Misinformation and leading-question paradigm

Description

An experiment showing an event, then feeding one group misleading information or loaded question wording, to measure how post-event suggestion distorts later memory reports.

Participants view an event, then answer questions in which one group receives misleading information, often embedded naturally in a question. Later memory is compared with a control group. Loftus, Miller and Burns (1978) showed a car at a stop sign; a question mentioning a yield sign raised false recognition of a yield-sign slide. Loftus and Palmer (1974) found that the verb 'smashed' versus 'hit' produced speed estimates of 41 versus 34 mph and, a week later, 32 versus 14 per cent false reports of broken glass. Lindsay (1990) showed the effect is a source-monitoring error: it was larger when misinformation came from the same narrator straight after the event than two days later in a different voice. The effect survives explicit warnings. For UX it is a direct test of how interview and survey wording contaminates users' recall of their experiences.

Attributes

Loss type
Unmigrated
Effort
Days

Steps

  1. Show all participants the same event, such as a recorded session, video or product walkthrough.
  2. Randomly assign participants to a misleading-information group and a control group.
  3. Embed the manipulation naturally, for example one leading question among 20, or a loaded verb in a single question.
  4. After a delay (minutes to a week), test memory with forced-choice recognition or open questions that include the critical detail.
  5. Vary source similarity if relevant, for example same versus different voice or medium, to test source confusion.
  6. Compare the proportion of misled reports or false recognitions between groups with a test of proportions.

Sources

  • Cognitive Psychology (Goldstein, 2019)
  • Cognitive Psychology (Sternberg & Sternberg, 2012)

Also known as

  • Misleading postevent information paradigm
  • Misinformation effect paradigm
  • MPI
  • Question-wording effect on eyewitness estimates
  • Loftus and Palmer paradigm

Categories

  • User Research
  • Cognitive Psychology & Perception