Early Socialization Experiences
and Intrafamilial Environment
A Study of Psychiatric Outpatient and Control Group Children
Martha S. Oleinick, M.A.; Anita K. Bahn, Sc.D.; Leon Eisenberg, M.D.;
Abraham M. Lilienfeld, M.D.
National Institute of Mental Health, Bethesda, Maryland;
Johns Hopkins University, Baltimore, Maryland
Archives of general psychiatry. Vol 15, Oct 1966, pages 344-353
Concomitant with the need for prompt and effective treatment of mental. disorders is the need for continuing research on the prevention of these disorders and the promotion of mental health. Knowledge of etiology is crucial, in the long run, for the attack on this highly prevalent health and social problem.
The present investigation, a retrospective case-control study, was designed to determine some of the ways the experiences and environment of children who are or were psychiatric clinic outpatients differed significantly from those of control group children. Data were collected that bear on possible etiologic factors suggested by theoretical formulations and the results of previous empirical research. Major areas of our investigation were: parental behavior and attitudes in early socialization of the child, the nature of family relationships, and separations of the child from either parent. Demographic variables and certain behaviors of the child were also considered.
Here we provide a brief resumé of the study methods and findings. A more detailed presentation will be found in a forthcoming monograph.
Submitted for publication Jan 6, 1966.
From the Register Development and Studies Section Office. of Biometry, National Institute of Mental Health, Bethesda, Md (Mrs. Oleinick and Dr. Bahn), and the Children’s Psychiatric Service, Johns Hopkins Hospital (Dr. Eisenberg) and the Department of Chronic Diseases School of Hygiene and Public Health, Johns Hopkins University, Baltimore (Dr. Lilienfeld).
Reprint requests to Register Development and Studies Section Office of Biometry, National Institutes of Health, Bethesda, Md (Mrs. Oleinick).
SAMPLE SELECTION AND INSTRUMENTS
The total clinic sample of 160 children consisted of 136 psychiatric clinic outpatients from the Children’s Psychiatric Service of the Johns Hopkins Hospital who were white, Baltimore metropolitan area residents, initially admitted to the clinic during the period from September 1959 to September 1960, and diagnosed other than organic brain syndrome or mental deficiency. An additional 24 “new” clinic cases had initial contact with the Children’s Psychiatric Service and the Outpatient Clinic of Child Psychiatry, Psychiatric Institute, University of Maryland Hospital, subsequent to June 1, 1961. An attempt was made to conduct the interview for these new cases prior to clinic contact, i.e., in the interim between the appointment and the first clinic interview.
The patient sample determined the race, age, sex, and to some extent the socioeconomic distribution of the two control groups. The first group, designated the population control, was matched on race, age, sex, and number of siblings to each clinic case from the population of Baltimore public school children. A rough control for socioeconomic level was provided by selection of the control from the same school as the clinic child, since schools draw from a relatively homogeneous population. A neighborhood control was matched on the same basis for 47% (75) of the clinic children who were either under school age (13), in private or parochial schools (23), or resided within the metropolitan area outside of Baltimore City (39). All except three clinic cases were thus matched, giving a population control of 157 children.
The second control group, the hospital control, matched to clinic cases on the basis of race, age, and sex, consisted of children from Johns Hopkins Hospital pediatric clinics, the ophthalmology refraction clinic, or children who had had an appendectomy or tonsillectomy during the same time period that the psychiatric clinic outpatients were seen initially. Because of (1) the limited number of children in the correct race, age and sex groups; (2) the exclusion of chronically ill children; and (3) the loss of children who could not be traced, only 130 hospital controls were obtained. (Number of siblings offered too great a matching problem; socioeconomic distribution was assumed to be similar in the psychiatric clinic and other hospital clinics.) The combined controls numbered 287.
The mother or mother substitute of each clinic and control child was designated as respondent for the interview schedule used in this study. Carefully selected and trained interviewers conducted the two-hour interview in the respondent’s home. Interviewing was “blind,” i. e., interviewers were initially unaware of clinic-control status of subject child, and remained so unless the mother mentioned hospital or clinic contact. An attitude scale (The Parental Attitude Research Instrument of Schaefer and Bell [Ref. 12,13]) was self-administered at the same time. The third instrument used in this study, a classroom teacher’s evaluation form designating the overall adjustment of the child as well as her certainty of the evaluation, was essential to eliminate disturbed children from the control group and thus avoid obscuring the differences. This form is almost identical with that developed by Goldfarb [Ref 4], but modified somewhat to include aspects of the Glidewell, et al, instrument [Ref 3]. Nineteen control children were classified as maladjusted (having “problems of considerable importance — serious” or “very serious problems — extremely maladjusted”; categories 4 and 5) and were excluded from the analyses.
The completion rate of 88% did not differ for clinic and control cases, nor did reason for failure to complete interviews differ. Seven percent of the respondents refused to be interviewed, 2% moved from the study area, and 2% were either never contacted despite frequent visits or were cooperative but unable to complete the interview prior to the termination of the study.
LIMITATIONS OF THE DATA
The nature and limitations of the data must be considered. The basic information is the mother’s report of what were the practices with and attitudes toward the child. This cannot be equated with the actual practices and attitudes, since, as in all retrospective studies, the mother’s reports are subject to the biases of memory.16 Matching on age of child maintains the same time intervals for clinic and control mothers to forget. However, accuracy of memory, conscious and unconscious distortion in reporting, and maternal motivation in the interview situation may be different for the two groups.
The interviewers’ impressions were, however, that for 78% of the interviews rapport was satisfactory or better and communication was adequate. While there were some communication problems in 13% of the interviews, in only 5% did the interviewer question the credibility of the respondent. Clinic and control groups did not differ significantly on any of these dimensions.
Finally there is the unknown effect of clinic treatment on recall, and more importantly, on attitudes.* The group of new clinic cases was to afford some measure of these effects. Unfortunately, because of the small size of this group, we have no conclusive evidence, but only some indication of what may be results of treatment. However, even had this group been sufficiently large, we would have had to raise the question of the effect of emotional distress in mothers awaiting the clinic appointment.
Analysis of clinic case data by diagnostic category was not possible due to insufficient numbers in most classifications.† Significant association of certain variables with specific diagnoses may be obscured in our analysis of combined clinic cases.
Childhood Nudity and Sexual Adjustment
Information from the interview schedule was coded for processing. The coder-pairs averaged 92% agreement. Statistical analysis was carried out by matched triad methods, employing the Cochran Q-test for dichotomized nominal data, and the Friedman two-way analysis of variance by ranks for ordinal data[Ref 14]. When significance was found for triads, pairs were considered (clinic vs population control; clinic vs hospital control; and population control vs hospital control). Relative risk was calculated using Mantel and Haenszel’s formula for matched samples.[Ref 8]
As stated previously, it had been impossible to match some of the clinic children, teacher evaluation classified some controls as disturbed, and other triads were incomplete due to refusals, or loss of respondents moving from the area. The sample size for analysis was reduced to 82 matched triads, representing 246 children.‡
* Amount of clinic contact was not great, in any event, as the median number of visits was eight. Thirty-two percent of the cases had only one or two visits and an additional 12% were classified as not treated, being seen only for diagnosis or evaluation.
† Forty percent were diagnosed as transient situational personality disorder and 40% as personality disorders of different types. Only 17% were classified as psychoneurotic and 2% as psychotic.
‡ An additional 77 triads were partially complete, providing information on 130 individuals. This “miscellaneous group” is considered in the monograph. Due to some matching of cases and controls in the miscellaneous group the usual tests of significance are inappropriate, and incomplete triads prevent use of matched triads methods. Therefore an analysis of differences between the clinic cases and the control cases of this miscellaneous group is impossible. However, the triad clinic cases can be compared statistically with the miscellaneous clinic cases, and similarly the population controls and the hospital controls in the triad group can be compared statistically with corresponding miscellaneous group controls. Despite the smaller size and the differing age and sex composition of the miscellaneous group, general agreement of the findings for these cases with those for the triad cases is seen, and in only a few instances were the χ2 significant.
The results are considered by six areas: (1) demographic and miscellaneous variables; (2) behavioral characteristics of the children; (3) early socialization experiences of the children; (4) family composition and relationships; (5) separations of parent and child; and (6) maternal attitudes.
For each area, we shall present the findings showing significant triad differences, and briefly indicate other variables which were considered. Differences between the two control groups are presented as well as case-control differences. Level of significance is given in parentheses, and where this value differs for comparisons of clinic and population control and clinic and hospital control, both levels are indicated. Table 1 shows percentages and relative risks for significant variables, while Table 2 lists the nonsignificant variables. (Findings on nonsignificant variables are available in the monograph.)
Clinic-control comparisons remain true with social class held constant, unless stated otherwise. Few clinic-control differences reach significance within class groups, however, due to the smaller subgroup size. Therefore, nonsignificant differences. within class in the same direction as total data differences are also designated as agreement. There were no reversals in direction of differences within class groups. Members of individual triads are not always of the same class, since class matching was by a rough index and only for population control; therefore, socioeconomic analysis was done on the total 374 completed interviews for which social class was ascertained. Table 3 presents the social class distribution of clinic and control cases. The distributions are not significantly different. Detailed findings of social class differences will be presented in another publication.
Demographic and Miscellaneous Variables. — While prevalence of mental illness is believed to differ by race, sex, age, and social class,[5,10] we were interested in case-control differences holding these factors constant. Matching eliminated these variables from consideration as etiologic factors. Race and sex were matched completely (the sample is 65% male, 35% female), and age and social class distribution did not differ significantly (Tables 3 and 4) indicating adequate matching. Census figures indicate that our clinic sample, and consequently the matched controls, are significantly better educated and from higher occupational groupings, while comparable in income to the general Baltimore white population. Our sample is also significantly higher in socioeconomic status than the national urban distribution, as estimated by Kahl.
Analysis of other demographic variables showed no difference for matched triads on the following for either mother or father: age, education, place of birth, farm residence, and intergenerational mobility. Father’s occupation, income, and classification as entrepreneurial or bureaucratic were not significant.
The single demographic variable showing significance was religion. The Q-technique for matched triads requires dichotomization, necessitating analysis in the form of: Protestant vs other; Jewish vs other; Catholic vs other; and “mixed” vs other. The clinic group and the population control did not differ, since homogeneity on religion was produced by the selection of the population control from the same school and/or same neighborhood as the clinic children. However the clinic group had significantly more Jewish children than the hospital control (0.001) and significantly fewer Catholic children (0.025). § (Population control had more Jewish children [0.01] and tended to have fewer Catholic children [0.10–0.05] than the hospital control.)
Complications of pregnancy with the subject child and prior or subsequent miscarriage or neonatal deaths were not significant.
§ Twenty-one percent of the clinic sample is Jewish, while only 11% of the total control sample is Jewish. However, teacher evaluation shows that Jewish control children and non-Jewish control children do not differ significantly in adjustment. It therefore appears that the difference between psychiatric clinic and hospital control cases in religious affiliation may be an indication of different subcultural attitudes regarding the seeking of psychiatric help with emotional problems.
Behavioral Characteristics. — The clinic children showed consistently more “problem” behaviors and “symptoms” than did the control group. Considering all “problem” behaviors, clinic children had a greater number occurring frequently (0.01, population; 0.025, hospital). With regard to specific behaviors, clinic children were reported more cruel (0.025, population; 0.01, hospital), more sulking and crying (0.10–0.05, population; 0.025, hospital), more whining than hospital control (0.025), and more irritable than population control (0.005). Clinic children also were reported to have more fears than the population control (0.005), but not more than the hospital control. Clinic children more frequently had nightmares (0.001, population; 0.05, hospital). Temper was more often a problem with clinic children (0.001). (The preceding specific behavior problems were not ascertained by social class.) Clinic children were reported more often to evidence dependency (not for social class V) by clinging and following (0.025, population; 0.005, hospital). Peer and sibling (not for social class V) relationships were poorer for clinic children (0.001, peer; 0.01, sibling). In addition to mothers’ reports, teacher evaluation showed the control group better adjusted (0.001), and school achievement, as indicated by school status, showed the clinic children more often below the school grade modal for their age ¦ (0.01, population; 0.005, hospital).
Hospital control children differed from population control children in having more fears (0.05), more problem with temper (0.01), and also a tendency to more frequent nightmares (0.10–0.05).
There was no significant difference in daydreaming, sensitivity, lying, occurrence of sex play with another child, eating problems, oral habits (nail biting, thumb sucking, etc), compulsive or “messy” habits, constipation, or enuresis.
¦ This difference occurred in spite of the fact that, in matching the population controls, school grade was used to determine which of two possible age matches would be considered first in family size matching.
Early Socialization Experiences. — The area of early socialization experiences showed very little difference between the groups. Spanking by the father was more frequent for clinic children (0.02, population; 0.055, hospital ¶). Reasoning as a technique of training and discipline (not for class I and II and class V) was used less often by clinic mothers (0.05 population; 0.10–0.05, hospital). (The Q-test was for “never used” vs other categories.)
Feeding method, scheduling, use of pacifier, weaning age and methods, toilet training age and methods, treatment of aggression and sexual exploration, exposure to sexual stimulation (familial nudity, sharing of bathroom, sleeping in same bed or bedroom), and most methods of training and discipline were not significantly different.
While there were no significant clinic-control differences in weaning and toilet training experiences, there were some immediate consequences of the methods of training employed. Combining all groups, severe weaning produced significantly more disturbance (0.025) at the time. Considering age of weaning, less disturbance occurred if initiation was early (8 months or younger) and most disturbance occurred if weaning was initiated late (18 months or older) (0.001). Severe toilet training produced more disturbance (0.001) and was associated with significantly more constipation (0.05). Early toilet training (initiated at 7 months or younger) was associated with greater “compulsiveness” as measured by more “neat and orderly” behaviors (0.05) and fewer “messy” behaviors (0.01). #
¶ The χ2, was significant (0.05), but the Q-test was not. Therefore, the Sign test was used for paired comparisons of clinic with each control and of the two controls.
# One is unable to determine whether the late weaning mothers weaned late because of anticipated disturbance and were “set” to perceive more disturbance. Similarly, the severe toilet training mothers might be more concerned with the area of toileting and perceive “constipation” where other mothers would not. Also the mother who initiates toilet training early may produce “compulsiveness” by demanding greater “neatness” in other areas, rather than by the toilet training itself.
Family Composition and Relationships. — This area showed the most numerous clinic-control differences. Clinic children had fewer natural mothers as opposed to other categories (stepmother, other mother surrogate, or no mother figure), (0.001, population; 0.05, hospital). The presence. of the natural father was also significantly less for the clinic than for the· population control (0.001), but not for clinic vs hospital control (0.10–0:05). The hospital control had significantly fewer children living with natural mother (0.05) and natural father (0.05) than did the population control.
As would be anticipated from the above findings, marital status of parents showed that more clinic children had a stepparent, or a divorced, separated, or widowed parent (0.001, population; 0.05, hospital).** Associated with this result is the fact that clinic children more often had step, half, or adoptive siblings than did the population control (0.025). The two controls differed, hospital control children having more siblings in these categories (0.05) and tending to have more disrupted families (divorced, etc 0.10–0.05).
Sibling relationships were significantly poorer for clinic children, as previously stated in the section on problem behaviors. When compared with the population control, the marital relationship of the parents of clinic children were significantly poorer: less satisfaction with marriage (0.01), less happiness in marriage (0.025), and mother had more often considered separation from father (0.01). Comparison with hospital control showed clinic parents having significantly more frequent serious disagreements during the past 12 months (0.005), and less satisfaction with marriage (0.05). ††
As a consequence of, or perhaps contributing to, the poorer marital relationship, differences between the parents regarding the discipline and training of the child (not for social class I and II, and V) were more frequent for clinic than for population control (0.005). While clinic and hospital control did not differ, the latter parents showed more conflict than the population control (0.005). Family composition variables which showed no significant differences were: birth order of subject child, sex-sibling configuration, number of siblings (a matching variable for population control only), age difference between subject child and adjacent siblings; presence of other individuals (extended family members and nonrelated individuals) in the household; maternal employment‡‡ and the influence of the father’s job on family relationships as shown by number of hours worked per week and absence from the home overnight.
** Only 74% of the clinic children were from intact homes compared with 95% of the population controls and 85% of the hospital controls. The 1960 census in Baltimore SMSA showed 90% of the white children under age 18 living with both parents, comparable to. our control groups.
†† Of intact homes, 82% control and only 67% clinic parents were reported to have good marital relationships (considering both “happiness” and “satisfaction”). Combining the variables of intactness and quality of marital relationship, 50% of the clinic children, 78% of the population controls and 70% of the hospital controls were in “happy-intact” homes (0.001, population; 0.01, hospital).
‡‡ Full-time maternal employment, however, is significantly more frequent in broken and unhappy-intact homes (0.001) (57% in broken homes, 38% in unhappy-intact, 23% in happy-intact).
Separations of Parent and Child. — With disruption of the marital relationship, one would expect to find more frequent separations of the child from one or both parents. Separations of the mother and child were more frequent for clinic children (0.001, population; 0.01, hospital). Considering only longer separations, of two months or more, clinic children were more frequently separated than population control (0.001), but not more than hospital control. Considering only separations for “disruptive” reasons (that is, excluding separations due to camp, trips, vacations), clinic had significantly more than both controls (0.005, population; 0.05, hospital). Clinic cases had significantly more long-disruptive separations (0.001, population; 0.05, hospital). (Hospital control children tended to have more maternal separations than population control [0.10–0.05], had significantly more longer separations [0.005], did not differ from population control considering only disruptive separations, and approached significance considering only long-disruptive separations [0.10–0.05].)
Separations of father and child also were significantly more frequent in the clinic group than in the population control looking at all separations (0.001), only longer separations (0.05), only disruptive separations (0.025), and only long-disruptive separations (0.005). Clinic and hospital control were not significantly different, but clinic tended to have more, considering all separations (0.10–0.05). (Hospital control had significantly more paternal separations than population control, looking at all separations [0.05], and approached significance considering long-disruptive separations [0.10–0.05].)
Maternal Attitudes. — Turning to maternal attitudes, mother’s general description of the child, evaluated as to “tone” showed the clinic mothers more “neutral or negative” (0.005, population; 0.01, hospital). Similarly, the overall evaluation of acceptance of the child showed clinic mothers to be more rejecting (not for social class V). (Triad analysis Q significant 0.025; 0.10–0.05, population; NS, hospital). Clinic mothers reported frequent inconsistency on their part in disciplining the child — threats were not followed through once a week or more (0.005, population; 0.001, hospital). Other maternal attitudes, such as choice of happiest and least happy times, autonomy-control, behaviors valued most highly, and most PARI scale variables §§ were not significantly different.
§§ Four PARI Scales gave significant χ2: Clinic showed less fear of harming baby, less exclusion of outside influences·, less deification of parents, and less intrusiveness. However, means were so close that one doubts any basic attitude difference or any difference in maternal behavior in these areas.
The data on the population control support its appropriateness as a control for the clinic group. The mothers of the population control perceive their children as relatively healthy and teacher evaluation supports the mothers’ reports. (School records often contain information regarding the child’s emotional problems; thus knowledge the teacher may have had of previous clinic treatment could have biased her judgment.) The mothers’ descriptions of the clinic children indicate many more areas of problem behavior, a finding which, while expected, was necessary if any reliability was to be attributed to the survey technique. Clinic mothers may of course be more willing to reveal problem behaviors or may report problem behavior more frequently than it occurs. However, the evidence of the Lapouse and Monk study shows correspondence between mothers’ and childrens’ reports of symptoms and Gildea and others note correspondence between teacher and parent evaluation of disturbed children.
The psychiatric literature has placed great emphasis on early socialization experiences as pathogenic factors in childhood and adult disabilities. Attempts to confirm this supposition by careful comparisons of clinic and control groups have, however, revealed quite conflicting evidence. The most notable aspect of our data is not that two socialization measures (spanking by father and use of reasoning by mother as a technique of discipline) do show differences between clinic and control but rather that the other 30 or so variables in seven areas showed no difference. Under such circumstances, differences on two measures at a 0.05 level cannot be regarded as reliable. Indeed, differences in socialization practices seem much more related to social class than they do to psychopathology (see monograph). (However, small numbers prevented analysis by diagnosis and significant association of certain socialization experiences with specific symptomatology could be obscured in analysis of combined cases.)
The clinic group was clearly separated from both the hospital control and the population control with respect to disruption of the nuclear family and marital discord in those families that were intact. Thus, some 78% of the families in the population control were both intact and described themselves as happy, as did some 70% of the hospital control, but only 50% of the clinic group were so classified. (Despite the many papers describing significant associations between the loss of a parent early in life and serious mental disorder in adulthood, a recent careful study by Pitts and others  failed to demonstrate any difference between adult admissions to a psychiatric unit and to a medical unit in the prevalence of parental loss in childhood.) Whatever the long range significance, our data along with those of other workers, demonstrate immediate consequences in terms of behavior disorder in childhood. This is completely in accord with clinical experience and expectations based upon most current theories of child development. What is not clear is the long-term consequence of these early events. Indeed, many studies have failed to separate the emotional consequences of the loss of a parent from the social consequences, including economic effects.
A striking and unexpected finding is the difference between the hospital control and the population control such that in a number of areas the hospital control tends to be similar to the clinic group though less severely affected. Hospital control patients were deliberately selected from those who had had minor surgical procedures, acute minor illnesses, and simple refractions. This was done in an effort to rule out the complications introduced by chronic illness. Yet loss of a parent and separation from a parent were more common as were certain behavior disturbances (fears, temper, nightmares, and teachers’ reports of severe maladjustment). It may be that there is greater likelihood of the child appearing at a medical or surgical clinic for elective treatment of a minor disorder if there is tension in the household. Child psychiatrists have considered that the behavior disorder of a child may be no more than an “admission ticket” to the clinic for a family in distress. Perhaps, over-concern with minor physical disorders in the child or urgent pressure for elective surgical procedures mayplay the same role in the presence of family disequilibrium. (Meyer and othershave recently reported on psychosocial factors in the pattern of appendectomies in an adult population.)
It is of some interest that social classes I and II were somewhat overrepresented in the clinic population and social class V somewhat underrepresented. This occurred in an administrative situation in which there was no ostensible selection by economic condition. Indeed, the pediatric clinics of Johns Hopkins Hospital serve a heavy portion of the indigent population in this area. This skewing of the psychiatric clinic population is similar to that which has been described for adults and in a study of the flow of children to outpatient psychiatric facilities in New York City. The covert screening may occur either at a level of referrals (with physicians and others being more likely to regard behavior symptoms in the lower social class as requiring solution at the level of social case work) or at the level of clinic intake at which, despite an overt commitment to service to all, subtle bias may result in admitting preferentially those cases which fit the clinic’s concept of the ones they can work with most effectively. It should be reiterated, however, that clinic patients selected for this study were all white and that census data used for comparative purposes were for white residents of the Baltimore SMSA. If the nonwhite clinic cases had been included and census figures for total population had been used, the findings might differ. Class attitudes toward seeking treatment may also vary.
This study of a group of 82 clinic patients matched against two control groups, a population control selected from the area of residence and a hospital control selected from cases admitted for minor medical and surgical procedures, revealed the feasibility of its methodology in that interviewers were accepted by a very high percentage of the families selected and were able to complete the lengthy two hour interview schedule. Major findings revealed that socialization experiences within the family did not differ significantly between the clinic population and its controls but correlated highly with social class, a subject covered in a separate monograph. Clinic patients differed from both control groups in the presence of symptoms of disturbed behavior and in teachers’ reports of aberrant behavior. They differed significantly in terms of the frequency with which the biological nuclear family had been disrupted and the frequency with which marital discord was present in intact families. Separations of the psychiatric child patient from his parents had occurred with far greater frequency than was true for either the population or the hospital control. An interesting but unexpected finding was the difference between the hospital control and the population control on broken families, separation experiences, and symptomatic behavior in the child. The clinic sample was of somewhat higher social class than the Baltimore metropolitan white population, indicating some bias in referral or intake procedures or in class attitudes toward seeking treatment.
The failure to demonstrate significant association between psychopathology and socialization experiences within the family cannot be said to have ruled out such an association, given the sample size and the fact that the information was retrospective and was dependent on mothers’ reports rather than actual observation. However, it is in keeping with the inconsistency of results in the reports of many other investigators.
1. Furman, S. S.; Sweat, L. G.; and Crocetti, G. M.: Social Class Factors in the Flow of Children to Outpatient Psychiatric Facilities, Amer J Public Health 55:385-392, 1965.
2. Gildea, M.; Glidewell, J. C.; and Kantor, M. B.: “Maternal Attitudes and General Adjustment in School Children," in Glidewell, J. C. (ed.): Parental Attitudes and Child Behavior, Springfield, III: Charles C Thomas, Publisher, 1961, pp 42-89.
3. Glidewell, J. C., et al: Behavior Symptoms in Children and Adjustment in Public School, Hum Organization 18:123-130, 1959.
4. Goldfarb, A.: Judgments by Teachers, Principals, and Mental Hygienists of the Relative Importance of Behavior Problems Occurring in Elementary School Boys, unpublished doctoral dissertation, Baltimore: The Johns Hopkins University School of Hygiene and Public Health, 1959.
5. Hollingshead, A., and Redlich, F.: Social Class and Mental Illness, New York: John Wiley & Sons, Inc., 1958.
6. Kahl, J. A.: The American Class Structure, New York: Holt, Rinehart and Winston, Inc., 1957.
7. Lapouse, R., and Monk, M.: An Epidemiologic Study of Behavior Characteristics in Children, Amer J Public Health 48: 1134-1144, 1958.
8. Mantel, N., and Haenszel, W.: Statistical Aspects of the Analysis of Data From Retrospective Studies of Disease, J Nat Cancer Inst 22:719-748, 1959.
9. Meyer, E.; Unger, H. T.; and Slaughter, R.: Investigation of a Psychosocial Hypothesis in Appendectomy, Psychosom Med 26:671-681, 1964.
10. Myers, J. K., and Roberts, B. H.: Family and Class Dynamics in Mental Illness, New York: John Wiley & Sons, Inc., 1959.
11. Pitts, F. N., et al: Adult Psychiatric Illness Assessed for Childhood Parental Loss, and Psychiatric Illness in Family Members-A Study of 748 Patients and 250 Controls, Amer J Psychiat 121: i-x, 1965.
12. Schaefer, E. S., and Bell, R. Q.: Development of a Parental Attitude Research Instrument, Child Develop 29:339-362, 1958.
13. Schaefer, E. S., and Bell, R. Q.: Informal Notes on the Use of PARI, mimeographed material, Bethesda: Laboratory of Psychology, National Institute of Mental Health, 1959.
14. Siegel, S.: Nonparametric Statistics For the Behavioral Sciences, New York: McGraw-Hill Book Co., Inc., 1956.
15. US Bureau of the Census: US Census of Population, 1960, Detailed Characteristics, Maryland, Final Report, PC (1) 22D, US Government Printing Office, 1962.
16. Wenar, C.: The Reliability of Developmental Histories, Psychosom Med 25:505-509, 1963.