Reliability- assessing and improving.

Reliability refers to the consistency of a test or measure and is linked with replicability. A test or measure is reliable if when it is repeated the same results are found providing consistency of results.

Assessed by: 

1) Test retest:

• A particpant repeats the study multiple times under the same conditions.
• There is time between the test and retest so that the participant will not recall their previous answers but not too much time so that a participant's view on a subject will change and alter their answers. 
• Once the results have been collected they are correlated. If there is strong positive correlation of +0.80 it indicates that the test has high reliability.

Also assessed by:

2) Inter observer/ inter-rater reliability:

• There is more than 1 researcher carrying out the study.
• The observers need to operationalise their behavioural checklist before than observing the study independently from one another to ensure they do not influence each others results.
• Once the study is completed the results are correlated, if there is a strong positive correlation of +0.80 it indicates that there is high reliability.

Improving reliability observations:

• Operationalise the behavioural categories so that there is no ambiguity for the observers and they will not need to make their own judgement whether the behaviuor they saw matches the behavioural category. This means the results gathered will have consistency because the observers will be looking for the same thing. Also the observers should discuss and decide each behavioural category before the observation occurs which may result in training in how to use the behavioural category.
• Ensure that the observations are recorded so that they can be reviewed as much as possible so the observers are certain their findings are consistent and therefore reliable.

Improving reliability questionnaires:

• Have more closed questions then open questions so that the respondent will not misinterprete a question. 
• If there is low reliability this could mean that some of the questions are too complex or ambiguous and need to be either taken out or altered to make them easier to understand.

Improving reliability experiments:

• Use standardised procedures and instructions to allow as much control over extraneous variables as possible and due to standardisation the environment will also be kept constant and not effect the experiment in any way.
• Conduct a pilot test before the study to see and alter any problems which might occur that can affect the replicability of an experiment. 

Improving reliability Interviews:

• Train interviewers to ensure they all ask the questions in the same way to limit interviewer effects.
• Use the same interviewer each time to ensure there is no difference in how the questions are asked. 
• Ensure that structured interviews are used so that all interviewers are collecting the same data and their behavouir does not change and is controlled as to not influence the participant in any way.