AQA 7182 / STATISTICAL TESTS / INDEPENDENT SUPPLEMENT
START HERE
Choose the test.
Explain the choice.
AQA A-level Psychology: statistical-test practice
A focused workbook for learners who know the test names but lose confidence when the study changes.
Practise extracting the aim, design and data type before naming a test.
20 original tasks 3 worked examples
12 test-choice scenarios Two contrasting study designs
4 sign-test datasets A full sign-test calculation
4 error-diagnosis prompts Separated solutions with reasons
How to use this workbook
Read the method map on page 2. Study the worked examples on pages 3-4 and the interpretation notes on
page 5. Complete pages 6-11 before checking pages 12-16. Use page 17 to trace the reference methods.
For every choice, write: aim + design + measurement + test. Then state why the closest alternative is
unsuitable. For a sign test, write the sign rule before calculating anything.
Scope and limits
Supports selected inferential-testing content in AQA 7182, with the specification for first A-level exams in
2027. It is not the whole research-methods topic or an official AQA resource. Only the sign test is calculated
here; the other named tests are selected and interpreted at an introductory level.
All studies and datasets are fictional. This is study practice, not a protocol for running research or advice
about real clinical decisions. No actual examination, assignment or test-bank questions are reproduced.
Created with AI assistance. Quantitative answers checked programmatically; no independent expert review or achieved grade is
claimed. Not affiliated with or endorsed by AQA. References and provenance: page 17.
Version 1.0 | September 2026 | First A-level exams 2027 onwards 1
,AQA 7182 / STATISTICAL TESTS / INDEPENDENT SUPPLEMENT
METHOD MAP
Read the study in layers
1. Aim: Is the question a difference between conditions, or an association between variables? The same
people can provide data for either aim.
2. Design: For a difference, are the observations related (same people or deliberately matched pairs) or
independent (separate, unpaired groups)?
3. Measurement: Nominal means categories; ordinal adds order; interval measurements have meaningful
equal units. Ratio measurements also have a meaningful zero and use the quantitative route here. Numeric
category codes remain labels. [1, 2, 8]
Question and evidence Conventional choice here
Difference; related; only directions retained Sign test
Difference; related; rankable differences Wilcoxon signed-rank
Difference; independent; ordinal outcome Mann-Whitney U
Difference; related; suitable quantitative data Related (paired) t-test
Difference; independent; suitable quantitative data Unrelated (independent) t-test
Correlation; ordered/rank data; monotonic pattern Spearman's rho
Correlation; suitable quantitative data; linear pattern Pearson's r
Association; independent categorical frequencies Chi-squared independence test
The map is a starting point, not a universal rule
Check assumptions before a final choice. A paired t-test concerns the distribution of differences. Wilcoxon
uses ranked magnitudes and needs suitable differences, including symmetry for its usual location
interpretation. A conventional pooled unrelated t-test assumes equal variances; real analysis may use
alternatives. [2, 4-6]
Chi-squared needs independent contributions and adequate expected counts. Ordinal does not mean assumption-free. Pearson
targets a linear relation; Spearman uses ranks for monotonic association. This workbook states key assumptions to avoid guessing.
[2, 7]
Version 1.0 | September 2026 | First A-level exams 2027 onwards 2
, AQA 7182 / STATISTICAL TESTS / INDEPENDENT SUPPLEMENT
WORKED EXAMPLES E1-E2
One detail changes the test
E1 | Same outcome, related observations
Independent volunteers use both a large-key and a small-key keypad. Each person gives an ordered ease
rating after each. Suppose the paired differences can be meaningfully ranked and are approximately
symmetric. The aim is a non-parametric comparison of the conditions.
Reasoning: a difference is being tested; each volunteer provides a pair; the analysis retains ranked
differences. Wilcoxon signed-rank is the conventional selection under the stated assumptions. Explain the
actual pairing rather than merely saying “related”. [2, 5]
Nearest wrong turn: Mann-Whitney uses independent groups. Calling the two keypad conditions “two
groups” does not remove the within-person connection.
E2 | Same outcome, independent observations
Now different volunteers use one keypad only. Each gives a single ordered ease rating. The aim is still to
compare the two conditions.
Reasoning: a difference is being tested; no person or matched pair links the conditions; the ratings are
ordinal. Mann-Whitney U is the conventional selection. The design changed, so the appropriate test
changed. [2, 6]
Nearest wrong turn: Wilcoxon cannot recover pairings that the design never created. Do not pair people
just because they appear on the same row in a spreadsheet.
Write a complete justification
“Use [test], because the study asks whether [specific outcome differs/is associated], the observations are
[design with evidence], and the data are [scale with evidence]. The stated assumptions [briefly identify them]
support this choice.”
If information is missing, say what you need. A test name without a defensible interpretation of the study is
not a complete explanation.
Version 1.0 | September 2026 | First A-level exams 2027 onwards 3
START HERE
Choose the test.
Explain the choice.
AQA A-level Psychology: statistical-test practice
A focused workbook for learners who know the test names but lose confidence when the study changes.
Practise extracting the aim, design and data type before naming a test.
20 original tasks 3 worked examples
12 test-choice scenarios Two contrasting study designs
4 sign-test datasets A full sign-test calculation
4 error-diagnosis prompts Separated solutions with reasons
How to use this workbook
Read the method map on page 2. Study the worked examples on pages 3-4 and the interpretation notes on
page 5. Complete pages 6-11 before checking pages 12-16. Use page 17 to trace the reference methods.
For every choice, write: aim + design + measurement + test. Then state why the closest alternative is
unsuitable. For a sign test, write the sign rule before calculating anything.
Scope and limits
Supports selected inferential-testing content in AQA 7182, with the specification for first A-level exams in
2027. It is not the whole research-methods topic or an official AQA resource. Only the sign test is calculated
here; the other named tests are selected and interpreted at an introductory level.
All studies and datasets are fictional. This is study practice, not a protocol for running research or advice
about real clinical decisions. No actual examination, assignment or test-bank questions are reproduced.
Created with AI assistance. Quantitative answers checked programmatically; no independent expert review or achieved grade is
claimed. Not affiliated with or endorsed by AQA. References and provenance: page 17.
Version 1.0 | September 2026 | First A-level exams 2027 onwards 1
,AQA 7182 / STATISTICAL TESTS / INDEPENDENT SUPPLEMENT
METHOD MAP
Read the study in layers
1. Aim: Is the question a difference between conditions, or an association between variables? The same
people can provide data for either aim.
2. Design: For a difference, are the observations related (same people or deliberately matched pairs) or
independent (separate, unpaired groups)?
3. Measurement: Nominal means categories; ordinal adds order; interval measurements have meaningful
equal units. Ratio measurements also have a meaningful zero and use the quantitative route here. Numeric
category codes remain labels. [1, 2, 8]
Question and evidence Conventional choice here
Difference; related; only directions retained Sign test
Difference; related; rankable differences Wilcoxon signed-rank
Difference; independent; ordinal outcome Mann-Whitney U
Difference; related; suitable quantitative data Related (paired) t-test
Difference; independent; suitable quantitative data Unrelated (independent) t-test
Correlation; ordered/rank data; monotonic pattern Spearman's rho
Correlation; suitable quantitative data; linear pattern Pearson's r
Association; independent categorical frequencies Chi-squared independence test
The map is a starting point, not a universal rule
Check assumptions before a final choice. A paired t-test concerns the distribution of differences. Wilcoxon
uses ranked magnitudes and needs suitable differences, including symmetry for its usual location
interpretation. A conventional pooled unrelated t-test assumes equal variances; real analysis may use
alternatives. [2, 4-6]
Chi-squared needs independent contributions and adequate expected counts. Ordinal does not mean assumption-free. Pearson
targets a linear relation; Spearman uses ranks for monotonic association. This workbook states key assumptions to avoid guessing.
[2, 7]
Version 1.0 | September 2026 | First A-level exams 2027 onwards 2
, AQA 7182 / STATISTICAL TESTS / INDEPENDENT SUPPLEMENT
WORKED EXAMPLES E1-E2
One detail changes the test
E1 | Same outcome, related observations
Independent volunteers use both a large-key and a small-key keypad. Each person gives an ordered ease
rating after each. Suppose the paired differences can be meaningfully ranked and are approximately
symmetric. The aim is a non-parametric comparison of the conditions.
Reasoning: a difference is being tested; each volunteer provides a pair; the analysis retains ranked
differences. Wilcoxon signed-rank is the conventional selection under the stated assumptions. Explain the
actual pairing rather than merely saying “related”. [2, 5]
Nearest wrong turn: Mann-Whitney uses independent groups. Calling the two keypad conditions “two
groups” does not remove the within-person connection.
E2 | Same outcome, independent observations
Now different volunteers use one keypad only. Each gives a single ordered ease rating. The aim is still to
compare the two conditions.
Reasoning: a difference is being tested; no person or matched pair links the conditions; the ratings are
ordinal. Mann-Whitney U is the conventional selection. The design changed, so the appropriate test
changed. [2, 6]
Nearest wrong turn: Wilcoxon cannot recover pairings that the design never created. Do not pair people
just because they appear on the same row in a spreadsheet.
Write a complete justification
“Use [test], because the study asks whether [specific outcome differs/is associated], the observations are
[design with evidence], and the data are [scale with evidence]. The stated assumptions [briefly identify them]
support this choice.”
If information is missing, say what you need. A test name without a defensible interpretation of the study is
not a complete explanation.
Version 1.0 | September 2026 | First A-level exams 2027 onwards 3
