124 PSYCHOLOGY IN HUMAN AFFAIRS
of norms, or answer frequencies, are available). To achieve objec-
tivity, test items are usually stated in some manner that requires no
writing. For example, true-false items are statements that the testee
recognizes and labels as being either true or false. (Columbus dis-
covered America in 1607.—True, False) Multiple choice items are
statements that may be completed by any one of a number of alternate
answers. The testee must recognize and choose the right one. (The
sum of 5 and 8 is—10, 17, 23, 13, 14.) Completion items are those
statements that are not finished and the testee must fill in the missing
concepts. (Columbus discovered America in------------) Matching
items are usually arranged in two lists (viz., a list of dates and a list of
battles) and the testee must connect the items in one list with their
mates in the other list.
Tests are standardized by being given to a large number of people.
Then, tables of frequencies are constructed so that a score' can be
compared to the scores made by others. In fact any test, whether
published or homemade, must have some sort of norms or the scores are
meaningless. This is likewise true with data in any field. A child
weighs 36 Ib. This is meaningless unless we know how much other
children of his age and height weigh. To standardize a test is to find
out what scores other people make on it. A high score is high only
when compared with the scores made by other people.
Also, published tests usually have satisfactory (or, at least, known)
reliability and validity. A high reliability means that a test can be
repeated with the-same results. If a pupil makes a high score on a
test one day and a low score on the same test the next day, it is then
unreliable. The reliability of a test may be computed in various ways.
It may be given to the same pupils on different occasions. However,
so many other variables (physical condition, intervening experiences,
etc.) may cause differences in scores that it is unwise to condemn, a
test on this basis. It is better to compare the score made on a part of
the test, usually the even-numbered items, with the score made on
another part of the test, the odd-numbered items. If half the test
gives relatively the same score as the other half, it is judged to be
reliable. Some tests have two forms, which are equated, and either
form may be used. When pupils make .the same score oa both forms,
taken, at different times, the test is then said to be reliable.
The validity of a test refers to its worth or value in doing what it is
supposed to do. If a history test measures historical knowledge, it
is then said to be valid. The validity of a test is usually determined
by comparing it with some other (presumably better) measure of the
same characteristic. If the scares made on a history test compare