PYB210 Lecture Three Notes
Reliability and Validity
Construct validity is the extent to which the operational IV and DV accurately reflect the factors
(concept of the experiment) they are intended to represent.
o For example, in Doug’s computer anxiety experiment, the concept was to explore the effect
of social learning on computer anxiety. Doug operationalised the social learning IV in the
form of either a positive of negative attitude instructor. However, there is a problem – does
a positive or negative attitude instructor equate to a computer anxious or non-anxious
instructor? No, it doesn’t. Therefore, the operationalisation of the experiment lacked
o Tests of construct validity are best conducted in a pilot study before commencing the actual
Method of measuring the construct validity in the Independent Variable:
Manipulation check – one way to check construct validity within the IV is a manipulation check. This
could, for example, involve asking the participants after the experiment how they felt about or rated
the IV. This will tell us if the IV was behaving as it should have been during the experiment.
Method of measuring the construct validity in the Dependent Variable:
Convergent validity – this is where we may test to see if there is a correlation between the DV and a
conceptually similar construct. A successful determination of convergent validity would show high
levels of correlation.
Divergent validity – this essentially works in the opposite way that convergent validity is tests. Here,
we measure the direction of our DV against the direction of a construct which is conceptually
different. A successful determination of divergent validity would show no correlation between the
In order to establish the reliability of our DV we need to ensure that the same results can be obtained
again and again. A good way of doing this is to utilise a repeated measures design in which the DV is
measured on several different occasions over time to determine consistency and thus reliability.
Hint: Epstein (1979) could possibly be a good reference for the reliability of an experiment.
Extraneous variables are essentially anything other than the IV which is capable of causing a change
in the DV. When conducting an experiment, we want to ensure that these potential extraneous
variables do not interrupt or in any way affect our experiment.
If these extraneous variables vary systemically with the IV then there is no way of saying whether it
was the IV or the extraneous variable which caused the change in the DV.
In this case, the EV is called a confounding variable because it literally confounds us. This confounding
variable needs to thus be controlled.
Often we may have extraneous variables which do not actually confound the IV. While this is good, it
does not necessarily mean that the EV has no effect on the experiment. This is because even though it
is not confounding, it still may affect the external validity of the experiment.
Remember – all confounding variables ARE extraneous variables, however not all extraneous
variables become confounding.
Types of Extraneous Variables to Look Out For:
History variables – this refers to the variation in the events experienced by different groups between
o E.g. women are asked about attitudes towards terrorism on the 10 of September 2001 and
men are asked on the 12 September 2001. We cannot attribute the difference to gender
because of the event – September 11 – that happened in between. Maturation – this refers to the natural changes that occur in a person across their lifespan in between
o E.g. in a study to see if certain lenses helped dyslexic first-graders read better, the study
found that six months later they indeed had better reading ability however a later study with
a control group found no difference. Here, maturation is a likely extraneous variable which
accounts for the increased reading ability.
Instrumentation variables – this refers to the systematic changes in apparatus between conditions.
o For example, in Doug’s font size experiment, the 10pt font could be shown first whilst the
projector is still dimmed. The 12pt is then show afterwards, when the