Validity refers to whether or not your results are valid. This can be done by examining your variables

“VAlidity = VAriables”

Reliability refers to how consist your results are.

“REliability = REpetition”

Reliability refers to the repeatability or reproducibility of measurements (Stallings & Gillmore, 1971). Reliable measurements are those that are similar in value over multiple experiments (conducted under similar conditions). Such data are consistent and stable (Elasy & Gaddy, 1998; Houser, 2009).

Reliability is the cornerstone of the scientific process. When publishing research, scientists must describe all relevant information about their experiments (materials, methods, and all analytical procedures) so that others may repeat them and confirm the reliability of the findings.

Some authors distinguish between two types of reliability measures - internal and external:

• Internal reliability: the consistency of measurements within an experiment. Internal reliability is synonymous with precision

• External reliability: the consistency of measurements over multiple, independent experiments. Thus, if data from two or more identical experiments are plotted on a graph, there should be a strong correlation between the data points.

In the natural sciences, reliability refers to external reliability.

The validity of an investigation refers to whether the procedures in it accurately measure what was intended to be measured (Eisenhart, 1968). It also describes the extent to which the findings of a study may be generalised or extrapolated to other situations. Thus, validity examines the ‘meaning’ or the ‘interpretation’ of the findings of an investigation (Streiner & Norman, 2006).

There are two aspects to experimental validity: internal and external.

• Internal validity: whether the experimental procedures measure what the investigators set out to measure (Thatcher, 2010). For an experiment to be valid, it must satisfy the following criteria:

o The hypothesis is properly constructed (Elasy & Gaddy, 1998).

o All variables (dependent, independent and controlled) are identified. A well-designed experiment can establish the relationship between the dependent and independent variables because all other variables are controlled.

o Systematic and random errors are minimised or eliminated.

o The experimental procedure incorporates steps to repeat measurements.

o The correct instruments or procedures are used in an experiment so that the measurements are relevant

o Where grouping is involved, investigators randomly allocates samples to the different groups.

o The findings of the investigation stem from the independent variable. In other words, there are no other plausible reasons for those results.

o All analytical tools used for the analyses of data, including data manipulation, must be relevant and appropriate for the data types generated in the investigation.

o All underlying assumptions must be carefully evaluated, including the settings in the software used for data analyses.

• External validity - the results of an investigation can be generalised or extrapolated beyond the immediate study. To accomplish this, the researcher must:

o Address all the criteria described for internal validity.

o Ensure that the samples used in the investigation are representative of the wider population.

Experimental outcomes that are internally valid allow researchers to establish causation (see below) – that is, establish the cause-and-effect relationship between the independent and dependent variables. Externally valid results allow researchers to predict future outcomes and extrapolate current findings to new conditions.