Study on Adaptation of the Snijders-Oomen Non

advertisement
Psychological Science China, 20, (1997), pp. 97-103
(published in Chinese language; translated by Wu Yingxiang)
Study on Adaptation of the Snijders-Oomen Non-verbal Intelligence Test
Zhang Houcan, Gongwen, Sun Yanqing & Tian Xiaoxun
(Psychology Faculty, Beijing Normal University)
Abstract
Snijders-Oomen Non-verbal Intelligence Test, published in 1965, has the advantage of being
independent on the use of language during the course of testing. We have been engaged in the study on the revision
of Snijders-Oomen Non-verbal Intelligence Test since 1995. During the course of the study, we selected and
pre-tested three groups of children from Beijing, Tianjing, Shandong and Hebei, which are: a). 6 years old, b). 11
years old and c). 14 years old. There are 334 children in the test totally. Based on the data from the test of SON-R, the
test is seen to possess a good inner structure and criterion-related validity. However, a few items in the test that are
not suitable for Chinese children due to different culture backgrounds, have been altered and adjusted.
Key words:
SON test
Intelligence assessment
Adaptation study
1. Foreword
At present the following tests are commonly used in China: 1. Wechsler Intelligence Test for children, 2. Wechsler
Intelligence Test for adults, 3. Raven’s Standard Progressive Matrices. These three tests are becoming more and more
widespread because of their own advantages and the different needs of different educational practices. However these
tests can no longer satisfy the growing social needs as time passes by. Besides, as copyright is being paid more and
more attention, it is not possible for us to go on revising foreign psychological tests without the permission of the
copyright owner. Therefore, we are supposed to increase the cooperation with foreign psychometric experts. Dutch
psychologists J.A. Laros and P.J. Tellegen contacted us for they would like to introduce SON-R5.5-17 Test in China.
Due to the nonverbal character of SON-R in both the testing process and test structure, we revised SON-R Test,
Chinese Version in March, 1995. We have made some elementary achievement by nearly one year of work.
Snijders-Oomen Non-verbal Intelligence Test (SON) was first published in 1965. SON-R5.5-17 Test, the latest
revised version of SON, was finished in 1989, which is suitable for the children at the age between 5.5 and 17. As a
non-verbal intelligence test, it contains 7 subtests---Categories, Mosaics, Hidden Pictures, Patterns, Situations,
Analogies and Stories, for a comprehensive valuation of intelligence by the assessment of concrete reasoning ability,
abstract reasoning ability, spatial ability and perceptual ability.
In the 7 subtests, except for Hidden Pictures, all other 6 subtests consist of 2 to 3 paralleled series of items,
ranking in an increasing degree of difficulty. This structure is especially suitable for its adaptive procedure, by
applying which not only the duration of the test can be shortened, but also the subjects are always motivated during
the course of the test for they skip some items that are either too simple or too difficult for them. According to the
instruction of the test, there should be a positive or negative feedback for the subjects while in traditional IQ tests
there is no feedback or merely positive feedbacks by saying “very good”, no matter what answer the subjects provide.
It is common in traditional IQ tests that the subjects misunderstand that the tester pays little attention to their answers
or their answers are correct. Thus, they can not find and correct their mistakes or adjust their strategies for solving
problems when they do not have a clear idea of what to do in the test or apply a wrong strategy. However, all
problems mentioned above can be avoided in SON-R tests by correct feedbacks.
Above all, the test is a non-verbal test with the independence on language ability. Since the test is hardly
affected by language knowledge or school knowledge, SON-R tests are suitable for testing the deaf or hearing
disabled children and others who have difficulty with verbal communication. Moreover, because the testing materials
in the SON-R do not need to be translated, the test can be used not only as an ideal measurement for children’s
intelligence, but for international and intercultural research, i.e. the research among various people and religions.
Given the situation in China, the use of the SON-R tests can be of great value. Firstly, there is great difference
between cities and countryside in the field of economy and education. Children in cities and those in the countryside
do not share the same knowledge background, which makes it hard to keep equity in verbal intelligence tests with a
great number of items related to school knowledge. SON-R tests mainly contain action items, free from school
knowledge or language knowledge, so that equity is guaranteed. Secondly, China is a large country with many
minorities who have their own languages. SON-R tests make it possible to test children from various areas all over
the country. Besides, one of the outstanding characteristics of SON-R tests is that the tests cover deaf and hearing
disabled children as well. As SON-R tests have specific advantages in testing materials, test administration and
analysis and presentation of results, the tests can be used for reference in the study of intelligence tests in China.
2 Method and Outcome
2.1 Selecting subjects
The SON-R is meant for the assessment of persons from 5.5 to 17 years of age. In the adaptation study, three
following groups are selected to be pre-tested: a 6-year-old group (6.0 years of age to 6.12 years of age), a
11-year-old group (11.0 years of age to 11.12 years of age), a 14-year-old group (14.0 years of age to 14.12 years of
age). The subjects are from big, medium and small cities, such as Beijing, Miyun, Tianjing, Guangrao Xian in
Shandong Province and Renqiu in Hebei Province. There are 114 persons in the 6-year-old group, including 52 boys
and 62 girls, while 115 persons in the 11-year-old group, with 64 boys and 51 girls, and 105 persons in the
14-year-old group, including 56 boys and 49 girls. Totally there are 334 persons taking part in the test as subjects,
with 172 boys and 162 girls.
2.2 Test course
The testers have been trained strictly. There are 2 programs of the test:
1. There are 263 subjects being tested according to the adaptive test procedure of the original test.
2. There are 71 subjects completing the items on the analysis of the test.
Table 1 Analysis on Difficulty of Categories
series
6-year-old group
a
b
11-year-old group
c
a
b
14-year-old group
c
a
b
c
1
.85
.74
.69
1.00
.88
.82
1.00
.56
.85
2
.39
.49
.44
.96
.81
.83
.96
.72
.74
3
.29
.34
.34
.69
.64
.49
.82
.70
.42
4
.21
.37
.15
.21
.64
.65
.28
.83
.79
5
.25
.00
.24
.51
.64
.42
.53
.78
.47
6
.16
.15
.44
.64
.53
.80
.77
.65
.81
7
.00
.00
.08
.13
.66
.37
.30
.86
.61
8
.00
.09
.00
.35
.25
.24
.59
.46
.63
9
.00
.18
.09
.08
.23
.37
.27
.50
.59
2.3 Outcome and analysis
2.3.1
Analysis of difficulty
Divided into the 3 age groups, difficulty of the items in the test is analyzed below, with uncompleted items as missing
values in the adaptive test procedure. Table 1 to Table 7 are the difficulty of the subtests.
Except for Hidden Pictures, the other 6 subtests are basically ranked in an increasing level of difficulty and
rationality of the adaptive test procedure is also seen according to the analysis above. Because of the different
cultural backgrounds, the positions of difficulty of a few items do not agree with the outcome of the test. The order of
those items will be rearranged basing on the test. In the subtests of Hidden Pictures, there is not any big difference
shown among the scores of the 4 items by the subjects, which corresponds to the structure of this subtest with
similar levels of difficulty.
Table 2 Analysis on Difficulty of Mosaics
series
6-year-old group
11-year-old group
14-year-old group
a
A
a
b
b
b
1
.97
.96
1.00
1.00
1.00
1.00
2
.93
.92
.99
1.00
.99
.75
3
.52
.71
.96
.98
1.00
.90
4
.23
.51
.92
.87
.96
.86
5
.14
.24
.78
.86
.87
.80
6
.06
.22
.82
.87
.85
.80
7
.07
.00
.60
.84
.85
.91
8
.00
.00
.33
.39
.72
.59
9
.00
.00
.24
.51
.56
.70
10
.00
.00
.53
.40
.84
.73
Table 3 Analysis on Difficulty of Patterns
series
6-year-old group
11-year-old group
14-year-old group
a
A
a
b
b
b
1
.91
.88
1.00
1.00
1.00
1.00
2
.72
.75
1.00
1.00
1.00
1.00
3
.35
.46
.97
.98
.97
1.00
4
.35
.31
.94
.95
.97
1.00
5
.20
.22
.79
.89
.81
.74
6
.03
.28
.68
.85
.83
.87
7
.06
.05
.64
.53
.88
.77
8
.00
.00
.05
.13
.17
.16
9
.00
.00
.06
.15
.09
.21
Table 4 Analysis on Difficulty of Hidden Pictures
6-year-old group
11-year-old group
14-year-old group
HID1
.174
.496
.594
HID2
.229
.561
.69
HID3
.273
.546
.643
HID4
.219
.552
.641
Table 5 Analysis on Difficulty of Situations
series
6-year-old group
a
B
11-year-old group
c
a
b
14-year-old group
c
a
b
c
1
.90
.59
.33
1.00
.95
.92
1.00
.80
1.00
2
.63
.55
.53
.95
.82
.92
.97
.57
1.00
3
.44
.60
.62
.83
.95
.81
.89
.89
.14
4
.37
.27
.45
.85
.92
.91
.89
.92
.67
5
.22
.17
.28
.61
.83
.92
.78
.73
.83
6
.21
.00
.26
.62
.62
.62
.76
.68
.75
7
.28
.08
.06
.57
.71
.43
.62
.78
.69
8
.15
.00
.00
.46
.48
.41
.54
.49
.57
9
.09
.18
.00
.44
.45
.64
.68
.55
.75
10
.00
.00
.09
.23
.11
.27
.48
.19
.46
11
.00
.00
.00
.07
.09
.13
.40
.27
.42
Table 6 Analysis on Difficulty of Analogies
series
6-year-old group
a
11-year-old group
b
c
a
b
14-year-old group
c
a
b
c
1
.84
.77
.61
.99
1.00
.97
1.00
1.00
1.00
2
.59
.61
.66
.96
1.00
1.00
.98
1.00
.50
3
.45
.55
.56
.99
1.00
1.00
.95
.80
1.00
4
.65
.43
.50
.96
.94
.91
.94
.89
.83
5
.32
.31
.32
.95
.86
.91
.91
.75
.82
6
.44
.48
.30
.73
.69
.81
.85
.82
.68
7
.72
.18
.35
.73
.75
.76
.70
.83
.84
8
.10
.35
.42
.56
.67
.66
.71
.66
.75
9
.17
.24
.15
.51
.38
.58
.68
.55
.73
10
.00
.00
.25
.36
.38
.58
.66
.60
.66
11
.00
.18
.08
.26
.51
.58
.54
.62
.58
Table 7 Analysis on Difficulty of Stories
series
6-year-old group
A
2.3.2
b
11-year-old group
14-year-old group
A
a
b
b
1
.74
.67
.97
.96
.96
.82
2
.35
.40
.96
.93
.95
.80
3
.48
.44
.72
.91
.82
.93
4
.26
.24
.71
.77
.84
.78
5
.18
.24
.57
.74
.76
.59
6
.11
.00
.54
.46
.74
.54
7
.08
.00
.57
.70
.60
.83
8
.00
.00
.39
.28
.75
.45
9
.00
.00
.47
.39
.56
.77
10
.00
.00
.00
.05
.16
.13
Analyses on discrimination
Discrimination analyses, i.e. calculation of the correlation between each item and rest-total score, are carried out by
groups. For more convenience of the calculation, all data by the subjects who completed all the items are transformed
into the ones that are consistent with adaptive test procedure. It is proved that the data stay more or less the same
after being transformed.
The discrimination analyses on the 3 groups by age are presented in table 8 to table 10.
Table 8 Distinction analysis of 6-year-old group
distinction>0.30
Distinction = 0
Categories
41% (a1~a6, b1~b4, c1)
29% (a7~a, b5, b7~b9, c8)
Mosaics
25% (a3, a4, b3~b5)
35% (a8~a10, b7~b10)
Patterns
44% (a1~a4, b1~b4)
11% (a8, a9)
Situations
21% (a1, a5, b2~b5, c4)
24% (a8~a10, b7~b10)
Analogies
61% (a3, a4, a6, a9, b1~b8, c1~c8)
15% (a10, a11, b10, b11, c11)
Stories
30% (a1, a2, a4, b1~b3)
45% (a7~a10, b6~b10)
Table 9 Distinction analysis of 11-year-old group
distinction>0.30
Distinction = 0
Categories
41% (a5, b1, b4~b7, c2~c6)
4% (a1)
Mosaics
60% (a5~a10, b5~b10)
15% (a1, b1, b2)
Patterns
33% (a6, a7, b4~b7)
22% (a1, a2, b1, b2)
Situations
42% (a4, a5, a7, b4~b9, c6, c8~c11)
6% (a1, c1)
Analogies
30% (a5, a6, a8, b7~b9, b11, c8~c10)
15% (b1~b3, c2, c3)
Stories
45% (a6, a9, b3~b7)
5% (a10)
Table 10 Distinction analysis of 14-year-old group
distinction>0.30
Distinction = 0
Categories
59% (a2, a5, a6, b3, b5~b9, c3~c9)
4% (a1)
Mosaics
60% (a5~a10, b5~b10)
15% (a1, a3, b1)
Patterns
44% (a5~a7, a9, b5~b7, b9)
33% (a1, a2, b1~ b4)
Situations
58% (a5~a11, b5~b9, c4, c5, c6, c8~c11)
9% (a1, c1, c2)
Analogies
82% (a2~a11, b4~b11, c2, c4~c11)
30% (a1, b1~b3, c1, c3)
Stories
75% (a3, a5~a9, b1~b9)
In general, except for the subtest of Hidden Pictures, the distinction of the other 6 parallel series of subtest items
appears to have the same tendency in all 3 groups. That is to say, to 6-year-old group, the distinction is low at the end
of the series while high at the beginning of the series; on the other hand, to 11-year-old group and 14-year-group the
distinction is low at the beginning of the series while high in the middle and at the end of the series. The item with
highest distinction moves backwards in the series with the age. Besides, in all 3 groups the items at the medium level
of difficulty appear to have high distinction while the ones that are too difficult or too easy appear to have the
contrary. In the subtest of Hidden Picture, there is no evident different distinction found in the 4 items. We found that
there were several items which contain diverse distinction tendency in each subtest. In table 1 to table 7 it is found
that the difficulty of most items does not correspond to the same tendency. Therefore, it is considered the reason why
the distinction of those items does not agree to the tendency is mainly due to the improper order of the items or
different culture background, because of which children can not understand the content of the test correctly. In order
to make the test adapt better to Chinese children, we are going to adapt some pictures and rearrange the items of the
test according to the rate of passing.
2.3.3
Comparison with the mean scores of each subtest in the Netherlands
The mean scores and the standard deviation of the 6-year-old group, the 11-year-old group and the 14-year-old group
in the 7 subtests are compared with those of the standardized tests in the Netherlands at the same ages. The results are
seen in table 11.
Table 11
Comparison of the children in two countries
6-year-old group
11-year-old group
China
mean
Netherl.
mean
Significance
China
mean
Netherl.
Mean
Significance
Categories
4.98
5.7
**
13.21
15.3
**
Mosaics
6.08
7.5
**
14.11
Hidden
Pictures
6.69
6.7
Patterns
4.83
3.7
Situations
7.20
Analogies
Stories
14-year-old group
China
mean
Netherl.
Mean
Significance
16.13
18.0
**
14.3
16.58
16.5
16.12
16.0
19.26
19.6
**
12.56
11.2
**
13.39
13.3
9.2
**
19.43
20.5
**
22.23
23.7
9.01
6.8
**
23.68
20.0
**
26.26
23.7
3.40
4.0
**
10.62
11.2
**
12.87
13.6
**
**
**
(China: 89 subjects in the 6-year-old group, 102 subjects in the 11-year-old group, 104 subjects in the 14-year-old
group; The Netherlands: 150 subjects in each group. * stands for the difference which is significant at the 0.05 level,
* * stands for the difference which is significant at the 0.01 level).
According to the table above, it is shown that in 3 age groups the mean scores for Dutch children are higher than
those for Chinese children in Categories, Situations and Stories, and the difference is significant at the 0.01 level; in
Patterns and Analogies, Chinese children performed better than Dutch children, and the difference is significant at the
0.01 level; in Hidden Pictures, the difference is hardly to be seen in both groups of mean scores. In Mosaics, the
6-year-old group and the 11-year-old group of Dutch children obtained higher mean scores than Chinese at the same
ages. The differences are significant at the 0.01 level and 0.05 level respectively, while the difference disappeared in
14-year-old groups.
Due to the characteristics of the subtests themselves, which was supposed based on the response of the subjects
in the test, Chinese children generally acquired low scores in Categories, Situations and Stories. The pictures in 3
subtests cover life and culture, and Chinese children made a number of mistakes for they were not familiar with the
content of the pictures. As a result, we adjusted the pictures that are unacquainted or easily misunderstood; moreover,
we have had the permission from the author of the test so that we can exchange the pictures in the future tests. It is
apparant that the test is more suitable to Chinese after the order of the items was rearranged and the pictures were
altered.
Chinese children generally obtained high scores in Patterns and Analogies. Since the test materials in both
subtests are geometrical figures, Chinese children have a good understanding on the demand of the test without much
influence of cultural background. Some relative research [3] also presents that Chinese children perform better than
American children at the same age on reasoning ability development.
2.3.4
Analysis on validity
Since Raven’s Standard Progressive Matrices is regarded as a well known test in the world, and it is non-verbal like
SON-R, we analyzed the correlation between these 2 tests in the 11-year-old group and the 14-year-old group. In 2
groups by age, the correlation between the total score in the SON-R and Raven’s Standard Progressive Matrices is
significant at the 0.01 level, the correlation between the SON-R and Raven’s Standard Progressive Matrices are
significant for the abstract reasoning test and the concrete reasoning test(at the 0.01 level or 0.05 level).
Because IQ takes an important place in scores obtained in school, the correlation study between the score of the
SON-R and that of the latest mid-term is carried out in the 11-year-old group and the 14-year-old group. It is found
that in the 11-year-old group, the correlation is significant at the 0.01 level between the total score of the SON-R and
the total score of mathematics and Chinese; in the 14-year-old group, the correlation is significant at the 0.05 level
between the total score of the SON-R and the sum of z scores of Chinese, English, algebra, geometry, and physics,
and the same happened between the total score of the SON-R and the scores of geometry and physics.
Based on the results above, it can be judged that the SON-R possesses good validity.
2.3.5
Comparisons of the mean scores in 3 groups by age
We examined the significance of the differences for the mean scores in each subtest of the SON-R by the 3 groups.
The results are shown in table 12 and table 13.
Table 12 Diversity significance examination for the mean scores (6-year-old group and 11-year-old group)
6-year-old group(N=89)
11-year-old group(N=102)
Mean
Standard
deviation
Mean
Standard
deviation
T test
Categories
4.89
2.94
13.21
5.07
47.15 ***
Mosaics
6.08
2.04
14.11
3.32
103.73 ***
Hidden Pictures
6.69
2.98
16.12
3.98
73.94 ***
Patterns
4.83
2.82
12.56
2.07
117.69 ***
Situations
7.20
3.51
19.43
4.93
64.93 ***
Analogies
9.01
5.39
23.68
4.43
55.25 ***
Stories
3.40
2.07
10.62
3.42
88.69 **
Table 13 Diversity significance examination for the mean scores (11-year-old group and 14-year-old group)
11-year-old group(N=102)
14-year-old group(N=104)
Mean
Standard
deviation
Mean
Standard
deviation
T test
Categories
13.21
5.07
16.13
5.88
3.82 ***
Mosaics
14.11
3.32
16.58
3.65
5.08 ***
Hidden Pictures
16.12
3.98
19.26
4.77
5.13 ***
Patterns
12.56
2.07
13.39
2.13
2.84 ***
Situations
19.43
4.93
22.23
5.48
3.86 ***
Analogies
23.68
4.43
26.26
4.75
4.03 ***
Stories
10.62
3.42
12.87
4.34
4.14 ***
Table 12 and table 13 show that the mean scores in each subtest in the SON-R are significant for diversity in the 3
groups, and the scores increase with the age, which is consistent with the development of children.
3 Conclusion
Compared with the current common intelligence tests, the SON-R broke with the limitation in content, structure, test
procedure and the explanation for the test results, and it can give comprehensive assessment from many dimensions.
The content of the test is novel, rich, and it focuses on stimulating children’s interest during the course of the test.
Children were thus always motivated though it took them long to be tested. Since the test is a non-verbal test without
depending on language knowledge or skills, it can be applied widely not only for normal children, but deaf or hearing
disabled children and others who have difficulty with verbal communication. Besides, the test does not consist of any
item related to subject teaching so that cross-cultural equity could be guaranteed in areas at different levels of
economy and education. In sum, the SON-R is an ideal intelligence measurement tool that is suitable for China.
Based on the data of the SON-R tested in China, it is proved to have good structure and validity. On the other hand,
because of the influence of culture background, a few items are found to be unsuitable for Chinese children or with
improper positions in the difficulty series of the test, which have been altered and adjusted. The revised version of the
test has been examined to adapt better to China.
4 References:
1. Zhang Houcan, Gong Wen, Current Psychometrics Development and Status. Psychology Exploration, 1995;15
(1)
2. Laros J A, Tellegen P J. Construction and Validation of the SON-R 5 1/2-17, the Snijders-Oomen Non-verbal
Intelligence Test. Woltes-Noordhoff bv Groningen, The Netherlands, 1991
3. Wang Xiaoping, Zhang Houcan, Lin Chuanding. Comparison of the Factors between Chinese Version and
American Version in Wechsler Intelligence Scale for Children –Comparison between Sample Groups by Age in
Both Countries. "Psychological Science," 1993;17(5):11- 16
Download