What validity is

1. Validity is how far a study measures what it set out to measure.
2. A valid study gives results that reflect the behaviour being studied.
3. Validity is about accuracy, and a measure can be consistent from one run to the next without being accurate.
4. A bathroom scale that reads two kilograms heavy every time is reliable and not valid.

Internal and external validity

1. Internal validity.
   1. Whether the study measured what it set out to, with no other factor producing the results.
   2. Weak control of other variables is the usual reason internal validity drops.
2. External validity.
   1. Whether the findings apply beyond the study, to other people, settings and times.
   2. A narrow sample or an artificial setting is the usual reason external validity drops.

Ecological, population and construct validity

1. Ecological validity: how far the results apply to real, everyday situations.
   1. Learning unrelated words in a silent cubicle is a different task from remembering a conversation, so the findings may not transfer.
   2. Using a real setting and a task people would carry out anyway raises it.
2. Population validity: how far the results apply to people beyond the sample tested.
   1. One Year 10 class in one Cardiff school cannot represent all UK teenagers.
   2. The less representative the sample, the less far the findings can be generalised.
3. Construct validity: how well the measure captures the idea it claims to measure.
   1. Counting how many words someone recalls measures recall, which is one part of memory.
   2. A questionnaire that asks about sleep and worry may be measuring stress rather than the anxiety it names.
   3. A measure that rests on the researcher's judgement is subjective, so an objective measure such as a timed score captures the idea more cleanly.

Threats to validity

1. A threat to validity is anything that makes a study measure something other than what it set out to measure.
2. Most threats work by changing how a participant behaves, so the data no longer reflects natural behaviour.

Demand characteristics

1. Demand characteristics are cues that let participants read the aim, after which they act to fit what they think is wanted.
2. Some try to help by giving the result they assume the researcher is after, and others deliberately do the opposite.
3. Either way the behaviour recorded is a reaction to the study rather than the behaviour the study wanted.

Social desirability

1. Social desirability is answering in the way that makes a person look good rather than in the way that is true.
2. It bites hardest on sensitive topics such as drinking, revision hours or how kind someone is to others.
3. The answers drift towards the socially approved response, so self-report data overstates the flattering behaviour.

The observer effect

1. The observer effect is behaviour changing because people know they are being watched.
2. Aware of the observer, participants act as they think they should, for example working harder or behaving more politely than usual.
3. It concerns what people do, while social desirability concerns what people say about themselves.

Investigator effects

1. Investigator effects are the researcher's tone, expression or choice of words leaking what they are hoping to find.
2. A warm nod at one answer and a frown at another steers the participant without a word being said.
3. The result then reflects the researcher's expectations as much as the participant's true response.

Extraneous and confounding variables

1. An extraneous variable is any factor other than the one being tested that could affect the results.
2. A noisy corridor, the time of day or brighter lighting in one condition are everyday examples.
3. Standardising the procedure, so every participant meets identical conditions apart from the independent variable, controls most of them.
4. A factor that varies systematically with the conditions becomes a confounding variable, an uncontrolled factor that changes alongside the independent variable.
5. The difference then cannot be put down to the independent variable rather than the confound, so validity falls.

Spotting a threat in a described study

1. A study whose measure does not match the behaviour it names has a construct validity problem.
2. A procedure that gives the aim away invites demand characteristics.
3. A question asked face to face about a sensitive topic invites social desirability.
4. A participant who knows an observer is present may show the observer effect.
5. Anything besides the tested variable that differed between the conditions is a possible confounding variable.

Turning this into marks

1. Name the type of validity at risk and then say which way the results are pushed, keeping the two as separate points.
2. Point at the feature of the procedure that causes the threat, because naming demand characteristics on its own does not secure the knowledge mark.
3. Give the threats their own names, so investigator effects belong to the researcher and social desirability belongs to what the participant reports.
4. The mark most often lost is filing a threat to validity under reliability, when consistency is a separate question from accuracy.