National Center for Health Statistics

This report presents basic data on parent ratings of certain behavioral patterns related to growth and development by sex and age f,Jr youths 12-17 years of age in the noninstitutionalized population of the United States. The data were obtained from the Health Examination Survey of 1966-70 and should provide information on the behavior of adolescents in the general U. S. population that previously has been unavailable or inadequate. These findings are based on responses given on a self-administered medical history quesUonnaire. A total of 6,768 youths were examined. A descriptive analysis has been made of the parents' ratings concerning the general health, peer relations, mental development, and emotional health of thfsir offspring. Certain general findings are summarized and compared with results from previous studies of the behavior and development of adolescents. Specific behavioral patterns are examined in relation to jeneral and emotional health evaluations and assessments of mental development. (Author)

In this report basic data on parent ratings of certain behavioral patterns related to growth and development are presented by sex and age for youths 12-1-years of age in the noninstitutionalized population of the United States. Other reports on various aspects of the behavior of youths as rated by parents, teachers, and the youths themselves are to follow. The data in these reports were obtained in the Health Examination Survey of 1966-70 and should provide information on the behavior of adolescents in the general U.S.. population that previously has not been available or has been inadequate. Reports on children 6..11 years of age in the United States which are similar except for the self-reported information have already been published. ' -1 The Health Examination Survey is one of three major programs of the National Center for Health Statistics, which conducts the National Health Survey as authorized in 1956 by the 84th Congress.; Tile Health Interview Survey, which is used to obtain information by household interview among samples of persons, is concerned primarily with the impact of illness and disability on the lives and actions of people. The Division of Health Resources Statistics cola acts health data as well as health res.lurce and utilization information through surveys of hospitals, nursing homes, and other resident institutions and the various persons in the health occupations.
In the Health Examination Survey (HES) data are collected through direct physical examination, tests, and measurements performed on the sample population selected for study. This is viewed as the most accurate way to obtain definite diagnostic data on the prevalence of certain medically defined illnesses. It is the most precise way to secure reliable information on unre%.ognized and undiagnosed conditions as well as on a variety of physical, physiological, and psychological measurements within the population. In addition it makes possible the study of relationships among the various examination findings and between these findings and certain demographic and socioeconomic factors.
HES is carried out as a series of separate programs referred to as "cycles." Each cycle is concerned with some specific segment of the total U.S. population, usually a particular age group, and with certain specified aspects of the health of that subpopulation. In the first cycle data were obtained on the prevalence of certain chronic diseases and on the distribution of various measurements and other characteristics in a defined adult populatior.'h In Cycle II a probability sample of the Nation's noninstitutionalized children 6-11 years of age was examined. The examination was directed primarily toward obtaining information on health factors related to growth and development, but it also included screening for selected diseases or abnormalities. An assessment was made by a dentist, tests were administered by a psychologis., and certain other measurements were made by technicians.' Cycle III, on which this report is based, covered youths 12-17 years of age at the time of survey. A comprehensive description of the survey plan, sample design, and examination content has been published.' Apart from age, the specifications of the program were similar to 1 those of Cycle IL Its target was the roughly 23 million youths 12-1" years of age (married or single) living In the United States outside institutions. Field collection operations started in March 10(4) and ended in December 19'0, During the period 69'68, or 00 percent, of the youths selected for the sample were examined. The examination focused on health factors related to growth and development and included medical examination of the eye, ear, nose, and throat, a check for goiter, a musculoskeletal and neurological evaluation, a cardiovascular examination, a dental examination, and a vision test. Several tests were administered by a psychologist, and a variety of other tests, procedures, and measurements were made by technicians.
A standard single-visit examination was given each youth by the examining team in mobile units specially designed for the survey. Prior to the examination, information including demographic and socioeconomic data on household members was obtained from the youth's parent or guardian. The parent also furnished a medical history and behavioral and related data on the examinee. Supplementary and supporting information was obtained from the youth himself; ancillary data, including grade placement and a teacher's rating of ability, performance, behavior, and adjustment to school, were requested from the school last attended. A birth certificate for verification of the youth's age and other information related to birth were also obtained. All information was collected under conditions of confidentiality. statistical comments on the survey design, reliability of data, and sampling and measurement error are included in appendix I. Standard errors associated with the percentages or rates presented are shown for the totals in the respective tables. 'the others may be derived from data presented in appendix I.

BEHAVIORAL DATA
Certain behavioral information related to the ex.rninev's growth and development was obtained in this survey from a parent, usually the mother, and from the school he last attended. In addition, each youth was given a questionnaire on health habits and history and asked to complete, it and return it on his visit for examination. Ile was 2 asked to complete another questionnairethis one on health-related behavior.-at the examination center. Large portions of the parent's and youth's questionnaires were designed to secure parallel views and attitudes from the two regarding selected types of behavior related to health. Ratings were given by a teacher who was thought to have sufficient knowledge to do this adequately.
This report covers only the evaluations and attitudes of the parent as expressed in his responses to the questionnaire, The parent gave information on a self-administered medical history questionnaire (see appendix II) left in the home by a U.S. Bureau of the Census interviewer to he picked up in about a week by the FILES field representative. At that time the I-ILS field representative reviewed it and tried to resolve any problems that the parent net in furnishing information.
The parent was asked to answer questions about the youth concerning his general health, specific health problems, mental development, school experience. eating habits, peer relationships, independence, values, educational goals, reactions to Illnesses, and cet win other related subjects.
These questions, shown in appendix II, were designed to elicit responses from which ratings of the general health, health-related behavior, mental development, social adjustment, and emotional health of the youth could he derived.
The principal-reason for including these behavioral questions in the survey was to be able to relate information on health, behavior, attitudes, and other questionnaire data to specific medical examination findings and to results of the psychological tests of performance, school achievement, and emotional adjustment. Another purpose was to provide a basis for comparing the expressed perceptions, attitudes, and values of the youths with those of their parents regarding a variety of topics, e.g., expectations concerning formal educational achievement, independence in decisionmaking, and standards of behavior. However, in this basic report of findings from the parents' responses, the patterns revealed are considered to have considerable value per se as baseline data for defining parental attitudes and describing their perceptions of their adolescent children with respect to the areas covered.

FINDINGS
In the iterpvetation of these findings, particularly with respect to differences by sex, attention should be given to the unplanned circumstance that the evaluations were made primarily by the female parent. Mothers completed 83 petcent of the questionnaires by themselves, and they were involved as respondents, either by themselves or with fathers or other persons, in 94 percent of the sample cases. Fathers completed 5 percent of these forMs by themselves and were involved, either as the only respondent or with other persons, in 8 percent of the cases.

General Health and Health-Related Behavior
Questions asked the parent regarding the youth's present health, rate of physical growth, relative body weight, reaction to illness, and eating habits were intended to elicit parental attitudes and evaluations of certain general health aspects of the examinee's growth and development.
Nine out of 10 parents responded that they thought the youth's physical growth had been occurring at "about the right rat;" (table 1). Parents of boys were slightly more likely to he concerned that physical growth was too slow than that it was too fast, while }.iris' parents tended to express the opposite concern. This difference was found not to be statistically significant.' But considered in another way, a significantly higher proportion of boys than girls were reported as growing too slowly (4.5 percent and 2.0 percent, respectively). Four out of five youths were thought to be at the proper weight (table 1), with slightly more concern being expressed about overweight than underweight, particularly with respect to older girls (figure 1).
The data show that nearly 15 percent of the parents were worried about what they considered a health problem affecting the south (table I ).
aThe tt,ntlustun IS ILISell on the Nlitiart'll is sample silt:. weighted pert. ellt.Iges. arid vart.Iflues computed by the pseutloreplit anon method and shown In the tables and ap pendix 1 of th!, riyort. i ur eAp1.111.it ion of method, see page 41 ut refereme I. In this report. the statistual mice ut a thtfereme in prtlpt)rtit.M1 was tutigeti at the 1. While the proportion of boys reported as having health problems increased with age, for girls this trend is not observed. With regard to general health, 96 percent of the youths were pronounced to be in good or excellent health (table 2), with remarkable consistency evident among data for the age and sex groups. Overall, only two out of five of the adolescents were described as never exaggerating illness (table 2). while about 4 percent were said to exaggerate illness "pretty often." Basic data on eating habits are shown in table 3, which shows that four out of five youths were reported to he eating the right amount of food; more of the remainder were considered to be eating too much than too little. Although a larger proportion of the girls were reported to be overweight, a slightly higher proportion of girls than boys were thought to be eating too little. For boys as well as girls a consistent relationship was observed between the estimate of how much the youth was eating and his weight status as seen by the parent (table 4). One-half of the adolescents were reported to be "not fussy at all" about food, and 7 to 8 percent were considered "very fussy.' There was a consistent pattern to the responses on attitude toward eating among the age and sex groups.

Social Behavior
To ascertain patterns of conduct .n certain social settings, questions were asked concerning the youth's ease in making friends, frequency of overnignt visits, the parent's acquaintance with the youth's friends, frequency of meals eaten with the family, and the amount of trouble the youth was to bring up.
Three out of five youths were said to usually eat at least two meals a day with the rest of the family (table 3). Eating with the rest of the family occurred less often among older youths, particularly older girls, lipys were more likely to eat with other members of the family.
Eighty-two percent of the youths were reported to make friends easily and 55 percent to have visited overnight with their friends "quite a few times" (table 5). A higher proportion of girls than boys were said to visit overnight frequently. As expected, frequency of overnight visits increased steadily for all youths as age increased. Three-fourths of the parents reported that they were well acquainted with most of their child's friends (table 6).
About 60 percent of the parents stated that they experienced no trouble in bringing up the youth (table 6). A higher proportion of boys than girls were reported to have been "a lot of trouble" to bring up, and a higher proportion of the parents said that they had had no trouble bringing up girls.
Youths who never visited overnight with friends and whose friends the parent knew well were less often considered difficult to rear than were youths who visited overnight more often and whose friends the parent did not know well (tables 7 and 8). 4

Early School Experience and Mental Development
To collect information on the effects of formal methods of fostering mental development, ques. dons were asked the parent concerning the examinee's attendance at nursery school and kindergarten, age on entering the first grade, and inidal reaction to the school experience. The parent was also asked to evaluate the overall rate of mental development observed in the youth.
Early school experience. -..Only 9 percent of these youths had attended nursery school, but twothirds had gone to kindergarten; one -half of them were sent voluntarily (table 9), Three-fourths entered the first grade between their sixth and seventh birthdays. More youths entered elementary school before the age of 6 than did so after becoming 7 years old. According to their parent's ecollection, three out of four youths were "quite happy" on entering the first grade (table 10). A higher proportion of girls were initially happy to be in first grade than were boys. Attendance at nursery school or compulsory kindergarten was found to have no significant association with the specific initial reaction to first grade (tables 11 and 12). However, a larger proportion of the girls whose parents voluntarily sent them to kindergarten were reported to have been happy on entering elementary school.
Mental development. Table 13 shows the distribution of responses to the item which asked for an evaluation by the parent of the youth's rate of mental development, About 95 percent thought that the youth had developed mentally at the proper rate. Among the remainder, more parents said that development had been too slow than too fast. Proportionately more than twice as many boys as girls (5 percent compared with 2 percent) were reported to be slow in development. A relatively high proportion (8 percent) of youths with health problems were reported to have developed too slowly in the cognitive area; for other youths the proportion was 3 percent, The right rate of mental development was reported for a lower proportion oi girls said to have health problems (92 percent) than for those without these problems (97 percent). For both sexes, but more so for girls, mental development was thought to be either too fast or too slow more BEST COPY AVAILABLE often ninon)); those vonsidered to have health prob. lems than among tho.,e without such problems. Overall, a tendency to exaggerate illness was not found to be significantly assocated with the reported rate of mental development (table 14). llowever, fewer youths considered to have shown the right rate of mental development were reported to exaggerate illness "pretty often." In table 15+ a positive association between rcsponses of 'right rate" of physical groWth and "right rate" of mental development may be obsers ed. .\ large proportion of those thought to be slow, in physical growth were reported to be slow in mental development too, Compared with other youths, a larger proportion of those reported as making friends vas-I Iy were considered to have developed at the proper rate mentally (," pereent), while one-third of those who were said to have a lot of trouble making friends were considered to have -too slow-.1 rate of mental development (table  The` association between slow mental development and difficulty in making friends was closer for boys than for girls; it was also closer for youths in the older age groups.
The data show that boys considered to have developed mentally too slowly gave parents more trouble to bring up than those thought to have de-'eloped at the right rate stable For either too fast or too slow mental development was positively associated with more trouble to bring up.

Emotional Development
Nvrpousness.one-half of the youths were reported -not nervous at all,-while 4 percent wvre said to he "very nervous" (table IS). These proportions vary little by sex and age. The prevalence ratios of reported nervousness among youths related to parental evaluations of their general health are shown in table lg. l'or all youths the proportion reported as nervous ranged from around 40 to 80 percent in a neat progression through the various health stavis groups, with more indr.:ation of nervousness among those considered to he in poorer health. The association between reported nervousness and health status was more prorounced for girls than for boys (figure 2). 'Fable 20 and figure 3 show the prevalence of nervousness among youths who reportedly had a health problem that worried the parent compared with those who did not. Two-thirds of those with health problems were said to be nervous, while less thin' one-half of those without such problems were so described. Nervousness tended to he associated with a "fussy-attitude toward foodmore so for girls than for boys (table 2i). Also, a statistically significant positive relationship was evident between the responses on nervousness and those on the tendency to exaggerate illness (table 22). Relating parents' responses on degree of nervousness and rate of mental development revealed that a larger proportion of youths who were thought to have too slow mental development were also reported to he nervous (table 23). This was true both for boys and for girls. However, among those youths reported as developing too fast intellectually, a lower proportion Of boys than girls were said to he nervous (figure 4).
With regard to peer relations, degree of nervousness was found to he significantly related to the amount of trouble the youth had making friends (table 241. However, a larger proportion of youths who never visited with friends overnight were said to be not nervous at all" than were those who did visit overnight (table 25). 'fable 26 shows that less nervousness was reported for youths whose parents thought they knew most of their friends well than for the whose parents did not. werall 4 percent of the youths wore reported to he "very nervous,-but in groupings according to certain parental evaluations the proportion was considerably higher. ousness and those selected characteristics that were most closely associated with such an evaluation. Rate of physical growth (table 27) and ability to make friends (table 2.1) are more closely related to reported degree of nervousness among the males than among the females.
Bedtvettirtgo.Five percent of the youths were reported to have wet the bed during the year pre. ceding their examination. A higher proportion of boys (6 percent) than girls (1 percent) were said to have done so (table 6). In continuation of the trend observed among the children 6-11 years of age in the previous survey.l there was a general decline in bedwetting with increasing age (table   In table 20, youths who wet the bed are compared with those who .id not with respect to their parent's ratings on degree of nervousness, A larger proportion of those of each sex who wet the bed were also reported to he nervous.
It may 1w seen in table 2t that of the 5 percent of the youths who were reported to have wet the bed in the past year, about one-fifth were said to he in excellent health. It is also evident that the prevalence of reported enuresis increased as the status of health was viewed less favorably, from three per 100 among those said to be in excellent health to I q per 100 among those thought to he in poor health. were reported also to have been treated by a psychiatrist or psThologist; i% slightly higher

7]
proportion of the boys made visits to mental health treatment facilities (3.6 percent, compared with 2.3 percent for girls), but the difference was not significant. However, a significantly higher proportion of boys (7.8 percent) than girls (4.4 percent) were reported to have been treated by Of youths said to have health problems, visits a parents indicated that 5 percent had mental hospital or guidance clinic and that 11 percent had been treated by a psychologist or a psychiatrist; lower proportions of and 5 percent, respectively, were found for those with no reportol health problems (table 2°).
Overall, 6.3 percent of the youths were reported to have been given mental health care either. in mental hospitals guidance clinics or h psychiatrists or psychologi.:ts outside of such facilities. The proportions report idly given mental health care under such circumstences among groupings of youths according to selected behavioral char acteristics are presented for comparison with the average rate for all youths in table

THEORETICAL CONSIDERATIONS
Some findings from this survey of parents' views cm the development and behavior of their adolescent children lend support to some con- In a positive sense the findings of Douglas and Ross In that advanced physical status in the adolescent is significantly associated with greater educational ability were generally ;supported by the collective parental assessment: covered by this ane!ksis. .1s in the !I S survey, liorowitil found that p.)pule rite with peers among yuth:4 is greater for th h e with the ' 'proper" rate of men ta 1 developmeet. The finding-from this survey with respect te parents' rseings of the youths' socializal ion tend :o confirm the (.6oncinf;ions of poavan and Adelson' and Musgrove '-' that girls evidenced more preoccupation with friendship end social relations, while boys were more con-8 cerned sith status (involving achievement-related skills) and independence. Thus, according to their theory, girls would be more concerned with evoiding obesity, visit more frequently, make friends more easily, tead to he nervous when their rate of mental deveopment seemed to be "too fast," be less trouble for parents to bring up, and be happier on entering the first grade in elementary school. For boys, on the other hand, underweight, slow physical growth, and slow mental development would be considered grave disadvantages; resistance to rules and authority at home or in school would he more evident. These conclusions, which the data in this report tend to supporteither as differing concerns of the female parent for the youth or as fair representations of the behavior of the adolescents involvedfit well with those drawn from a world sample made up from various studies and reported by, among others, Barry, Bacon, and Child 1 I and Hallworth and Waite. I SUMMARY This report presents by age and sex estimates of the distributions of parental evaluations of selected behavioral characteristics of youths aged 12-1' years in the U.S. noninstitutionalized population. These findings are based on the responses given on a self-administered medical history qaestionnaire concerning the youths examined in the Health Examination Survey of 1066-"0. A total of h."68 youths were examined.
They comprised a sample drawn to he closely representative of American adolescents with :espect to age, sex, race, region, and certain other available demographic and socioeconomic factors.
Data ire presented on various aspects of the growth and development of children during the adolescent period as seen by their parents. A descriptive analysis has been made of the parents' ratings concerning the e,eneral health, peer relations, mental development, and emotional health of their offspring. Specific behavioral patterns are examined in relation to general and emotional health evaluations and assessments of mental development. Certain general findings are summarised and compared with results from previous studies of the behavior and development of ad-. olescents.
The responses of the parent (in most cases the mother) to questions on the health, growth, and development of the examinee tended to be of a favorable nature by a ratio of 6 to I. Among the various aspects of health-related behavior and development of the adolescents studied, some-what higher proportions of definitely unfavorable responses were expressed with respect to their tendency to exaggerate illness, the existence of health problems, ability to make friends, reactions to entering first grade, and difficulty involved in their bringing up. Percent distributions of youths by amount of trouble they were to bring ap and parent's acquaintance with their friends and percent of youths who wet bed during the past 12 months, according to age and sex, with standard errors for totals: United States, 1966-70 17 7. Percent distribution of yQuths by amount of trouble they were to bring up, according to frequency of overnieht visits with friends, age, and sex, with standard errors for totals: United States, 1966-70 18 8. Percent distribution of youths by amount of trouble they were to bring up, according to how many of their friends parent knows well,age, and sex, with standard errors for totals: United States, 1966-70 19 9.

REFERENCES
Percent of youths who att...nded nursery school, percent who attended kindergarten and percent distribution by whether attendance was voluntary or compulsory, and pereent ciftri'"utlon t whih thc:      Percent distributions of youths by amount of trouble they were to bring up and parent's acquaintance with their friends and percent of youths who wet bed during the past 12 months, according to age and sex, with stuadord errors for totals:

2.71
Boys 12-17 years--              Percent of youths who wet bed during the past 12 months, percent distribution of these youths by present health status, and prevalenoe rate of youths who wet bed per 100 youths by present health status, according to age and sex, with standard errors for totals: United States,   This percentage differs slightly from the corresponding one shown in table 6,which includes a few youths for whom health status was not reported.

STATISTICAL NOTES
The Survey Design The sample design for each of the first three programs of the Health Examination Survey (Cycles I-III) were essentially similara multistage, stratified probability sample of clusters of households in land-based segments. The s iccessive elements for this sample design were primary sampling unit, census enumeration district, segment (a cluster of households), household, eligible person, and finally the sample person.
The 40 sample areas and the segments utilized in the design of Cycle III were the same as those in Cycle II. Previous reports describe in detail the sample design used for Cycle II and in addition discuss the problems and considerations given to other types of sampling frames and whether or not to control the selection of sibling 8. T.16 Requirements and limitations placed on the design for Cycle III, similar to those for the design in Cycle II, were that: 1. The target population be defined as the civilian, noninstitutionalized population of the United States, including Alaska and Hawaii, between the ages of 12 and 17 years, with the special exclusion of children residing on reservation lands of the American Indians. The latter exclusion was adopted as a result of operational problems encountered on these lands in Cycle I.

The time period of data collection he limited
to about 3 years, and the length of each individual examination within the specially constructed mobile examination center be between 2 and 3 hours. 3. Ancillary data be collected on specially designed household, medical history, and school questionnaires and from birth certificate copies. 4. Examination objectives he related primarily to factors of physical and intellectual growth and development.
5. The sample be sufficiently large to yield reliable findings within broad geographic regions and population density groups as well as age, sex, and limited socioeconomic groups for the total sample.
The sample was drawn jointly with the U.S Bureau of the Census, starting with the 1960 decennial census list of addresses and the nearly 1,900 primary sampling units ( PSU's) into which the entire United States was divided. Each PSU is either a standard metropolitan statistical area (SMSA), a county, or a group of two or three contiguous counties. These PSU's were grouped into 40 strata, with each stratum having an average size of about 4.5 million persons, in such a manner as to maximize the degree of homogeneity within strata with regard to the population size of the PSU's, degree of urbanization, geographic proximity, and degree of industrialization. The 40 strata were thenclassified into four broad geographic regions of 10 strata each and then, within each region, cross-classified by four population density classes and classes of rate of population change from 19C0 to 1960. Using a modified Goodman-Kish controlled-selection technique, one PSU was drawn from each of the 40 strata.
Further stages of sampling within PSU's required first the selection of census enumeration districts (ED's). ED's are small, well-defined areas of about 250 housing units into which the entire Nation was divided for the 196n population census. Each ED was assigned a "measure of size" equal to the rounded whole number resulting from a "division by nine" of the number of children aged 5-9 years in the ED at the time of the 1960 census. A sample of 20 ED's in the sample PSU was selected by systematic sampling, with each ED having a probability of selection proportional to the population of children 5-9 years at the time of the 1960 census. A further random selection by size of segments (smaller clusters of housing units) within each ED was then made, I3ecause of the 3-year time interval between Cycle II and Cycle III, the Cycle III frame had to he supplemented for new construction and to compensate for segments where housing was partially or totally demolished to make room for highway construction or urban redevelopment.
Advanced planning for the examinations at the various locations, or stands, provided for about 17 days of examinations, which limited the number of examinees per location to approximately 200.
In Cycle III, as 4n Cycle 11, twins who were deleted in the sample sel' ion were also scheduled for exam-41 illation, time permitting, as were youths deleted from the Cycle Ill sample who had been examined in Cycle II. the sample was :elected in ('vole Ill, as it had been for the children in Cycle 11, so as to contain the correct proportion Of youths from families haying only one eligible youth, ,two eligible youths, and so on to be representative of the total target population. llowever, since households were one of the rime; its in the sample frame, the of _elated youths in the resultant sample was greater than would result Iron a design which sampled youths if 12 -17 yea rs m it how regard to household. the resultant estimated mean measurements or rates should be unbiased, but their sampling variability-will be som..what greater than, those from a more costly., tineconsuming, systematic sample design in which every kth youth Uould be selected. the total probability sample for Cycle III included 7,314 youths representative of the approximately 22.7 million noninstitutionalized 1'.S. youths aged 12-17 years. Me sample contained youths from 25 different States and had approximately 1,1100 youths in each single year of age.
l'he response rate in Cycle Ill was 90 percent, with 6,768 youths examined out of the total sample. These examinees were assigned weights to make the group representative of the entire population studied with respect to age, sex, race, region, population density, and population growth in area of residence.
Measures used to control in general the quality of the data from these surveys have been described in previous reportsr additional measures specifically related to the particular examinations, tests, or measurements are outlined in the analytic reports describing and presenting the respective initial findings.
Reliability bile measurement processes in the surveys were carefully standardized and closely controlled, the correspondence between true population figures and survey results cannot be expected to he exact. Survey data are imperfect for three major reasons: (1) results are subject to sampling et ror, (2) the actual conduct of a survey never agrees perfectly with the design, and (3) the measurement processes themselves are inexact even though standardized and controlled. l'he first report on Cycle Ill' describes in detail the faithfulness with which the sampling design was carried out.
Data recorded for each sample youth are inflated in the estimation process to characterize the larger universe of which the sample youths are representative, ale weights used in this inflation process area product of the reciprocal of the probability of select ing the youth, an adjustment for nonresponse cases, and a post-stratified ratio adjustment which increases precision by bringing survey results intocloser alignment with known 1'.S. population figures by color and sex within single years of age 12 through 17 for the youths' survey. 42 to the third cycle of the Health Examination Survey' (as for the children in Cycle 11) the samples were the result of three principal stages of selectionthe single PSI' from each stratum, the 20 segments from each sample and the sample youth from the eligible persons. the probability of selecting an individual youth is the product of the probability of selection at each stage.
Sinee the strata are roughly equal in population size and a nearly equal number of sample youths were examined in each of the sample 1'S1''s, the sample design is essentially self-weighting with respect to the target population, that is, each youth 12 through 17 years had about the same probability of being drawn into the respective stages of selection.
The adjustment upward for nonresponse is intended to minimize the impact of nonresponse on final estimates by imputing to nonrespondents t he characterist ics of "similar" respondents. Here "similar" respondents were judged to be examined youths in a sample rsu haying; the samo age (in years) and sex as youths not examined in that sample PS12.
l'he post-stratified ratio adjustment used in the third cycle achieved most of the gains in precision which would have been attained if the sample had been drawn from a population stratified by age, color, and sex and makes the final sample estimates of population agree exactly with independent controls prepared by the U.S. Flureeu of the Census for the U.S. noninstitutionalized population as of March 9, 1968 (approximate midsurvey point for Cycle 111) by color and sex for each single year of age 12-17. The weight of every responding sample youth in each of the 24 age, color, and sex classes Is adjusted upward or downward so that the weighted total within the class equals the independent population control for the survey.
In addition to the sample youths who were not examined, for a small fraction of the group (U.t) percent), questionnaires containing the parent's ratings were not received. In terms of population estimates, for one-half of 1 percent of the youths there was no response to this questionnaire. The rate of item nonresponse was relatively low, around 1 percent, except for one question, which concerned the initial reaction of the youth on being enrolled in the first grade (number 40, appendix II), and for which the nonresponse rate was 7 percent (table 10).

Standard Error
In the present report reference has been made to efforts to minimize bias and variability of measurement techniques.
The probability design of the survey makes possible the estimation of standard errors. The standard error is primarily a measure of sampling variability, that is, the variations that might occur by chance because only a sample of the population is surveyed. As          40 SO calculated for this report, the standard error also reflects part of the variation which arises in the measurement process. It does not include estimates of any biases which might be in the data. The chances are about 68 out of 100 that an estimate from the sample would differ by less than the standard error from the value obtained from an examination of all persons in the population. rite chances are about 05 out of 100 that the difference would he less than twice the standard error and about 99 out of 100 that it would le less than 2'2 times as large.
Generally the rates or percentages shown In the detailed tables for the entire group, all the males, or all the females are accompanied by their respective standard errors. In the interest of simplicity and brevity, specific standard errors for estimates by single year of age are not presented; however, an approximate standard error for each can be estimated from the curves in figure I. The curve labeled 4.0 (population base in millions) provides estimates of standard errors for percentages or rates cited for all the youths (males and females) grouped by single year of age, e.g.., all 12-year-olds. Similarly, the 2.1) curve gives estimates foc either sex class grouped by single year of age, e.g., 16-year-old girls. Table I shows population base estimates for those.
percentages that pertain to less than all youths in an age or sex-age class, e.g., those 14-year-old boys who had attended nursery school.
Employing the information contained in table II, the following example shows how the graph (figure 1) and the guide (table I) may be used to obtain estimates of standard errors for percentages based on these subpopulations. The first estimated standard error shown in table II (1.8) was obtained by locating the appropriate percentage value (11.3) on the hori- zontal scale of figure I, reading vertically on the 1.0 and 2.0 (million) curves, and interpolating for 1.6 million, using the scale to the left. Values for the errors related to 5.4 percent and 73.6 percent may be read directly from the appropriate curves, using for 73.6 percent the value of its complement, 26.4. An approximation of the standard error of a difference -vof two statistics x and y given by the formula (s, where S, and ,s are the standard errors, respectively, of x and y Of course, where the two groups or measures are positively or negatively correlated, this give an overestimate or underestimate, respectively, of the actual standard error.
Certain tests of the statistical significance of the association between responses to related questions in this report made use of Pearson's classic chisquared test with modifications to adapt the origin..1 procedure for use with the complex sample design of the survey. These adaptations, which follow an approach suggested by McCarthy, " are explained by Baird in a previous series 11 report.

Small Values
In some tables magnitudes aic shown for cells for which the sample size is so small that the sampling error may be several times as great as the statistic itself. Obviously in such instances the statistic has no meaning in itself except to indicate that the true quantity is small. Such numbers, if shown, have been included in the belief that they may help to convey an impression of the overall story of the table.