Methodology

Check the claims

Critical reading does not mean rejecting everything. It means asking every claim the same transparent questions, regardless of whether we agree with it.

Seven steps for verification

1. Define the claim

Be clear about exactly what can be checked. A claim that is too broad or shifts in meaning must first be narrowed.

2. Identify the source type

Primary research, reviews, opinion pieces, news reports, lectures and legal documents answer different kinds of questions.

3. Examine the measurement method

Look at operational definitions, sampling, instruments, data processing and exclusion criteria.

4. Look for an appropriate control

A control should show what happens without the factor under study or under a relevant alternative.

5. Assess uncertainty

Consider sample size, margins of error, possible biases and alternative explanations.

6. Look for replication

A single result is rarely final. Ask whether independent groups can reproduce it using comparable methods.

7. Compare the conclusion with the data

Titles, abstracts and media interpretations should not claim more than the results actually support.

Quick checklist

  • Can you find the original source?
  • Are the main terms clearly defined?
  • Are the sample and measurement procedure described?
  • Was there an appropriate control or comparison?
  • Are limitations and uncertainties disclosed?
  • Have other studies reached similar results?
  • Is the conclusion proportionate to the strength of the data?

Common warning signs

Certainty without a source

A confident claim is presented without a reference that can be checked.

Shifting definitions

The same word changes meaning at different points in the argument.

Cherry-picked evidence

Only supportive data are presented while conflicting results are omitted.

Authority as evidence

A person’s status or popularity substitutes for verifiable methods and data.

Put it into practice

Put it into practice

Choose an article and identify its claim, source, method and the limits of its conclusion.