Component 2: Inferential statistics (knowledge, appropriate application and interpretation of) — Eduqas A-Level Psychology

Inferential statistics are used in psychology to determine whether the results from a sample can be generalised to the wider population. This topic covers the selection and interpretation of non-parametric statistical tests, including Spearman's rho, Wilcoxon signed-rank, Mann-Whitney U, and Chi-square. You need to understand the criteria for choosing each test based on experimental design, level of measurement, and hypothesis direction, as well as how to interpret probability values and critical values.

This topic is essential for the Eduqas A-Level Psychology exam because it assesses your ability to apply statistical knowledge to research scenarios and to evaluate the significance of findings. It builds on your understanding of research methods and data types, and it is often examined through multiple-choice questions, short-answer calculations, and extended response questions requiring interpretation. Mastering inferential statistics will also support your critical evaluation of psychological studies and your own practical work.

• →The choice of statistical test depends on three factors: whether the design is independent groups or repeated measures (or correlation), the level of measurement (nominal, ordinal, or interval), and whether the hypothesis is directional (one-tailed) or non-directional (two-tailed).
• →Non-parametric tests are used when data are nominal or ordinal, or when assumptions of parametric tests (e.g., normal distribution, interval data) are not met. The four tests you must know are Spearman's rho, Wilcoxon signed-rank, Mann-Whitney U, and Chi-square.
• →The observed value is calculated from the data, and the critical value is found in a table based on the significance level (usually 0.05), the number of participants (N) or degrees of freedom (df), and whether the test is one-tailed or two-tailed.
• →For Spearman's rho, Wilcoxon, and Mann-Whitney U, the result is significant if the observed value is equal to or less than the critical value. For Chi-square, the result is significant if the observed value is equal to or greater than the critical value.
• →The significance level (p<0.05) indicates that there is a less than 5% probability that the results occurred by chance. If significant, the null hypothesis is rejected and the alternative hypothesis is accepted.

• Identifies chi-square as a test for nominal (categorical) data based on frequency counts.
• States that chi-square compares observed frequencies with expected frequencies.
• Recognises that chi-square is used to test for a difference or association between categories.
• Selects chi-square when the data are independent and each participant falls into one category.
• Explains that the calculated value is compared with a critical value at a stated significance level to decide whether to reject the null hypothesis.
• Identifies Mann-Whitney U as a test for ordinal (ranked) data.
• States that Mann-Whitney U compares two independent groups in a test of difference.
• Recognises that scores are ranked and the ranks of the two groups are compared.
• Selects Mann-Whitney U when participants are in separate (independent) groups rather than repeated measures.
• Explains that the calculated U value is compared with a critical value at a stated significance level to decide whether to reject the null hypothesis.
• Identifies the sign test as a non-parametric test suited to nominal data from a repeated measures or matched pairs design
• States that each participant contributes a plus or minus according to the direction of the difference between the two conditions
• Explains that pairs showing no difference are excluded and N is the number of remaining usable pairs
• Defines the test statistic S as the smaller of the total number of plus signs and the total number of minus signs
• States that S is compared with the critical value for the chosen level of significance and the relevant N
• Concludes that the null hypothesis is rejected when the observed value of S is equal to or less than the critical value
• Identifies Spearman’s rank order correlation coefficient as a test of the relationship between two co-variables rather than a test of difference
• States that the data must be ordinal, or interval data converted to ranks, with each set ranked separately from 1 to N
• Explains that d is the difference between the two ranks for each pair and that each d is squared before summing to give Σd²
• Applies the formula ρ = 1 − (6Σd²)/(N(N² − 1)) with N equal to the number of pairs of scores
• Interprets the sign of the coefficient as the direction of the relationship and its magnitude as the strength, with values near 1⁺ or 1⁻ indicating a strong correlation
• Compares the observed coefficient with the critical value for the chosen significance level and N to decide whether to reject the null hypothesis
• Identifies the Wilcoxon test as a non-parametric test for related or paired data, such as repeated measures or matched pairs designs.
• States that differences between paired scores are ranked by absolute size with signs retained, and that the smaller sum of positive or negative ranks gives the test statistic T.
• Compares the observed value with the critical value for the chosen significance level, N and one- or two-tailed hypothesis, rejecting the null hypothesis when the observed value is equal to or below the critical value.
• Recognises that the test requires at least ordinal data and is chosen when a parametric test such as the related t-test is not appropriate.
• Defines a probability value as the likelihood of obtaining the observed result, or a more extreme one, if the null hypothesis is true.
• States that a result is significant when the probability is equal to or less than the chosen significance level, commonly 0.05, leading to rejection of the null hypothesis.
• Explains that the significance level sets the risk of a Type I error, rejecting a true null hypothesis.
• Distinguishes statistical significance from effect size, noting that a small probability value does not by itself show a large or important effect.
• States that the significance level is a pre-set probability threshold used to decide whether to reject the null hypothesis.
• Identifies 0.05 as the conventional level in psychology and explains it as a 5% chance of the observed result arising if the null hypothesis is true.
• Explains that adopting a stricter level such as 0.01 reduces the risk of a Type I error while increasing the risk of a Type II error.
• Distinguishes the significance level, chosen in advance, from the probability value calculated from the data and compared against it.
• Defines the observed value as the statistic calculated from the sample data by applying the chosen inferential test.
• States that the observed value is compared with the critical value from a table at the chosen significance level.
• Identifies that the direction of comparison depends on the test, with some tests requiring the observed value to be at least the critical value and others requiring it to be at most the critical value.
• Explains that the decision about significance follows from the comparison and determines whether the null hypothesis is rejected.
• States that the critical value is read from a published table for the specific inferential test used.
• Identifies the need to know N or degrees of freedom before locating the correct row in the table.
• Distinguishes a one-tailed from a two-tailed test when selecting the correct column.
• Compares the observed calculated value with the critical value to reach a decision about the null hypothesis.
• Checks the direction rule for that particular test rather than assuming larger always means significant.
• Selects the equals sign when stating that two values are identical.
• Uses the less-than-or-equal sign to express a significance level such as p ≤ 0.05.
• Applies the greater-than sign when the observed value must exceed the critical value for significance.
• Matches the symbol direction to the rule of the specific test rather than reversing it.
• Reads a printed comparison correctly by identifying which quantity is the larger.

• 💡Read the scenario carefully and check the level of measurement before choosing chi-square.
• 💡Link your choice to the design: nominal data, independent groups and frequency counts.
• 💡When interpreting, state the significance level and whether the calculated value exceeds the critical value before concluding.
• 💡Check whether the design uses independent groups before selecting Mann-Whitney U.
• 💡Confirm the data are ordinal and can be ranked before choosing the test.
• 💡In interpretation, name the significance level and state whether the calculated value reaches the critical value.
• 💡State the design and level of measurement first, because these determine whether the sign test is the appropriate choice
• 💡Show the plus and minus tally and the resulting N before quoting S, so the working is visible
• 💡Quote the critical value alongside the significance level and whether the test is one- or two-tailed, then state the decision in terms of the null hypothesis
• 💡Show the ranking of both variables and the d and d² columns so the calculation can be followed
• 💡State N and substitute it carefully into N(N² − 1), checking the arithmetic before quoting ρ
• 💡Report the direction and strength in words alongside the numerical value, then compare with the critical value to reach a decision about the null hypothesis
• 💡Name the design and level of measurement before justifying the choice of test, so the examiner sees the reasoning.
• 💡State the significance level, N and whether the hypothesis is one- or two-tailed when interpreting the critical value.
• 💡Write the conclusion in terms of the null hypothesis and the probability of the result occurring by chance.
• 💡State the chosen significance level explicitly before deciding whether a result is significant.
• 💡Use the phrase 'equal to or less than' when describing the threshold for significance.
• 💡Link the decision to rejecting or retaining the null hypothesis rather than simply saying the result is 'proven'.
• 💡Read the stem carefully to see whether it asks about the level itself or the error it affects, then answer only that.
• 💡Use precise wording such as 'reject the null hypothesis' rather than 'prove the hypothesis'.
• 💡Check the direction of the error: stricter level means fewer Type I errors and more Type II errors.
• 💡Name the test before deciding the direction of comparison, because the rule differs between tests.
• 💡Quote the observed value and the critical value side by side before stating the conclusion.
• 💡State the conclusion in terms of rejecting or retaining the null hypothesis at the stated level.
• 💡Name the test, N, tail direction and significance level before quoting any critical value.
• 💡Write the comparison explicitly, for example observed value against critical value, then state the decision.
• 💡Learn the direction rule for each test you might meet rather than memorising one rule for all.
• 💡Say the symbol aloud as a sentence to check the direction is correct.
• 💡Remember the open end of the symbol faces the larger quantity.
• 💡State the significance level with the correct inclusive symbol, for example p ≤ 0.05.
• 💡Always state the three criteria for choosing a test explicitly in your answer: design, level of measurement, and hypothesis direction. This ensures you cover all marking points.
• 💡When interpreting a statistical test, always compare the observed value to the critical value, state whether it is significant, and then relate this back to the experimental hypothesis (accept or reject) and the null hypothesis.
• 💡Use the correct terminology: 'significant at p<0.05', 'critical value', 'observed value', 'one-tailed', 'two-tailed', 'degrees of freedom'. Avoid vague language like 'the results are good' or 'it works'.

• Treats chi-square as suitable for ordinal or interval data; restrict it to nominal categories and frequency counts.
• Confuses chi-square with a test of correlation; use it for differences or associations between categories, not for measuring the strength of a relationship between two continuous variables.
• Assumes a significant chi-square proves the experimental hypothesis is true; recognise it only indicates whether the null hypothesis can be rejected at the chosen level.
• Applies Mann-Whitney U to a repeated-measures design; use it only for two independent groups and choose a related test for matched or repeated designs.
• Treats Mann-Whitney U as suitable for nominal data; restrict it to ordinal data that can be ranked.
• Concludes significance without comparing the calculated value with the critical value at a stated level; always state the comparison before deciding.
• Treats the sign test as suitable for ordinal or interval data without ranking; reserve it for nominal or directional data and use a rank-based test when the data are ordinal
• Counts pairs with no difference as a plus or minus; discard them and reduce N to the number of usable pairs
• Uses the larger of the two sign totals as the test statistic; take the smaller total as S
• Accepts the null hypothesis when S is below the critical value; reject the null when S is equal to or less than the critical value
• Describes the coefficient as showing a cause-and-effect relationship; report only the strength and direction of an association between co-variables
• Ranks the two sets of scores together as one combined list; rank each variable separately from 1 to N
• Forgets to square each rank difference before summing; square every d and then total to obtain Σd²
• Treats a coefficient near zero as evidence of a strong relationship; interpret values near zero as little or no correlation
• Treats the Wilcoxon test as suitable for independent groups; use the Mann-Whitney test for unrelated data and reserve Wilcoxon for paired scores.
• Ranks the raw scores rather than the differences between pairs; rank the absolute differences and then restore each sign.
• Rejects the null hypothesis when the observed value exceeds the critical value; reject it only when the observed value is equal to or less than the critical value.
• Claims a p-value gives the probability that the null hypothesis is true; it gives the probability of the data assuming the null hypothesis is true.
• Treats any probability below 0.5 as significant; compare against the chosen level, usually 0.05.
• Equates a significant result with a large effect; report effect size separately from significance.
• Treats the significance level as the probability that the experimental hypothesis is true; correct this by describing it as the chance of the result under the null hypothesis.
• Confuses Type I and Type II errors when discussing stricter levels; correct this by linking a stricter level to fewer false rejections and more missed true effects.
• Assumes a result below 0.05 proves the hypothesis; correct this by noting it only permits rejection of the null hypothesis at that threshold.
• Treats the observed value as a figure looked up in a table; correct this by describing it as the value calculated from the data.
• Applies the same direction of comparison to every test; correct this by checking the rule for the specific named test.
• Ignores the number of tails or sample size when selecting the critical value; correct this by matching all relevant table criteria.
• Treats the critical value as the same as the observed value calculated from the data; keep the two separate and compare them.
• Ignores whether the test is one-tailed or two-tailed; select the column that matches the stated hypothesis direction.
• Assumes a larger observed value always means significance; check the direction rule printed with that test's table.
• Reverses the direction of an inequality so the larger quantity sits at the closed end; place the open end toward the larger value.
• Writes a strict inequality where an inclusive one is meant; use the or-equal form when equality counts.
• Uses the equals sign to compare an observed value with a critical value when a decision rule is intended; choose the inequality that expresses the rule.
• Students often think that a significant result proves the alternative hypothesis is true. In fact, it only indicates that the results are unlikely to have occurred by chance, not that the hypothesis is definitively proven.
• Many students confuse the direction of the comparison for significance: they think a higher observed value is always better. For most tests (Spearman's, Wilcoxon, Mann-Whitney), the observed value must be less than or equal to the critical value, while for Chi-square it must be greater than or equal to the critical value.
• Students sometimes believe that non-parametric tests are always less powerful than parametric tests. While they are less powerful when parametric assumptions are met, they are more appropriate when data are not interval or normally distributed, and they make fewer assumptions.

1. 1Week 1: Learn the criteria for choosing each non-parametric test. Create a decision tree or table summarising the design, level of measurement, and hypothesis direction for each test. Practice with past paper scenarios.
2. 2Week 1: Memorise the direction of significance for each test (less than or equal to for Spearman's, Wilcoxon, Mann-Whitney; greater than or equal to for Chi-square). Use flashcards to test yourself.
3. 3Week 2: Practice interpreting observed and critical values using past exam questions. Write full model answers for 3-4 mark interpretation questions, ensuring you state the comparison, significance, and hypothesis decision.
4. 4Week 2: Complete timed exam questions on inferential statistics, including multiple-choice and extended response. Review mark schemes to identify common pitfalls and refine your answers.
5. 5Ongoing: Use online quizzes and revision apps to test your recall of key terms and criteria. Explain the concepts to a study partner to reinforce your understanding.

• 📋Multiple-choice questions asking which statistical test should be used for a given scenario. Advice: read the scenario carefully and apply the three criteria (design, data type, hypothesis) before selecting your answer.
• 📋Short-answer questions requiring you to calculate or interpret a statistical test. Advice: show your working, state the observed and critical values, and clearly explain the conclusion in terms of the hypothesis.
• 📋Extended response questions asking you to design a study and justify the choice of statistical test, or to evaluate the use of non-parametric tests. Advice: use specific terminology and link your justification to the design, data, and hypothesis.
• 📋Questions requiring you to identify the level of measurement or the experimental design from a description. Advice: look for keywords such as 'independent groups', 'repeated measures', 'ordinal', 'nominal', and 'correlation'.