One of the most common definitions of validity relates this concept to the answer to the question of how well the research procedure allows to study the subject for which it is used, or how well the research conclusions are justified by the procedures for obtaining and processing data.

Validity is high when the independent and dependent variables are best operationalized, accurately measured, no confounding variables affect the dependent variable, all measurements are correctly statistically processed, and conclusions are made considering all additional variables. Low validity means that the results obtained do not contain any information about what the researcher wanted to study, and if they contain information, then about something else.

As a joke example of a reliable, but absolutely disabled research, you can cite an anecdote about a cockroach. The researcher assumes that the cockroach has ears on its feet. If you put a cockroach on a table and clap your hands, the cockroach will run away. If you tear off the cockroach's leg, put it on the table and clap your hands, the cockroach will also run away, but more slowly. According to the logic of the researcher, this is because the cockroach has become worse to hear. If you tear off two legs of a cockroach, he will run away even more slowly, apparently because he began to hear even worse. If you tear off all the paws, then the cockroach will not run away at all, but the same logic - because it does not hear the pops. Such an "experiment" is absolutely reliable - it will be reproduced in 100% of cases, by any experimenters on any arthropods and not only arthropods. But it is absolutely not valid, since the researcher did not have the right to draw conclusions about the cockroach's hearing, while actually affecting its motor organs.

The concept of validity applies not only to an experiment, but to any research procedure. For example, it happens that, having filled out a psychological test, people are perplexed: the result obtained does not at all correspond to their ideas about themselves. Then they turn to the test questions and ask themselves whether, for example, such actions as "going crazy", "talking to myself", "running or walking", "teasing someone" really help. overcome stress? Does it really mean that if they don't do all these things, they don't know how to overcome their stress and can't cope with it? Such questions raise doubts about the validity of the test.

Depending on the violation committed by the researcher when obtaining or evaluating data and, accordingly, when formulating conclusions, different types validity. Some violations lead to the fact that it is impossible to draw any conclusions about the hypothesis from the data - neither confirming nor refuting it. Other violations only impose restrictions on these conclusions or areas and ways of applying them.

In the following, we will discuss four types of validity: internal, external, construct, and statistical. We will also discuss the threats associated with their violation, recommendations for avoiding them, and specific findings from the study when each type of validity is threatened.

Internal validity

Internal validity reflects how much the results of the measurement of the dependent variable are due to a change in the independent variable, and not to some other uncontrolled factor. Such an uncontrollable factor, as noted above, is called a side variable.

Thus, the more side variables the researcher thought through and tried to avoid, the higher the internal validity, i.e. the higher the confidence in the conclusions of the study about the reasons for the change in the dependent variable. If the hypothesis of a valid experiment is not confirmed (confirmation of the counter-hypothesis), the researcher can be quite sure that

that the independent variable being tested in his study does not affect the dependent variable.

For example, in psychological research in a psychiatric clinic, it is very common for a secondary variable that threatens internal validity to be the drugs the patients are taking. Let's say a researcher wants to find out if schizophrenia affects attention productivity. To do this, he recruits three groups for comparison: healthy people, patients with a short history of the disease, and patients with many years of experience. Thus, his independent variable is the experience of schizophrenia, which has three levels: zero experience, small and long-term. The dependent variable is the success of the attention test.

In order to prove that the disease of schizophrenia affects the productivity of attention. the researcher must show that the three groups are very different on the attention test. Suppose the results of the study show that this is the case: healthy people perform the test most successfully, patients with little experience are significantly less successful, and patients with a long history of the disease are even less successful. Does this mean that schizophrenia affects how attention works? Not necessary. Perhaps it is not the disease that provokes the deterioration in the attention of the subjects, but the medications they take. Healthy subjects perform well on the test, not because they are healthy, but because they do not take psychotropic drugs. Short-term patients perform worse on the test, not because they are ill, but because they are taking such drugs, but they do better than long-term patients because they take weaker drugs and at lower dosages.

As a result, the researcher does not know the exact cause of the deterioration in the attention of the subjects: the disease of schizophrenia or the effect of drugs. Both factors with equally likely can be such a cause, since they act simultaneously and in the same way.

AT this example the effect of drugs is an incidental variable not controlled by the investigator. The researcher cannot draw any conclusions about his hypothesis. He cannot claim that the hypothesis about the impact of schizophrenia on attention productivity has been confirmed, since it may not be the disease of schizophrenia that affects, but the medications taken for this disease. But he also cannot claim that the hypothesis was not confirmed, and the disease of schizophrenia does not affect the productivity of attention, because it is possible that the drugs still have nothing to do with it, and it is the disease that affects the productivity of attention.

Since the conclusions about the hypothesis in this study it is impossible to do, the value of the results obtained in it tends to zero, but because of the low internal validity of the experiment, which was the result of a systematic mixing of the effects of a variable considered as an independent variable (experience of the disease) and a side variable (the effect of drugs).

Side variables are especially troublesome in that it is often impossible to predict in advance which feature of the subjects or which circumstance may affect the dependent variable instead of or together with it. The researcher may not even be aware of the presence of such a circumstance. Therefore, it is impossible to develop a system of measures, the observance of which would guarantee protection against side variables.

To increase the internal validity of a study, two things can be done first. First, the literature on the research topic should be examined in detail in order to identify factors that may affect the dependent variable as well as the independent variable. Second, the known and most common threats to internal validity need to be controlled. Following Campbell, there are usually seven threats to internal validity, which are different sources of side effects on the dependent variable. Let's consider them in a little more detail.

Influence of history (background). Recall that the general structure of the experiment necessarily includes a preliminary measurement of the dependent variable, the impact on the measured variable from the independent variable, and the final measurement. A big problem what the researcher does is to eliminate any side effects on the dependent variable between two measurements, other than the effect on the independent variable. Such influences are called the influence of history or the influence of the background. The subjects cannot be isolated from the outside world for the period of the study, they are exposed to many influences, a significant part of which the researcher does not even suspect. And if one or more of these events affect the state of the dependent variable, then the internal validity of the study will be violated, since the change will not be caused by the independent variable, but by the side variable.

Since it is impossible to eliminate all exposures to subjects, researchers use a control group of subjects to control for the effects of history. The subjects in the control group are subjected to the same measurements (preliminary and final) as the subjects in the experimental group, but do not receive the influence of the independent variable.

Thus, the researcher first assesses whether the expected effect occurred in experimental group- whether the results of preliminary and final measurements differ. If the results differ, then some event affected the dependent variable between these measurements. After that, the researcher compares the results of the experimental and control groups. If these results differ, then the independent variable is responsible for this difference, since it is assumed that any other factors influenced both groups, and the independent variable - only the subjects of the experimental group. In the event that some uncontrolled event became the cause of the change in the dependent variable, then the changes between the first and second measurements in the experimental and control group will not differ significantly.

For example, the researcher is interested in whether good mood on the ability to solve problems. He fixes the severity of different emotions in the subjects, for example, using the "Health. Activity. Mood" questionnaire, and asks them to solve several problems, then shows them a funny video about animals, then again measures the emotions of the subjects and again asks them to solve problems. The results show that subjects are better at problem solving after watching the video. But the quality of problem solving could be affected not only by video, but also by any other factors - training, the sounds of conversations from the corridor, the desire to show oneself from the best side. Therefore, the researcher needs to recruit a control group that will solve the same problems, hear the same noise from the corridor, strive to positive evaluation and in the same way will be interrupted for a while, corresponding to the duration of the video, but will not view it. Only if other equal conditions the subjects of the experimental group will show the best increase in solving problems, the researcher has the right to conclude that his hypothesis was confirmed. If the results of the experimental and control groups do not differ significantly, it means that it was not a good mood that led to changes in the quality of problem solving, but some other factors.

It has a similar effect natural development a person, which is otherwise called the effect of growing up. If the experiment is long enough or falls on such moments of a person's life when maturation proceeds faster, then the researcher has every reason to doubt for what reason the dependent variable has changed: as a result of the influence of the independent variable or as a result of the natural development and maturation of the studied mental function.

Unlike the story effect, a side variable in this case is not some specific event that affected the dependent variable, but the changes in the subjects due to the passage of time. So, for example, in developmental psychology, periods of rapid and fairly global changes in the personality and psyche are known, which are called crises; there are more changes in childhood than in adults.

Certain circumstances operate similarly, such as adaptation to new conditions. For example, students in the first months of training in any educational institution, an employee recently hired to a new job, a patient who has been hospitalized, a conscript in the army adapt to a new place, regime, rules of conduct. The difficulties that they experience at first fade over time, and it is difficult to name a specific event that leads to this. It's just that a person has changed, adapted to new circumstances and rules.

Campbell also refers to less global changes as factors of growing up, such as changing physical condition due to fatigue, hunger, satiety with the activities carried out, etc. Like growing up, these factors affect the performance in general, the functioning of all psychological processes, and it is rather difficult to establish the moment of the beginning of their influence, since all people have fatigue, hunger, etc. arise at different speeds, and the people themselves, being busy with an experimental task, may not notice them immediately.

The maturation effect violates internal validity, since it is not known whether the experimental exposure or the change in the subjects themselves during the experiment affected the dependent variable. Resolving the question of which of these factors turned out to be decisive again helps. For example, evaluating a workplace coping program suggests that we measure the level of mental tension and stress in people who have just joined an organization. After that, we work with half of them under the adaptation program, and half adapt on their own, without the help of a psychologist. If, at the end of the program, both groups show the same decrease in the level of mental tension and stress, then the program does not work, and the decrease in negative experiences is associated with the effect of growing up. Only if the experimental group working on special program, will show a greater reduction in negative experiences, credit can be attributed to the effectiveness of the program.

The effect of testing is the influence of the fact of participation in the study on the results of the study. Often people want to show their best side, and if, for example, they are told that their learning ability is being studied, they mobilize and try to do all the tasks as best as possible. Therefore, in the experiment, they act as effectively as they never act in everyday life.

Especially often, the testing effect occurs in situations of control and evaluation: when hiring, psychological selection of personnel, evaluating the effectiveness of employees, comparing them in terms of some psychological qualities that are important for the area in which they work, during psychological examination in a clinic, in court, etc. As a consequence, the researcher is faced with the question of why the subjects showed certain results: because the experimental effect is so effective, or because the subjects tried to show their best side?

Another source of testing effect is the researcher's use of unusual procedures and tasks. So, potential subjects are already used to questionnaires, questionnaires, interviews, tests, they come across them on the pages of magazines, at public events, on the Internet, they see them on TV. If the researcher uses more complex procedures, gives unusual tasks, then the very fact of performing such a task or a series of tasks can lead to a change in the dependent variable. Examples of such tasks might be a request to complete habitual action in an unusual way, or to compose a human figure from a limited set of simple geometric shapes, to solve a tricky problem that is not similar to those that the subject has ever solved, etc. In this case, the dependent variable will change not as a result of the experimental treatment, but because of the unusualness of the pre- and post-testing or the unusual modes of treatment, and any other equally unusual treatment will lead to the same effect.

Using a control group to eliminate this effect does not always help. Pre-testing and post-testing are carried out in both groups, and this fact can lead to the fact that the results in the control group will change in the same way as in the experimental group under the effect of testing. But if this effect accompanies the experimental effect, then it is absent in the control group, and the subjects of the experimental group are faced with two inseparable effects, one of which is the experimental effect itself, and the second is the unfamiliarity, novelty of the method of influence. In this case, the control group will not help against the testing effect.

The way to combat the effect of testing is to pre-train the subjects. They are given many tasks, similar to those that will later be used in the study, in order for people to get used to them, gain experience in handling them, so that the effect of novelty or fear, uncertainty before the tasks with which they will be assessed disappears. Such training will make it possible to equalize all the subjects in terms of experience in handling research materials and, in the case of using new unfamiliar tools, will eliminate the influence of skill and speed of learning of the subjects.

However, this method of struggle does not allow assessing the presence or absence of the testing effect, as well as its strength. In order to determine if it affects this effect in the course of the study and how strong its influence is (should it be eliminated or it can be neglected), the study is built in accordance with the plan of Solomon, named after the American researcher R. Solomon. This plan involves the use of four groups of subjects - two experimental (EG1 and EG2) and two control (CG1 and CG2) (Table 11.1).

Table 11.1

Solomon's plan

Using this plan allows us to evaluate the effectiveness of the experimental impact in the presence of preliminary testing and in its absence, since the effect of testing is equally manifested in the experimental and control groups, and the influence of the independent variable is manifested only in the experimental group.

Instrumentation effect. The influence of tools (those procedures, specific methods and materials that the researcher uses in working with subjects) is a particularly painful problem in psychological research. Therefore, research problems related to the influence of instruments are given much attention when planning research. The problems associated with the influence of novelty and unusual research methods on internal validity have already been discussed above. A special concept is devoted to the discussion of specific distortions of the research results associated with the influence of the instrumentation - "construct validity", which will be discussed below.

Also, instrumentation issues form an important issue when discussing the reliability of a study, but reliability should not be confused with this kind of validity. A technique can be very robust in the sense that it produces very consistent results when reproduced, but does not actually measure what the researcher intended to measure if the internal validity is low, in part due to instrumentation effects.

The effect of the instrumentation and its influence on the level of internal validity of the study appear when the instrumentation is not accurate enough and allows for different interpretations of the same or similar results, if this instrumentation is changed, refined, calibrated in the course of the study or gives different results depending on the specific measurement conditions. . In this case, the side variable that threatens internal validity is measurement fluctuations. The researcher cannot make a definitive conclusion about why the dependent variable changed - as a result of experimental influence or as a result of instrumentation error, which periodically gave different data.

For example, if in the study of factors influencing perceptions of social justice, the researcher changed the questions of the questionnaire, the range or type of assessment scales, added and removed answer options, then there is nothing strange in the fact that the results of the preliminary and final measurements will differ. If the researcher gives instructions to the subjects in different ways, even without changing the essence, changes the way the same tasks are presented, gives additional comments on one measurement and does not give them on the other - all this can lead to differences in the data caused by non-experimental influence. , but by the influence of inaccuracies of the instrumentation.

To combat the toolkit effect, it is recommended to first use reliable, well-proven psychological techniques and procedures. The size of the sample of subjects is also important: the larger the sample, the smaller the impact of each individual error on final result. When the tools are new, it is necessary to formalize the research procedure as strictly as possible: formulate instructions to the subjects in advance, information about the goals and objectives of the study, clearly formulate all the actions that are required from the subject, so as not to deviate one step from the procedure. You should think in advance of several evasive answers to the questions of the subjects, such as "act as you think is right, there are no right or wrong answers." wrong reactions"," which wording (action) is closer to you personally, then write (and do) ", etc., so that the subjects do not differ in the amount of information about the study. These actions are directly aimed at reducing the number of measurement fluctuations that are the source of the effect tools.

Regression to the mean, or statistical regression in accordance with the definition of F. Galton, is the effect of returning the system to the mean state. The influence of the regression effect can be shown in the following example.

Imagine two students, one of whom spent the whole semester preparing for seminars, answered all questions and received high marks, and the second did not study anything and worked at half strength. When the time came for the final exam, the first student caught a cold, fell ill and, as a result, passed on a three, and the second student sat well at the books all night and got an five.

Both of these scores reflect the deviation from the average performance of each student. Will these grades adequately reflect the knowledge of both students? Of course not, since the first student, who studied throughout the semester and received fives, proved his high knowledge and, if he got into the exam on the next day or in a week, in healthy condition would definitely get an A. The second student who learned the material overnight will forget it just as quickly, and in a week he will have no chance of getting a high score again. The regression effect will already appear during the repeated assessment of students' knowledge, in which each of them will receive a grade that is closer to his usual (average) performance: an excellent student - a higher one, a three-year student - a lower one.

The same problem arises if we select for the study groups of subjects with extreme indicators (very high or very low) according to the results of a single measurement. For example, if the researcher believes that the creative abilities of the subjects can be enhanced with the help of jigsaw puzzles, he should measure the creativity of his subjects, select those who demonstrated low ability.

Suppose a repeated measurement will show more high results but the test of creativity in the experimental group. Perhaps this is the result of experimental influence. But it is possible that there was a regression to the mean. It is possible that the subjects selected for the study did not show very high creative abilities at the first measurement, because they did not have breakfast in the morning, fell ill, danced in the club all night, quarreled with relatives, received shocking news, etc. In fact, these people have much higher creative abilities, and if it were not for specific circumstances, they would have shown them in the test. Upon re-examination, these circumstances no longer affected the subjects, and they demonstrated their usual level of creative abilities. At the same time, the results of the repeated measurement did not increase under the influence of the independent variable (positive experiences of the subjects), but returned to the normal average result, which is usual for these subjects.

We repeat that the danger of regression to the mean is especially great in the case when subjects with extreme severity of the quality under study are selected for the study on the basis of a single measurement. If an equivalent control group is recruited, then it will show the same change in the dependent variable as in the experimental group, due to regression to the mean. A non-equivalent control group will not show the effect of regression to the mean (in the above example, a non-equivalent group will be assembled from subjects not only with low, but also with medium and high levels of creativity).

You can reduce the likelihood of regression if you select subjects on the basis of multiple measurements of the quality under study - for example, on the basis of the average level creativity but the results of running several equivalent subtests at different points in time. This is difficult and time-consuming, but in some cases, for example, when assessing school performance or if the subjects participate in psychological monitoring, multiple measurements of their performance and the qualities being monitored are available in sufficient quantity before the start of the study.

Finally, in those cases where the level of the dependent variable was below the mean before the experimental exposure and became higher after (and vice versa, it was above the mean, and after the exposure it became lower), the effect of statistical regression can be excluded, since the effect of regression to the mean does not imply a transition for average values.

The subject selection effect suggests that the change in the dependent variable can be affected not only by experimental exposure, but also by the specific features of the sample of subjects. This effect occurs when the experimental and control groups, for reasons beyond the control of the researcher, included subjects who differ in certain qualities.

For example, a teacher wants to conduct research on his students and promises to pay extra points. He makes an announcement about this at the beginning of the semester, before all tests are taken. The most anxious students who want to insure against failure, shortfall on the test, as well as the most anxious students will respond to this invitation first. motivated students who are interested in trying on themselves the methods of work of a psychologist. Later, closer to the middle of the semester, those students who realized that the subject is difficult for them, that there is a risk of not getting points for the desired grade, will catch up. The last students to take part are those who have already tried to write a test and realized that without additional points they will not pass the subject.

If the teacher collects all the students who came first into the experimental group, and all the students who came last into the control group, he will even get two side variables. The subjects of the experimental group will differ from the control group not only in the presence of experimental exposure, but also in higher anxiety and motivation to participate in the experiment. Therefore, the researcher will not be able to draw an unambiguous conclusion, because of which the dependent variable changed - as a result of experimental exposure or as a result of differences in the motivation of the subjects.

The selection effect affects not only by itself, it often interacts with other effects. So if the non-equivalence of groups is not itself a side variable, then the interaction of selection with other effects can act as the latter. The interaction of selection and the maturation effect occurs if people are selected into one group who develop faster or slower, get tired, begin to experience hunger, etc. than the subjects of another group. Let's say that the more anxious students were the first to volunteer to participate in the teacher's experiment on the formation of secondary stress as a result of watching TV news, and calm students were selected in the control group. Anxious students will more quickly fall under the influence of secondary stress, but not because the influence of the researcher is so effective, but because they panic faster when they meet with a stressor than calm students. In other words, it is not the influence of the researcher (watching TV news), but the selection of subjects, coupled with the effect of maturation, that leads to a change in the dependent variable.

The interaction of the selection effect with the history effect can manifest itself in the fact that the same event will have a different effect on the subjects of different groups. For example, the researcher is interested in whether the silent presence of the teacher affects the success of problem solving. Let's say that fans of rock music were selected randomly in the experimental group, and lovers of the classics were selected in the control group, i.e. there was a selection of subjects, which in itself would not have affected the outcome of the experiment. During the experiment, the subjects heard a few bars of a rock tune from the street (a random event that would affect members of the two groups differently). This will lead to positive impressions in the experimental group and no such impressions in the control group. As a result, it will not be possible to draw an unambiguous conclusion about what exactly influenced the change in the quality of the decision - the presence (absence) of the teacher or positive impressions caused by a random event (i.e., the combined effect of selection and history effects).

In the same way, the selection effect can interact with the testing effect if subjects are selected into one group who are more prone to fear of assessment and have less experience in participation in psychological research than in the other. The selection effect and the instrumentation effect can lead to a violation of validity if one of the groups is selected for subjects on whom the instrumentation gives fewer failures, and the other group includes more complex, critical or marginal subjects for whom the instrumentation gives more failures.

How the problem of non-equivalence of groups, which underlies the selection effect, arises and how it is solved, is discussed in more detail in the next chapter when discussing intersubject experimental designs. Here we emphasize the obvious: in order to control for violations of the internal validity of the study due to the selection problem and subsequent problems of the interaction of selection and other effects, it is necessary to equalize the experimental and control groups as much as possible.

The attrition effect (dropout) is such a threat to internal validity when subjects, for some reason, stop participating in the study between pre-test and final testing. As a result, the question arises before the researcher: is the change in the dependent variable from the first measurement to the second due to the action of the independent variable, or did the results of the second measurement change because some of the people dropped out, and the rest have some unknown to the researcher common psychological characteristics that influenced the result.

The situation when the subjects did not like the preliminary testing, they were offended, disappointed and left before the end of the experiment, happens quite rarely. But if enough time elapses between pre-test and post-test so that the subject and the researcher need to meet several times, then the threat of an attrition effect increases many times over.

In order for the exhaustion effect to occur, it is not necessary to conduct lengthy studies when the same subjects need to be found after a few months. A simple survey on junior schoolchildren it is impossible to carry out at one time if the methods require perseverance and attention from the subject for more than 40 minutes. To carry out all the methods, the researcher will have to meet with schoolchildren several times. And in this case, there is always a certain number of children who fell ill and for other reasons did not come to school (there was an exhaustion of the sample).

Children who pass all the methods of such a study are distinguished by at least better school attendance compared to those who dropped out. Better attendance may be related to health, higher motivation to study, better relationships with classmates, etc. These are all possible confounding variables that arise as a result of screening, are mixed with experimental influence and lead to a change in the dependent variable, as a result of which the researcher is unable to determine the true reasons for its change.

The control group could help against the attrition effect if the researcher had guarantees that the dropout of the subjects occurs equally and evenly, i.e. psychologically the same subjects remain in the experimental and control groups, and all the rest drop out in both groups at the same rate. But, of course, there are no such guarantees. AT different groups different people and in different numbers drop out, and therefore the control group does little to save from the effect of attrition. Another possible method of struggle is the exclusion from the sample of the results of those subjects who participated in the primary testing and did not participate in the second. But in this case, the effect of selection of subjects will be added to the effect of exhaustion: the researcher himself in literally will select from its subjects only those who, for some reason, went through the entire study from beginning to end.

Obviously, it is difficult to deal with the effect of attrition, because if the subjects do not want to continue participating in the study, forcing them to do so would be at least unethical. It is possible to reduce the risk of dropout by creating additional motivation to participate in the study. This motivation is not necessarily payment; it can be how the researcher will present the importance of the results of the study, its specific benefits for people in general and the subjects personally, affect the subjects' curiosity about some aspects of the study, etc. The main thing is that the researcher motivates the subjects in the same way, so that differences in motivation do not act as a side variable.

Also, a preventive measure of dropout will be the creation of conditions so that it would be convenient for the subject to come to the re-examination. For example, the probability of coming to a second study will be higher if the subject does not need to specially go to the laboratory, if the study is carried out at the place of his regular presence.

These are the threats to internal validity. There are quite a lot of them, and different ways of constructing a study (experimental plans) are subject to them to a greater or lesser extent. lesser degree. When planning a study, the author must take into account and control the impact of all these threats so that the study is not in vain, and based on its results it is possible to draw reliable meaningful conclusions about the causes of the studied phenomena. In table. Table 11.2 summarizes all of the threats to internal validity discussed above, the nature of their impact, the features of the study that increase the likelihood of their occurrence, and actions to control them.

Table 11.2

Threats to internal validity, features of the research procedure that increase the risk of these threats, and ways to prevent them

		Features of the research procedure that contribute to the emergence of affect	Control methods
	Random, uncontrolled exposure of subjects between pre-test and post-test	In any procedure, this effect is possible.	Using the control group
growing up	General changes in subjects' personality, not related to a specific event, between pre-test and post-test	The young age of the subjects, the change in their living conditions, requiring rapid adaptation, the rate of change in functional states (fatigue, hunger, etc.)	Using the control group
testing	The unaccustomedness of the subjects to participate in research, the desire to show their best side, the fear of showing their shortcomings	The use of methods that provoke socially approved behavior, research in a situation of evaluation or competition of subjects (certification, examination, etc.)	Use of Solomon's plan. Introduction of the training series
tool	Errors, inaccuracies of any methods used in the study	Research procedures with new, previously untested tools or ways to use them	Use of reliable, psychometrically verified instrumentation. Maximum formalization of the research procedure

The end of the table. 11.2

	Threat to internal validity (side variable)	Features of the research procedure that contribute to the appearance of the effect	Control methods
regression to the mean	Errors in estimation psychological abilities and qualities in the selection of subjects on the basis of a single test	Test groups are selected on the basis of extreme indicators	Use of an equivalent control group. Assessing changes in the dependent variable relative to the mean
test subjects	Unaccounted for psychological differences between the subjects of the experimental and control groups	All studies in which the distribution of subjects into groups depends not on the researcher, but on uncontrollable circumstances	Use of all possible group adjustment procedures (see chapter "Types of experimental plans")
exhaustion	Psychological differences between the initial sample (participating in the preliminary testing) and those who came to the final testing	All sufficiently long studies in which the researcher is forced to meet with the subjects repeatedly	Measures to prevent the dropout of the examinees (motivation, convenient place and time, etc.)

• Campbell D. Models of experiments in social psychology and applied research. S. 65.

Internal validity b is related to the question of whether it was this experimental exposure that actually led to changes in this experiment?

Internal validity concerns the relationship between dependent variables and independent variables. This validity is associated with special procedures that allow you to determine how reliable the conclusions made in this study are. After the existence of a relationship between variable X and variable Y has been established, it is necessary to decide which of the variables is the cause and which is the effect, that is, to determine the direction of this relationship. If Y is observed after X, then X can be said to be the cause of Y.

However, it may turn out that the relationship of dependence between X and Y is caused by the third variable C. To establish internal validity, it is necessary to consider all the possibilities of the influence of the third variable C on variables X and Y and exclude them. A study is considered to have internal validity if it can be shown that there is a causal relationship between dependent variables and independent variables.

External validity refers to the possibility of generalizing the output to populations and, situations, other independent variables.

These two criteria are important, although often they are in conflict with each other in the sense that the increase validity and one type may threaten the validity of another type. The ideal is to choose experimental designs that provide both types of validity. This is especially important for psychological and pedagogical research, where it is highly desirable to extend data to practical situations of a certain nature.

There are eight different classes of external variables related to internal validity. If these variables are not controlled experimentally, they can produce effects that are confounded by the effect of the experimental exposure. These variables represent the influence:

1) background- specific events that occur between the first and second measurements along with the experimental exposure;

2) natural development- changes we test th, which are the result of the passage of time (not related to specific events), for example growing up, increased hunger, fatigue, etc.;

3) testing effect- the impact of the performance of the tasks used for measurement on the results of the repeated test;

4) instrumental error, instability of the measuring instrument, in which changes in the calibration of the instrument or changes characterizing the observer or performance indicators, may cause changes in the measurement results;

5) statistical regression, which occurs when groups are selected on the basis of extreme indicators and ratings;

6) selection of subjects - non-equivalence of groups by composition, causing the appearance systematic error in the results;

7) dropout during the experiment- uneven elimination we test th from the compared groups;

8) factor interactions selection but with natural development and others, which in a number quasi-experimental designs with multiple groups are mistaken for the effect of the experimental variable.

Factors that threaten the external validity b, or representativeness of the experiment, relate:

1) reactive effect, or testing interaction effect, is a possible decrease or increase in susceptibility we test s to experimental exposure under the influence of pre-testing. The results of persons who have passed the preliminary testing, will be unrepresentative of those who have not been pre-tested, that is, those of whom the population from which the subjects were selected;

2) factor interaction effects selection and experimental influence;

3) conditions for organizing the experiment, causing a reaction we test s on the experiment, which does not allow to extend the obtained data about the effect of the experimental variable on persons exposed to the same effect in non-experimental conditions;

4) mutual interference experimental influences , which often occurs when the same subjects are subjected to several influences, since the influence of earlier influences, as a rule, does not disappear.

(See: Reader. Campbell D. Models of experiments in social psychology and applied research. M. Progress, 1980.)

End of work -

This topic belongs to:

Research methods in psychology

Section of research methods in psychology. topic modern methodology scientific knowledge.. methodology of scientific knowledge postulate of an active subject..

If you need additional material on this topic, or you did not find what you were looking for, we recommend using the search in our database of works:

What will we do with the received material:

If this material turned out to be useful for you, you can save it to your page on social networks:

tweet

All topics in this section:

The main postulates of modern theory of knowledge
Traditionally, within the framework of psychological problems, two most general topics are distinguished: - the study of motivational processes (motivational-need field of activity;

Pink glasses
Suppose all people began to wear rose-colored glasses that only let through pink. Accordingly, the world itself can no longer be described by them in any other colors than pink. True, by

The origin of the concepts of "idiographic" and "nomothetic"
In order to imagine the main framework in which “classical” science developed (which manifested itself primarily in connection with the scientific revolution of modern times), we consider several episodes of the development

The problem of measurements in psychology
To see the possibility of such a breeding of concepts, it is worth referring to - in the very general view- to Marxist philosophy, on which the cultural-historical psychology of L.S. Vygots

stone ax
In order to know the properties of objects, one must try to influence them and fix what happens to them as a result of this influence. Starting point similar analysis of things

The birth of introspection
The main patterns of self-observation were formulated within the framework of subjective, or introspective (from Latin introspecto - I look inside), psychology

Types of introspection
In the process of development of introspectionist psychology, in particular, such types of introspection were developed as: - analytical introspection; - systematic introspection;

Essence of observation
Another basic method of psychology is observation. Psychology, like any other objective science, actually begins with observation, i.e. with direct perception real

Milestones in the development of the method of observation in psychology
The method of observation was often used in the period of the birth of experimental psychology, most actively - within the framework of developmental psychology. At the same time, many scientists who observed the development of their children,

Basic observation parameters
The method of observation, which is used mainly in genetic psychology (developmental psychology and zoopsychology), unlike the experimental method, does not involve the manipulation of independent

General Observation Procedure
The general observation procedure consists of the following processes (see [Yadov V.A. Sociological research: methodology, program, methods. M .: Nauka, 1987, p. 110-166]):

Object of observation
The object of observation is defined in the context research program and depends on hypotheses, empirical indicators of the selected concepts, research strategies in general. Plan

Sample types
There are different types of samples used in observations. An unstructured sample is a sample in which no restrictions are placed on

Types of observation
Classification of observations is made on various grounds. Back in 1982 Bailey proposed a classification of different types of observation based on two criteria:

Simple, or structureless, observation
In the absence of clear hypotheses, uncontrolled observation is used. Its purpose is to provide hypotheses for a more rigorous description of the observed object. Here

Controlled or structural surveillance
For the first time, the method of controlled observation was used by the American social psychologist R. Bales in 1950 to study the successive phases of group activity. Compiled

natural observation
With natural observation, the behavior of people (or animals) in their natural conditions is studied. Observation in natural conditions involves conducting research in such a way

Non-Included and Included Surveillance
According to the position of the researcher of observation conducted to describe behavior, one can distinguish between non-involved observation (or ethological (see Fig. 30)) and included observation (or ethnographic

Non-included observation
Non-participant observation is carried out in natural conditions by a detailed and objective description of what is happening. Such an observation takes place without an observer - usually he hides and

Included Surveillance
When observation is turned on, the researcher is a participant in what is happening and tries to achieve admission to the group. Members of the group know they are being watched. Observer

Lack of control
When observing, as a rule, there is no control over the situation when the observer has the opportunity to influence it. Therefore, conclusions based on observations must be made with great caution.

Observer Influence
An essential problem of observations is the distortions introduced by the observer. This implies the presence of a preconceived opinion about the object, observation through the prism of this opinion. The distortion may

Place of observation among other data collection methods
Observation - a research method, which is the identification of certain characteristics of a particular process, with the aim of determining its invariant features, without active inclusion

Verbal and non-verbal components of communication
The most important and developed area psychological observation is the observation of external, bodily manifestations of emotional processes. Modern approach to express analysis

Expressive manifestations of emotional reactions and states
As already mentioned, emotional markers play a special role in observing behavior. Sometimes emotional reactions (affects) are violent and sudden, occurring almost immediately after the action.

Mimic expression identification
The study of mimic expression of emotions began in the 19th century. Back in 1859, the German anatomist Piderit formulated the position according to which mimic expression can be characterized using

Universal and unique in facial expression
When a researcher has to communicate with a person for a long time, it is determined that such and such a facial expression means irritation in him, while another means delight: in addition to common language emotions, you need

Comprehensive observation of gestures and postures
When performing observation, it is important to interpret gestures and postures in their unity, in the context of a holistic non-verbal message. Thus, a person who experiences anxiety

Features of observation in psychotherapeutic conversation
Within the framework of a psychotherapeutic conversation, the opportunity for included observation naturally opens up. In order to have a positive impact on the client when conducting a psychotherapist

Interaction between verbal communication and non-verbal communications
The interaction between the source of the message and the level of awareness is described in six categories of non-verbal behavior. These are: - selection, - facilitation, - image,

The essence of the document analysis method
Often, previously recorded information, for example, in the form of texts, is used to test research hypotheses. However, document studies are often part of larger studies.

Types of documents within the document research method
According to the method of fixing information, there are: handwritten and printed documents; recordings on film or photographic film, on magnetic tape. From the point of view of the intended purpose, materials are distinguished,

Reliability control in document analysis
The reliability of information from personal documents is determined by many factors. This is the verification of the authenticity of the document, the analysis of motives, motives, the conditions for its preparation, the purpose of the author,

Evaluation of the document research method
Document analysis is an important method of collecting information in psychological research, serving primarily to put forward hypotheses and a general definition of a topic, as well as at the stage of working on one or another topic.

Qualitative and quantitative approach to the analysis of documents
Qualitative analysis of documents is a necessary step in the analysis of documents. In some cases, the study ends there, without proceeding to the quantitative stage. In like

The essence of content analysis
Let us consider in more detail the quantitative approach to the analysis of documents, since it is he who provides the most reliable data. Often in researching documents, it is required not only

Basic elements of content analysis
Content analysis begins with the identification of semantic units, which can be used: concepts, topics, names, etc. Concepts expressed in words and individual terms (often

Counting Procedures for Content Analysis
In general, the counting procedures for content analysis are similar. standard methods classification by selected groupings, ranking and scale measurement. For example, when studying the subject of gas

Reliability of the content analysis procedure
The reliability of information obtained by content analysis is ensured in the following ways. Substantiation of the completeness of the allocated semantic units. Originally

The essence of the survey
The survey method is based on a simple idea: if you want to know what people think about a certain topic, you should simply ask them about it. A survey is a structured collection of questions.

Interview
Interview - a conversation conducted according to a specific plan, involving direct contact between the interviewer (interrogator) and the respondent (interviewee; from Latin responsare - to answer).

Types of interview
There are many types of interviews. According to the content of the conversation, they distinguish: - the so-called documentary interviews (study of past events, clarification of facts), - opinion interviews

telephone survey
This type The survey is based on the selection of phone numbers and calling people. The obvious advantage of a telephone survey is the lower cost of conducting it compared to

Written survey
Written survey, or questionnaire (from French anquete - a list of questions) - a research method based on receiving answers from respondents (usually written, but also possible

Types of questionnaires
Questionnaires are classified primarily by the content and design of the questions asked. The questionnaires differ in the wording of the questions. Yes, questions can be open or closed.

Poll by mail
Polling by mail is distinguished from face-to-face questioning on the spot: - in the first case, a return is expected questionnaire(for example, by prepaid postage),

General assessment of the capabilities of the survey method
When studying people's subjective states, motivation, opinions and attitudes towards events, the survey method has significant advantages over other methods of collecting empirical data. So,

Sample
For the results of the survey to be valid, the sample must reflect the characteristics of the population, that is, it must be representative. In the survey method, the adequacy of the conclusions drawn directly depends

Developing questions for respondents
The reliability of the data essentially depends not only on the content of the planned information, but, of course, on the design of the question itself, the expediency of which is dictated by the specific task and

Poll vocabulary
In mass surveys, a language that is too difficult and, therefore, obscure for respondents is unacceptable. But adjusting to the style of a certain audience group also brings little benefit (see [Yadov V.A.

Vocabulary of the expert survey
If an expert survey is conducted, then specialists act as respondents. For expert interviews, only those who deserve the status of a competent person in a given

Questions on motivation, assessments and opinions represent the most difficult part polling procedures. When compiling the questions of the questionnaire, one should be guided by the following considerations: 1. Especially

Conducting a questionnaire
The questionnaire is filled in by the respondent himself, so its design and all comments should be extremely clear to the respondent. The basic principles for constructing a questionnaire are as follows.

pilot study
Any research tool is tested for its validity. For this, a pilot study is used. In the usual "pilot" research

Expert survey
Survey of specialists - special variety polling method. Such surveys are not anonymous, they exclude "traps", as they involve the active cooperation of the respondent in clarifying the set

Features of interviewing
The ideal interview resembles a lively and casual conversation between two people who are equally interested in it. But the interviewer remembers that in this situation he acts as a professional

The essence of focus groups
Within the framework of social psychology and personality psychology, the method of focus groups, which can be considered as a form of conversation, has recently become especially widespread (see [Dmitrieva E.

Stages of conducting focus groups
Conducting focus groups involves three stages: - at the first, preparatory stage, the goal, object and subject of research is determined, the research team is trained

Research team training
The first step is to train the research team. Such a team consists of: - a researcher leading a focus group (moderator), - stenographers,

Determining the number and size of focus groups
Several factors influence the number of participants in a focus group: 1. Because the data collected through a focus group is descriptive and

Topic selection
Based on the limits of psychological involvement in the discussion, focus group participants show interest in it for no more than one and a half, maximum two hours, and in a longer discussion

Writing a Focus Group Plan
If the focus group is conducted according to a pre-written plan, then such a group is designated as a formalized focus group. The plan begins with the formulation of

Typology of questions
There are various classifications of questions in a focus group: according to the degree of importance, formality, order of appearance in the discussion, etc. In order of importance, the following

The role of the facilitator in the focus group
During the field study great importance assumes the role of focus group leader. It can be a researcher who organizes the work and interprets the results, as well as a specially invited

Form of holding and organization of focus groups
The focus group begins with an introduction, introduction of the facilitator, research team, participants, a description of the purpose of the study and the duration of the work. During the course of the meeting, the respondent usually

The essence of the conversation method
Unlike surveys conducted through interviews or a survey, a conversation, which is also implemented on the basis of verbal relationships between people, is primarily focused on the individual.

Types of conversation
The most widespread are: - a clinical conversation, or a clinical interview, organized around the painful experiences of the patient (see Fig.

Ability to listen actively
For effective interaction with the subject or client of a psychotherapeutic session, it is extremely important for a psychologist to be able to “listen actively”. The ability to "listen actively" involves the following

Milestones in the development of the conversation method in psychoanalysis
Based on the use of conversation as a method psychological research, lie rich tradition of development rational therapy in psychiatry. special role in the development of the method of persuasion, as one

The essence of the structured interview
Conducting a conversation in the framework of psychotherapy and psychoanalysis has a number of features. One of the techniques is called the structural interview [Kernberg O. Structural interview // Zhurna

The initial phase of the structured interview
The psychotherapeutic conversation begins with the fact that the client is asked to briefly talk about: - the reasons that made him see a psychotherapist, - expectations regarding psychosis

The main phase of the structured interview
The main subject of a structural interview is the study of pathological character traits, which is important not only to assess the type of character, but also to assess the severity of these

Final phase of the structured interview
Having completed the study of neurotic symptoms and pathological character traits, assessing the main defense mechanisms, feelings of reality and key symptoms of psychoses or organic disorders, the psychotherapist

Scaling in experimental psychology
As already mentioned, the natural-scientific methodological orientation in psychology is manifested primarily in two orientations. Firstly, in the reliable identification of certain psychological phenomena

Fundamentals of psychophysics
For the first time, the problem of the threshold of sensations was posed by I. Herbart (1776 - 1841) (see Fig. 38), a German philosopher, psychologist and teacher, the founder of empirical

Multidimensional scaling
The psychologist now has access to what was previously considered the prerogative of only the ideographic approach. We are talking about the application of mathematical procedures, with the help of which, for example,

The use of measuring tools in psychology
Conversation and observation, as a rule, is carried out directly, without the use of any special means of fixing and classifying the identified psychological phenomena. But, unlike neither

Brain scintigraphy
Physical essence: the method is based on the ability of special radioactive chemicals to concentrate in pathologically altered tissue in greater quantities than in normal

Single photon emission computed tomography
Physical essence: the method is based on layer-by-layer visualization of the distribution of special radioactive chemicals (ceretek) in the brain. These substances are introduced

Positron emission tomography
Physical essence: the method is based on the registration of extremely low concentrations of ultrashort-lived radionuclides, which mark the brain structures whose metabolism is being studied.

Some data illustrating the possibilities of computed tomography
1. The brain of patients suffering from severe depression was studied. An increase in the metabolism of brain regions responsible for the production of stress hormones has been recorded. After taking patients

The essence of psychological diagnostics
Psychological diagnostics, or psychological testing (from the English test - test, research), is a scientific discipline about methods for classifying individuals according to their inherent and

Tasks of psychological testing
Within the framework of the classical theory of testing, the basic procedures for assessing the quality of the developed psychodiagnostic tool are formulated. The main characteristics of effective personality surveys

Types of psychological diagnosis
Based on differences in the degree of understanding of the internal laws of one or another psychological sign distinguish the following types of psychological diagnosis: - symptomatic, or emp

Use of scales in psychological diagnostics
The classical theory of testing is based on the concept of subjective scaling, which implies the use of quantitative indicators to determine the quantitative severity of certain

Systematics of psychological tests
Test (from the English test - test, research) - an experimental research tool, which is a standardized procedure for psychological measurement on OS

Personality Questionnaires
Personality questionnaires are a group of psychodiagnostic methods designed to determine the degree of severity of certain personality traits. They are characterized by a set

Planning the stages of developing a personality questionnaire
Construction personality questionnaire involves going through a number of stages. These are the following stages: 1. Theoretical analysis, on the basis of which a psychological construct is built, as a theoretical

Development of a psychological construct
Methodological approaches in the psychodiagnostics of a personality largely predetermine which aspects of the empirical manifestations of a personality must be involved in the created personality questionnaires.

Development of a psychological construct based on theoretical analysis
This path is the most common and is now recognized as mandatory in the development of any personality questionnaires. First studied special literature relating to soda

Development of a psychological construct based on correlation studies
The use of statistics, including those based on the identification of various correlations, is mandatory when designing any tests. What statistical procedures and, first of all

Development of a psychological construct based on the criterion
A completely different approach than with factor analysis is implemented when constructing a personality questionnaire using an external criterion. The most developed strategy for determining external criteria

Operationalization of the psychological construct
After it turned out to be decided to resolve the issue of the content of the constructed questionnaire, a plan is outlined for checking the construct put into the personality questionnaire. how

Choice of the main area of ​​localization of self-evaluative judgments
Based on descriptions taken from the psychological literature, on the basis of previously conducted empirical research, based on expert assessments, etc. must be received verbalization of certain psi

Rules for formulating items of a personality questionnaire
When formulating tasks of a personality questionnaire, you can follow the following rules. 1. In most cases, it is necessary, if possible, to avoid the penetration of the subjects into the essence of something

Choice of response scale for questionnaire items
A separate task of constructing a personality questionnaire is the formulation of meaningful answers to its items and the choice of an appropriate response scale. There are two types of survey

Examination of the psychological construct
After all the items of the personality questionnaire have been formulated and the compiled list of all items is arranged in order, an examination of the questionnaire can be carried out, which acts as a separate

Methods for analyzing the interpretation of the questionnaire by the subjects
The first group of such methods involves additional questions to the subject. He is required to say how he understands a particular item and formulate an answer to it, i.e. reflect on

Normative and pilot samples
One of the most important decisions in the design of the questionnaire is the question in the sample of subjects, which are carried out through the procedure of psychological diagnostics. How to rule

Normative sampling
In statistics, a distinction is made between a sample (the group that actually was tested) and a population (a larger group, but having the same composition as the sample, which may be subject to testing).

Aerobatic sample
pilot study- pilot preliminary study, conducted, as a rule, on a small sample, which serves to clarify the adequacy of the use of certain researchers

Evaluation of the final score of a separate scale
Discriminativity of a test is a psychometric characteristic of a test, which is its ability to distinguish between subjects. The more high level has discree

Evaluation of a single item on the scale: level of difficulty
One of the main decisions that are made when designing personality questionnaires is the decision on the effectiveness of using a particular item of the questionnaire. It depends on how

Assessing the difficulty level of an item with a dichotomous response scale
If we are talking about a dichotomous scale of answers to the items of the questionnaire, then for each of these items two numbers can be obtained: the percentage of those who answered “yes” to this item (or gave another similar

Assessing the difficulty level of an item on a Likert response scale
Typically, a Likert response scale is used for items on a personality questionnaire. Therefore, the distribution of responses to a specific item of the questionnaire of the following type can be obtained: “very often” =

Testing Reliability
Test reliability is a psychometric characteristic of a test, which is the reproducibility of measurement results under similar conditions. It shows to what extent an individual

Size and composition of the sample
Since the standard error of the correlation coefficient used to estimate reliability is related to the size of the sample on which it was obtained, then sufficiently large values ​​should be used.

Methods for determining the reliability of the questionnaire
There are two main methods for determining the reliability of a questionnaire. Based on a single test, internal consistency can be determined, which is calculated by the formula

Internal Consistency
In traditional psychometrics, a test is said to be reliable if it is internally consistent. Internal consistency is the degree of homogeneity of the composition of tasks in terms of measurable

Calculation of the index of internal consistency for each item of the questionnaire
Internal consistency can be determined by the correlation between the final score on the scale of the questionnaire and each of its individual items. Based on calculations of the internal consistency of all points

Calculation of the indicator of internal consistency for each item of the questionnaire in factor analysis
When using factor analysis when constructing a questionnaire, there is a specificity in determining the reliability by the criterion of internal consistency. So, to get an indicator

Computing the Internal Consistency Score for the Entire Scale
When the question of points that strongly and weakly correlate with the final scale has already been resolved, i.e. it became clear which items work on the scale and which do not, the indicator of internal consistency is calculated

Retest reliability
Retest reliability is the consistency of results obtained by the same individual when tested repeatedly with the same test (or its equivalent form).

Retest Reliability Calculation
When calculating the retest reliability, the correlation coefficient is determined between the results for the entire questionnaire obtained on the same subjects, but in different time(r

Reliability of Parallel Forms
A subspecies of retest reliability is the reliability of parallel forms. This assumes testing with a certain time interval of the same sample with n

Test Validity
Validity (derived from the English valid - valid) of the test is a psychometric characteristic, which is the actual ability of the test to measure that psychological construct, d

Apparent validity
Obvious validity is not, in the proper sense, a psychometric indicator of the test, it characterizes only the impression of the test as something understandable, “transparent”. This is the ability

Content validity
Content validity is characteristic of tests that fully model a particular activity, primarily in its subject aspect. Those. the content of the test itself reflects the key

Criteria Validity
Criteria validity is determined with a single, extremely important for practice, goal: to assess the individual predictive value of the test. To do this, the test results are compared with

Criteria Validity Calculation
The best and most in a simple way determining criterion validity is the calculation of a dichotomous correlation (there is a sign - there is no sign) between the final score on the scale and belonging

Criteria validity of individual items of the questionnaire
Criteria validity can be calculated not only in relation to the entire scale of the questionnaire, but also in relation to each of its points. At the same time, if a personality questionnaire is created only on

predictive validity
A special type of criterion validity is predictive validity, which can be obtained by comparing test scores with an external (pragmatic) criterion that is used

Construct validity
Constructive validity, substantiated in 1955 by L. Cronbach, provides information about the degree of ability to measure such a trait by the test, which was identified by the theory

Competitive Validity
Competitive validity is assessed by the correlation of results this test with the results of other similar tests, the validity of which was established earlier. This procedure makes sense in

Expert empirical validity
Of particular importance is empirical validity, based on a comparison of the final scores on the scale of the questionnaire with the assessments of the subjects by experts. Empirical coefficient value

Test Validity Calculation
Quantitatively, the validity of a test can be expressed through correlations of the results obtained with its help with other indicators, for example, with a pragmatic criterion, in particular, with success.

Basic provisions of factor analysis
Factor analysis is based on the hypothesis that directly observed variables (for example, signs of behavior) only indirectly reflect the essence of the phenomenon under study. By

Factor analysis methods
There are two main methods of factor analysis: in one of them, correlated factors are distinguished (oblique solution used, for example, by R.B.

Factor analysis requirements
1. The sample must be representative so that the estimates of the matrix pairwise correlation coefficients are reliable. Thus, the sample size in factor analysis should be at least 100 subjects.

Normalization of indicators
In order for the personality questionnaire to be used practically, i.e. make, on the basis of its filling by an arbitrarily taken subject, a forecast of his behavior in new situations

Linear and non-linear transformations of raw scale values
Standard indicators can be obtained both by linear and non-linear transformation of primary indicators. Linear transformations are obtained by subtracting from the primary indicator constants and d

Requirements for a personality questionnaire
In preparing for the publication of a personality questionnaire, a number of requirements must be met. 1. The subject, scope and purpose of the test must be unambiguously formulated. Item as diagn

Description of the personality questionnaire
The structure of the description of the psychodiagnostic technique for publication has next view: 1. Subject and target population: - social significance (social need

The essence of the experiment in psychology
The essence of the experimental method lies in the fact that the experimenter deliberately creates and in a controlled way changes the conditions in which the subject under study operates, puts before him

Limitations of the experimental method
In general, an experiment is a piece of research in which the researcher manipulates variables and observes the effects that manipulation has on other variables. Expert

Types of natural experiment in various research fields
One of the options for a natural experiment is a psychological and pedagogical experiment, or experimental learning, where the study mental characteristics schoolchild, subject

Conducting experiments
The experimental method has special benefits distinguishing it from other research methods. First of all, it is the experimenter's ability to manipulate one or more variables,

Probability sampling
This strategy is used if the task is to learn something specific about certain group of people. The group as a whole is called a population, and any of its subgroups is called

Experiment progress
The pilot study is generally characterized by three phases. In the first phase there is a pre-test: one or two variables are measured before applying the experimental

Experimental plans
AT pilot study there may be one or more independent variables. If there is only one independent variable, then that variable must have

Types of validity
Experiments are a means of testing hypotheses about causation. Traditionally, there are three necessary conditions to assert with a certain degree of certainty that

Reliability of experimental data
The measurement process is the more reliable, the weaker the influence of random factors and the more constant results it gives. When creating methods designed to measure variables

Validity
It is assessed whether the characteristic under consideration is measured or, on the contrary, the measurement reflects the influence of other factors. The validity of the measurement tool is assessed depending on the degree of conformity

Study Validity
Research validity refers to establishing the accuracy of research findings and highlights the relative nature of the truth that can be achieved in psychology. In any scientific study

Validity of statistical inferences
This type of validity corresponds to testing the statistical significance of the relationship between two variables. Such conclusions are always probabilistic. At the same time, it is possible to

Validity of procedures
The third type of validity is the validity of procedures that allow variables to be varied and measured. Even the need to define in operational terms the conceptual variables

There are several main methods of control
One of the most commonly used control methods is to conduct an experiment with a group of subjects who are not affected by the variable under study and who are compared with the test.

Quasi-experimental schemes
The term "quasi-experimental scheme" itself was introduced in 1966 by Campbell and Stanley. Quasi-experimental schemes are understood as schemes in which the random method is not used.

Pre-experimental schemes
Pre-experimental research schemes, as well as quasi-experimental ones, do not provide for random selection of subjects into experimental groups. This is a schema

correlation method
With the help of the correlation method, the relationship between two variables is studied, not mediated by the intervention of the researcher. Thus, there are no manipulations of variables,

Methodological approaches to personality analysis in various psychological schools
Personality in psychology is considered as a relatively stable system of attitudes and behavior of the individual, which is characterized, first of all, by the features of the inclusion of this individual in the social

Introspective Approach
Back in 1734, H. Wolf gave the following definition: "Personality is that which retains memories of itself and perceives itself as the same both before and now." This

Psychodynamic approach
A completely different model of personality, important in the context of the discussed problem of psychodiagnostics, was developed in depth psychology, primarily in psychoanalysis (S. Freud

Behavioral Approach
Actually introspective and psychodynamic approaches are essentially close to each other, if we pay attention to the role that is attributed to introspection in them. On the contrary, in the model l

Physiological approach
A special direction, in its origins quite closely related to behaviorism, is biological, or physiological, psychology (see Fig. 76). Here, too, personality analysis is based on

Humanistic approach
As part of humanistic psychology, where the individual is seen primarily as making responsible decisions in specific situations his life, most often it is concluded that h

Activity approach
In Marxist psychology, personality is defined as the result of historical development individual, which is carried out primarily within the framework of a joint labor activity(A. Vallo

Integration of methodological approaches to personality analysis
At present, there are all the prerequisites for using all the achievements of psychology, obtained regardless of the original methodology, in personality analysis. Yes, definitely needed.

• - See Validity...

  Great Psychological Encyclopedia

• - V. in. refers to the degree of confidence with which one can judge the alleged causal relationship between variables ...

  Psychological Encyclopedia

• - - conformity of the test to the purposes of diagnostics...

  Pedagogical terminological dictionary

• - English. validity; German Validitat/Gulltigkeit. Validity and adequacy of research instruments...

  Encyclopedia of Sociology

• - English. validity, empirical; German Validitat, empirische. The degree to which variables and indicators fit empirical data...

  Encyclopedia of Sociology

• - English. validity, logical; German Validitat, logische. The degree of interrelation and mutual deducibility of variables and indicators...

  Encyclopedia of Sociology

• - from fr. valide in statistics - the legality and reliability of the source information, the reliability of the methodology for collecting, obtaining data, including for economic research ...

  Glossary of business terms

• - Degree of conformity of variables and indicators with empirical data...

  Glossary of business terms

• - A kind of theoretical validity, expressed in correlation ...

  Glossary of business terms

• Big Economic Dictionary

• - in relation to the initial statistical information used in economic research: reliability of information, the absence of errors in it due to the inaccuracy of the chosen data collection methodology ...

  Economic dictionary

• - shaft "...

  Russian spelling dictionary

• - validity Absence of errors related to the definition of initial assumptions in the development of research methodology; reliability of information...

  Explanatory Dictionary of Efremova

• - VALIDITY and, f. validate f. lat. validus. Validity, legitimacy, weightiness, validity; possession legal force. Komlev 1992. || The ability to form a coherent sign system...

  Historical Dictionary of Gallicisms of the Russian Language

• - 1) weightiness, validity; 2) reality, legality; legal power...

  Dictionary of foreign words of the Russian language

• - ...

  Synonym dictionary

"Internal validity" in books

4. Truthfulness and validity

From the book Ways to Create Worlds author author unknown

Validity

From book Philosophical Dictionary author Comte Sponville André

Validity (Validit?) Used in logic, a synonym for truth, more precisely, its formal equivalent. A conclusion is considered valid (true) if it represents a transition from true to true (from the truth of the premises to the truth of the conclusion) or remains true regardless of

RELIABILITY AND VALIDITY

From the book Psychology author Robinson Dave

RELIABILITY AND VALIDITY Amy's parents are concerned that she is not as good at reading as she is at adding and subtracting numbers. After discussing the matter with her teacher, they decided to take Amy to a psychologist. The psychologist suggested that the girl take tests that should find out her

Validity or Reliability

From the author's book

Validity or reliability is the most important criterion for the quality of measurements, which means that a method, technique or test measures what it is intended for, i.e. the measure of compliance of the tools used and the results obtained with the tasks set. This criterion

Validity

From the book Overclock your site author Matsievsky Nikolai

Validity The resulting CSS file is absolutely valid (since all mhtml inserts occur in comments). Every CSS declaration is valid under CSS 2.1, and the little tricks that effectively compress data for all browsers are not reflected in

3. Factor validity

author Luchinin Alexey Sergeevich

3. Factor validity Factor validity. direct relation to the characterization of construct validity has factor analysis, which allows strictly statistical analysis of the structure of relationships between the indicators of the test under study with other known and latent

4. Differential validity

From the book Psychodiagnostics: Lecture Notes author Luchinin Alexey Sergeevich

4. Differential validity Differential validity is a type of construct validity that considers the internal relationship between psychological factors diagnosed using psychodiagnostic methods. The content of the differential

2. Apparent validity

From the book Psychodiagnostics: Lecture Notes author Luchinin Alexey Sergeevich

2. Apparent validity Apparent validity is the perception of a test, scope, performance, and predictive value that a test subject or other person has with no specific knowledge of use and purpose.

24. Factor validity

author Luchinin Alexey Sergeevich

24. Factor validity Factor validity. Directly related to the characterization of construct validity is factor analysis, which allows strictly statistical analysis of the structure of relationships between the indicators of the test under study with other known and latent ones.

25. Differential validity. Validity by age differentiation

From the book Psychodiagnostics author Luchinin Alexey Sergeevich

25. Differential validity. Validity by age differentiation Differential validity is a type of construct validity that considers the internal relationships between psychological factors diagnosed using psychodiagnostic

26. Criteria validity

From the book Psychodiagnostics author Luchinin Alexey Sergeevich

26. Criteria validity Criteria validity is a set of characteristics that includes the current and prognostic validity of the method and reflects the compliance of the diagnosis and prognosis with a certain range of criteria for the phenomenon being measured. As a criterion

29. Predictive validity

From the book Psychodiagnostics author Luchinin Alexey Sergeevich

29. Predictive validity psychological quality later certain time after measurement. predictive validity

Validity

by Miller Scott

Validity Every study has variables and relationships between them. So when we want to describe research, the variable construct is central: what kind of differences are being explored, and in what ways? If we want to go beyond the description and go to the assessment

Validity

From the book Developmental Psychology [Research Methods] by Miller Scott

Validity When evaluating the validity of a test, the following question is asked: Does the test measure what it is supposed to measure? If this is, for example, an IQ test, does it really measure differences in the level of intelligence or the differences in the performance of the subjects are due to something

Validity

From the book The Intelligence of Success author Sternberg Robert

Validity The first important information of this kind, which will be discussed throughout this book, is often referred to as the "criterion of validity." The question here is to what extent test results measure what they should measure in accordance with their

(internal validity) V. c. refers to the degree of confidence with which one can judge the alleged causal relationship between variables. The experiment has V. century. to the extent that the observed effects can be attributed to the influence of independent variables, and not to.-l. other extraneous factors. Attribution of causality usually assumes that two variables - cause and effect - will change conjugately, but before inferring about C. c., it is necessary to demonstrate that A does cause B (i.e., the occurrence of A alone leads to B ). It is possible, for example, to mistakenly assume that learning to solve perceptual-motor problems at home a) leads to an increase in the performance of children at school, b) whereas in reality such a desired result is due not so much to perceptual-motor learning itself, c) how much parental care and attention. Psych. Experiments are particularly susceptible to misleading and unintended side-variable effects, since these experiments can rarely be carried out under such perfectly clean conditions as are created in physics. laboratories. Therefore, the primary concern of researchers in psychology is the planning of experiments, so that the positive results obtained in them allow you to draw conclusions about the existence of causal relationships. V. in. differ from the external validity relating to a possibility of distribution rez-t beyond that specific context, in Krom V. of century was established. For example, if under carefully controlled conditions it was possible to prove that overcrowding of cages in which laboratory animals are kept leads to aggressive behavior white rats (V. v.), we would have grounds for extending this conclusion to people living in the ghetto ( external validity)? V. in. should be distinguished from construct validity, which is theoretical. explanation of the alleged causal relationship. In most cases, however, when there are all reasonable grounds to consider V. c. established, the researcher wants to formulate a conceptual framework or construct to explain these relationships. Besides, V. century. should not be confused with internal consistency, a testing term referring to intercorrelations within a set of tasks. When establishing V. century. it is necessary to verify the existence certain sequence events. Obviously, if there is a relationship between A and B, and A always precedes B, it is logical to assume that A causes B, and not vice versa. In addition to the organization of the maximum possible control over the conditions of the environment and experiments. research situations. psychologists are engaged in the selection of subjects in such ways, to-rye exclude systematic errors in their distribution in experiments. and control groups. They also resort to carefully developed. experimental Designs (taking into account covariance and including balancing) to implement statistical methods for controlling the influence of variables so that reasonable judgments about causal relationships can be made. What follows is an abbreviated analysis of the problems encountered by Cook and Campbell in conducting quasi-experiments. or field research. in psychology, each of to-rykh represents a potential threat to V. century. Specific events that occur between the first and second measurements can change the reactions of the subject. For example, if in research the effect of a new drug on reducing anxiety is being tested, and in the period between measurements a tornado swept over the area, the results of measuring the level of anxiety can be seriously affected by the fear caused by this catastrophe. In addition, changes that occur naturally over time and are related to the development of the subjects can also cause consequences that contaminate the experiment (i.e., improvements in perceptual-motor skills in children can be, rather, a consequence of simple growing up, than teaching). In research involving repeated testing, repeated use of the same instrument may lead to a cumulative effect that contaminates the study exposure (for example, an improvement in IQ scores as a result of "enrichment" of the environment may be a consequence of familiarity with this intelligence test ). Mechanical or made by people. changes in the instruments used for measurements may cause effects other than those under investigation. Over time, the mechanical spring may lose its elasticity, or the examiner may change its orientation. Neglecting the phenomenon of statistical regression (trends of high and low scores on retesting to change towards the mean range) can lead to an overly positive and misleading assessment of the relationship and, as a result, to erroneous conclusions about causation. An unintentional error in the selection of subjects for a control and/or experiment. groups can cause results associated with this unaccounted for, but non-random distribution of individual subjects. In this case, differences between groups, due more to the selection of subjects than to the effects under study themselves, can lead to erroneous conclusions. Any dropout of subjects during the experiment (conscious withdrawal, illness, or even death) can distort the results of the experiment if such a decrease goes beyond random dropout. Uncertainty as to whether A is the cause or result of B inevitably affects V. century. Does it play football team better due to the presence of more fans in the stands, or vice versa, more fans gathered for the game because this team began to play better? Random and unwanted communication between research groups destroys the experiment. the naivety of the subjects and can form attitudes and expectations that are not related to the goals of this study. In turn, the experiment itself can cause compensatory rivalry, resentment, indignation and demoralization in individual subjects and / or their groups - conditions that are not related to the goals of the experiment and can distort its results. Since the experimenters do not work in a vacuum, it sometimes happens that according to administrative or social. For reasons, groups receive a compensatory effect, which leads to an equalization of effects and works against the achievement of valid results. For example, providing at least minimal therapy to a control group of patients who should not have received any therapy at all might. justified with t. sp. humanity, but no doubt defeats the purpose of such an experiment. There are no ways to conduct a perfect experiment (i.e., one in which such complete control is achieved that it would be impossible to classify the result obtained or the relationship as artifacts). In addition, usually the context itself is psychol. experimentation makes them extremely prone to interpretive ambiguity and drawing erroneous conclusions. However, deep knowledge in this area and attention to those numerous traps that lie in wait for a psychologist when conducting quasi-experiments. research., can help him plan experiments that make it more likely to obtain true conclusions regarding V. century. See also Confidence Bounds, Experimental Designs, Hypothesis Testing, Mill's Canons, Probability, Statistical Significance E. E. Wagner

Other related news.

External and internal validity.

Borovkova Victoria I-01

VALIDITY(validity) - the reliability (or degree of reliability) of the conclusion, which is provided by the results of a real experiment in comparison with the results of an impeccable experiment. “B.” is a central concept in the experimenter's vocabulary: it combines the main objectives of the study with ideal standards for their achievement and actual procedures for conducting experiments.

The global goal of any experimental study is to generalize the results obtained and the conclusion about the experimental hypothesis. However, the full achievement of this goal is possible only in a thought, impeccable experiment, impracticable in practice. A real experiment in one way or another represents (represents) an impeccable one, and the better this representativeness, the higher the V. of the experiment. Thus, the increase in V., i.e., the planning of an experiment in accordance with its impeccable model, is a specific task of the researcher, the success of which depends, firstly, on the nature, real conditions and, secondly, on the adequacy of the choice of means. Thus, the sources of violation of V. (first of all, unreliability and confusion) move the real experiment away from the perfect one, and the methods of their control make it possible to approach it, that is, to ensure high V. for generalizing experimental results. The V. of the data obtained can be evaluated statistically: for example, the V. of a test (in a correlation study) is determined by the degree of correlation between the results of its performance and the type of activity being studied by the subjects.

In accordance with the different types of impeccable experiment, internal and external V. are distinguished:

- inner B.- the reliability of the conclusions, which is provided by the results of a real experiment in comparison with the results of an ideal infinite experiment. An increase in internal V. is associated with the elimination of the results of the action of side variables and with the averaging of their variability and instability. Internal V. is the first and necessary requirement for experimental conclusions: an experiment that does not possess internal V. is called simply unsuccessful by the author.

In an internally valid study, the researcher is confident that the results obtained by measuring the dependent variable are directly related to the independent variable and not to some other uncontrolled factor.

However, in fact, in science (especially in psychology) it is impossible to say with absolute certainty that internal validity is observed. For example, it is impossible to study any mental process apart from the psyche as a whole. Therefore, always in any psychological experiment, a scientist can only maximally (but not absolutely) remove or minimize various factors that threaten internal validity.

Example of violation of internal validity

Let's say we want to test a drug that will make people taller. Let's say for our research we select 13-year-old teenagers as subjects, measure their height, and give them medicine. Two years later, we return to the now 15-year-olds and record their current height. There is no doubt that they became taller, but there is no doubt that we cannot conclude from this that the growth effect was caused by the drug, since we did not take into account the natural processes of growing up in our theoretical experiment. Here, internal validity is violated in the following way: we did not take into account the influence of other (obvious in this case) factors, which were incidental to our study, while they should have been taken into account.

- external B. - the reliability of the conclusions, which is provided by the results of a real experiment in comparison with the results of a full compliance experiment. An increase in external V. is ensured by achieving correspondence between the levels of additional variables in the experiment and the levels of these variables in the reality being studied. An experiment that does not have external V. is incorrect, not corresponding to the hypothesis posed (but, perhaps, suitable for testing another hypothesis).

External validity determines the extent to which the results obtained in the experiment can correspond to the type of life situation that was investigated, and the extent to which it is possible to generalize these results to all such life situations. For example, criticism of experimental psychologists that they know a lot about sophomores and white rats, and very little about everything else, can be seen as a criticism of external validity.

As with any other validity, it is probably impossible to say about external validity in a study that it is absolutely observed, only a violation of it can be indicated. Absolute compliance with external validity would be considered when the results of the study can be generalized to any population under any conditions and at any time, so scientists do not talk about compliance or non-compliance with external validity, but about the degree of its compliance.

External Validity Criteria

For a study to have the highest degree of validity, it must be possible to generalize its results to:

Other populations

Other conditions,

Another time.

An example of violation of external validity

Lawrence Kohlberg's study of the moral development of children has been criticized for having a low degree of external validity. In 1963, Kohlberg challenged adolescent boys to resolve various short-formulated dilemmas. Based on the answers he received, he developed a theory of six stages of moral development.

Criticism of the external validity of this study, in particular, was that the scientist did not take into account gender differences in the peculiarities of thinking, ways of making moral decisions and, therefore, unduly generalized his model to both sexes.

Kohlberg also proceeded from the premise that there are some "universal principles of justice and respect for the rights of the individual", but he did not take into account intercultural differences, therefore, the conclusions he made cannot be perceived as universal.

Compliance with external validity

To increase the degree of external validity, studies may not be limited to the experimental method, but also use others - for example, observation, cross-cultural studies, etc.

With the modern scientific approach in psychology, the observance of external validity in research plays a leading role, since the most valuable is the creation of general theories, models and algorithms applicable to the psyche as a whole, regardless of the context - that is, models with extremely broad predictive capabilities; therefore studies, the results of which can be applied only to some particular limited situation, have a descending value.

It should be noted that any variable (factor) that is different from the independent variable, which also affects the dependent variable, is simply called “other” by the author. In translation, two different terms are used to designate such variables: “side” - in relation to a variable that violates internal V., and “additional” - a variable whose level is included in the experimental hypothesis and must be adequately represented in order to achieve external V. Sometimes one and the same factors can act in the experiment both as secondary (violating the internal vein) and as additional (affecting the external vein). Elimination and unification of side factors, which increases the internal V., can lead to a mismatch in the levels of additional variables, which reduces the external V. (see Correspondence).

Operational V. - a kind of external V. for a laboratory experiment; correspondence (or degree of correspondence) of the applied methodological procedures to those theoretical concepts that are included in the experimental hypothesis.

Threats to internal validity

"Threats to internal validity" is an expression related to research design that data interpreters use to describe a range of issues that arise due to possible inaccuracies in interpreting causal relationships between the variables under study. Some of these threats have already been mentioned, in particular, when describing quasi-experiments. The expression regression to the mean is used to describe the effect of inaccurate measurement of current and future observations, and it poses a serious threat to the internal validity of the psychol design. research

Another threat is natural development, since the subjects are subject to natural processes of maturation and development over time, regardless of the manipulations that the researcher performs on them. This last hypothesis, in particular, explains the need to use research. control groups, allowing comparisons with naturally developing subjects.

Another threat is dropout. A systematic reduction in the quantitative composition of groups due to dropouts, absenteeism, people moving to another city, unfilled criterion questionnaires, etc., can make research less noticeable in the results. the role of those remaining participants, to-rye are less subject to experiment. impact. But the more noticeable the improvement in the criterion, the better the effectiveness of the studied impact looks.

The threat of selection arises at the stage of formation issled. the composition of the groups of participants. Systematic selection or passive selection (self-selection) of participants for the exposure or condition under study may lead to differences among some groups before the start of the study. These groups will continue to be different after the impact has been realized - even when they themselves have not been affected. Therefore, in order to achieve high internal validity of (causal) statements regarding the relationships of variables, it is necessary to formulate and test alternative hypothesis about what those effects might be in the absence of that exposure or condition.

Background refers to those changes in the context of an ongoing study that may be confounded by the impacts being studied. Such confusion makes it impossible to attribute changes in subjects to the test exposure or external changes.

Instrumental error is a threat to the internal validity almost exclusively inherent in psychology, because it arises from changes in the observer's standards or conditions of observation over time. These changes cause differences in estimates. The problem is that such discrepancies in measurement standards can lead to erroneous estimates of the changes that result from the effects being studied.

Testing effects are an example of another, almost exclusively psychological, threat to internal validity. It is related to the effect of improving the performance of subjects simply as a result of previous test taking.

Instability, as noted above, results from attempts to interpret the impacts that occur immediately after the experiment or after the c.-l. time natural changes in subjects and measurements.

There are also four affective threats to internal validity: imitation, compensatory rivalry, compensatory leveling, and demoralization. They are also inherent exclusively in psychology. science, because they arise as a result of emotional changes in subjects and personnel conducting measurements as a form of adaptation to the fact that they were not included in the number of "chosen" participants in the study receiving and performing the experiment

External validity and meta-analysis

Finally, there is the concept of external validity - the sum of the characteristics of studies, which allows generalization or distribution of causal statements obtained in studies (as a result of high internal validity) to similar categories of people, influences and criteria outside of this study.

Meta-analysis. The external validity of the results of a study is greatly enhanced when a number of studies of the same influencing variable are combined and summarized. This procedure is called meta-analysis. In a meta-analysis, the magnitude of the effect of the experiment, the impact and the main conditions of the study on the criterion variables are determined. The latter are chosen depending on the theoretical interests of the researcher conducting the meta-analysis. Thus, if the main analysis involves comparing the relationships of all pairs of variables, then the set of relationships of interest to the metaanalyst is a set of condition variables and criterion variables, which allows one to judge the magnitude of the effect of the studied impact on these criterion variables. The effect size for each study is determined by simply dividing the difference between the mean scores for the criterion variable in the experimental groups and the control group by the measure of variability in the individual scores in those groups. As a result of this analysis, it can be shown that a number of variable conditions studied do not affect the set of criteria, which allows researchers to generalize the results of their studies without looking back at these extraneous conditions, as found in the meta-analysis.

Finally, it is possible to determine the degree of influence of each variable-impact and their various combinations on the criteria. This is of particular interest in application areas where it is important to know not only that one variable influences another, but also how large such an influence is.