reliability

reliability

the extent to which a test produces consistent results

internal reliability

test is consistent in itself

external reliability

consistent over time

reliability issues

- only one researcher

- research only conducted once

- instructions not given in same way

- variables not clearly defined

- p's not asked q's in the same way

solution for one researcher

use more than one researcher

solution for research only being conducted once

repeat the study (external reliability)

solution for instructions not given in the same way

standardise the procedure

solution for variables not being clearly defined

operationalise the procedure

solution for p's not being asked questions in the same way

standardise the procedure

three methods for assessing reliability

split-half method

test-retest method

inter-rater reliability

split half method

- assessing internal reliability by comparing two halves of the same measure

- an individuals performance on two compared through correlation

- can only be used when the questionnaire measures the same thing throughout and when answers are easily quantifiable

test retest method

- involves running the same test again (same design + P's) to see whether the results are the same over time and place

- if two tests achieve the same results on both occasions there with be a high correlation between the two sets of scores

inter rater reliability

- used during observational research or whenever researchers have to rate or categorise behaviour

- researchers should discuss and agree on the behavioural categories before and to ensure consistency