4.2.3.1 - Validity

What validity means

In psychological research, validity asks whether a study, measure or conclusion is genuinely about the thing it claims to be about. A memory test should measure memory, not reading speed. A questionnaire about anxiety should measure anxiety, not willingness to admit anxiety. A laboratory study of helping should tell us something meaningful about helping, not only about how people behave in a strange artificial task.

AQA now refers to the measurement of validity. This means researchers should not simply say that a method "seems valid"; they should gather evidence about validity. The evidence might come from informed judgement, correlations with established measures, replications in different settings, or checks that findings still apply over time.

Validity applies across all methods of investigation:

Reliability and validity are related but not the same. Reliability is about consistency. Validity is about accuracy or truthfulness. A bathroom scale that is always 5 kg too heavy is reliable, because it gives consistent readings, but it is not valid, because the readings are inaccurate. In psychology, a questionnaire can produce consistent scores while still measuring the wrong construct.

The four types named by AQA are:

Strong AQA answers do not just name a type. They explain what is being checked, how it could be measured, and how the researcher could improve it.

Face and concurrent validity

Face validity and concurrent validity are especially useful when researchers are judging a measurement tool, such as a questionnaire, interview schedule, psychological test, observation checklist or coding system.

Face validity is usually measured by inspection. The researcher might ask experts, teachers, clinicians, target participants or independent judges to look at the items and rate whether they seem relevant to the construct.

For example, a questionnaire claiming to measure exam anxiety has good face validity if its items clearly relate to anxiety around exams, such as worry before tests, physical tension, avoidance of revision and fear of failure. It would have weak face validity if most items asked about general hobbies or favourite lessons.

Face validity is easy to check, but it is weak evidence on its own. A measure can look valid while still being misleading. A participant may also work out the aim of a high-face-validity questionnaire and answer in a socially desirable way.

Concurrent validity is usually measured statistically. The researcher gives the same participants the new measure and an established comparison measure, then calculates the correlation between the two sets of scores. If higher scores mean more of the construct on both measures, a strong positive correlation supports concurrent validity. If one scale is reversed, the expected relationship is negative. A relevant comparison matters more than the sign alone. The numerical coefficient measures association, while a significance test asks whether the association is unlikely under a no-association null hypothesis; this lesson does not teach choosing or calculating inferential tests.

For example, Dr Okafor writes a new 20-item social anxiety questionnaire for sixth-form students. To measure concurrent validity, she asks the same students to complete her new questionnaire and a well-established social anxiety scale in the same week. If students who score highly on the established scale also score highly on Dr Okafor's scale, this supports the concurrent validity of the new questionnaire.

Consider two new measures compared with an established anxiety scale. Scale A has a correlation of +0.86 and Scale B has +0.18; higher scores mean greater anxiety on all three. A has stronger evidence of concurrent validity in this sample. This does not establish that anxiety causes either score, or that A works equally well for every population. A strong negative correlation could be appropriate if the new measure instead scored calmness, so always read what high scores mean.

However, concurrent validity depends on the quality of the comparison measure. If the established measure is outdated, culturally biased or unsuitable for the sample, agreement with it is less convincing. A new measure may also capture a slightly different part of the construct, so a moderate correlation does not automatically mean it is useless.

Ecological and temporal validity

Ecological validity and temporal validity are about whether findings travel beyond the original study. They are types of external validity: they ask whether the conclusion can be generalised.

Ecological validity is not just "was the study in a laboratory?" A laboratory study can have reasonable ecological validity if the psychological process being measured is the same as in everyday life. A field study can still have low ecological validity if the task is unusual or the participants know they are being observed and behave differently.

For example, a study of helping behaviour may have lower ecological validity if participants press a button in a cubicle to indicate whether they would help a stranger. It may have higher ecological validity if researchers observe actual helping behaviour in a public place, as long as ethical issues are handled appropriately.

Researchers can measure ecological validity by checking whether findings are replicated in more natural settings, with more realistic tasks, or across different real-world contexts. They can also compare the study procedure with the everyday situation it is meant to represent.

Temporal validity matters because social norms, technology, education, family life and cultural attitudes change. A study of conformity from the 1950s might not generalise perfectly to people in the 2020s. A study of online relationships from the early 2000s may not apply well after smartphones, video calls and social media platforms became normal parts of daily life.