ON THE PROBLEM OF CORRELATION OF RESULTS OF DIFFERENT TYPES OF TESTS

Национальный исследовательский университет в системе непрерывно образования - Н.Н. Пахомов

T.S. Oznobikhina, E.K. Henner

Perm State
University, Perm, Russia

ON THE PROBLEM OF CORRELATION OF RESULTS OF
DIFFERENT TYPES OF TESTS

The
use of testing technologies for knowledge control has become an integral part
of the modern education process. Out of many types of tests the most frequently
used in practice is the cover test. The cover test is a number of alternative
assignments aimed at choosing one correct variant out of the offered ones.
However, it is the cover test that is severely criticized by many specialists.

In
the paper the comparative analysis of the results of two types of tests passed
by the first-year students of the three faculties of Perm State University, who
studied the same subject (“Computer Science”; the final test on the topics
“Word Processor”, “Tabular Processor”, “Data Base”, “Radix” and “Information
Measurement”) is carried out. The purpose of the experiment was to test the
same students two times between which there was a short interval (2-3 days), in
two variants of the same test presented for the first time in the form of the
cover test and in the second - in the form of the free response test (a test
when a student fills in the space with an answer).

Consider the question of the cover test: “What can be an example of
graphic information?”

1)       
lyrics

2)       
table
of values of trigonometric functions

3)       
image
on the screen

4)       
poem
in a book

5)       
telephone
talk.

The
free-response test contains only the question “What can be an example of
graphic information?” and the space for the student to fill it in with his\her
answer (not all the questions are so primitive, this is just an example). It is
obvious that it is not necessary for the student in the second type of the test
to write down only the third answer out of the mentioned above; any essentially
correct answer is regarded as a correct one. Moreover, the same questions in
the two variants of the tests are sometimes formulated differently to be
grammatically and stylistically correct in Russian.

149
students in all passed 8 variants of the tests; each test contained 30 test
assignments. Each test assignment was assessed according to the binary scale (0
- false, 1 - true).

The
aggregate result of the two types of the tests is presented by the diagrams in
Fig.1 and Fig.2 which show the correlation between the number of testees and
the total number of the scores.

In the
abstract the authors will not describe the results of statistical processing            and further
details      of                                                                                                                           the ex­

Н.Н. Пахомов: Национальный исследовательский университет в системе непрерывно образования. Часть 2.