Inferential statistics
IB Psychology· Research methodology· 14 min read
1. Core concept
Inferential statistics test whether a result is likely to be real rather than chance, so a conclusion can extend beyond the sample. Tests of difference compare two conditions or groups (for example chi-square, related and unrelated t-tests, Mann-Whitney, Wilcoxon); tests of relationship measure how two variables co-vary (correlation coefficients). Each yields a significance level, and researchers also report effect size to show how large the effect is. Significance says 'probably not chance'; effect size says 'this much' — a strong conclusion needs both.
2. Key studies
IB Psychology answers must be supported by named studies. Learn these to cite as evidence:
Kuhn et al. (2019)
A randomised controlled trial found no effect of eight weeks of violent video-game play on aggression or related measures. — A test of DIFFERENCE between conditions returning a non-significant result — a reminder that inferential testing can, and should be able to, come back null.
Rosenquist, Fowler & Christakis (2011)
Depressive symptoms were significantly correlated between people up to three degrees of separation in a social network. — A test of RELATIONSHIP (a significant correlation) — but a correlation, however significant, does not by itself establish cause.
3. Evaluation (AO3)
Strength — disciplined inference — Significance testing puts a probability on chance, and pairing it with effect size prevents both false alarms and over-claiming.
Limitation — not causation, not importance — A significant test of relationship shows co-variation, not cause; and significance depends on sample size, so it can flag trivial effects in very large samples.
4. Scope
5. Common Pitfalls
Wrong move:
A significant correlation coefficient shows that one variable causes the other.
Why:
A frequently penalised misconception in IB Psychology exams.
Correct move:
Tests of relationship quantify how two variables move together; causation needs an experiment or ruling out confounds. Significance is about chance, not about cause or size.
Going deeper
- Descriptive statisticsdescriptive statistics are the input to these tests (RM_5).
- External variables and causalityexternal variables are why a significant correlation still is not cause (RM_15).