Validity and reliability

Internal validity

• outcomes observed in experiments are due to manipulation of IV and not other factors

Assessing internal reliability- split half method

• compare 2 halves of a test, questionanire

• Randomly selecting half of test items and placing them on Form A, other had in Form B

• End up with 2 forms of same test

• Each form should yield same score if items on test were consistent

• Compare 2 scores with correlation coefficient

Assessing external reliability= test-retest

• give same test, questionnaire to same person on 2 separate occasions

• If there’s no treatment between tests, questionnaire should give same results

• If the results aren’t similar then there’s low reliability

• Interval between tests must be long enough that participants can’t remember past answers but not to long that their feelings have changed

Internal validity

• if results show internal validity, researcher measured what they initially set out to measure and haven’t been confounded by internal variables like demand characteristics

• Where results down to manipulated IV causing change on DV

External validity

• is results show external validity they can be generalised

Threats to internal validity

• Paticipant effects- demand characteristics, social desirability. Overcome by single blind and lie scales

• Investigator effects- leading questions. Overcome by double blind and standardised instructions

• Situational variables- time of day, temp, order effects

• Participant variables- age, intelligence, motivation

Face validty

• considering if tests looks as if it measures what its meant to measure

• e.g. reaction times measured reactions to nearest second, would show all participants had reaction of 1 second

Concurrent validity

• looking at correlation between results from a proven test and a new one

• If new one is valid we expect a high correlation

• E.g. new IQ test on groups that’s already been tested on an old IQ test- expect new results to correlate if new test is valid

Predictive validity

• If test has good predicative validity, expect to predict future observations with high degree of accuracy

• E.g. company look at predictive validity of test used in selection procedures to select employees who score well in professional exams

Construct validity

• whether a test actually measures the underlying psychological construct it claims to measure.

• E.g a questionnaire claiming to measure intelligence would have construct validity if its scores correlate with other established measures of intelligence