A Report on the Standardization of the Witmer Cylinder Test
- Author:
Franklin Cressey Paschal, Ph.D.,
University of Pennsylvania.
It has been three years since the Witmer Cylinder Test took its present form and was adopted for use in the Psychological Clinic of the University of Pennsylvania. The important part it has played among the mental tests there, has been reflected in the frequent references that have been made to it in The Psychological Clinic, particularly in the Clinic Reports. But heretofore there have been no data available to those who have not worked with the test. It is now possible to remedy this situation and it is the intention of this paper to present a report of the results of the standardization which has recently been completed.1 When the revision of the formboard had been completed, resulting in the Witmer Formboard, attention was turned to the need of a test of a similar nature but more difficult, which would present a problem to children older than those with whom that test was of the most value. Among other things, the Montessori cylinders were brought into use and out of these grew the present Cylinder lThe complete report has been published in monograph form by the author, and may be obtained from Thb Psychological Clinic. Price 75 cents.?Editor.
Test. In the early form, the presentation of the three separate sets of cylinders was followed by that of the three placed together in a triangular position, the thirty cylinders being removed and mixed within the enclosure. While this was not entirely satisfactory, the method gave promise of being of value and the attempt was made to construct a piece of apparatus which would possess the advantages without the disadvantages of the three individual sets of cylinders. The piece of apparatus which has resulted, is a circular board having a series of recesses about its outer edge into which are fitted eighteen cylinders corresponding in size and order to those of the Montessori apparatus. There are no duplications of sizes, however, and the largest and smallest cylinders have been omitted. The knobs on the cylinders have also been eliminated. There is a central compartment into which the blocks may be thrown and mixed. The largest cylinder is two and an eighth inches in each dimension. Starting from this point in a counter-clockwise direction, there are seven blocks of constant height but decreasing steadily in diameter to one inch. Then the next six retain this same diameter but decrease in height to one inch. Beyond this point there is an increase in both height and diameter up to the original block.
The method of procedure could not be made as simple as in most performance tests, as this apparatus is not self-correcting. While we may expect all but the youngest children to return every block to some recess, even a considerable proportion of college students do not return them to their proper positions without some positive suggestion. Coupled with this is the necessity of having a procedure which will give the least advantage to language ability, since performance tests must be our principal source of data in examining deaf children or those who lack the ready use of our own tongue. If we give the most simple instructions possible in beginning, followed where necessary by a series of steps of instruction until the subject has correctly replaced all the cylinders, we may bring out in the first trial the ability to comprehend instructions and to solve a new problem. In the remaining trials, since the end to be attained is now known, we have a purely performance test in which the effects of language deficiency have been eliminated. We have, then, two tests within one. The first trial is a qualitative one in which the time of performance is of value only as a qualitative indication. The remaining trials (and the first trial as well in the better cases in which the nature of the whole problem is grasped before starting) are quantitative, and the time is the only valid measure of the value of the type of performance. The children tested for this standardization were pupils in five of the public schools in the city of Philadelphia, these schools being chosen in sections of the city which would give a fair sampling of the different social classes. There were 867 boys and 855 girls tested in these schools. The records of adults were obtained from several sources. The 154 women and 123 of the men were students in the course in General Psychology at the University of Pennsylvania during the past two years. In addition to this, the records of 54 aviators from one of the army training fields were available. Following the method of Simpson1, who holds that the median performance of two widely divergent groups should give an indication of what would be the average of the entire population, 177 records of prisoners of the Indiana Reformatory were added and the distribution for adult men figured from the resultant 354 cases. Owing to the fact that our results for women are taken from but one group, there being no records available to combine with those of the university women, these figures cannot be considered indicative for adult women in general.
In undertaking this investigation, the intention was to examine a hundred subjects of each age group from six to fifteen years inclusive, fifty to be taken from each half year. In its final form, this plan was adhered to for the ages eight to thirteen inclusive, and was closely approached for ages seven and fourteen, but the sixth and fifteenth years were given little attention after the work was well under way. It was necessary in an investigation of this kind to limit the amount of time to be devoted to any individual. In Dr H. H. Young’s work on the formboard, three minutes was the limit set. For this test, it was decided that if the first trial had not been completed at the end of five minutes, regardless of the point to which it had progressed, the subject should be dismissed. This proved to be sufficient time for practically all children above the age of eight, only twenty-five boys and the same number of girls failing who were over that age, and half of these were within the ninth year. Only eight boys and eighteen girls of the eight year group failed to complete the trial within this time. The percentage of failures for the two lower ages, however, was so high as to make it evident that a special investigation would be needed for these years in which any amount of time necessary might be allowed for the completion of the first trial.
Preliminary investigations, one of them a correlation with mechanical ability or manual ability as shown in shop rating, has indicated that the shortest time trial is the best quantitative measure 1 Simpson, B. R.: Correlations of Mental Abilities. Teachera College, Columbia University, Contributions to Education, No. 53 (1912).
of performance ability for this test. Therefore, the figures used here are those of the shortest trials. While the shortest of three trials has been used, in no case is a trial accepted where the task was not satisfactorily completed or where additional instructions were necessary. Three trials were found advisable; less than this gives an inadequate result, while four trials tend to fatigue the subject as well as to require too much time in the case of the younger children.
The accompanying tables present the shortest time trials of all the subjects examined, distributed by quintiles and with the inclusion of the means and standard deviations. The number of failures (F.) indicates the number who failed to complete the first trial within the five minutes alloted. There are also three cases included under this head of failures to give a satisfactory shortest trial after the first trial had been completed.
The failures are included within the distribution under the heading DNC (did not complete). It may seem that the grouping of all the failures within the slower end of the distribution might be a source of error, as these individuals have had no opportunity to obtain a shortest trial. In other words, it might be said that they are being classified on a basis of learning ability rather than performance ability. A number of control cases and the distribution of those barely successful within the time limit indicates, however, that practically all these cases would fall within the fourth or fifth quintiles had they been permitted to complete three trials, this meaning that our 80 per cent point alone would be displaced to any extent. This, then, is a much better means of distribution than our only alternative, forming a distribution of the successful cases, omitting failures. In the ages six and seven, this distribution of the checking cases did not hold true, which supports our conclusion that the standardization for these ages must allow sufficient time for the first trial to be satisfactorily completed.
The tables show that there is, without exception, a continuous decrease in the time of the means, quintiles and medians with increasing age for each sex. The standard deviation shows the same general tendency as does the minimum, though the latter is particularly affected by the single exceptional cases. This statement does not take into consideration the youngest ages, nor the fourteenth and fifteenth years. The figures for these two years are totally inadequate since there were only available the pedagogically “at age” or “over age”, the better half of the boys and girls of these ages having completed the grade schools. Our experience with the test does not lead us to believe that an adequate sampling of these ages would show them to be slower than the thirteen year group. The figures for the ages six, seven, fourteen and fifteen, have been included for whatever they may be worth, but we do not ascribe to them the same validity as to the ages eight to thirteen inclusive. Particularly indicative is the decreasing range of distribution with increasing age. This is at a minimum in the groups of college men and aviators, where the upper and lower quartiles are but seven seconds apart. In only one instance in both tables is there a variation from the steady decrease in the standard deviation. This investigation has shown the Witmer Cylinder Test to possess the requisite qualities of a performance test. It is, first of all, applicable to a wide age range, since it has been shown that with a uniform method of procedure it is at once a satisfactory test for adults and for children as young as six years. In clinical work it has been found to be useful even with children of five years, and it is not improbable that the same method may be extended to the year below that. The test has, in the second place, been shown to give an increased performance with increasing age. The figures quoted have shown a steady increase in the rate of performance from the sixth year to adult, with the exception of the two years in which the data was insufficient. Then again, a graded series of stimuli is offered, in that the three sets of Montessori cylinders, themselves graded in difficulty, may be used in those cases in which our own test is too difficult. Under the method we have adopted, there is also an absence of the factors which favor language ability, thus making this test applicable in those cases in which we must depend largely upon performance tests. Another requirement satisfied by the Cylinder Test is that it can be used at subsequent examinations of the same individual if an interval of several months has elapsed. While the qualitative aspects of the test will be changed, so far as the performance is concerned, the first trial at the second examination will be a relearning trial and it alone will be affected to any considerable extent by having been previously performed. Finally, so important from a clinical standpoint, there are to be found qualitative differences of performances between the various mental types. The especial value of the Witmer Cylinder Test lies in the degree to which the mental steps are represented by the physical steps. There are eighteen moves to be made, at the least, and the cylinders are so similar in appearance that in all but the most careful and slow performances the number of moves is much in excess of this. The plan, as shown by the blocks picked up; the estimation, as shown by the positions in which they are placed; the moves in making corrections,?all these furnish considerable material for the observation of behavior. The subject is moving almost continually, and any idea which arises is immediately followed by an attempted placement. In many performance tests, the better subjects do a great deal of their planning and checking without making a move. One placement may show them the failure of their whole plan without any of it becoming evident to the examiner. But in this test it is possible to follow very closely the mental activity of the subject. For diagnostic purposes, we must require that in a performance test, the physical processes shall adequately portray the mental processes involved.
Quintile Distribution. boys. 10 11 12 13 14 15 Ad. No. 58 75 100 100 100 100 100 100 94 40 354 Mean S. D. 82.43 73.98 66.54 52.48 49.49 45.26 43.16 38.66 39.60 38.87 34.20 26.57 29.69 23.36 11.98 10.00 11.73 11.48 8.40 7.43 6.35 7.34 Min. 46 41 33 26 32 25 26 23 23 29 18 20% 40% 103 69 59 49 48 42 40 35 39 36 32 Med. DNC 76.0 63.5 52.9 49.8 43.5 41.6 37.7 39.7 37.3 33.5 60% DNC 87 68 57 52 46 43 40 43 39 35 80% DNC DNC 88 72 58 55 51 46 46 44 40 Max. DNC DNC DNC DNC DNC DNC DNC DNC 61 53 GIRLS. Age 10 11 12 13 14 15 Ad. No. 43 83 100 100 100 100 100 100 82 47 154 27 31 18 11 8 3 1 0 1 2 0 Mean 75.75 80.79 65.77 58.51 52.07 49.50 47.06 45.88 44.10 45.71 35.41 S. D. 14.23 31.58 14.79 13.56 12.53 10.97 13.50 12.29 10.43 14.81 9.29 Min. 55 44 41 33 32 30 29 25 27 26 18 20% 77 65 55 47 42 40 37 36 36 35 28 40% DNC 80 62 54 47 47 42 40 41 40 32 Med. DNC 89.0 68.0 58.9 50.5 49.7 44.5 42.3 42.2 42.4 33.2 DNC 177 75 65 56 51 47 46 46 46 35 80% DNC DNC 101 74 67 58 54 53 51 56 42 Max. DNC DNC DNC DNC DNC DNC DNC 94 DNC DNC
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.