Stanford Binet “Indicators” of Mechanical Ability
- Author:
Riley
- Role:
Research Fellow
- Affiliation:
The Training School, Vineland, N.J.
One of the common practices in clinical psychology is the use of the results from one test to indicate the advisability and the direction of further testing. The development of this method is largely a matter of individual clinical experience and is rarely validated quantitatively. This study is an attempt to make a statistical evaluation of the items in the Stanford Binet and, incidentally, of certain performance tests that might indicate the presence of mechanical ability.
The theory that general intelligence and mechanical ability are distinct traits is consistently confirmed by the low correlations found between tests of intelligence and mechanical ability. Most of these studies are based on the total score or verbal “pencil and paper” type of intelligence test. Because of the variety of tests in the Stanford Binet, which presumably taps many functions, some might be sufficiently prognostic of the presence of mechanical ability to be of practical clinical value. The Stanford Binet has a further advantage in that it is more frequently used as the base of a test battery than any other test.
Performance tests seem to require, in common with tests of mechanical ability, speed of activity, motor coordination, manipulative skill, and a certain ability “to see relationships of a concrete nature.” From this overlapping in the functions necessary for these two types of tests a higher correlation between them might be anticipated than with Binet and mechanical ability, and so be of greater prognostic value.
The Arthur Performance Scale1 was used in this study. It is composed of nine tests (in this experiment the Kohs was omitted to conserve time) : Knox Cube, Seguin, Two Figure, Casuist, Manikin and Feature Profile, Mare and Foal, and Healy I?Picture Completion. The scores were obtained by applying a differentiating formula to the norms of Pintner and Paterson2. This weights the 1 Arthur, Grace: A New Point Performance Scale?Journal of Applied Psychology, 1925, 9: 390. ‘Pintner, B., and Peterson, D. G.: A Scale of Performance Tests. tests according to their ability to separate an age level from the succeeding one. The tests are combined into a scale by taking the Median mental age.
As a test of mechanical ability the Minnesota Assembly3 was Used. This is a revision of the Stenquist Assembly. It is arranged in three boxes containing thirty-three common mechanical objects. There are to be put together correctly, as rapidly as possible. The time is cumulative for each box and the score is from zero to ten for each object.
The reliability and validity of this test are comparable to the Results obtained for intelligence tests. The correlation of odd and even items raised by Brown’s formula gives a reliability coefficient for the total scale of plus 94. The criterion of mechanical ability used was the quality of the boy’s work in shop graded in a semi objective way against certain standard objects. This correlation was plus 55.
The subjects of this study were sixty-five boys from the Hennepin County Probation Office. The usual procedure in giving an individual test battery was employed. Special care was taken to establish rapport because of the type of subjects and to avoid fatigue because of the length of the examination (about three hours). The life ages of these boys extended from 10 to 17 years (median 14.0 years). Their mental ages on the Binet ranged from 8.0 to 18-5 years (median 12.1 years) and on the Arthur Performance Scale from 9.0 to 15 plus years (median 13.0 years). On the Assembly they made scores from 160 to 320 (median 255).
Their scores on all tests were in line with other findings, keeping in mind the selection and small number of cases. Although the distribution is not normal, this would not materially vitiate the results for the purpose of this study. A wide variation might do so, but this as indicated by the standard deviation of the Binet I.Q.s is 13, which is similar to Terman’s result and identical with the pupils ?f one of the city high schools.
A summary of the correlations computed from these data is given in Tables I and II.
As shown by the intercorrelation of the various boxes, the Minnesota Assembly shows a reliability comparable to the coefficients at its formation. In its standardization two boxes of ten items each 3 Anderson, D., Paterson, D. Gr., et al: Tests of Mechanical Ability (manuscript).
gave a correlation of 63. The average in the present study for the three boxes is 66 (Table I).
The low correlation (r 13, life age constant) between the Binet M. A. and the Minnesota Assembly in consistent with other findings and confirms the “unique trait” theory.
As was anticipated, the Arthur Performance Scale gave an appreciably higher correlation than the Binet with the Assembly (r 44, life age constant). This is 11 points higher than the correlation with the Stanford Binet. If the validity and reliability of the Arthur Performance Scales were known, this might have certain theoretical implications regarding the nature of intelligence. The individual tests of the Arthur Performance Scale show some range in their correlation with the Minnesota Assembly (r 42 to 13). The Knox Cube (r 52) and the Casuist (r 40) show the highest coefficients and therefore are the most valid indicators of mechanical ability as measured by the Minnesota Assembly.
Considering the separate items on the Stanford Binet, the scores (passes and failures) form a series of dichotomous measures which are to be compared with a series of graduated measures, and so the biserial r method was used. Correlations were not calculated for tests having less than ten passes or failures. The results (Table II) show a rather wide range (58 to ?14). In general the verbal and memory tests show little relationship with mechanical ability while the problem and perceptual tests are fairly close, although the clock test is an exception to the latter statement.
The ten tests in the Binet showing the highest correlation with the Assembly (starred in Table II) were combined into a scale by adding up the number of mental ages represented in the scattering in each case. This was expressed as a per cent of the possible starred tests in the scattering and the results correlated by the product moment method with the Assembly. As this gave a correlation of only 45, the complexity of its formation would not warrant the slight gain in validity.
The application of these findings to clinical procedure may be illustrated by the situation which originated the study. Part of the subsidiary work for the Minneapolis Child Guidance Clinic consists in a brief study of certain cases from the County Probation Office. The minimum psychometric datum is a Stanford Binet. The schedule allows about an average of forty-five minutes for additional testing. Studying the Binet scores on some two hundred cases shows there is a bimodality in the distributions of I.Q.s. The high points occur at 85 and 105. This, with other data, suggests that one of the chief factors in the delinquency among these boys is that they have reached their academic limit. In such cases the advisability of their transfer to a vocational or trade school is the paramount problem. This in turn raises the question of their mechanical ability. As there is insufficient time to give every case the Assembly Test, which takes an hour, it can only be used in cases where is will be of positive prognostic value. This will be indicated by comparing the “scattering” on the Binet with the correlations in Table II. If the highest tests passed are those showing the highest correlations (the starred tests), or if a high score is made on the Knox Cube and Casuist, the Assembly Test should be given to obtain confirmation of the presence of mechanical ability.
Table I Correlations N-65 Inter-correlations of the Minnesota Assembly Box A?Box B r equals 61? .05 Box A?Box C r equals 63? .05 Box B?Box C r equals 73? .04 Life Ages and Tests Life Age?Stanford Binet r equals 55? .07 Life Ago?Arthur Performance Scale r equals 49? .07 Life Age?Minnesota Assembly r equals 42? .07 Partial Correlations (Life Age Constant) Stanford Binet (MA.)?Arthur Performance Scale (score) r equals 39? .07 Stanford Binet (MA.)?Minnesota Assembly r equals 13? .08 Arthur Performance Scale (score)?Minnesota Assembly r equals 44? .07 Arthur Performance Tests?Minnesota Assembly Knox Cube?Minnesota Assembly r equals 52? .07 Casuist?Minnesota Assembly r equals 40? .07 Two Figure?Minnesota Assembly r equals 34? .07 Feature Profile?Minnesota Assembly r equals 24? .08 Knox Cube?Minnesota Assembly r equals 18? .08 Seguin?Minnesota Assembly r equals 17? .08 Healy Pictorial I?Minnesota Assembly r equals 13? 08
Table II Biserial r’s between Stanford Binet (test items) and Minnesota Assembly Scores Test Year Name r 10 *1 Vocabulary .38 2 Absurdities ? *3 Designs .58 4 Beport ? 5 Comprehension .03 6 Sixty Words .14 12 1 Vocabulary .25 2 Abstract Words .26 *3 Ball and Field .31 4 Dissected Sentences .11 *5 Fables .37 6 (Six) Digits Backwards .16 *7 Pictures .40 8 Similarities .18 14 1 Vocabulary .20 *2 Induction .33 3 Difference of President and King .24 4 Problems .14 *5 Arithmetic .33 6 Clock Test .21 16 1 Vocabulary ? 2 Fables .28 3 Differences of Words .21 *4 Problem Boxes .42 5 (Seven) Digits Backwards .09 *6 Code .30 18 *2 Binet?Paper Cutting .33 3 (Eight) Digits .11 Mean .24 Scale of starred tests?Minnesota Assembly r equals 45? .07 * Tests used in forming scale to indicate mehanical ability.
Disclaimer
The historical material in this project falls into one of three categories for clearances and permissions:
Material currently under copyright, made available with a Creative Commons license chosen by the publisher.
Material that is in the public domain
Material identified by the Welcome Trust as an Orphan Work, made available with a Creative Commons Attribution-NonCommercial 4.0 International License.
While we are in the process of adding metadata to the articles, please check the article at its original source for specific copyrights.