Lecture
A variable is:
1. any reality whose observable changes (in specific parameters or indicators of a technique) can be recorded and measured on some scale (Kornilova).
2. Any reality that can change, and this change manifests itself and is recorded in the experiment.
1) The independent variable (IV) – a manipulated variable, that is, one actively changed by the researcher; in other words – a functionally controlled variable; represented at two or more levels. In the experimental hypothesis it is understood as a causally acting factor.
2) The dependent variable (DV) – the «response», or the variable measured in the experiment, whose changes are causally conditioned by the action of the independent variable. In psychological research it is represented by indicators of the participant's activity, any forms of assessment of their subjective judgments and reports, psychophysiological parameters, and so on. Another name – the «measured variable».
3) Additional variables (also extraneous variables – they influence the conditions of the experiment, but we cannot control them): «third» variables, among other things, set or limit the breadth of the extension of the conclusions from the experiment to other situations, in which a new level of the third variable will cause a change in the relation between X and Y. Such variables enter into the hypothesis usually as conditions for which the dependency holds, and are called additional (AV). Their influence is not statistically evaluated, unless they are considered in factorial schemes as an independent variable in their own right. The level of the additional variable determines the possibility of subsequent generalizations to reality and is usually indicated in the experimental hypothesis. The researcher always chooses the criterion with respect to which the established dependency can be transferred to other kinds of reality. In doing so, the chosen levels of the additional variable will always limit these possibilities of transfer, but making them more demonstrable.
4) The basic variable – the independent variable exerts its action on it (according to Gottsdanker).
And according to Campbell (the basic extraneous variable): this is a variable of internal conditions (factors of between-individual differences) that is confounded with the main basic process under study (represented by the main basic variable).
5) The latent variable – a hypothetical variable that cannot be measured in the study, but that, in a model of the connections between variables, characterizes the source of the measured variables – unrefined influences («disturbing» factors) acting on the measured variable.
6) The accompanying variable – the one determining the confounding of the basic experimental effect of the experimental factor with the influence of the manner of presenting its conditions. The effect of the influence of its active level may be represented by the placebo effect. It is controlled by a general scheme of setting the active levels of the accompanying variable at all levels of the independent variable.
Three types of confound are distinguished:
1. Nonsystematic confounding arises when any of the factors (or their combinations) irregularly interferes with the dependency under study. The source of confounding variables connected with the influence of the time factor may be either internal causes (changes in the states of the participant themselves, background fluctuations of the dependent-variable indicator) or external ones (a chance distraction by noise in the corridor, a colleague's shout, a telephone call, and so on). If they have been distributed unevenly across the compared conditions of the independent variable, then a distortion of the experimental effect (as the difference in the dependent-variable indicators) will occur.
One of the consequences of such irregular influences of confounding variables is the unreliability of the data, that is, with a different scatter of the levels of the confounding variable – across trials over time – a different connection is established between the values of the dependent variable and the levels of the independent variable. Usually this threat to the conclusion about the experimental dependency is controlled in two opposite directions. On the one hand, the experimenter strives to reduce the number of trials in the overall experimental sequence to a minimum, in order to conduct the individual experiment in as short a time as possible, neutralizing the time factor. On the other hand, the experimenter must ensure a sufficiently large number of trials, that is, try to approach an infinite experiment, so that all the confounds with fluctuations on the part of confounding variables are distributed randomly – and in this sense equally – among the levels of the independent variable.
There are also other sources of unreliability of the data. Thus, there may be variability of the independent variable itself, when the experimenter considers trials to belong to the same level, but in fact in one or some of them something occurred that does not allow the conditions to be regarded as identical. In the example of physical stimuli, one may suppose fluctuations of voltage in the network; in the example of participants' acceptance of problem situations – changes in their subjective understanding or personal acceptance of the problem, and so on.
2. Systematic confounds – the main type of threat to internal validity.
- the confounding of the levels of the independent variable and the levels of the confounding variable, in which there exists a regularity in the combination of active and inactive levels of these variables; as a result, one cannot conclude that the experimental effect is connected with the action of the independent variable and not the confounding variable. The main factors threatening from the side of systematic confounds: experimenter effects, factors of tasks, time, sequence, and between-individual differences.
3. Accompanying confounds - The emergence of accompanying confounds is connected with the very manner of setting the independent variable, the recording of the dependent variable, or the actualization of the basic variable under study together with another basic process.
Accompanying confounds are distinguished by the immanent connection of the confounding variable with the methodological conditions of controlling the variables or with other variables.
Comments