Skip to content
AP Psychology · Mental and Physical Health

Causal inference

Causal inference appears in 3 reviewed AP Psychology questions and sits in Mental and Physical Health. Below are worked examples showing exactly how the exam tests it, with the answer marked and every distractor explained.

Causal inference means concluding that one variable actually causes a change in another. Only an experiment can support it, because the researcher manipulates the independent variable and randomly assigns people to conditions.

Why it matters

Claims about what causes disorders or which therapies work depend on it. A correlation alone cannot rule out reverse causation or a third variable.

How the exam tests it

You get several designs and must pick the one that best isolates a cause. The answer is the one with a manipulated variable and a comparison condition, like Asch testing people with a unanimous wrong majority versus alone.

Don’t confuse it with

Correlational and naturalistic studies show that two things vary together, but they cannot show which one causes the other.

Causal inference in real questions

Example 1Medium
A researcher studying PTSD treatment compares veterans who chose to attend group therapy (n = 40) versus those who did not attend (n = 40) and finds that the therapy group has significantly lower PTSD symptoms. What is the primary reason the researcher CANNOT conclude that therapy reduced PTSD symptoms?
  1. APTSD symptoms are subjective and cannot be meaningfully measured with the standardized instruments researchers typically use
  2. BThe study lacks a placebo condition, so any improvement could be attributed to general attention rather than the therapy
  3. CThe sample size is too small to detect significant differences between groups
  4. DWithout random assignment, the groups may have differed at the start; those who chose therapy may be more motivatedCorrect

Why D is right

Without random assignment, this is a quasi-experimental study. Veterans who chose therapy may be more motivated to improve, less severely affected, or more socially supported — all of which could explain lower PTSD symptoms independent of therapy. Pre-existing group differences (self-selection confound) prevent causal conclusions.
Why the other 3 choices miss
A
Incorrect because PTSD symptoms can be reliably measured using validated standardized scales (e.g., PCL-5); measurement validity is not the primary limitation here.
B
Incorrect because a placebo condition addresses expectancy effects; the more fundamental problem is that groups self-selected, creating pre-existing differences that confound the comparison.
C
Incorrect because n = 40 per group is a reasonable sample size for detecting moderate effects; the primary limitation is the absence of random assignment, not sample size.

Studied alongside

Other concepts from Mental and Physical Health.