Primary Control Schemes - General Psychological Practice

Primary Control Schemes

There are different approaches to how equivalent groups can be created. The main methods used here are random distribution and equalization.

One can not arbitrarily treat the attribution of subjects to different groups, since this threatens the appearance of different types of mixing. For example, if the experimenter has given an announcement about the search for participants in the study and will subsequently determine the first who came in one group, and those who came later - in another, then the groups are likely to be nonequivalent with a high probability. It is possible that subjects responding earlier have some similar qualities (for example, performance or anxiety) that those who came to research later will not have. In this case, according to these characteristics, the groups will differ, which may affect the results. Or if the experimenter decides which group to relate to the subject, after acquaintance with him, it is possible that the researcher will involuntarily determine in one group (for example, the experimental one) those who will appear to him more successful in the studied activity, which will also make groups are not equivalent.

The distribution of subjects in groups should be done according to predetermined rules that would level out the impact of such difficult to detect and control factors. Each participant must have a equal chance to be in any of the experimental conditions. Ideally, this should ensure a uniform distribution of all possible individual characteristics across all groups, and therefore, neutralize their influence on the characteristics under study.

If you use a random distribution to create equivalent groups, you usually do not resort to a random number table, tossing a coin or throwing dice, since these actions do not guarantee that an equal number of subjects will fall into all conditions.

With this, one can cope with such a method as block randomization. It helps to ensure that the number of participants in each group remains the same, and that the subjects are randomly assigned to the groups. The idea is that at first one subject is randomly assigned to each group, then - the second, etc., so that the second subject does not get into any group until one person is identified in each group.

To use block randomization, it is necessary to prepare a scheme in advance, according to which each next subject will be attributed to the condition. One block in this case will be the number of subjects needed to ensure that each group has one person each. If, for example, five conditions are studied in the experiment, five people will participate in one block. Each condition should be numbered or denoted by letters (for example, A, B , C, D, E). Then you need to mix the conditions randomly, for which you can use a random number table or a computer program, or you can take small sheets of paper, each of which will be written one of the letters, mix them and pull them one at a time.

Suppose that as a result of any of these procedures, the following sequence is obtained: A, B, E, D, C. This means that the first visitor the study of the subject must be determined by the condition A, the second by the condition B , the third by the condition E . When each group is typed but one person, the block will be closed. Next, you need to re-mix the conditions, you will get a new sequence (say D, A, C, B , E). Thus, the sixth subject needs to be determined on the condition D of the seventh - on condition A, etc.

The most convenient way is to create a table with a random order of conditions for the desired number of blocks. Each row of such a table will be one block, and it will contain a sequence in which each next subject should be assigned to the group. You can calculate how many blocks you need (according to how many subjects in each group you want to see), and make the appropriate number of rows in the table. After the participant took part in the experiment, you can delete the corresponding letter in the appropriate line to avoid confusion.

Block randomization is very suitable for cases where you do not know the full list of all subjects in advance.

If a complete list of all the subjects is known to you in advance (for example, if you conduct a study on a group of schoolchildren, or if the subjects registered in advance to participate in the experiment), you can determine in advance which of them will be assigned to which condition. To do this, you can simply take a list of participants, mix their names randomly (using, for example, the function of generating pseudo-random numbers RAND in the application MS Excel), and then simply divide them in accordance with the mixed list for the necessary number of equal groups, assigning the first part to one condition, the second to the second, and so on.

If the subjects participate in your research simultaneously and at the same time come to the experiment, you can invite everyone to draw lots - a sheet with the designation of the group to which it is defined,

Prepare the necessary number of leaves, indicating the desired number of groups.

So, the concrete realization of a random distribution of subjects may depend on the conditions under which you are an experimenter. The main thing is that eventually it will create equivalent groups.

However, it is important to understand that the use of random processes does not necessarily lead to the desired result. Random processes are random and random, which can lead to any outcome. So, as a result of the simple application of random numbers to create equivalent groups, it may happen that groups are not equivalent. This method is very effective, and its effectiveness increases with the increase in the number of subjects. The more samples there are, the more likely the groups created in this way will be equivalent.

In order to cope with these difficulties, you can use another way to create equivalent groups - equalization. The idea underlying this method, in general terms, is as follows. If we take variables strongly related to the dependent variable, i.e. such that the differences in which the subjects can most significantly affect the results of the experiment, and put each subject from one group of a subject from another group with the same level of expression of these variables, then there will be a situation similar to the intra-entity plan: if one the subject participates in different experimental conditions, since there will be no individual differences in the important secondary variable for a given pair of subjects. Variables that are taken into account in this process are called adjustment variables.

If the number of subjects in the study is not large enough for the random distribution to work reliably, then the adjustment is a more successful method.

Of course, the procedure is more complicated. Usually it is necessary to carry out the measurement of the adjustment variables in advance for a larger sample of the test subjects, the results of which will be used to compile the groups, and only then to carry out the main part of the experiment.

For example, in the study of methods for memorizing poems, memory characteristics are important, since subjects with poor memory will in any case learn rhymes more slowly than subjects with good memory. You can first test the effectiveness of memorization on some other material (for example, use word lists), select subjects in different conditions based on these measurements.

Ideally, each subject of one group should be assigned one subject from another group (or other groups) with exactly the same degree of equalization - roughly speaking, who scored an identical score in the preliminary testing. Of course, most often this is not possible, and in pairs just select people who scored in the preliminary testing the closest points to each other.

Equalization is carried out as follows. All potential participants are measured by the variable chosen for the adjustment. Further, all the subjects are ordered according to the increase in the indices obtained with this measurement. After that, depending on the required number of groups, pairs of subjects (or triples, if three levels of the independent variable are examined, or four, if there are four levels, etc.) are selected with the lowest values ​​of the measured characteristic, and each of them is randomly assigned to one from the experimental conditions. Further, the subjects with more expressed indices of this characteristic, which are similar to each other, are selected and again are randomly determined in the conditions. This happens until the norm, until in each group is not typed the required number of subjects.

Suppose, to study the influence of the method of formulating the conditions of the problem (detailed verbal description or schematic representation) on the speed of its solution, you want to equalize the subjects in groups by the intelligence factor. You conducted a preliminary test using the Amthauer test and transferred the raw scores to the IQ scores. Table. 12.4 shows the results obtained.

Table 124

Points IQ, measured in all subjects in the preliminary

Testing

Subjects

Score IQ

1

105

2

110

3

122

4

113

5

120

6

112

7

98

8

111

9

115

10

102

After ordering the points IQ in ascending order, we get the sequence shown in Table. 12.5.

Table 12.5

The list of subjects located in accordance with the increase in the score IQ

Subjects

Score IQ

Test pairs

7

98

First pair

2

100

10

102

The second pair

End of the table. 12.5

Subjects

Score IQ

Test pairs

1

105

8

111

Third pair

6

112

4

FROM

Fourth pair

9

115

5

120

Fifth pair

3

122

The lowest scores were scored by subjects 7 and 2, this will be the first pair of subjects. We assign randomly one of them to the first condition, and the other to the second. The next two subjects - 10 and 1 - scored a slightly higher score; this will be the second pair. Similarly, we randomly assign each of them to one of two groups. In the same way we will do the rest of the subjects. As a result, the following filling of groups can result (Table 12.6).

Table 12.6

Subject division by groups based on adjustment

points IQ

Group 1

Group 2

Subjects

Score IQ

No. of the examinee

Score IQ

7

98

2

100

1

105

10

102

8

111

6

112

4

FROM

9

115

3

122

5

120

Average score IQ

109.8

Average score IQ

109.8

Thus, each subject in one research group has a "similar" by the key parameter the subject in the other group, and the average degree of variable adjustment in both groups is almost the same.

If more than two levels of the independent variable are studied in the study, then the equalization procedure turns out to be similar, except that the subjects should be selected not in pairs, but in triples, quads, etc.

As a result of such a procedure, we can accept the assumption that there are no individual differences in the equalized characteristic of the groups being compared.

In order for the adjustment to be effective, it is necessary that two basic conditions be met.

First, the adjustment variable must be adequately matched. This should be the sign that correlates with the dependent variable and which can be easily reflected in the results of the experiment. Usually this is just one variable, because the more variables you try to call, the more difficult it will be for you to find the corresponding subjects for each of these parameters. For the adjustment, the most important and most important characteristic should be selected. It is not necessary to try to equalize by such features, which are weakly related to the dependent variable. For example, in the above example, it is correct to try to equalize the subjects according to the level of intellectual development and it is completely pointless - by the amount of monthly earnings.

Secondly, there must be reliable and valid methods that allow you to measure the equalized characteristic. Otherwise, all your efforts will be meaningless. If you do not know how to correctly measure the characteristic by which you want to equalize the subjects, or doubt that the method you selected for measuring it is good enough, you should abandon the adjustment in general. Otherwise, it will take a lot of time, and the equivalence of groups will not be achieved.

Carrying out the equalization procedure complicates the organization of the study, since the subjects must take part in an additional stage of measurements. To facilitate the work of the subjects and themselves, it is best, when possible, to use some indicators that can be obtained without the direct participation of the test subjects. For example, indicators of achievement that can be requested from teachers at school, socio-demographic characteristics, which can be elucidated by indirect evidence, etc.

As you can see, the use of the adjustment procedure turns out to be quite complicated and can not lead to the desired result, therefore researchers often prefer to increase the sample and use a simple random distribution.

There are other methods for creating equivalent groups. For example, you can use a method that is somewhat average between the use of random distribution and equalization. It is possible to check the adjustment variable so that it is approximately the same for all subjects who participate in the study.

So, if you want to equalize the intelligence groups, after the preliminary testing you can choose to participate only the subjects with the mean values ​​ IQ in 100-110 points, and all the rest simply do not take in the main stage of work. Then you can distribute the people so selected in groups randomly. This method is convenient, but it must be understood that the data obtained in this study can not be generalized to the whole population, the results obtained will be applicable only to those individuals who have similar indicators of the secondary variable. On the other hand, having obtained the desired effect on this group of subjects, you can conduct similar studies with the same selection method, but using higher and lower values ​​of the secondary variable (for example, separately for IQ within the limits of 130-140 points and separately for IQ within 90-100 points). Thus, you can understand how much the effect is characteristic of the whole population.

Another way to create equivalent groups is to use stratification, or clustering. In this case, the ranges (strata) of the variable adjustment are initially allocated, and the same number of subjects from each of these ranges is recruited to different experimental groups. For example, you can type in 20 experimental groups of 20 people with IQ within 130-140 points, but 20 people with IQ within 100-110 points and 20 people with IQ within the range of 90-100 points. In each group, people from different strata of the equalized variable will be equally represented. This as a whole will equalize the groups for the controlled parameter and make them equivalent without performing a pairing procedure.

Most often this method is used when socio-demographic characteristics (for example, gender, age, income level, etc.) are subject to control.

So, in the intergroup experiments, each level of the independent variable is presented to different groups of subjects. At the same time, the main threat to internal validity is the non-equivalence of the groups under study, and to overcome it it is necessary to use various methods to neutralize the effect of individual differences of subjects falling under different experimental conditions, such as random distribution or equalization.

thematic pictures

Also We Can Offer!

Other services that we offer

If you don’t see the necessary subject, paper type, or topic in our list of available services and examples, don’t worry! We have a number of other academic disciplines to suit the needs of anyone who visits this website looking for help.

How to ...

We made your life easier with putting together a big number of articles and guidelines on how to plan and write different types of assignments (Essay, Research Paper, Dissertation etc)