Assessing and improving reliability Flashcards
What is reliability?
Reliability is a measure of consistency, if the results are not consistent then they are not reliable. Researchers want study’s to be reliable between participants and over time.
Test-retest reliability
This is when the same group of people are asked to undertake the research measure such as a questionnaire on different occasions. The same group of participants must be used both times so demand characteristics need to be factored in. This means the study should not be repeated quickly after the last one, but if to much time is taken between the studies as participants will change and grow as people over time which will effect the results of the study.
The two scores of each occasions are correlated and if the correlation is significant then the measures have good reliability. A perfect correlation is 1 and the closer the score is to 1 the more reliable it is.
Inter-observer reliability
This is when two or more observers are observing and recording behavior in a consistent way. This is useful for creating reliability in situations where subjectivity is a risk. It is also important when behavioural categories are being applied as it checks if the behavioural categories are being used in the same manner. An example is in the strange situation when there was a 94% agreement between observers so the findings are considered more meaningful.
Improving reliability in questionnaires
Identify which questions affect reliability the most and modify them if necessary. If they are important to the questionnaire then reword them so they are easier to interpret or if they are an open question turn them into a closed one.
improving reliability in interviews
You can make sure all the interviews are conducted by the same interviewer as it reduces researcher bias as different people will say questions in different ways and if the same one cant be used training is needed limit the potential bias. Making the interview a structured one will also limit researcher bias.
Improving reliability in experiments
The reliability in experiments can be influenced by the level of control they have over variables, laboratory experiments have high levels of control making them easy to replicate. Taking more control of extraneous variables reduces the chance they become confounding and improves reliability.
Improving reliability in observations
Observations can lack objectivity as it relies on the researcher’s own interpretations, if behavioural categories are used then they must be applied correctly. One way to ensure this to operationalise the behavioural categories making them clear and specific. There should be no overlap between them or need for personal interpretation on the meaning.
