All subjects · Biology · WBI16 Practical Skills in Biology II

Choosing the right statistical test

Edexcel IAL Biology · WBI16 Practical Skills in Biology II

In planning questions you must name a suitable test for the data you would collect. The January 2024 report noted tests proposed that did not suit the expected raw data.

Practise this in the app: instant marking and a fix for every mistake. Free, no email needed to start.

Practise this lesson

Three tests, three questions

Question you are asking	Data	Test	Graph
Is there a difference between two means?	Continuous measurements (lengths, masses, rates) from two groups, roughly normal	Student's t-test	Bar chart of means with SD or error bars
Is there a correlation between two variables?	Paired measurements of two variables, at least 5 pairs	Spearman's rank correlation	Scatter graph
Do observed counts differ from expected counts?	Frequencies (counts) in categories	Chi-squared	Bar chart of counts

Matching the test to the plan

Two light intensities in CP10, several repeats each: t-test comparing the two mean rates (the October 2023 mark scheme: bar graph and t-test for two light intensities).
Five or more light intensities: scatter graph and a correlation test such as Spearman's rank.
Abundance of a plant along a transect against light intensity at each quadrat (CP11): Spearman's rank.
Numbers of animals choosing damp or dry, or offspring in a 3 : 1 ratio: chi-squared (raw counts, never percentages or means).
Comparing mean zones of inhibition for two antibiotics (CP14): t-test.

Conditions and pitfalls

The t-test compares two groups only; it needs measurements, not categories.
Spearman's rank needs at least 5 pairs (the critical value table in Pearson papers starts at n = 5 for p = 0.05); more pairs give a more powerful test.
Chi-squared uses actual counts, with expected values that are not too small.
Correlation shows association, not cause.
Name the test and say what you would compare (calculated value against critical value at p = 0.05).

Worked example

A student measures the length of 20 limpet shells on an exposed shore and 20 on a sheltered shore. Which test should be used and why?

Two groups, continuous measurements (length), comparing means.
So use a Student's t-test, with H₀: no significant difference between mean shell length on the two shores.

Answer: Student's t-test, because it compares the means of two sets of continuous data.

Key idea: Difference between two means: t-test. Correlation between paired variables: Spearman's rank (5+ pairs). Observed vs expected counts in categories: chi-squared. Match the graph to the test.

Check yourself

Which test compares the mean height of plants grown with and without fertiliser?
- Spearman's rank correlation
- Student's t-test
- Chi-squared
- Range
Show answer
Student's t-test. Two means of continuous data.
Which test is used for the relationship between distance from a footpath and number of daisies in each quadrat?
- Student's t-test
- Chi-squared
- Mitotic index
- Spearman's rank correlation
Show answer
Spearman's rank correlation. Paired data, looking for a correlation.
Which test checks whether 120 offspring fit a 3 : 1 ratio?
- Student's t-test
- Chi-squared
- Spearman's rank correlation
- Standard deviation
Show answer
Chi-squared. Observed and expected counts in categories.
Which graph pairs best with a Spearman's rank test?
- Bar chart of two means
- Pie chart
- Scatter graph
- Histogram of one variable
Show answer
Scatter graph. Correlation is seen on a scatter graph.
What is the smallest number of pairs of values for which the Pearson Spearman's rank table gives a critical value at p = 0.05?
Show answer
5. The table gives 1.000 at n = 5 for p = 0.05; fewer pairs cannot show significance at that level.
Why should chi-squared not be done on percentages?
- Percentages are always too large
- Chi-squared only works with means
- It needs actual counts; percentages hide the sample size
- Percentages cannot be compared
Show answer
It needs actual counts; percentages hide the sample size. The size of the counts affects the test.
In CP10, a student uses six light intensities with three repeats at each. Which analysis suits the data?
- Scatter graph and a correlation test such as Spearman's rank
- A t-test between all six groups at once
- Chi-squared on the mean volumes
- No analysis is possible
Show answer
Scatter graph and a correlation test such as Spearman's rank. More than two values of a continuous IV: correlation.

Practise with instant feedback

Sources

Last updated 8 November 2026. Spotted a mistake? Tell us.