The Relation of Fenuresis to Intelligence, to Conduct and Personality Problems, and to other Factors.1
Luton Ackerson and Muriel Highlander, Institute for Juvenile Research, Chicago.
This article gives the results of a preliminary statistical study of some aspects of enuresis or bed-wetting in 3,000 consecutive clinic cases. These cases were children of both sexes, over three years old, on whom adequate psychometric ratings were obtained. Enuresis was counted as a problem if it persisted beyond the beginning of the third year of age, except for occasional’’ accidents.’’ The age at which dry habits should become established is, of course, largely a matter of opinion and in fixing upon the age of three as the minimum age at which it would be viewed as a clinical problem we are following the standard of Woolley and of Thom. Of the total 3,000 cases, about 700, or 23 per cent, have a history of enuresis. In 206 cases the histories of this enuretic group stated specifically the age at which bed-wetting ceased. The remaining 500 odd cases were composed, for the most part, of children who were still enuretic at the time of examination, together with a small number of cases in which the history reports a former enuresis but is indefinite as to its time of ceasing.
Thus, among our 3,000 cases, which are fairly representative of the intake of a children’s behavior clinic, over one-fifth of the cases have a history of enuresis persisting into and beyond the third year of life, and more than one in every seven children is enuretic at the time of examination. This high incidence, which is unquestionably higher than the frequency of bed-wetting among children over three years of age in the general population, establishes the importance of enuresis as a clinical item. In Figure 1 is shown the comparison of the intelligence quotient distribution (usually Stanford-Binet) for the enuretic and non-enuretic groups. The fact is conspicuous that enuresis is present in all ranges of intelligence in our clinic population. Such 1 Studies from the Institute for Juvenile Research, Chicago. Herman M. Adler, M.D., Director. Series C. No. 140. Portions of this paper were read at the Fifth Annual Meeting of the American Orthopsychiatry Association at New York, February 24-25, 1928. differences in I.Q. which do exist, however, are in favor of the nonenuretics. The average I.Q. of the enuretic group is 74.6 and of the non-entireties, 77.5. This difference is small but “statistically significant.” This relation may also be shown by the familiar method of Pearsonian correlation. The value of a Pearson biserial coefficient of correlation2 on these data is - 078?.017. This value is small, but significantly negative in that the coefficient is more than four and one-half times its probable error. An examination of this chart shows other characteristics of the enuresis-I.Q. correlation which are not apparent in the mere statement of group averages and correlation coefficients. At the lower end of the I.Q. scale, we find the incidence of bed-wetters very high. Up to about 15 I.Q. a half of the cases have not yet established “dry habits;” from about 15 I.Q. to about 45 I.Q. a third are still enuretic, while for the great mass of our cases lying between about 65 I.Q. to about 95 I.Q. the proportion of enuretics is less than one-fifth. At the upper end of our I.Q. ranges we find again that the incidence of enuresis rises. Above 95 I.Q., the ‘ Biometrika, 1909, VII, pp. 96-105.
COMPARISON OF THE IQ.5 OF THE ENURETIC AND KON-ENUKETIC GROUPS, N= 3000 500 400 300 C <u d Cr J> 200 100 Non- enuretic ? Enuretic Biserial-r =-.078 ?.017 C = .129 ?.012 . P= .000012, or one chance in over 83000 T” 1 0 a OOOOQOQOOOO ?<j-iri>or~0DoO ? Intelligence Quotient FIGURE 1 FIGURE 1
proportion is over one-fourth. Therefore, if we consider only the lower half of the distribution we find a correlation of negative sign, i.e. the lower the intelligence quotient, the greater the tendency toward enuresis; but within the upper half, the correlation shows an interesting reversal of sign, i.e. at these levels, the higher the I.Q. the greater the tendency toward enuresis.
Now a biserial-r coefficient states the general tendency over the distribution as a whole, with the result that the force of a negative correlation in one portion of the distribution may become neutralized by a positive correlation in another portion. In cases of distribution of this sort in which regression is markedly nonrectilinear, to resort to its technical description, coefficients of the product moment type are not adequate to express the underlying relationship. Upon our data an instructive calculation is a tworow mean square contingency coefficient for which Pearson suggests the name of “coefficient of class divergence or heterogeneity.”3 By the way of explanation, if the two curves in this chart represent strictly homogeneous groups with respect to I.Q., that is, if these distributions could have arisen out of each other purely through a random sampling such as would be obtained by dice throws or card draws, the value of this coefficient would be zero. If, on the other hand, these two groups, the enuretic and nonenuretic, should have shown extremely divergent distributions with respect to intelligence quotient with no overlapping whatsoever (which is obviously not true of our data) the value of such a coefficient would be unity, denoting a complete class divergence or heterogeneity.
Upon our data the value of such a two-row mean square contingency is .129 ? -012, a value which is low, but nevertheless large enough in relation to its probable error to assure us that these groups are not homogeneous with respect to I.Q. This coefficient does not have a sign since it does not discriminate between positive and negative correlation, but has the effect of cumulating correlations regardless of sign throughout the distribution. Thus it affords a sort of maximum measure of the relationship between characteristics within a given series of data. A supplementary contingency calculation, Pearson’s “chi-square test of goodness of fit,”4 demonstrates further the improbability that our enuretic 3 Biometrika, 1906, V, pp. 198-203. 4 Philosophical Magazine, London, Series V, 1900, L, pp. 157-175; Biometrika, 1911, VIII, pp. 250-254. 122 THE PSYCHOLOGICAL CLINIC group is undifferentiated from the non-enuretic group in respect to I.Q. The value of P was found to be .000012, which means that the odds are only twelve in a million, or one in over 83,000, that such a divergent pair of distributions could have occurred by chance alone. So much for the mathematical discussion. With these findings should be compared the results obtained by Rosenow6 upon cases from the same clinic studied several years ago. Eosenow’s data were selected on a somewhat different basis than ours. He fixed upon the age of six instead of three years as the age at which urinary incontinence was counted as a clinical problem. Furthermore, his group showed a larger selection from the lower I.Q. levels, and also he was more strict in excluding cases in which the history of enuresis seemed inadequate. He noted, however, a similar tendency for bed-wetting to be higher among the low I.Q.’s and again among the higher I.Q.’s, or in other words, a negative correlation among the lower I.Q. ranges and a positive correlation among the higher I.Q. ranges within a clinic population. He reported the high biserial correlation of -.33 with a probable error of less than .06. This value is definitely higher than ours, even in relation to its probable error. This difference may well arise because of the different set of conditions of the investigation. In the main these studies corroborate each other. Let us consider now the relation of enuresis and the severity of behavior difficulties. We have attempted in our indexing of case record items to distinguish between conduct problems and personality problems. Our classification is as yet tentative and arbitrary and is based only on the current usage of these terms. In so far as a criterion is possible, a behavior item is designated as a “conduct problem” if in an extreme form the typical reaction of society toward it is commitment in a correctional or penal institution among older persons, and in the case of younger children, a spanking or similar corrective response. Thus, among conduct problems are listed stealing, truancy from home or from school, lying, fighting, incorrigibility, destruetiveness, and the like; sex delinquency, masturbation and other sex offenses; and such undesirable habit problems as bed-wetting, thumb-sucking, finicky food habits, laziness, or talking out loud in school.
A behavior item is considered to be a “personality problem” if in an exaggerated form it usually indicates a psychiatric diag? Journal of Delinquency, 1920, V, pp. 41-55. nosis of psychosis or psychopathy or similar mental disorder. Thus, among the personality problems are listed apprehensiveness, abused or injured feeling or attitude, day-dreaming, sleeplessness, irritability, seclusiveness, over-suggestibility, unpopularity, and the like. The purpose of making this differentiation was to permit a separate totalling of the conduct and personality items for each case to serve as measures of the severity of the patient’s delinquency and mental abnormality. For the validity of the “conduct-total” we already have some evidence. Some preliminary correlations were computed on about a thousand eases, of which 629 were boys and 382 were girls, between the conduct-total and the notation in the case history of a police arrest or court commitment to a correctional institution. For the boys the biserial-r was .57 ? .04, and for the girls, .45 ? .05. The corresponding correlations for the “personality total” with the fact of a police arrest were only .16 ? .4 for the boys and .11 ? -06 for the girls. The fact that the conduct-total yields the substantial correlations of .57 and .45 with the “outside criterion” of a police arrest or court commitment to a correctional institution indicates that even in its present tentative, unrefined form it amounts to a plausible measure of the patient’s delinquent tendencies. Furthermore, the fact that the personality-total yields only the low correlations of .16 and .11 with the criterion of arrest or commitment indicates that these behavior items show a meaningful differentiation into conduct problems and personality problems on the criterion of social attitude toward them.
In Figures 2 and 3 are shown the frequency distributions of the conduct-totals and personality-totals among our 3,000 cases who have passed their third birthday. It will be noted that the curves are markedly skewed. The average numbers of conduct and personality problems are 6.6 and 4.5, respectively, while the extreme cases show a total of fifty-four conduct items and forty-four personality items. The biserial-r calculations of enuresis with the numbers of conduct and personality problems yielded values of .109+ -017 and .131 ? .017, respectively. (For the sake of those interested in the technical aspect, I wish to say parenthetically that enuresis itself was excluded from the conduct-total, so that the coefficient of .109 ? .017 represents the correlation of enuresis with the total number of conduct problems other than enuresis.) These coeffi124 THE PSYCHOLOGICAL CLINIC cients are both positive in sign and of sufficient magnitude in comparison with their probable errors to justify the conclusion that within our clinic population the enuretic patients as a group show a slightly larger number of conduct and personality problems than the non-enuretic patients. The correlation with the personalitytotal is somewhat larger than with the conduct-total. Whether this indicates that bed-wetting is associated with personality aber
FREQUENCY”DISTRIBUTION OF CONDUCT PROBLEMS N = 3000 15 20 25 30 35 40 45 50 Number of Conduct Problems FIGURE 2 FREQUENCY DISTRIBUTION OF PERSONALITY PROBLEMS N= 3000 15 20 25 *30 35 40 45 .50 Number of Personality Problems FIGURE 3
rations to a greater degree than with conduct disorders cannot be answered definitely at this time since the difference between the coefficients of .131 and .109 is probably not “statistically significant” upon our preliminary group of 3,000 cases. Table 1. Tetrachoric correlation op Enuresis (Presence or Absence, 3000 Cases) With Underweight condition Discord between parents Underweight Not underweight Totals Discord between parents No discord between parents Totals Enuretic 105 3% 599 20% 704 23% 160 5% 543 18% 703 23% Nonenuretic 296 10% 2000 67% 2296 77% 446 15% 1851 62% 2297 77% Totals 401 13% 2599 87% 3000 100% 606 20% 2394 80% 3000 100% rt = .053? .027 rt = .068?.024
In Table 1 are shown the tetrachoric or four-fold correlations with two other factors, underweight condition and discord between parents. The tables are read in this manner: of the total 3,000 cases, 704 cases or 23% were enuretic, and 2,296 cases or 77% had no history of enuresis. Of the enuretic group, 105 cases, or 3% of the total group of 3,000 cases showed an underweight of about 10% or more according to the Baldwin age-height-weight norms, and 599 cases or 20% of the total showed no marked underweight. Of the non-enuretic group, 296 cases or 10% of the total were underweight, and 2,000 cases or 67% of the total showed no marked underweight. The magnitude of this relationship is expressed in the Pearson tetrachoric correlation coefficient6 of .053 rt .027.
In the next table, which shows the correlation of enuresis with a case record notation of discord between parents, we find that in 80 per cent of the cases there was no notation of discord between ” Philosophical Transactions, Royal Society, London, Series A, 1900, CXCY, pp. 1-47.
parents, and in 20 per cent there was a positive history of discord between parents. Here again the relative incidence of bedwetting is about equal for both groups, with a slight apparent superiority for the group with discord between parents. This relation is expressed in the tetrachoric correlation of .068 ? -024. Both of these coefficients are very low, so low in comparison with their probable errors that one cannot safely conclude that the relation is greater than zero within our representative clinic population of 3,000 cases. Table 2. Summary of Correlation Coefficients Correlation with enuresis (presence or absence, 3000 cases): Correlation with enuresis (age of ceasing, 206 cases): Intelligence quotient (IQ) Number of conduct problems Number of personality problems Underweight condition Discord between parents bis-r= - .078 + .017 bis-r = .109 ?.017 bis-r = .131+ .017 r?= .053+ .027 r,= .068 ?.024 r = .045 ?.047 r = .037 ?.047 r= - .034 ?.047 bis-r = .129 ?.071 bis-r = ? . 123 ? .065
At the beginning of this article mention was made of a group of 206 cases no longer actively enuretic whose age of ceasing was definitely noted in the case histories. A separate study was made of this group on the supposition that a child who is enuretic until the age of, say, twelve years presents a more serious degree of enuresis than a child who establishes dry habits at, say, four years, so that we may have here a measure of relative severity of enuresis. The correlations of this measure with the five factors which have just been discussed are shown in Table 2. These coefficients are too low in relation to their probable errors to permit any interpretation beyond the fact of their small size.
Conclusions
About 23 per cent of a representative group of 3,000 clinic cases had a history of enuresis continuing beyond the third birthday. As measured by its presence or absence among 3,000 cases, enuresis showed a small negative correlation with intelligence quotient, i.e., the enuretic group showed a slightly lower average I.Q. than the non-enuretic children, especially among the children who would be rated among the mentally defective by psychometric findings. There was a low positive correlation with both the number of conduct problems and the number of personality problems, but no significant correlation with underweight condition or with discord between parents.
As measured by the age of the establishing of dry habits among 206 cases formerly enuretic, enuresis showed no significant correlation with intelligence quotient, with the total numbers of conduct problems and personality problems, or with underweight condition or with discord between parents. This study is being extended to a group of 5,000 consecutive cases in order to investigate these relationships more closely upon different sex and age groupings.
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.