Experiments (Experimental Studies)

Intervention/Treatment vs Control Group

  • They should have similar characteristics
  • Use random assignment, so that all variables will be equally present in the two groups
  • Only difference should be the intervention/treatment
  • Subjects should be blind (they should not know if they are in the treatment or control group)
    • This is to minimise human bias

Random Assignment

  • Must be random with respect to everything
    • e.g. choosing people with names starting with letters is not random

Single Blind vs Double Blind

  • Single blind: either participants or evaluators don’t know whether they are in/evaluating the treatment or control group
  • Double blind: both participants and evaluators don’t know which group they are in/evaluating

Observational Studies

  • When there are ethical issues performing the experiment
  • Cause and effect issue
    • the study can only show an association between two things
    • it cannot provide evidence for cause and effect
  • Randomisation is lost
    • Participants assign themselves to the groups (they report their own data)