2. Statistical nature of test scales. How to increase the share of the
constant component and reduce the share of the random component in the
total score on the test scale.
4. Evaluation of the type of distribution of test scores and checking the
stability of the distribution. The following parameters are used: arithmetic mean,
standard
deviation, skewness, kurtosis, Kolmogorov's criterion. The general
logic for checking the stability of a distribution is based on inductive reasoning:
if a “half” (obtained on half of the sample) distribution well models the
configuration
In the representativeness of test norms, the following are considered:
integer distribution, then we can assume that this integer distribution will model
well the distribution of the population.
5. Test standards (or test norms).
Problems:
Proving the stability of a distribution means proving
would be constant carriers of a given value of the measured
The sample on which the statistical
test norms are determined,
Percentage scale. Percentile - The percentage of subjects in the
standardization sample who received an equal or lower score than that subject's
score. Percentiles indicate the relative position of an individual in the
standardization sample. They can be considered as rank gradations, the total
number of which is one hundred, only (unlike
called the standardization sample. Its population is usually
less than 200 people.
properties. The role of indirect standards in psychometrics is performed by the tests themselves.
Representativeness means that, with some predetermined or
3. The problem of measure in psychometrics. In differential psychometrics
The raw scale itself may have practical meaning.
With
a certain statistical error, we can assume that the distribution of the studied
characteristics presented in the sample population corresponds to their real
distribution. The representativeness error is the difference between the
characteristics of the sample and the general population.
there are no physical standards: we do not have individuals who
Standardized scales: IQ scale, T-scale, Stanine scale (standard nine), Stan
scale (standard ten).
representativeness of the rules. The traditional way to prove stability is to find a
good approximation of the empirical distribution to some theoretical one (for
example, the normal distribution, although it can be any other).
measured properties characteristic of the subjects of a particular group. They
are changed every 5 years.
Representativeness (from French - indicative) of test norms is the property of a
sample population to represent the general population.
1. Standardization of the scale.
Machine Translated by Google
distribution of results in the sample of subjects.
;
The diagnostic technique differs from any
research one in that it is
standardized. As A. Anastasi notes, standardization is the uniformity of the
procedure for conducting and evaluating the performance of the test.
Standardization is the uniformity of the procedure for conducting and evaluating
the performance of a test. Standardization as the development of uniform
requirements for the procedure of the experiment and as the definition of a single evaluation criterion
Z=
diagnostic test results.
Z-score (standard score on a standard Z scale).
Socio-psychological standard.
Independent of test results and objectively specified. The socio-
psychological standard is implemented in the totality of tasks that make up the
test. Therefore, the test itself in its entirety is such a standard. To analyze the
data regarding their proximity to the
socio-psychological standard, considered
as 100% test completion, the subjects are divided into 5 subgroups. For each of
the subgroups, the average percentage of those who correctly completed the
tasks is calculated. 10% - the most successful, 20%
- close to successful, 40% -
average, 20% - little successful, 10% - the least successful.
criterion. High efficiency is shown by highly specialized
The initial total score calculated using the key is not
an indicator that can be
interpreted diagnostically. It is called the "raw test score". To apply the test
norms, it is necessary to transfer the test scores from the "raw" scale to the
"standard". This procedure is called "test score standardization".
diagnostic techniques targeting very specific and narrow criteria. Well-established
in education (achievement tests and criteria-based tests).
With the simplest linear standardization, it is first calculated -
ranking) is counted from the bottom. Therefore, the lower the percentile, the
worse the position of the individual. Percentiles are different from percentages.
Percentage indicators fix the quality of completed tasks. Percentile is a derived
indicator that indicates the share of the total number of members of the group.
criteria norms. The target is used as a reference.
The standardization of the experimental procedure implies the regulation of the
procedure, the
unification of instructions, examination forms, methods for
recording results, the conditions for conducting the examination, and the
characteristics of the contingents of subjects (the scope of the test is indicated).
In the second case, standardization is understood as the transformation of a
normal (or artificially normalized) rating scale into a new scale based not on the
quantitative empirical values of the indicator under study, but on an assessment
of
its relative place in
Do'stlaringiz bilan baham: