A Clinical Study of Twenty-Six Pairs of Twins
By Kenneth B. White, M.S.
Rutgers University Introduction The first serious study of twin resemblance was made exactly fifty years ago (1883) by Sir Francis Galton.1 On the basis of written and verbal reports of thirty-five pairs of similar twins and twenty pairs of dissimilar twins, Galton concluded that heredity is more important than environment in determining adult characteristics. Thorndike in 1905 2 first attempted to discover the extent of twin resemblance by means of a number of mental and physical measurements. Since that time, a growing realization of the importance of studying twin resemblance to determine the relative influence of heredity and environment, has led a large number of investigators to turn their attention to this problem. Rexroad,3 in a recent article, has assembled the results of fifty-seven studies of twin resemblance, which have been conducted within the past decade.
In measuring twin resemblance in intelligence, none of these recent studies makes use of the Kuhlmann-Anderson Intelligence tests, nor do they give any data on form board performance. It is the object of this study to supply this data in a tentative way, and to augment the evidence on twin resemblance already amassed, by means of other clinical tests, and information gained from replies to a questionnaire and from personal interviews. The subjects for this study were twenty-six unselected pairs of twins living in or near New Brunswick, New Jersey. Their ages ranged from five years to sixteen years and five months, with a median age of nine years and one month. They all attended elementary school with the exception of one pair who were in the 1 Galton, Francis, Inquiries into Human Faculty. New York: Macmillan, 1883.
2 Thorndike, E. L., Measurement of Twins. J. of Phil., Psychol, Scientific Methods, 1905, 2, 547-553. 3 Rexroad, Carl N., Eecent Studies of Twin Resemblance. Psychol. Bull. 1932, 29, 204-217. 243
tenth grade. They represented eight different nationalities and a great disparity of economic and social advantages. Twenty-one pairs were of like sex; in twelve pairs both members of the pair were girls, in nine pairs both were boys. The remaining five pairs were of unlike sex. It is a matter of regret that a larger number of pairs of unlike sex were not available.
Procedure
In every case both members of a pair of twins were examined at the Rutgers Psychological and Mental Hygiene Clinic on the same afternoon. In so far as the age and ability of the twins allowed, the following tests were administered and were used for comparative purposes in this study: The Stanford-Binet, Kuhlmann-Anderson Intelligence tests, Witmer Form Board, Witmer Cylinder, and Healy A. For the most part, the examinations were conducted by two examiners. Dr Anna S. Starr administered the form boards to each twin individually and the Binet to one of them; the writer gave the Kuhlmann-Anderson to both twins at the same time and the Binet to the other twin.
In addition to these clinical tests, specimens of handwriting were secured from fifteen pairs of twins. Further information concerning the twins was secured from the parents, by means of a standard questionnaire, and in many cases, by an interview with them. Twin resemblances will be reported in terms of correlation coefficients. It was ascertained from the parents which of the twins was born first, or was the older. This twin is referred to as “twin A.” The younger or last born in each pair is known as “twin B.” Scores made on a given test by twins A can then be correlated with scores on the same test made by twins B. Since all correlation coefficients must be looked upon as only suggestive due to the small number of cases, all except in the case of memory span and quality of handwriting were determined by the method of rank-differences rather than by the product-moment method. It would be desirable to divide the twins into identical and fraternal and to report the resemblance found in each group. Among our cases there were only ten pairs considered, on the basis of sex and physical appearance, to be identical twins, and seven pairs fraternal. The others we were quite uncertain about. Consequently, we will not consider this interesting question in this paper.
Intelligence Quotients The average intelligence quotients and the variability as determined by the Stanford-Binet and the Kuhlmann-Anderson Intelligence tests, for all of the twins studied, for twins A, and for twins B, are shown in Tables IA and IB.
Table IA Intelligence Quotients of Entire Group of Twins No. Cases Average Range S.D. Binet Kuhlmann-Anderson 52 48 99 ?1.3 103.8?1.5 62-141 70-135 14.4 15, Table IB Intelligence Quotients of Twins A and B Twins A No. Cases Average Range S.D. Twins B No. Cases Average Range S.D. Binet KuhlmannAnderson 26 24 98.2?2. 103.6?2. 62-141 70-129 15.4 14.3 26 24 99.7?1.7 104 ?2.2 67-122 70-135 13.2 15.7
An analysis of Tables IA and IB suggests some conclusions regarding the intelligence quotients of the sample of twins included in this study.
1. The twenty-six pairs of twins, considered as a group, were of average intelligence when compared with Terman’s study of the intelligence of a large number of school children. The average intelligence quotients, of the twins here reported, was in close accord with the findings of Merriman, Lauterbach, Wingfield and Holzinger, all of whom used a larger number of subjects.4 2. The average I.Q. and the variability of the twins who were born first (Twins A) were approximately equal to that of the twins who were born last (Twins B). The Binet and the KuhlmannAnderson both indicate this equality. In order to take account of the slight difference in the average between the two groups, Pearson’s Coefficient of Variation was used to get the relative variability of each group. (“F = 100 S.D./Average.) It turned out 4 Bexroad, Carl N., op. cit.
that this measure was almost equal for twins A and twins B on both intelligence tests. For this group of twins, then, order of birth seemed to have no consistent effect on the average intelligence quotient or the variability of the group. 3. The Kuhlmann-Anderson gave a slightly higher average I.Q. for the twins as a whole, and for each of the smaller groups, than did the Binet.
The degree of resemblance in intelligence quotients between the twins as a whole, between members of the younger thirteen pairs, and between members of the thirteen older pairs, may be seen in Table II. In both the younger and older groups there were five pairs who were judged to be identical twins. The results of both intelligence tests are shown in Table II, resemblance being indicated by coefficients of correlation.5
Table II Resemblance of the Twins in I.Q. All Twins No. Pairs Younger Twins Age 5 to 9-1 No. Pairs Older Twins Age 9-1 to 16-5 No. Pairs Binet Kuhlmann-Anderson. 26 24 .82?.05 .93?.02 13 12 .52?.14 .94?.02 13 12 .97?.01 .96 ?.02
The correlation coefficients presented in Table II are based on only a few cases, and must therefore be considered only suggestive of the following tendencies.
1. There was a marked correlation between the twins with respect to intelligence quotients, when determined by either the Binet or Kuhlmann-Anderson tests. These coefficients were slightly higher than most investigators have reported for all twins taken together. They indicated a higher degree of resemblance between twins than is usually reported for ordinary siblings.6 s Rank order coefficients (p’s) were changed to corresponding values of r by reference to Table XX in H. E. Garrett’s Statistics in Psychology and Education. The probable error of the r’s found from the p’s was determined by the formula PZ?r = .7063(1 ? r*)//N.
e Hildreth, in summarizing the results of several studies of sibling resemblance, states that “siblings show on the average less resemblance than twins. The amount appears to be on the average about .50, though it may vary greatly from this figure.” Hildreth, Gertrude H., The Resemblance of Siblings in Intelligence and Achievement. Teachers College Contributions to Education, 1925, No. 186, p. 20.
2. The correlation between the twins was higher on the Kuhlmann-Anderson than on the Binet. This may be due to the fact that the Binet was given and scored in most cases by two different examiners, for each member of a pair separately. The KuhlmannAnderson, on the other hand, was given to both twins at the same time, thus assuring standard conditions, and is not subject to individual interpretations in scoring. 3. The comparison of the correlations found in the younger and older twin pairs, each group containing five pairs judged to be identical twins, gave conflicting results. The Kuhlmann-Anderson correlations suggest that there was no significant difference between the resemblance of the younger and older twin pairs. The results of the Stanford-Binet, on the other hand, would indicate that the members of the older twin pairs were much more alike than the members of the younger twin pairs. In view of the small number of cases in each group, and the high probable error of the Binet correlation, these results are not very significant.
Form Boards
The correlation between the twins as shown by their performance on three form boards is shown in Table III. Correlations are based on time in seconds for each trial. Only those individuals who finished the trial without help are included. For this reason there are more cases considered in the second trial of the Witmer Cylinder and the Healy A, than in the first.
Table III Twin Resemblance on the Form Boards Name of Test First Trial No. Pairs Second Trial No. Pairs Witmer F.B. Witmer Cyl. Healy A…. 25 19 17 .59?.09 .31?.15 .33?.15 25 22 20 .82 ?.05 .75 ?.07 .06?.15
In considering the first trial on these form boards, it is to be pointed out that the subjects were here confronted with a new problem. The common factor of school knowledge, which is involved in the above mentioned intelligence tests, is of little consequence in form board success. The results show a much lower correlation between the twins on the form boards than on the Binet or the Kuhlmann-Anderson, a condition which might be expected from the nature of the problem. In fact, the correlation on the Witmer Cylinder and the Healy A might even be negative, in view of the high probable error. The Witmer Form Board, which is standardized at the four year level, was quite easy for most of the twins. On this test we find a fairly high degree of resemblance between the twins, when the time taken to solve the problem is taken as the criterion of success.
One would expect to find a greater degree of resemblance between the twins on the second trial of the form boards. The element of novelty and strangeness is then absent, and the time taken by a pair of twins might feasibly be more nearly alike. On the second trial of the “Witmer Form Board and Witmer Cylinder, we find this to be the case. The correlations are much higher than on the first trial and are quite significant. In the case of the Healy A, on the other hand, the correlation on the second trial is not only insignificant, but even considerably less than on the first trial. This contrasting result may be due to chance factors in the solution of this problem, which are not taken account of when the time for a single trial is used as measure of the subject’s ability. Qualitatively, the performance of some of the twin pairs on the form boards was very much alike. Their method of attacking the problem, their attitude toward it, and even the errors they made were often strikingly similar. There were other pairs whose performance was qualitatively very dissimilar. The individuals of the same pair showed a wide difference in the rate of discharge of energy, in foresight and analytical ability, and the hand they preferred to use. We were able to observe no general resemblance characteristic of all the twin pairs, with respect to qualitative performance on the form boards.
Memory Span for Digits
The forward memory span for digits of the twenty-six pairs of twins considered, ranged from 4 to 8. In twelve pairs the forward memory span of each member was the same. In the remaining fourteen pairs it differed by one or two digits. The correlation between the memory span for digits of the twenty-six pairs of twins (Pearson formula) was .68 ? .07. Members of the younger twin pairs were found to have an identical forward memory span for digits twice as often as members of the older twin pairs. Eight of the thirteen younger pairs had the same memory span; in only four of the thirteen older pairs was this true.
The Reverse Memory Span for digits shows even a closer resemblance between the twins than the forward span. Both members of three pairs of twins and one member of a fourth pair failed to grasp the idea of reversing the digits, so were unable to repeat even two correctly in the reverse order. The Reverse Memory Span of the remaining twenty-two pairs ranged from 2 to 7 digits. The members of sixteen pairs of twins had the same reverse memory span for digits, including the three pairs who were not able to reproduce any. In only ten pairs, did one member have a different reverse memory span than the other. The correlation between the reverse memory span of the twenty-two pairs of twins who were able to do the problem (Pearson Formula) was .87 ? .04.
The reverse memory span was more often alike for the younger twin pairs than for the older. In only three of the thirteen younger pairs was the reverse memory span of the two individuals unequal. The remaining seven cases of unequal reverse memory spans were found among the thirteen older twin pairs, ranging in age from 9 years, 1 month to 16 years, 5 months. This observation is in close agreement with the result already noticed in connection with the forward memory span.
Handwriting
A short sample of handwriting was obtained from fifteen pairs of twins. After the specimens were well mixed, each was given a quality rating by comparing it with the samples given in Thorndike’s handwriting scale.7 In four pairs, the handwriting of both members received the same quality rating. Individuals in most of the pairs differed by only one point and in two pairs by two points. The correlation between thes<? quality ratings of the fifteen pairs of twins, although only suggestive, was .68 ? .09. Another method was used to get a rough measure of the similarity of the handwriting of the twins. Six judges were asked to match up the handwriting of the twins on the basis of similarity of general appearance and size and shape of the letters. By means of a number system, unknown to the judges, the writer was able to check their accuracy. The number of specimens successfully matched ranged from five to eight, the average being six. This result indicated that in slightly less than half the cases, there was enough similarity between the handwritings of the twins to enable inex7 Tliorndike, E. L., Handwriting. Teachers College Eecord, 1910 11 83-175.
perienced judges to pick them out. A certain similarity was further indicated by the fact that after the correct selections had been removed, all of the judges were able to select from two to four more pairs which were similar. In all, there were eleven different pairs which were correctly matched by the judges as a whole.
In six of the pairs consistently matched, the age of the twins was above nine and a half years. This might suggest that the handwriting of older twins is more similar than that of younger twins, possibly due to the longer period of similar training and practice in this trait, enjoyed by the former group.
General Observations Concerning the Twins
In fourteen of the twenty-six pairs of twins, there is one twin who seems to be more aggressive than the other. In five pairs they take turns being leader, and in one of the pairs it is reported that they exchange this trait about every six months. In six pairs, neither twin is reported as being the more aggressive. “Where one twin is more aggressive than the other, there seems to be no causal connection with priority of birth, intelligence or health. Of the twenty-six pairs of twins, members of seventeen pairs first walked at the same age and of four pairs, within a month of each other. Data on this point are not available for the other five pairs. Individuals of seventeen pairs first talked at the same age, and of three pairs, within three months of each other. One pair is reported as having talked a language of their own, which no one could understand but themselves, before they began to talk in the usual way.
In none of our cases was there any report of similar diseases or illnesses, other than contagious diseases such as measles, afflicting both twins at the same time?a report which is quite frequent in the literature of twins.
In three pairs of twins, or 11.5 per cent, one is righthanded, and the other lefthanded. This percentage showing symmetrical reversal of handedness is considerably lower than that reported by Danforth, 19 per cent8 and Lauterbach, 35 per cent9 and may be due to the small number of cases investigated or inadequate determination of handedness. Of the fifty-two individual twins, five 8 Danfortli, C. H., Resemblance and Difference in Twins. J. of Hered., 1919, 10, 399-409.
9 Lauterbach, C. E., Studies in Twin Eesemblance. Genetics, 1925, 10, 525-568. or 19+ per cent were lefthanded. Hirsch found that 25 per cent of 202 individual twins were lefthanded.10 AVhen it is realized that only about four per cent of the population at large is lefthanded and that twins comprise less than two per cent of the general population, it is suggested that lefthandedness is in some way connected with twinning.
The median age of the mothers at the time of the birth of the twins, calculated with respect to twenty-one pairs, was 30 years. The ages range from 19 to 38 years.
The average number of children in the twenty-five families represented in this study, including the twins, is four. The birth of twins seems to be in no way connected with the number of previous births in the family or to have any influence on the number of subsequent births. With seventeen pairs there were from one to seven previous births and 11 pairs have from one to seven younger brothers and sisters.
In all but seven cases there are twins reported in either the maternal or paternal side of the family, or both. This observation would suggest a definite hereditary tendency toward twinning in certain families.
Conclusions
Since the statistical treatment of data obtained from only twenty-six or fewer cases is apt to lead to unwarranted conclusions, due caution should be employed in referring the conclusions of this study to the twin population in general. With the results of tests, used to measure twin resemblance by other investigators, our results are in fairly close accord. With respect to the other tests used in this study, concerning the results of which there are few if any other reports available, such as the Kuhlmann-Anderson and the form boards, the conclusions presented here must be considered as only suggestive until these tests are applied to many more cases. 1. The group of twenty-six pairs of twins studied, were as a whole, of average intelligence, when measured by the StanfordBinet or Kuhlmann-Anderson Intelligence tests. The order of birth of the twins seemed to have no effect on the average intelligence quotient or on the variability of the group.
2. There was a marked resemblance in the I.Q. between the twins when measured by the Binet or Kuhlmann-Anderson Intelligence tests. There was also a high resemblance between them in io Hirseh, Nathaniel, Twins, Heredity and Environment. Harvard University Press. 1930. the case of the forward and reverse memory span for digits. There was much less similarity between the twins on the first trial of three form boards, the Witmer Form Board, Witmer Cylinder and Healy A. On the second trial of the first two form boards, the resemblance was marked, while on the latter test there was no significant resemblance whatever.
3. The Kuhlmann-Anderson showed a higher average I.Q. and a closer resemblance between the twins than did the Binet. 4. The handwriting of some of the twins was very much alike.
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.