Reliability and validity

Reliability refers to how consistent the outcomes of research are, which means that similar results are found when similar research studies are carried out. This is important because it shows that any results found are not due to chance or to a one-off incident. Reliability is increased by standardised procedures, which, if not used, may mean that any results that are found are unreliable.

• Reliability in sampling methods: this ensures that the sample or technique being used in unbiased. If the sample or technique is biased, this means that any findings may not be reliable as the study would have a different outcome if it were replicated
• Reliability in experimental designs: when an independent measures design is used, the participants are different in each condition. This means that findings may be unreliable due to participant variables

Issues of validity

Validity is different to reliability and you will need to be able to clearly differentiate between the two. Validity refers to the extent to which research measures what it has actually set out to measure. For example, if a study was set up to measure how much people communicate via body language, then the way in which the research is set up must ensure that it is body language which is being tested; this is known as internal validity.

If the same research were looking at body language amongst women then they are the target population. A study which focused on men would not be valid as it is not measuring what it sets out to measure – it would be seen to lack external validity. Therefore, it is vital that a research study is designed in a way that will not affect either internal or external validity.

• Validity in sampling methods: if there is an error in the sampling method then the results of the research will not accurately reflect the target population and this will mean that there is a problem with the study’s external validity
• Validity in experimental design: the way in which participants are allocated to a study’s condition may affect the validity of what is found. This can happen in a repeated measures design if counterbalancing is not used – participants ‘learn’ how to carry out a task and therefore their true ability is not being tested. Similarly, in a repeated measures design, participants are more subject to demand characteristics because some are able to work out what the researcher wants to find and may change their behaviour to meet this

Reliability and validity of qualitative methods

When qualitative methods are used, this means that researchers are trying to collect lots of rich, detailed information in order to explore a subject or theory in more depth. Examples of this type of method include case studies and unstructured interviews both of which do not really test a hypothesis like in a laboratory study but enable researchers to find out things like why someone behaves in a certain way or how they feel about a subject.

Qualitative methods are not easy to replicate because much of what is found will be specific to a participant due to how the interviewer, for example, responds to their answers. Someone else might give a completely different answer to a similar question and therefore will be asked something different in return. This puts the reliability of findings into question. It also means that because findings will not be generalisable to a wider population, they may lack external validity.

In qualitative methods, researchers are usually more heavily involved than when a laboratory experiment takes place and this can result in researcher bias where the researchers manipulate what they have found to match what they were looking for. This questions the validity of the study but researchers can work with other researchers so that their professional objectivity is not brought into question.

Reliability and validity of quantitative methods

Quantitative methods are more scientific and usually begin with a testable hypothesis, which is tested using a laboratory experiment or a closed interview or questionnaire.

This method collects facts and information, which measures behaviour and can be applied to a larger population, which increases its external validity. As the researcher is not usually directly involved, this also lessens the chances of bias and therefore the method is seen as more professionally objective.