RELIABILITY Flashcards
Reliability
reliability is a measure of consistency.
external reliability
extent to which a measure is consistent when repeated.
internal reliability
internal consistency of a measure, whether different questioonaires in a same construct are neasuring the same construct.
assessing internal reliability
split half method
assessing external reliability
test retest
inter observer
split half method
- split the participants into two parts
- participants complete the both parts
- test the strength of correlation
- a string correlation indicates internal reliability
inter observer reliability
the extent to which an observer agrees on the observation they record is called inter observer reliabilty.
how to improve reliability in behavioural categories
- need to be operationalised
- some observers may need more practice so as to respond quickly.
test retest reliability
the same test or interview is given to same participant on two occassions to see if same results are obtained
assessing reliability for interviews ( inter interviewer reliability)
in interviews, a researcher could assess by comparing answers from one person on a occasion with same interviewer and then with the same person and interviewer a week later.
- or using the same method as with two observers.
reliability of experiments
dependent variable in a n experiment is often measured using rating scales or behavioural categories. eg during rutters study on romanian orphan study iq scores were one of the dependent variable they used.
improving reliability standarisation
making sure procedures are same every time so as to compare the performance of participants.
improving reliability of observation
- providing appropriate training to the observers can improve their accuracy
- pilot studies can be carried out to identify poorly define behavioural categories which can later be operationalised.
improving reliability of interviews
- use structured interview rather than unstructured interview
- using a same script for all participants by the interviewer.
- using the same interviewer
improving reliability of questionnaires
- use closed questions
- if there is a lready an established questionnaire for what you want to measure, use that instead of creating your own questions.