Reliability and Validity Flashcards
What’s universalism?
Assumption that principles in psychology won’t vary
Findings will apply anywhere irrespective of their geographical location, culture and gender
What’s a two tailed significance level?
2 outcomes in a positive and negative direction
This is for non directional hypothesis
What’s a one tailed significance level?
This is for a directional hypothesis
Only one outcome is predicted
What is statistical significance?
.05 (5%)
Due to chance 5 times out of 100
This value or lower to be significant
What’s generalisability?
Means that we can take the results of a study and apply it to the broader population of interest
What’s reliability?
the consistency
the same answer every time
What’s validity?
the test measures what it is intended to test
the accuracy
What are the types of reliability?
Over time
Internal consistency
Across researchers
What is over time reliability?
How stable our measures are over time
e.g. a measure at time A should give a similar result to time B
What is internal consistency in reliability?
How well items on an individual scale measure the same construct
e.g. items on a questionnaire measuring levels of depression should give similar answers
What is across researchers reliability?
How consistent ratings are across different individual researchers
Results should be the same irrespective of how they’re recording the data
What is test retest reliability?
Useful for constructs that are stable over time
Constructs that aren’t stable across time can’t be assessed for test retest reliability as they’re unstable by their own nature e.g. mood
Measure at time 1, then again at Time 2, do the tests product similar answers at both times?
What correlation indicates good reliability?
> +.80
How can we test internal consistency?
split half reliability (calculate a score for the first half of the items and the second half, check if the scores in each half are similar)
odd even reliability (calculate a score for the odd numbered items and for the even numbered items, check if the scores are similar)
Chronbach’s alpha (check for correlations of every possible half with every other half, take the average, positive correlation means more reliable)
How can reliability across researchers be tested?
Researchers observe ps and rate the behaviours
Check if different researchers rate the behaviours in the same way
Percentage agreement
