What are the types of validation? What it is? You will find answers to these and other questions in the article. Validity is inherently a single characteristic that includes, on the one hand, data on whether the technology is suitable for measuring what it was developed for, and on the other hand, what is its effectiveness, practical usefulness, efficiency. Checking the validity of a technology is called validation. We will consider it in more detail below.
Views
Many people ask: "What is validation?" It is said that the refined mind has no boundaries. Methodologists seem to compete with each other in who will discover or invent more types and types of validity. What only names have not arisen lately! It turns out there is:
- external and internal validity;
- convergent;
- discriminant;
- obvious;
- factorial;
- predictive;
- constructive;
- criterial;
- meaningful and so on.
There is no way to understand, and even more so to sensibly distinguish one variation from another. Confusion in definitions and classifications leads to the fact that different authors attribute the same form of validity to completely different ways of improving it.
External criteria
To perform the pragmatic validation of the methodology, that is, to assess its practical significance, effectiveness, efficiency, an external independent criterion is usually used - an indicator of showing the quality being studied in everyday life. Such a criterion can be both production achievements (for technologies of a professional orientation), and academic performance (for tests of intelligence, achievements or learning abilities), and the effectiveness of actual activities - modeling, drawing, and so on (for tests of special skills), personal assessments (for identity verification).
Types of external validation criteria are as follows:
- measures of performance (these may include such as the amount of work implemented, the time spent on training, academic performance, the rate of growth of qualifications, and the like);
- physiological signs (used when studying the impact of the environment and other situational variables on the human psyche and body);
- blood pressure, pulse rate, fatigue symptoms, skin electrical resistance and so on are measurednext;
- subjective measures (include various types of answers that reflect a person's attitude to someone or something, his views, opinions, preferences; as a rule, such measures are obtained using questionnaires, questionnaires, interviews);
- signs of accidents (used when the purpose of the study concerns, for example, the problem of selecting for work such people who are less prone to accidents).
Empirical validity. What is it about?
Few people know what empirical validation is. In the case of content validity, the test is evaluated by experts (establishing the compliance of the test tasks with the content of the object of measurement). And the empirical one is always measured using statistical correlation: the correlation of two types of values is calculated - test scores and indices for an external parameter chosen as a reliability criterion.
Constructive
Not everyone knows the types of validation methods. What is construct validity? It has to do with the theoretical construct itself, and consists of looking for factors that explain test performance behavior.
As a specific type, construct validity is legitimized in an article by Mil and Cronbach (1955). Using this form of validity, the authors evaluated all test surveys that were not directly aimed at predicting some significant criteria. The research contained information about psychological constructs.
Content Validity
You stillyou ask: "Validation - what is it." Consider content validity. It requires that each problem, question, or task belonging to a given area has an identical chance of becoming a test item.
Content Validity evaluates the suitability of the test entity to the measured area of behavior. Verifications created by two groups of developers are carried out on a sample of those being checked. The reliability of tests is calculated by splitting questions into two zones, resulting in an index of content validity.
Predictive
We continue to consider validation methods. Predictive validity is also established by an external, fairly reliable criterion. But information about him is collected some time after the verification.
The external yardstick is usually an individual's vocation, shown in any assessment, for the type of occupation for which he was selected on the basis of the results of diagnostic tests.
Although this method is most suitable for the task of diagnostic tools - the prediction of future success, it is very difficult to apply. The accuracy of the forecast is inversely dependent on the time assigned for such a prediction. The more time passes after the measurement, the more factors need to be taken into account when evaluating the predictive value of the technology. However, it is almost impossible to take into account all the factors that affect the prediction.
Retrospective
Agree, validation is a very intricate process. It is known that retrospective validity is revealed on the basis of the criterion,reflecting the state of a quality or event in the past. Can be used to instantly obtain predictive technology sources. So, in order to revise the extent to which good results of a skill test correspond to rapid learning, one can compare previous expert opinions, performance assessments, and so on in individuals with low and high diagnostic indices at the moment.
Discriminant and convergent
Types of validation are of interest to many. Let's find out what discriminant and convergent validity are. The strategy of introducing established items into the test depends on how the psychologist identifies the diagnostic construct. If Eysenck defines the quality "neuroticism" as independent of introversion-extroversion, then this means that his questionnaire should equally represent positions that will be approved by neurotic extroverts and introverts.
If in practice it turns out that the items from the "introversion-neuroticism" quadrant will prevail in the task, then from the position of Eysenck's theory, this means that the indicator "neuroticism" is loaded with an irrelevant indicator - "introversion". An identical effect appears when there is a bias in the sample - if there are more neurotic introverts in it than the same extraverts.
To avoid such complications, psychologists are willing to deal with empirical items that inform only about a single factor. But in reality, this requirement is never fulfilled: each empirical index turns out to be determined notonly by the factor that we need, but also by others - irrelevant to the problem of measurement.
Thus, for factors that are conceptually defined as orthogonal to the measured (occurring with it in all combinations), the test creator is obliged, when selecting items, to use the strategy of inauthentic balancing.
The correspondence of points to the measured indicator guarantees the convergent validity of the test. Consistency of items with respect to irrelevant sources provides discriminant validity. Empirically, it manifests itself in the absence of significant correlation with the test, which measures the conceptually unique quality.
Tool kit
In the general set of validation methods, authors usually include:
- non-formalized (from simple tricks to carefully review the list of alternatives in the questionnaire to more sophisticated procedures for theoretical step-by-step analysis);
- formalized, which include procedures and technologies of mathematical statistics: testing of statistical hypotheses, calculation of estimates, correlation analysis, construction of confidence intervals, assessment of relationships between variables, dispersion, factorial, regression and structural analyzes, and so on.
Creating tools
And yet, what is validation? Sophisticated validation tools were first created by psychologists. Back in 1959, a special technique was developed by D. Fiske and D. Campbell (USA). She received a completely natural for English, but untranslatable intoour language, name: multi-method-multi-damn matrix (MTMM). This matrix was a table of correlations. It consisted of two very attractive inventions, one of which was intended to reveal convergent truth, and the other for discriminant truth.
Its authors argued that anyone can prove internal acceptability if:
- values, between which a high level of theoretical connection is assumed, will find an identical level in the empirical (convergence);
- values that are theoretically unrelated will turn out to be unrelated empirically after the test is completed (discrimination).
Roughly speaking, convergent validity should state that there is much more similarity between two teams of employees, for example, construction and assembly workers, in terms of labor market chances than between owners and employees. If the theoretically intended relationship is found empirically, your sample is valid.
Discriminant validity shows the degree of identification of various phenomena. To take the same example of the labor market, one would expect that a well-formed theory could, by its means, distinguish between the opportunities of owners and employees in the labor market. You can't confuse them, and your theory can tell them apart.
If you created a scale that measures mathematical abilities, then in case of validity of the convergent indexes of mathematical talents should correlate well with the general skills of a person, if at the theoretical level such a relationshipthere is, and badly - with aesthetic abilities that require completely different talents from a person than counting skills, if, of course, a low correlation is proclaimed by your theory.
Types
Validation is the process of tuning, improving the created psychodiagnostic methods. Its basic task is to ensure that the technology diagnoses exactly what the developer needs. A distinction is made between theoretical and pragmatic validation.
For the first type, the cardinal problem is the relationship between mental phenomena and their indices, with the help of which these phenomena are trying to be known. It shows that the results of the methodology and the abstract intention of the author are the same.
To establish abstract validity, it is very difficult to find any independent criterion that is outside the methodology. In the history of psychodiagnostics in the early stages, therefore, the reliance was on the instinctive concept that the test measures:
- A method was considered valid if it was simply "obvious".
- The proof of validity was based on the prospector's belief that his technology could "understand the testee".
- The scheme was only considered valid because the theory behind the technology was "pretty good".
Next began the search for evidence justified by science. It should be noted here that the imperceptible accumulation of an arsenal of technologies with an already substantiated and knownvalidity. If a psychodiagnostician creates a scheme for assessing a quality and it is known that other, valid technologies are aimed at the same assessment, then you can study the correlation, compare the results according to someone else's and your own method.
If the correlation coefficient is too high, then the generated scheme has impressive abstract validity. If you are in doubt about what exactly a technology evaluates, compare the results for it with the results for someone else's valid schemes that recognize adjacent (suspicious) properties. If the correlation values turn out to be unexpectedly large, it can be concluded that the method evaluates not what was expected.
Thus, if there are already other schemes aimed at measuring exactly the same norms or related ones, we can determine discriminant and convergent validity.
Nuances
So, what is validation? In simple words, this is a revision of the product to see how it matches the declared characteristics. That is, any smartphone validation will not pass until then. Until customers make sure it has the camera and storage they are willing to pay for.
The validation criterion is a measure of mental quality that is independent and direct from the test being validated, and the psychodiagnostic scheme is aimed at studying.
Current validity - a feature of the test, reflecting its ability to distinguish between subjects on the basis of a diagnostic feature that is the object of study in thismethod.
Competitive validity is estimated by the correlation of the created test with others, the validity of which relative to the measured parameter is established. Differential validity can be explained by the example of interest tests.
