validity

• what is validity?

  • Is the research accurate? 
  • ‘Describe 2 ways of assessing validity’ 
  • How can validity be improved? 
• external validity

  • If research findings can be applied to real life situations 
  • Can be broken down into ecological and temporal validity  
• ecological validity

  • part of external validity
  • whether it can be generalised to real life situations.
  • Lab experiments DONT have ecological validity as they're so controlled so lack mundane realism
• temporal validity

  • part of external validity
  • achieved if findings can be generalised across time e.g. are the findings still valid in 10 years, if so, they have high temporal validity
  • Asch's line study doesn't have temporal validity as it was a ‘child of its time; due to the 1950s being a conformist time but wouldn't be valid now
• internal validity

  • If the experiment measures what it intends to measure (is the IV, the only thing impacting the DV as extraneous variables prevent the cause-and-effect relationships meaning the research won't be valid) 
  • Can be broken down into face and concurrent validity  
• face validity

  • part of internal validity
  • does the research measure what it intended (must make sure a depression questionnaire is measuring depression and not demand characteristics, it can be assessed by checking with a specialist who will say if it has high face validity, or a pilot study could be used) 
• concurrent validity

  • part of internal validity
  • if research gets similar concurrent results to an already trusted study (if a new depression questionnaire got similar results to Beck's depression inventory, it is concurrent) 
• improving validity in experiments

  • have a CONTROL (placebo) group which shows if the IV affects the DV  
  • Single blind control – don't tell participants what group they're in  
  • Double bind control – researcher also doesn't know who is in which group 
  • Standardised instructions- all ppts get same instructions in the same format to reduce investigator effects and bias  
• improving validity in questionnaires

  • Anonymity – ppts are more truthful if they know their answers are anonymous 
  • Lie scale- checks consistency of results. Writing 2 questions in different ways but both measure the same thing 
• improving validity in observations

  • Covert- won't change behaviour if they don’t know they're being observed 
  • Behavioural categories– reduces researcher subjectivity as behaviours are ticked when seen are defines