What you need to know
- Unit 6Comment on the trend, determine a relationship from the graph, and suggest realistic modifications to reduce error or improve the experiment.
Do not invent a critical value or a probability. Chi-squared, a t-test and a correlation coefficient are mathematical skills you use when the question asks, not a slogan for every graph. The arithmetic of those tests is in the statistics lesson. No treatment advice follows from a school practical. The unit also expects conclusions to respect measurement uncertainty. These notes show that without pretending every table has an uncertainty printed on it.
By the end, you should be able to…
- Describe the trend or relationship and stop at the evidence.
- Select chi-squaredA statistical test of whether observed numbers differ significantly from the numbers expected., a t-test or a correlationA relationship in which two variables change together. A correlation does not prove that one causes the other. coefficient only when the data match, and do not call a result significant without a supplied critical value.
- Calculate a percentage uncertainty when an absolute uncertainty is given, and propose a realistic modification.
Build your understanding
The pattern first
Say what the graph does: as the independent variable increases, the dependent variable increases, decreases, stays the same, or increases and then levels off. Quote a pair of values if you are asked for evidence. “It changed” is too vague. A constant from a straight section is the gradient, with its unit, not a second adjective about the same line.
Do not write “significant” or “proves” unless a test and a critical value were supplied and you have compared them. Overlapping ranges, or two means that look different, are not that comparison. An explanation that uses biological knowledge should be labelled as an explanation, not as extra data.
Choose the test from the data, then stop if no critical value is given
Categories with observed and expected counts suit chi-squaredA statistical test of whether observed numbers differ significantly from the numbers expected.. You need counts, not percentages. Two means, with a spread, suit a t-test. Paired measurements of two continuous variables suit a correlationA relationship in which two variables change together. A correlation does not prove that one causes the other. coefficient. Standard deviations that do not overlap are none of these tests by themselves.
Illustrative counts are 50 and 30, total 80, tested against a 3:1 expectation. Expected counts are 60 and 20. Chi-squared = (50 − 60)² / 60 + (30 − 20)² / 20 = 100 / 60 + 100 / 20 = 1.67 + 5 = 6.67 to three significant figures. No critical value was supplied here, so this lesson does not call 6.67 significant and does not tell you to reject a null. If your question gives a critical value, compare the statistic with it and state the comparison. The full method sits in the statistics lesson.
Uncertainty and a modification that could be done
If an absolute uncertainty is printed, percentage uncertainty = absolute uncertainty ÷ measurement × 100. An illustrative length of 18.0 mm ± 0.5 mm has percentage uncertainty 0.5 ÷ 18.0 × 100 = 2.8% to one decimal place. Do not invent an uncertainty the question did not state.
A realistic modification names the limitation and the change. “Use a gas syringe instead of counting bubbles, so volume is measured” is a modification. “Use more accurate equipment” is not. A wider range, intermediate values, a blank, a stated quadratA square frame of known area used to sample organisms. area, or repeats are realistic. A change that also alters the independent variable has not improved the original test. One plate, or one unpaired zone, is not a medicine and is not a significance test.
Make your reasoning visible
Claim, evidence, limit
One sentence for the pattern, one for a value if it was asked for, one for what the data cannot show.
No critical value, no significance word
You may still calculate a statistic the question asks for. You may not borrow a critical value from memory and treat it as given.
Avoid this misconception
The claim: A clear trend means the result is statistically significant.
The correction: A trend is a description. Significance needs a named test and a comparison with a critical value supplied in the question. These notes do not give you that value.
Check your understanding
Try each question before opening the suggested answer.
Practice question
Retrieval and explanation · suggested answer
An illustrative graph shows the rate increasing from 10 °C to 30 °C and falling at 40 °C and 50 °C. Write a conclusion that does not overclaim.
Show suggested answer
The rate increased as temperature increased from 10 °C to 30 °C, then decreased at the two higher temperatures tested. The data do not fix an exact optimum between 30 °C and 40 °C, and they are not a significance test.
Practice question
Retrieval and explanation · suggested answer
A length is given as 18.0 mm ± 0.5 mm. Calculate the percentage uncertainty to one decimal place.
Show suggested answer
0.5 ÷ 18.0 × 100 = 2.8%. The percentage uncertainty is 2.8% to one decimal place.
Practice question
Retrieval and explanation · suggested answer
Which test, if any, compares mean leaf length in two fields, and which tests a 3:1 ratio of counts? What do you still need before using the word significant?
Show suggested answer
Mean leaf length: a t-test. A 3:1 ratio of counts: chi-squared, using expected counts. You still need the critical value supplied for that test. Do not treat non-overlapping standard deviations as either test.
Take the idea further
Apply the same specification knowledge to an unfamiliar problem. Any extra assumptions are supplied in the question.
Stretch question
An unfamiliar extract produced a zone 6 mm wider than a water control on one plate. No repeat and no identified chemical are given. What conclusion is justified, which test would you not force onto these two numbers, and what modification would you want?
Show a hint
One plate is a comparison of two zones, not a set of categories and not a clinical trial.
Explore the reasoning
The extract inhibited the test bacterium more than water under these conditions, by 6 mm on this plate. Two single measurements are not a chi-squared test and not a t-test until you have repeats or counts that match those tests. A realistic modification is to repeat the plates and record a mean difference. You must not claim a medicine, name a chemical that was not identified, or call 6 mm significant. The assumption is that the control was water and that the organism and conditions were those in the question.
Recall this lesson
Lesson recall questions
Question 1 of 5: A logarithmic plot may be used when bacterial numbers grow ________; the y-axis would show ________ bacterial count.
Unit 6 analysis
Practice question
Retrieval and explanation · suggested answer
A logarithmic plot may be used when bacterial numbers grow ________; the y-axis would show ________ bacterial count.
Lesson: A2: Scales, log axes and gradients
Try this from memory first.
Put your biology to work
Explore where these ideas appear in the world. Enrichment, not extra required recall.
A test statistic is not a decision until it is compared
A3 · Patterns, tests and realistic changesThe NIST engineering statistics handbook describes rejecting a null hypothesis when the test statistic falls in a rejection region set by a critical value, or when a p-value…Specification Unit 6 · AnalysisRead & use the scienceClose article
Why this matters
The NIST engineering statistics handbook describes rejecting a null hypothesis when the test statistic falls in a rejection region set by a critical value, or when a p-value is compared with a level chosen in advance. The choice of that level is a convention, not a result hiding inside your graph. School questions that want a decision give you the critical value. The test statistic is a summary calculated from a particular data set under assumptions about that test. It is not a verdict printed inside the raw number. A decision depends on the correct test, degrees of freedom and threshold or critical value supplied for the comparison. Even a statistically significant difference does not by itself show a large or medically important effect. Start with the pattern, then state exactly what the data allow.
Back to the specification
Pearson Edexcel IAL Biology · Issue 2, February 2021 · Unit 6, pp. 39–42. The application extends beyond required recall.
- Unit 6 · Analysis Tabulate, calculate, graph and interpret data, including possible logarithmic scales, gradients, uncertainty and realistic modifications.
Use the science
A calculated chi-squared value is 6.67 and the question prints no critical value. Why is “the result is significant” not yet justified?
Explore a possible response
Significance here means the statistic has been compared with a critical value, or a stated p-value threshold. 6.67 by itself is only the result of the sum. It does not contain the comparison.
Explore the source
A small p-value is not a large biological effect
A3 · Patterns, tests and realistic changesThe American Statistical Association cautions that a p-value does not measure the size or importance of an effect. It also does not give the probability that the null…Specification Unit 6 · AnalysisRead & use the scienceClose article
Why this matters
The American Statistical Association cautions that a p-value does not measure the size or importance of an effect. It also does not give the probability that the null hypothesis is true. In Unit 6, a test statistic and supplied critical value may justify calling a result significant, but the biological pattern still needs description. A tiny difference measured very precisely may be statistically detectable yet unimportant for an organism. Conversely, a small study can miss a meaningful effect. The graph, design and uncertainty belong beside the formal decision.
Back to the specification
Pearson Edexcel IAL Biology · Issue 2, February 2021 · Unit 6, pp. 39–42. The application extends beyond required recall.
- Unit 6 · Analysis Tabulate, calculate, graph and interpret data, including possible logarithmic scales, gradients, uncertainty and realistic modifications.
Use the science
Two treatments differ by only 0.1 mm in mean growth, with a statistically significant result. What else should be reported?
Explore a possible response
Report the effect size, variation or uncertainty, sample design and biological context. Statistical significance alone does not show that 0.1 mm matters in practice.
Explore the source
- ASA statement on statistical significance and p-valuesAmerican Statistical Association