts Checklist
Exam Tips
What mark schemes look for
A plan that names the statistical test and collects data suited to it. Naming the test is often worth a mark on its own.
The trap
Recording percentages when a chi-squared test is intended. The data cannot be recovered, and this is a favourite examiner scenario.
Easy marks
"A pilot study showed…" justifies the range, the concentration and the timing in one sentence each.
Worth remembering
Unless you manipulated the variable yourself, you have a correlation. Saying so — and naming a plausible confounding variable — reads as judgement, not weakness.