Read an argument
Before deciding which side you favor, make the disagreement small enough to test. These reading routes connect substantive disputes to the inference each one requires.
From a score to a claim
- Start with construct validity: is the task measuring the attribute named in the conclusion?
- Read delay of gratification: distinguish an association, incremental prediction and an intervention effect.
- Continue to generalizability: which people, tasks and circumstances does the evidence actually cover?
When predictions disagree
- Read replication to ask whether proposed boundary conditions were predicted in advance.
- Use preregistration and exploration to separate an advance test from an explanation developed after seeing results.
- Examine the consciousness collaboration: agreed tests can challenge parts of more than one framework.
A worked reading exercise
Original hypothetical example
A study finds that volunteers who score higher on a short “resilience” quiz report less stress one week later. A headline says that building resilience prevents stress.
Which inference needs more evidence?
The quiz needs a defensible construct interpretation. The observed association does not establish that changing the score would prevent stress. Shared self-report methods, prior stress and recruitment could matter. A useful next study would define a target population, test an intervention and measure outcomes with more than the same questionnaire.
Where does a value judgment enter?
Even if an intervention improves an average score, deciding who should receive it depends on costs, burdens, alternatives and whose outcomes count. An effect estimate informs those choices; it does not determine them.
Keep the questions separate
- Empirical
- What happens, to whom and under which conditions?
- Theoretical
- Which mechanism or representation explains that pattern?
- Conceptual
- What do the key terms mean, and can rival accounts use them differently?
- Ethical or practical
- What should be done, given consequences, rights and competing priorities?