When conducting psychological testing, the following rules must be observed:

Testing should be done with the participation of a person familiar with the basics of psychology.

A person should not be subjected to testing fraudulently or against his will.

The test subject must be warned that during testing he can unwittingly communicate such information about himself, about his thoughts and feelings, which he himself is not aware of.

The test subject has the right to know the results of his testing. In this case, it is imperative to indicate that the results are probabilistic, and depending on the test method, the probability can be 60-80%.

Test results should not injure the subject and (or) reduce his self-esteem.

The subject should be informed about the purpose of testing and the forms of use of its results.

The tester must ensure an impartial approach to the test procedure and results.

Information about test results should only be provided to those to whom it is intended.

The tester is obliged to keep professional confidentiality.

For people involved in psychodiagnostics, special qualification requirements. The main ones are the following: good theoretical training, thorough knowledge of psychodiagnostic techniques and the rules for their application, the presence of sufficient experience in the practical use of the relevant techniques.

The users of psychodiagnostic techniques can be specialists in fields related to psychology: teachers, doctors, sociologists, social workers, engineers, economists, etc.

Specialists-non-psychologists have the right to use only some well-theoretically and psychometrically grounded methods that do not require special knowledge when interpreting the results (for example, tests of educational achievement or professional skill). Techniques that are not provided with an unambiguous standard instruction, the necessary indicators of reliability and validity, requiring the parallel use of highly professional expert methods, cannot be used by non-psychologists.

It is important to take into account requirements for psychodiagnostic techniques:

the goals, subject matter and scope of the methodology must be unambiguously formulated;

the theoretical concept underlying the methodology should be outlined;

should be clearly marked area of ​​application of the technique- a special social environment or area of ​​social practice;

there must be a clear algorithm for carrying out the method;

the processing procedure should include statistically sound methods for calculating and standardizing the test score;

test scales should be checked for representativeness, reliability and validity;

the self-report procedure should include means of control, reliability;

periodically, the normative data should be corrected. The non-psychologist user must first

consult with psychologists working in this practical field about what kind of techniques can be applied to solve the assigned tasks.

The user who gets access to psychodiagnostic techniques automatically assumes the obligation to comply with all the requirements of professional secrecy, follows all ethical standards in conducting the survey in relation to the subject and third parties.

TO non-psychological users the following requirements:

there should be a consultation with a psychologist or general knowledge in psychodiagnostics;

if the technique requires special training, then the following options are possible: it is necessary to choose another technique; get trained; invite a specialist in this technique, or refuse the technique;

the user automatically takes full responsibility for compliance with all requirements for compliance with professional secrecy;

ethical standards must be observed in the conduct of the survey;

officially released methods should be used.

The user of psychodiagnostic techniques must have test competence, which includes:

comprehensiveness of assessment - collection of complete, integrated information about a person (testing, observation, conversation);

correct use of the test;

psychometric knowledge, i.e. knowledge and correct use of statistical measurement principles;

understanding the limitations of test indicators;

the accuracy and correctness of the assessment;

appropriate use of norms;

the ability to present the results to the test subject, so as not to cause harm.

1. Informing the subject about the purpose of testing.

2. Familiarization of the tested with the instructions for performing test tasks and the achievement of the researcher's confidence that the instruction is understood correctly.

3. Providing a situation of calm and independent performance of tasks by students; maintaining a neutral attitude towards the tested, avoiding prompts and help.

4. Compliance by the researcher with the guidelines for processing the data obtained and interpreting the results, which accompany each test or corresponding task.

5. Preventing the dissemination of psychodiagnostic information obtained as a result of testing, ensuring its confidentiality.

6. Familiarization of the subject with the test results, providing him or the person in charge of relevant information, taking into account the principle "Do no harm!" (in this case, it becomes necessary to solve a series of ethical and moral problems).

7. Accumulation by the researcher of information obtained by other research methods and techniques, their correlation with each other and determination of the consistency between them; enrichment of their experience of working with the test and knowledge about the features of its application.

Testing quality indicators are: validity, reliability, representativeness, economy.

Validity(from English - valid, suitable, having a topic) of the test answers the question of what exactly the test reveals, how suitable it is for identifying what it is intended for. For example, ability tests often reveal something different: fitness, relevant experience, or lack of it. In this case, the test does not meet the validity requirements.

There are various types and ways of establishing the validity of a test. Most often, validity is determined by comparing the results of a psychological test with expert assessments of this quality in subjects.

Reliability the test is determined by how stable the obtained indicators are and how they do not depend on random factors. A reliable test should have consistency in test scores obtained from repeated testing of the same subjects.

Representativeness test means that the sample should adequately reflect the characteristics of the contingent of subjects being surveyed. If you have test norms obtained on children of 7 years old, then they cannot be used to assess (for example, personal abilities) of children of 10 years old, although they are also younger students.

Under cost-effectiveness understand the feasibility of using the test, determined by the presence of an acceptable ratio of costs for testing (time, labor, finance) and the benefits of it.

The testing method is widespread in personality research, but at the same time it is the most controversial, because:

1) Successful solution of the test under normal conditions does not mean the success of similar mental efforts in difficult conditions.

2) Knowing the nature of the test procedures, one can prepare in advance for solving a similar test, and in some cases, find out the result in advance.

3) The results of testing at the moment may differ significantly from the results of similar tests after a considerable period of time due to the uneven development of abilities in people.

4) Most of the parameters that are reliably diagnosed by test methods are not the leading, defining personality traits.

Thus, the capabilities of a testing method should not be overestimated. The tests should be used in conjunction with all other research methods. It is advisable to use not a single test, but a combination of them, that is, a battery of tests.

Assignment to topic 2.4. Testing

1. Who sets the testing standard:

a) those who conduct the test,

b) the developer of the test,

c) those who check the test.

2. What is not an indicator of the quality of the test:

a) validity,

b) standardization,

c) representativeness.

3. Tests of aptitude reveal.

Let's consider the basic rules of testing for testers and test subjects.

When testing, the teacher (test methodologist) must first fulfill the following basic conditions:

1. Define the purpose of testing
2. Define a testing hypothesis.
3. Determine the training model of the test taker (trainee).
4. Determine the pedagogical situation (scenario) of testing.
5. Develop a test plan
6. Determine the structure of the test.
7. Structure the test according to didactic goals and connections.

Then the following rules of thumb must be followed.

The time for testing must be determined by the complexity and complexity of the test tasks.

The average testing time should be no more than 60–90 minutes. For heterogeneous tests, this value can increase several times, but, as a rule, with damage to the psychological and physiological state of the subject, which is not always justified.

Measures are needed to limit the selection of answers "at random", since the probability of guessing is usually high (in particular, for tests of a closed form and four variants of equally attractive answers, it is equal to 0.25).

The group of subjects should be selected according to the same principles of motivation and this condition should not be violated throughout the testing.

The test should not be focused on finding out the opinions of individual subgroups of trainees.

Educational, motivational testing should provide for the possibility of re-passing the test. It should also be borne in mind that the repetition of the test task has the effect of verbal assistance and is in no way acceptable for certification and control testing.

To obtain sufficiently reliable indicators of the quality of test items, a representative sample (complete from the point of view of reflecting different levels of education by tests) is required.

A test corresponding to the level of preparedness minimizes the time for individual testing, but this is often a difficult situation to implement, which complicates the development of high-quality assignments.

The computer testing process should ensure user friendliness and multimedia interface, interactivity, adjustability, logging, automatic calculation of the test result, integration with other forms and systems of control and training, visualization, etc.

During proficiency testing, trainees must be familiar with the standard forms of test items, the technology of work in a particular testing system no later than one to two weeks before the start of the test.

In addition to these basic requirements for testing, there are rules for testers as well.

The subjects should have objective information about the upcoming testing in advance (the allowed time reserve, the didactic units covered, etc.).

It is necessary to focus on a reasonable personal level of training (for this purpose, it is possible to conduct frequent self-test and learning testing).

It is important to learn how to plan time during testing, to allocate time rationally between tasks of various groups of complexity throughout the testing period.

Finding the answer should start with easier test items and gradually move towards more difficult tasks, for example, according to the scheme: A - B - C.

Difficult tasks must be performed iteratively.

Complete each section, each topic not only by solving problems, but also by teaching, motivational and diagnostic self-testing using similar tests.

Not only individual quanta of knowledge are important, but also the ability to quickly find the right choice, to connect these quanta. Tests linking different didactic units are useful.

It is important to analyze the test results and adjust the learning process in order to improve it.

The correct choice of educational and methodological literature is important in preparation for testing - from textbooks to Internet resources. It is necessary to have a sufficient set of high-quality test items of a training nature.

Unfortunately, the (mostly "unwritten") rules of testing ethics are often violated.

You can not disturb neighbors with your questions - this will reduce their own points.

Do not ask the teacher questions about the assignments (he has already made sure that the prepared subject does not have such questions).

In extreme situations, you can ask a question about incomprehensible moments of testing technology and presentation of results.

In the course of creating a website, you often have to face the question - which style is better to choose, how to place the elements so that they attract attention and bring more profit? Such a marketing tool as A / B testing will help solve this problem.

In simple words, A / B testing Is a measure of effectiveness by comparing two options. In general, the method looks like this: option A is created, after which one parameter changes in it (font, element location, etc.) and this will be option B. Both options are equally shown to site visitors, after which a comparison is made - from which page users more often subscribed, ordered, called.

But despite the seeming simplicity, A / B testing has its own rules, which must be carefully followed in order to get a good result.

What can be tested?

A split test is a tool that is appropriate to use to evaluate any item. The choice depends on the tasks that are set at this stage of time. So for example, if a large number of visitors come to the page, but the bounce rate is high enough, this is a sign that something is wrong with the design, then it’s time to change the design or content. With the help of testing, you can optimally adjust the color content, fonts, and the arrangement of elements.

It is also done if the site gives low conversion rates - in this case, you should pay attention to the elements of the subscription or checkout. By changing the color and size of the button, you can achieve significant efficiency gains.

If we talk about the elements being tested, then most often they are:

• design, font, text and layout of subscription forms;
• titles and descriptions;
• design of key page elements;
• registration of the price of goods;
• images, including goods;
• text in descriptions.

How do I conduct A / B testing?

Testing requires the use of specialized tools. Google Experiments can be considered one of the most common. It is located in the Google Analytics interface. To start a test, you need to set its parameters - name, purpose (to define it yourself or choose from the proposed ones), percentage indicators and the method of notification of the results. Next, information about the site is filled in and you can run the test. The first test results will appear in a day.

Another good tool to help marketers is RealRoi. It is easy to work with it, and if you have any questions, there are explanatory videos on the site itself. The resource is also free.

Test rules

Before embarking on A / B testing, it is worth familiarizing yourself with the key rules for conducting it.

Defining goals

Before doing A / B testing, you need to clearly define your goal.

In addition, the site being tested must meet the following criteria: a large and constant flow of visitors, frequent conversions - calls, orders, subscriptions, a well-developed analytics system. If these conditions are not met, then testing will not show results, but will only take time and effort.

Construct a clear hypothesis

A / B testing should have a clear hypothesis. Comparing two elements out of curiosity is a waste of time. First, you need to identify the problem of the page, as well as the desired solution. For example, the problem might be a small number of mailing list subscriptions. In this case, you should pay attention to the subscription form and its motivation. Having made an assumption about the design and filling of the form, it is already possible to conduct a split test.

If you do not know how to formulate a hypothesis and what the problem may be, then you can contact the site visitors with questions - what they like here, what they would like to see, you can analyze the statistics of calls to the technical support service.

In this way, you can identify key user problems. Pages with poor behavioral characteristics will also indicate this. You can analyze user behavior using Webvisor in Yandex.Metrica. This is a great tool that will show - where the user clicked, how he worked with the mouse and which elements caught his attention and which went unnoticed. After analyzing the results, you can begin to identify the weak points of the site.

Set the indicators you are looking for

What are you focusing on during the split test - do you want to increase conversion by 5%, extend a user's visit to the site by 1 minute, or increase the number of calls by 2%?
Depending on this, an indicator is taken, which the system is guided by during testing. At the same time, it is important that statistics on this page are correctly and clearly configured, which will subsequently help to identify the rate of change.

Testing is carried out on only one element

You only need to choose one detail - the font color or the position of the title, the increase in the font on the button, or the change of the image. You cannot test with changing several elements at once. Because if there are several variables at once, it is impossible to determine what exactly influenced user behavior. Thus, A / B testing will rather turn from a working tool into a regular lottery.

Set test parameters

Set the parameters of the tests - in particular, the number of people on whom it will already be possible to draw conclusions. This indicator should be significant enough and it depends on the number of website visitors as a whole. It is impossible to draw any conclusions by analyzing the actions of only 10-20 people. So in conducting A / B testing, it is important to operate on accurate data and be patient while waiting for a set number of users to visit the site.

At the same time, it is important to establish - to which users the tested pages will be shown? Will they be regular visitors or new users? Or will it be a certain part of the visitors that you have established? It all depends on what the site is focused on and what specifics it adheres to. If this is an online store of popular goods (cosmetics, electronics, etc.), then you can draw a conclusion on any visitors. Otherwise, it is worth considering this issue carefully.

Set a test date

Set the date for testing - it is better to be a week. Even if the required number of visitors was recruited after the expiration of the term, you should not interrupt the experiment. Since users show different behavioral characteristics on different days of the week, it is impossible to draw an accurate conclusion based on only 1-2 days. It is also worth considering how long it takes a consumer to make a purchase on your website. Perhaps it is 5 days, and maybe 2 weeks.

Control your visitors

For the experiment to be clean and correct, the visitors must be identical. That is, there should not be much difference between the two groups in basic social indicators. To test this, before testing, divide your audience in two and show them the same page. If the conversion rates are different, it means that the audience is not equal, in this case, the test will be meaningless, since it is impossible to understand what exactly led to the result - audience features or page changes.

Consider external factors

It seems like a waste of time to do testing in the middle of the New Years or during the summer vacation season. It is best to choose quiet days for tests, when most of the audience is not on the road, and the social situation in the world does not cause great excitement.

Monitor the site during testing

You can't just run a split test and relax. Be sure to check if everything works, if the buttons are clickable. If, during the test, the system shows things that seem incredible at first glance, then it is worth checking again if everything is running correctly. Even if you get clearly good results, do not interrupt the test - only after going through the full cycle can you focus on the data obtained.

Re-test

After the test is complete and the results are pleasing to the eye, technicians often do another retest to make sure the results are correct.

In conclusion, I would like to note that A / B testing is a useful tool. But first of all, it is important for the marketer to be able to identify the problems of the site and suggest possible solutions.

So, for example, if a site owner suddenly wants to change the splash screen on the main screen, this is unlikely to clearly affect the quantitative indicators. It is possible to improve conversion rates only by conducting thorough preparatory work by analyzing statistics and behavioral factors of users. It is important to understand which elements of the site are not working and how you can change them. If you just make unreasonable assumptions and check them on A / B testing - of course, in the end it can and will give some results, but most likely insignificant and at the same time will take a lot of time and effort.

It is also worth noting that A / B testing is a tool that must be used constantly. With the passage of time and fashion trends, user behavior, their interests and preferences change, so that on any site you can constantly improve and change something. And A / B testing is exactly the tool that will help you understand the behavioral characteristics of visitors and increase the effectiveness of your pages.

In addition to the requirements for test standards, there are certain, strict rules for conducting testing, processing and interpreting its results ... The most important of these rules are as follows:

1. Before applying this or that test, the psychologist needs to get to know him and try it on himself or on another person. This will allow in the future to avoid possible errors associated with testing and due to insufficient knowledge of its nuances.

2. It is important to make sure that before starting the test tasks, the subjects understand them well and the instructions accompanying the test.

3. During testing, it is necessary to ensure that all subjects work independently, independently of each other and do not influence each other that could change the test results.

4. For each test, there must be a reasonable and verified procedure for processing and interpreting the results to avoid errors that arise at this stage of testing. This, in particular, concerns the methods of mathematical and statistical processing of primary data, which must also be strictly and in advance established.

Test classification

Classification of tests by purpose:

· General diagnostic (personality tests of the type of Cattell or Eysenck questionnaires, tests of general intelligence by Veksler, Binet-Simon, edited by Termen and Merril, etc.);

· Professional aptitude (many general classification tests - general army classification test, Flanagan's ability classification tests, etc.);

· Special abilities (technical, musical, tests for pilots, radio operators, etc.);

· Achievements (for example, pronunciation, quality of writing, etc., ie, tests designed to assess the results achieved by students in the learning process.).

By means used in the testing process:

· Blank (tests "paper and pencil", which use test books or forms in which the subjects mark or write down the correct answers - personality questionnaires, Bennett's mechanical intelligence test, most achievement tests, etc.);

· Subject (in which it is necessary to manipulate material objects, the effectiveness of these tests depends on the speed and correctness of the tasks - Koss cubes, the Stenquist test for assembling structures and parts, the test "Addition of figures" from the Wexler set, etc.);

· Instrumental (tests using devices for studying the peculiarities of attention, perception, memory and thinking - a "lie detector", an apparatus for determining brain encephalograms, etc.);

· Practical (appeared relatively recently, these tests are similar to the laboratory works known in our country - in chemistry, physics, biology, etc., but they are supplied with appropriate instructions and have test equipment);

· Software (tests using computer programs).

By the number of simultaneously examined people:

· Individual (Wechsler tests, Luscher test, etc.);

· Group (most of the standardized questionnaires that assess the basic characteristics of personality, the set of Flanagan tests, most tests of achievement, etc.).

The tests are divided according to the form of the answer:

· Oral;

· Written;

· With the use of a computer.

By leading orientation:

· Speed ​​tests containing simple problems, the solution time of which is limited so much that not a single test subject has time to solve all the problems at a given time (Landolt's, Bourdon's rings, "encryption" from the Veksler set);

· Power or performance tests involving difficult problems, the time to solve which is either not limited at all, or gently limited. The success and the way of solving the problem are to be assessed. An example of this type of test would be the assignments for the written final exams for the course of the school;

· Mixed tests that combine the features of the two above. In such tests, tasks of various levels of complexity are presented: from the simplest to the most difficult. The test time in this case is limited, but sufficient for solving the proposed problems by the majority of the surveyed. The assessment in this case is the speed of the tasks (the number of tasks completed) and the correctness of the solution. These tests are most commonly used in practice.

By the degree of homogeneity of tasks:

· Homogeneous, having, as a rule, one scale, which allow one to assess one property or quality of personality and include tasks that are similar in nature, but differ in specific content;

· Heterogeneous, having several scales that allow assessing various personality characteristics and include tasks that differ both in nature and in content.

By the nature of the actions:

· Verbal (associated with the need to perform mental actions - verbal and logical tests, questionnaires to test knowledge, establishing patterns, etc.);

· Non-verbal (associated with the practical manipulation of objects - cards, blocks, details).

Directional, i.e. by what exactly is supposed to be studied using this test:

· Tests of intelligence, revealing the features of the latter;

* Personality tests (sometimes called temperament tests), with the help of which the personality traits of the subject are studied; mood and state tests, which look at temporal states such as emotions;

· other.

By type of rationing:

· Focused on statistical norms - tests, the basis for comparison in which are appropriately justified statistically obtained values ​​of the performance of a given test by a representative sample of subjects;

Criterion-oriented - tests designed to determine the level of individual achievements of the test subject relative to a given criterion that exists in real practice and a previously known level of knowledge, skills, and abilities necessary to perform a certain type of activity. The criterion can be determined on the basis of an expert assessment (for example, the criterion of school success can be determined by interviewing teachers working in this class or with this child) or the practical activities of the subjects (the criterion of school success can be determined by grades for a quarter or a year);

· Predictive, focused on the success of further activities;

· Irregular.

By the nature of the answers to the questions:

Open type (with free answers - when the subject needs to independently add a word, phrase, sentence, sign, formula, etc.)

· Closed type (with prescribed answers - when the subject needs to choose one or another option from the offered answer options).

In addition, there are the following possible answer options used in open-type tests:

· Tasks of the addition - the subject must formulate the answers, taking into account the restrictions provided for in the task;

· Free presentation - the subject must independently formulate the answers, because no restrictions are imposed on them in the task.