4.1. Sample sizes and comparison group

Bald numbers tell you very little. They assume meaning as part of the whole population that they are drawn from. ‘Four out of five doctors’ may sound impressive, but suppose only 20 doctors out of the thousands practicing in India participated in a survey. A total of 16 doctors is hardly representative of how doctors across the entire country might think or behave.

There are more criteria a representative survey has to meet. In the doctor example, journalists should check:

  • > Whether the survey was conducted at various cities or hospitals in order to capture different opinions around the country.
  • > Which method was used to collect the answers – telephone, online or a face-to-face interview?
  • > How does the chosen survey method influence the answers of the participant?
  • > Have all age and gender categories been captured with the survey?

These are just a few examples; the criteria that journalists should consider when questioning the survey validity will vary depending on the topic and results.