It is the simplest and cheapest experimental design. It is usually used when the hypothesis is exploratory. Only one independent variable with two values is used, one of which is usually the absence of treatment. This design consists of two groups: An experimental one, to which the One control treatment to which it is not applied (or a placebo is applied) It is also possible to use two nonzero values of the independent variable and thus have two experimental groups.
The unifactorial designs are characterized by studying the influence of a single independent variable on a dependent variable in two or more equivalent groups. The researcher only manipulates one independent variable, which must have at least two values.
Classification of one-factor designs of random groups
Of two groups
- With post-treatment measure only
- With pre and post treatment measure
Multi-group
- With post-treatment measure only
- With Solomon pre and post treatment measure
Both the assignment of the subjects to the groups and their assignment to the treatments must be done randomly to guarantee the equivalence of the groups. The logic of two random group designs is based on this initial equivalence: if the groups are the same before treatment, the differences found after treatment will be due to the effect of this.
For the groups to be equivalent, in addition to the random assignment of the subjects to the groups, it is necessary that the sample be large enough for chance to act. classification In two-group randomized designs, the measure of the dependent variable can be taken before and after after treatment or only afterwards, resulting in these designs being subdivided into following:
- Two-group randomized design with only post-treatment measures
- Only one measure of the dependent variable is taken from each group of subjects after application of the treatment.
This design consists of two groups formed at random:
- Treatment is applied to one of them (experimental group)
- Treatment is not applied to another (control group)
- A non-zero value of the independent variable can also be applied to each of the groups, giving rise to two experimental groups.
The symbolic representation of the design of two random groups with post-treatment measures and control group would be the following: See attached image.
Provided that the initial equivalence of the groups is guaranteed, it is better to use this design than that of the random groups with pre- and post-treatment, since taking a pre-treatment measure can produce a sensitization of the subjects to the pre-treatment media and distort the results.
The process To follow to carry out this design would be: Select from the population of interest a sample of subjects large enough for chance to act. It is not necessary that the selection of this sample be done randomly, although doing so increases the external validity.
Randomly assign the subjects to the two groups or conditions. We also randomly assign these groups to the two values of the independent variable. Apply the treatment and measure the behavior of the subjects in the two groups. Compare the results of the two groups using the most appropriate data analysis technique. Draw the conclusions, generalize the results and write the research report.
The advantage of this design are:
The random assignment of the groups guarantees the equality of the subjects before the treatment.
Great control is exercised over threats to the internal validity of history and maturation, since that very little time elapses between the application of the treatment and the measurement of the behavior of the subject.
Nor can the pretreatment measure be sensitized (since it does not exist) or statistical regression, since the subjects have been randomly assigned to the groups. The threats to the validity of this design are: Threats to internal validity: Instrumentation, if we use different devices to measure behavior in the two groups Differential selection, if the sample of subjects is small or the random assignment of the subjects to the groups.
Threats to external validity: Interaction of selection and treatment, if the samples are not random and therefore not representative of the population, preventing the generalization of the results to it. Reactive effects of the experimental devices, due to the artificiality of the experimental situation. Two-group randomized design with pre- and post-treatment measures
In this design, once the groups are formed, a measure of the dependent variable or of a variable closely related to it is taken before administering the experimental conditions or treatments. This measure is called a pre-treatment measure.
The purpose of the pretreatment measure is to verify that the groups are equivalent in the variable under study and, in this way, to be able to attribute the differences or equality found between groups in post-treatment measures for the effect of the variable Independent. The design of two randomized groups with pre- and post-treatment measures has the same characteristics as the design above, except that two measurements are taken in each group of subjects: one before and one after the application of the treatment.
In the following table we see the symbolic representation of this design when two different levels of the independent variable other than zero (2 experimental groups), although there could also be an experimental and another of control.
The process To carry out this design is the same as the previous one except in the following points: When the groups are already formed, takes a measure of the dependent variable or another variable closely related to it in the two groups of subjects and is checked, by the adequate statistician, if there are differences in the pretreatment measures of the two groups. If there are none, continue with the design. If there are differences, the blocking technique can be used to assign subjects to groups or certain statistical techniques such as analysis of covariance to control the effect of that variable strange. Treatments are randomly assigned to each group and the treatment is applied.
The behavior of the subject under the effect of the treatment is measured in the two groups. Data analyzes corresponding to this design are performed. In addition to comparing the two pretreatment means to verify the equivalence of the groups, other comparisons must be made:
- To see the influence of the treatments within each group, we compared O1 with O2 and O3 with O4 using a contrast statistic of mean differences.
- To check if the hypothesis has been fulfilled, it is necessary to compare the post-treatment measures (O2 and O4) of the two groups using a parametric or non-parametric hypothesis test statistic for two tights.
The advantage of this design are: Thanks to the pretreatment measure, we can study whether the two groups of subjects are equivalent. Almost all threats to internal validity can be controlled. Between the drawbacks They are: If the measuring instruments and the experimenter are different in each group, threats to the validity of the experimenter effect and the instrumentation may arise.
Among the threats to validity are:
Threats to internal validity: Pre-measure awareness, which consists of the subjects being able to become familiar with the type of tasks, guessing the research objectives, etc. influencing their responses and, therefore, the results of the experiment. Statistical regression when the scores in the pre measure are very extreme, but since the groups are equivalent, the same probabilities are that these threats influence the dependent variable of the two groups.
Threats to external validity:
Interaction between pre-measure and treatment. This interaction occurs when the effect produced by the treatment depends on the sensitization produced by the pretest in the subjects. Interaction between selection and treatment can occur to the extent that the sample is or is not representative of the population to which the results are intended to be generalized. Artificiality of the experimental situation.
This article is merely informative, in Psychology-Online we do not have the power to make a diagnosis or recommend a treatment. We invite you to go to a psychologist to treat your particular case.
