Double list experiment
A double list experiment is a survey method for sensitive questions.
Contents
Description
The motivation of this experiment is that subjects modify their responses in the presence of a sensitive question. This is expected to lead to a social desirability bias, i.e. people respond according to what they believe is expected of them rather than according to what they themselves think or feel.
To encourage genuine response, respondents are shown a list of statements, then asked only for how many they agree with. This creates a degree of anonymity.
Two lists are utilized, A and B. Both have a control version and a treated version. If a respondent is shown the treated list A, then they will see the control list B. Vice versa, if a respondent is shown the control list A, then they will see the treated list B. Generally the control lists simply exclude the sensitive statement, rather than attempting to replace it.
By comparing the means across control and treatment cohorts, the 'agree-ers' can be inferred.
Inference depends on:
- Randomization
- No design effects
- No liars