Lecture
Isolating any IV does not rule out the possibility that the basic process is influenced by random and non-randomly accompanying changes in other variables that are not part of the hypothesized relationship between the IV and the DV. These are confounding (extraneous) variables.
An extraneous variable is a variable that is not part of the experimental hypothesis but that, under the conditions of the experiment, may either influence the basic process under study or become confounded with the IV or the DV, which will distort the form of the relationship being investigated.
The greatest threats to internal validity arise from systematic and non-systematic confounding with extraneous variables.
Internal validity is a criterion for evaluating the planning and conduct of a study that makes it possible to regard as reliable the conclusions concerning precisely that (and not some other) relationship which is formulated in the experimental hypothesis.
Systematic confounding is a non-random mixing of the levels of the IV and an extraneous variable, when a particular level of the IV is regularly combined with the active or inactive level of the extraneous variable, as a result of which one cannot conclude that the established experimental effect is connected with the action of precisely the IV and not the extraneous variable.
Experimental control must include the stabilization of other confounding variables or their random variation across levels, in order to exclude a correlation of the IV with the extraneous variable.
Control of the confounding of the IV with extraneous variables is the principal task of planning in the development of experimental schemes.
Extraneous variables are usually controlled by means of their stabilization or the averaging of the influences they exert, or by other methods that are usually determined by the chosen experimental scheme (for example, this may be the setting of an extraneous variable at all levels of the IV, as in the experiment with the little monkeys).
Non-systematic confounding….
arises when any of the extraneous variables or their combinations irregularly intervene in the relationship under study.
Accompanying confounding….
Comments