Page:Kohs-Block-Design tests-1920.pdf/18

From Wikisource
Jump to navigation Jump to search
This page has been validated.
374
S. C. KOHS

It will, no doubt, be acceded without much question that the block-design tests are less affected by school training than the Binet.

At any rate the total correlational evidence seems to indicate that the block-design tests possess a high degree of reliability.

(5) Conformity of Intelligence Quotient Distribution with Normal Probability

A very necessary index in weighing the reliability of any standardized test is to determine the extent to which an actually found distribution conforms to its theoretical distribution.

In the following table are presented the I. Q.-range distributions for the Binet and the block-design tests. The respective percentage values are compared with what one should theoretically expect.

Table VII
Intelligence Quotient Ranges
26
to
35
36
to
45
46
to
55
56
to
65
66
to
75
76
to
85
86
to
95
96
to
105
106
to
115
116
to
125
126
to
135
136
to
145
146
to
155
156
to
165
166
to
175
Stanford
Binet
Obtained
1.7 5.5 16.5 22.7 28.2 13.8 8.3 2.1 1.0
Theoretical
Expectation
  .16 1.6   8.5 23.42 32.64 23.42 8.5 1.6   .16
Block-
Design
Obtained
.034 .034 1.4 6.5 6.2 14.4 15.1 18.9 14.4 10.0 5.2 5.2 .07 1.0 .034
Theoretical
Expectation
(Median at 99)
.49 1.28 2.78 5.21 8.67 12.05 14.66 15.30 13.84 10.69 7.25 4.11 2.08 .88 .33

The average deviation from theoretical expectation for the Binet I. Q. ranges is 3.3 per cent, per I. Q. group. The average deviation for the block-design tests is only 1.4 per cent, per I. Q. group.

In conclusion, one may state that the evidence presented seems to indicate not only that the tests measure intelligence,