7—
Settings, Treatments, and Social and Personality Characteristics by Sibsize

Sociologists have become increasingly concerned with identifying the familial behavior that is indicated by, or results from, family background measures such as parental education, occupation, income, and marital cohesion. Spaeth (1976) has suggested that differential parental socioeconomic status is an indicator of variability in cognitively stimulating home environments. He has also suggested the usefulness of making a heuristic distinction between "settings" and "treatments" as aspects of the home environment—settings being the various cognitively stimulating features of the environment that are available to be engaged but do not actively engage the child, such as books, musical instruments, and so forth; and treatments being purposive parental interventions in the child's intellectual development, such as active teaching, correcting, encouraging, goal setting, and the like. The importance of environmentally stimulating settings to the development of intellectual ability has been a focus of much research by psychologists (Milner 1951; Dave 1963; Bloom 1964; Wolf 1964; Hess and Shipman 1965; Lewis and Goldberg 1969; Hess 1970a and b ; Marjoribanks 1972a and b ; Williams 1976). Sociologists have demonstrated the significance of differential treatments, in the instance of differential parental encouragement to educational achievement among high school students (Sewell and Hauser 1980, for example). We have just discussed parental encouragement in Chapter 6.

If the home environment contributes (or fails to contribute) cognitive stimulation at an early age and continuing through adolescence, the school increasingly takes over this function as the child advances in age. It is hardly surprising, therefore, that consid-

erable interest has been generated in the character of the school environment, and particularly in the types of extracurricular activities engaged in by young people in high school.

Finally, it is fruitful to ask about aspects of young people's characteristics that seem relevant to educational attainment, and the ability to engage progressively more stimulating situations and challenges.

In sum, this chapter will present information from surveys concerning (1) the cognitive stimulation of two types of settings (home and school); (2) what we call parental "treatments"—specific parental interactions relevant to cognitive stimulation; and, finally, (3) personal characteristics of young people that would seem relevant to successful educational advancement. Our focus will be on sibsize differences in these elements. Very little data gathering has been directed to the topics of cognitively relevant settings and treatments by sibsize, although there are, by contrast, numerous questions on psychological characteristics and moods. The purpose of this chapter is not, therefore, to present a definitive picture. Rather, it should be viewed as a stimulus to research of all kinds to fill in the enormous gaps in our knowledge.

To make a hard and fast distinction between stimulation that is simply "passive" and stimulation that is actively directed toward the child by a parent would not be productive. Parents may make a great deal of effort to provide settings that are stimulating. There is nonetheless a worthwhile distinction to be made between stimulation that is a constant, undirected feature of simply living in a particular setting, and stimulation that involves targeted interaction by the parent with regard to the child's behavior—what we will call "treatments."

Settings at Home

After controlling for parents' socioeconomic characteristics, do the types of cognitively stimulating "settings" in the home differ by sibsize for the young people in our samples? We will use the 6- to 11-year-olds in Cycle II of the Health Examination Survey, the 12-to 17-year-olds in the same survey, the high school sophomore boys in Youth in Transition, and the sophomore and senior boys and girls in High School and Beyond. The data differ among the studies,

and some of the information might be considered as equally relating to "treatments" and "settings."

Although much heat has been expended on the unsociability and loneliness of the only child (see, Bossard 1954; Bossard and Boll 1956; Terhune 1974; Thompson 1974; Claudy, et al. 1979; Blake 1981b ), little systematic research has been done on the differential environments of only children, and children from other small families, compared to children from large families. One could entertain the hypothesis, therefore, that some of the widespread perception of only children as "less sociable" may actually be a function of their class position and intellectual levels.^[1] Such children tend to come from milieus that value privacy, time alone, and intellectually demanding uses of time—reading, listening to music, playing an instrument, and so on. And, on average, such children have the intellectual ability to respond to these environments, as well as the desire.

Claudy and colleagues (1979) have addressed the issue of differential uses of discretionary time in their analysis of only children and children from two-child families in the Project Talent Study. They have controlled for indicators of the parents' socioeconomic status, but not for IQ in these particular tabulations. They found that only children participated to a greater extent than children with siblings in extracurricular activities generally—in reading for pleasure, in collecting, hobbies, raising animals and pets, in acting, singing, and dancing, and in music, photography, and clubs. Children with siblings participated more in sports and other team activities, high school leadership roles, hunting/fishing, woodworking, crafts, and cooking. In effect, only children were engaged in activities that were often solitary, intellectual, and artistic; whereas "nononlies" were engaged in group-oriented and practical activities.

Another study of the effect of the family settings of youngsters, by number of siblings, is represented in the work of Heise and Roberts (1970). Studying children in elementary school, the investigators were concerned with children's extent of knowledge of common family and community roles and with the correlates of role knowledge. Of relevance here is the counterintuitive finding that an increase in number of siblings was negatively correlated with role knowledge, and that this negative correlation increased with the age of the child. Interestingly, the negative effect of siblings was

greatest for what the investigators called "peer focus" roles (filial roles, sibling roles, and a role-complex called "High School Boy"). The investigators interpret this finding as being a consequence of the parochialism of children in large families. Such children are more inclined to confine their playing and their interaction to within the family because the family is a more self-sufficient unit. It turns out, therefore, that children with numerous siblings do learn peer-focus roles, but ones that have little social currency or validation. In effect, these roles have been mis learned by paying too much attention to the behavior of family members exclusively. The authors caution that their findings, based on intensive study of 93 children, are suggestive only. However, their results are worth citing to represent a level of theoretical sophistication that is relatively rare in available research on the effects of family size.

A principal source of our data on family settings is Cycle II of the Health Examination Survey. The Cycle II data are particularly important because early childhood has been targeted by psychologists as a critical time for cognitive and linguistic development (Passow, Goldberg, and Tannenbaum 1967; Jensen 1968; Whiteman and Deutsch 1968; Gewirtz 1969; Bernstein 1970; Bijou 1971). Moreover, it would seem that as children get older the influence of the family is diluted by that of schools and peer groups (Campbell 1969).

Cycle II obtained information from parents on a variety of the child's (age 6 to 11) uses of time—Scouts, sports, cultural activities, watching television, listening to radio, reading newspapers, reading books, and playing by him or herself. The reader is cautioned that these parental replies are not systematic time-use data. Their value inheres in the fact that they add to a body of prior information (for example, the Project Talent data analyzed by Claudy), and they provide a stimulus to future, more rigorous, data gathering.

Taking all formalized extracurricular activities (Scouts, sports, church groups, music and other cultural pursuits) as a group, we have coded the children in terms of whether they engaged in any or none. Table 7.1 indicates that children from small families, including only children, are more likely to be involved in one or more such activities even after controls for the child's age and sex, parents' education, family income, family intactness, region of the country, community size, and the mother's usual activities. Among

those engaging in such activities, table 7.1 shows the results for music and other cultural pursuits. Respondents were coded "1" if they participated in no activity in each category, and up to "4" if they participated in three or more activities in each category. We see that children from small families are more participatory. In the unadjusted data, children in small families also participated more in sports and Little League. However, the difference disappeared where data were adjusted.

For watching TV, listening to radio, and reading books, responses ranged from zero to six or more hours per day, and reading newspapers ranged from zero to four or more hours per day (see table 7.2). There are no sibsize differences in TV watching, but time spent listening to radio, reading newspapers, and reading books shows declines as sibsize increases. Only children, in particular, spend the most time on these pursuits. The same table shows that "playing by self" (zero to six or more hours a day) is related to sibsize, and is particularly marked for only children.

These results, while bearing out the Claudy findings that were confined to differences between only children and those from two-child families, suggest that preferences for relatively solitary and/or intellectual, as well as cultural and artistic, uses of time characterize the entire small end of the sibsize continuum versus those from large families.

Is this intellectual/solitary tendency of children from smaller families a function of being less popular and well liked by other

children? The data do not justify such a conclusion. Table 7.3 shows the teacher's evaluation of the frequency with which the child is chosen when other children are choosing sides. These results suggest that only children and those from small families are, if anything, more popular than children from larger families. Parent responses to questions concerning shyness and getting along with other children show no difference by sibsize.

If enforced isolation due to unpopularity does not account for differences in children's activities by sibsize, does the association of higher ability with sibsize (as well as with reading, playing by oneself, and other intellectual and cultural activities) account for the sibsize association with these activities. Controls for intellectual ability not shown here reduce somewhat, but do not eliminate, the sibsize difference in reading shown in table 7.2, suggesting that the intellectual-ability differences by sibsize conjoin with discretionary use of time to set young children from small families on a path toward academic success. Moreover, although playing alone is asso-

Figure 7.1.
Items Included in Index of Richness of Past Experience (Youth in Transition).

ciated positively with intellectual ability, using the WISC as a control does not reduce the overall relationship of playing alone with sibsize—a relationship that is obviously enforced by the absence of siblings among only children and the ubiquitousness of siblings in large families. Since studying and other intellectual activities typically require concentrated periods of solitude, the development of a tolerance for being alone at an early age may also be helpful to the academic development of children from small families.

Finally, the picture that emerges from these data may account for some of the popular prejudice against only children, especially, as overprivileged, asocial, and royally autonomous. These children clearly are something of an intellectual and social elite, as well as being relatively less numerous than the mass of children from larger families. Cumulated over a number of activities and uses of time, young children from small families, and only children particularly, appear to lead rather different personal lives from those residing with numerous siblings. Children born into large families who complain of lack of privacy and the constant need to adjust to other people may well be told by their parents that conditions in small families "spoil" children, making them self-centered, aloof, and overly intellectual. Since so many more people come from large

families than from small ones, it is not too surprising that prejudices against small families have been widely transmitted.

Turning to the Youth in Transition study, tenth-grade boys were asked to respond to a 46-item Index of Richness of Past Experiences, answering whether they had or had not done each of the itemized activities at least once in their lives (see fig. 7.1). Although some of the items would seem to favor children from large families (such as having a part-time or full-time job), an analysis of the overall index of richness of past experience (see table 7.4) indicates that those from small families have had richer and more varied lives. Other information from this survey, although indirect, suggests that only children and those from small families experienced socially stimulating and "upbeat" home environments. Nothing in our data suggests depression, isolation, sadness, hopelessness, and the like. For example, table 7.4 gives results for a social skills index

(boys were asked to place a value on various skills that help people get along with others); an anomie index (8 items measuring respondent's feeling that the world is uncaring and "going to the dogs"); and a resentment index (7 items measuring feelings of being cheated, getting a "raw deal," and that others "get the breaks"). All show relatively small differences by sibsize, but differences favoring those from small families.

Both the sophomores and the seniors in High School and Beyond were queried concerning various aspects of the home environment. Among these aspects was whether and how frequently they were read to before they started attending school. The code went as follows:

Table 7.5 demonstrates for sophomores and seniors of both sexes that reports of being read to are associated negatively with number of siblings, even after controls for parents' educations, father's occupation, and whether the family was intact. Only children and children from small families were allegedly more likely to have been read to before they started school than were children from larger families, and the relationship, although not large, is quite linear.

We must remember that by the time children become sophomores and seniors in high school, there has been selective dropping out that may well affect the strength of this relationship. In any event, this type of cognitive enrichment is negatively associated with number of siblings, in spite of the fact that a potential contribution of large families might be the presence of older siblings who could read to younger ones. These results do not, therefore, offer support to Zajonc's notion that older siblings in large families provide substitutes for parents.

Sophomores were questioned concerning whether they had had a variety of advantages—music lessons, dance lessons, and travel outside the United States. Children from small families were more likely to have had one or more of these advantages, whereas those from large families were more likely to have had none. Table 7.6 shows the adjusted mean average score on the three activities by sibsize for boys and girls. The difference between boys and girls results from a lower frequency among all boys with regard to taking dance lessons. There is a substantial and linear sibsize difference even after controlling for family background.

Settings in School:
Extracurricular Activities

Discretionary behavior in school can be regarded as an extension of the settings/treatments dichotomy to another phase of the life cycle. In this case, as Spaeth (1976) has suggested, youngsters are participating actively in choosing settings in which to place themselves and in the self-administration of "treatments." At least two studies have found a relationship between educational attainment (years of school completed) and participation in extracurricular activities in high school independent of parental SES, academic ability, and academic performance (Spady 1970; Otto 1975). Extracurricular activity has also been suggested as a mechanism by which parental SES is transmitted to offspring (Otto 1975). Otto has postulated that extracurricular activity provides socialization for attitudes and skills that "pay off" later in life.

A rather different perspective has been offered by Coleman in The Adolescent Society (1961). Coleman concentrates on the types of activities in which students are engaged (rather than on the level

of participation), and on the degree to which activities are congruent with high levels of academic achievement and scholarship. Our own analysis will lean more toward the Coleman type of research question. Are there sibsize differences in the types of activities in which young people are engaged after controlling for major family background characteristics? Are young people from small families, including only children, more likely to engage in extracurricular activities that are intellectually and culturally stimulating, in contrast to community-oriented activities, vocationally oriented activities, or sports? In considering the results we will present, the reader should bear in mind that extracurricular activities—especially school activities—are not always perfect indicators of students' principal interests and diversions. Some activities require particular talents and cannot be enjoyed passively, and some may be selected as a change of pace from the student's usual activities, or as an experiment. However, as will be seen, our analysis attempts to deal with this problem.

We will concentrate our analysis on seniors in the High School and Beyond data. Students were asked the questions in figure 7.2. Although it is possible to distinguish between students who had no extracurricular activities during the past school year and those who had one or more such activity, it is not possible to count activities because, in the initial questions, items were grouped (for example, types of sports were grouped, as were activities like debating or drama, chorus or dance, and so on). We thus turned our attention to the types of activities in which students were engaged according to sibsize, rather than to the number of activities.

As a beginning, students could be grouped into those who had no activity, only sports, only nonsports, and both sports and nonsports. For boys and girls, the percentage distributions in these categories are shown in table 7.7. Clearly, few seniors have no activity. Most boys are engaged in sports, or sports and another activity, and girls are engaged modally in a nonsport activity.

The next question was whether the activities of students checking more than one nonsport category formed a pattern. For example, we believed that the activities of dance, band and orchestra, and drama would go together; and that student government, service on school newspapers and magazines, and membership in honor clubs would be cognate. We also saw a congruence between church activi-

Figure 7.2.
Questions Asked of Seniors Concerning Extracurricular Activities
(High School and Beyond).

ties and youth organizations such as Scouts and the "Y." A factor analysis showed that we could combine students engaged in sports and nonsports with nonsports only.^[2]

The analysis showed that we could divide the various nonsport activities into five factors: (1) "Intellectual," (2) "Artistic and Cultural," (3) "Community," (4) "Hobbies," and (5) "Vocational." These factors were then used to assign students to one of the five groups. Students with one nonsport activity were assigned to the relevant group, students with more than one such activity were assigned to the group in which they had the most activities. And if number of activities in all groups was equal, then the student was assigned to the group with which any given activity had the highest correlation. The final distribution is shown in table 7.8.

Among those who have activities (table 7.8), the single most important type of nonsport pursuit for both boys and girls is some form of community involvement—church, youth groups, and so

forth. After community activities, boys and girls divide quite markedly. Boys are more likely to be engaged either in intellectual activities (school newspapers, student government, yearbook, etc.) or in hobby clubs. Girls are more likely to be involved in artistic and cultural activities, or in vocational pursuits. There is a particularly sharp gender discrepancy with regard to hobby clubs and artistic and cultural pursuits—the one almost exclusively a male activity, the other primarily a female activity. It is also true that very few girls are engaged solely in sports, whereas this is true for 11 percent of the boys.

If we assume that the pursuit of "intellectual" activities (and intellectual activities plus sports) provides students with training and experience that will help to further their academic careers, are there sibsize differences between those who engage in such activities versus those who engage in sports only, community only, or community and sports; vocational only, or vocational and sports; or no activity at all? Turning first to boys, the top tier of table 7.9 shows the results of four multiple classification analyses (MCAs). The dependent variable in each case is a dichotomy—" 1 " for intellectual activities, and "0" for only sports, or for community, vocational, or no activity, respectively. The proportions shown have been adjusted for the parents' education, the father's occupation, and whether the family is intact. We see that, in general, boys from small- and medium-size families are more likely to engage in intellectual activities (versus any of the other activities) than are those from large families (sibsizes five and over). Relative to community and vocational activities, only children are particularly likely to engage in intellectual pursuits. The findings are similar if we combine artistic/cultural activities with intellectual activities and redo the MCAs as in the second tier of table 7.9. Boys from small- and medium-size families are also more likely to engage in intellectual (and artistic) activities than no activities at all. However, in this case, only children are slightly less participatory than children from two-child families.

As for girls, since artistic and cultural activities are modal, we have in table 7.10 performed the same analysis as for boys (table 7.9) using Artistic (which includes cultural) in the top tier instead of Intellectual, and combining both Intellectual and Artistic in the second tier. As between artistic and community activities, only

children are clearly more likely to engage in artistic pursuits, whereas there is no difference among the other family sizes. The same pattern appears if we include intellectual activities in the dependent variable. Interestingly, however, only girls are slightly more likely to engage in no activities (than artistic or artistic/intellectual activities) compared with those from families of two to four children. In this respect, only children appear somewhat more like children from large families than those from two- to four-child families.

Children from large families are the most likely to have no activity (when this category is compared with artistic and intellectual).

In sum, only children are particularly unlikely to engage in "community" type activities relative to young people from other sibsizes, and those from small families generally are more likely to engage in intellectual and artistic pursuits than are those from large families. Children from large families are more likely to have no

activity at all than an intellectual or artistic one, and only children are very slightly more likely to have no activity (compared to an intellectual and artistic one) than those from two- to four-child families. Except for this last-mentioned anomaly among only children, it thus appears that coming from a small family versus a large one is associated with extracurricular activities that are congruent with and facilitative of academic and developmental skills. By contrast, those from large families are more likely to engage in sports only, in church activities, youth groups and other community efforts, in vocational activities (boys only), and in no activity at all.

Turning to whether students participate (or do not participate) at all in extracurricular activities, are there major sibsize differences in this behavior? Theoretically, we expect that school grades and type of school program will be major determinants of participation since both schools and parents tend to limit involvement when students have low grades, and schools that are heavily oriented toward vocational and general programs may have limited opportunities for extracurricular activities. It is also true that youngsters from farm families (although often in vocational and general tracks) tend to have uniquely high levels of extracurricular participation because of their involvement in church, community, and farm-related vocational and hobby activities. Those from farm backgrounds also come from larger families.

Accordingly, we limited our multiple classification analysis to senior boys and girls from nonfarm families, and included as predictors high school program, grades, parents' education, father's occupation, whether the family is intact, and sibsize. In addition, we introduced a control for ability as measured by the age-standardized vocabulary test. No major sibsize differences in overall participation for either boys or girls were evident (results not shown here). The principal determinants of participation/nonparticipation are grades and high school program. The major program difference is between academic (higher participation) and other programs (all other variables in the model equal), and there is marked variability in participation by grade performance. The relative importance of program (other variables controlled) is slightly greater than the relative importance of grades. Given controls for grades and program, parents' education is not salient, nor is family intactness.

Parental "Treatments"

What do we already know about sibsize and parental behavior toward children as these relate to intellectual development? Actually, very little work has been done that considers sibsize as an influence on parental behavior, and the latter in turn as an influence on intellectual achievement. It is nonetheless worthwhile to trace out some of the theoretical and empirical background for our own effort. This background stems primarily from educational psychology, social psychology, and sociology.

Beginning with the work of Bloom at Chicago (1964) and Plowden in Britain (1967), there has developed a major research literature on the relation of intrafamilial environments to children's cognitive development. This research has concerned active parental interventions (treatments) in the process of the child's intellectual development such as expressed and reinforced aspirations and expectations, rewards for intellectual development, pressures for language development (correcting diction and grammar, presenting opportunities to enlarge vocabulary), as well as providing opportunities for learning (settings). An informative review of this literature and the various theoretical frameworks involved is provided in the first two chapters of Marjoribanks's Families and Their Learning Environments (1979). In general, this literature suggests that specific measures of environmental influences—"treatments" as well as "settings"—are important in explaining intellectual development, and are more powerful than the simple use of parental social status as an indicator of family environment. In particular, these intrafamilial environmental measures have been strongly associated with verbal performance scores, moderately associated with mathematics scores, and minimally associated with scores on nonverbal intelligence.

Rather than asking how background variables (parents' education, occupation, income, etc.) and sibsize affect parental behavior, most research in the intrafamilial-environment tradition has concentrated on examining the importance of environmental measures (settings and treatments) versus the importance of family background and number of siblings. However, an exception may be found in an article by Marjoribanks and Walberg (1975) which, unlike much of their other work, deals specifically with whether

sibsize affects the social-psychological environment in the family. Using some of the usual "environmental" measures (in this case, treatments) like press for achievement, press for activeness, press for intellectuality, press for independence, and so on, Marjoribanks and Walberg found correlations ranging from –.30 to –.45 between these variables and sibsize, indicating that as sibsize goes up parental interaction and attention declines.

Sociologists have had a major interest in differential socialization practices and values of parents (what the intrafamilial environment school would call "environmental influences"), but the sociological focus has been very largely on the social-structural antecedents of parents' childrearing behavior. One school of thought has actually derived the characteristics of presumed parental treatments of children from differential family size. Another school has derived a very similar kind of paradigm from the influence on parents of the demands of their occupations. It is worthwhile to consider each briefly.

Bossard (1953) and Bossard and Boll (1955) explicitly emphasized the role of family size in how parents actively relate to their children. Parents of small families are characterized as driven by ambition for their children, exhibiting great overt concern for performance in school, and as expressing strong affect in response to the child's successes and failures in relation to standards of excellence. The large family, by contrast, is described as emphasizing cooperation, obedience to parental authority, conformity, and self-reliance in situations of self-caretaking (but not in situations relating to standards of excellence). Competition and rivalry in such families is said to be discouraged by parents as divisive and disruptive to the family group. More recently, a number of researchers have suggested that parents in small families are less restrictive of children's autonomy and more encouraging of inquiry and self-direction (Rosen 1961; Elder and Bowerman 1963; Clausen 1965; Nye, Carlson, and Garrett 1970; Clausen and Clausen 1973).

Clearly, this literature relating family size and parental concerns for achievement implicitly assumes that achievement-oriented parents are motivated to have fewer children in whom they will invest enormous effort and that large families structurally require that children be less individuated and more obedient and conforming. In effect, no matter how achievement-oriented parents in large families

might be, life in such families tends to involve more conformity and less self-direction than life in small families, and less of a tendency to pressure any given child for achievement. Hence, whatever antecedent factors affect both family size and achievement orientation of parents, this formulation sees life in small and large families as having an independent, structural effect on parental behavior (if not parental values).^[3]

A very different chain of reasoning concerning the relation of self-direction versus conformity as parental values for children is represented by the work of Melvin Kohn. Kohn's research, resting on a long tradition in sociology of deriving childrearing values from the values that are salient in the parents' socioeconomic status (Duvall 1946; Kohn 1959, 1963, 1969, 1971, 1976, 1977; Miller and Swanson 1960), is based on a scale measuring self-direction versus conformity in parental values. Kohn's main thesis is that individuals' orientations to life—and, as a consequence, the values they emphasize in childrearing—are a result of socialization in the occupations they perform. Middle-class job conditions involve self-direction and autonomy, whereas working-class job conditions socialize people to value conformity over autonomy. Although other investigators working in this area have increasingly emphasized that factors such as education and religion may actually be more important than jobs (Lenski 1961; Wright and Wright 1976; Alwin and Thornton 1984), Kohn believes that the association of these variables with self-direction and conformity is still simply due to the principal "cause"—the association with occupation, and consequently with job conditions. All investigators following the Kohn tradition, insofar as they are interested in family size, simply control for all background variables related to parental values, and they report that family size has no effect (see Kohn 1969; Alwin and Thornton 1984). Actually, Alwin and Thornton (1984) have found a very small effect of all background variables, education, occupation, religion, and so forth, on the "thinks/obeys" scale (a modification by Lenski of the Kohn scale, autonomy versus conformity).

We should note parenthetically, as well, that the self-direction–conformity items in the Kohn scale may be suffering some contamination as a consequence of the popularization of the developmental childrearing literature since the items were first constructed. We shall come back to this point shortly. It is also true that the self-

direction versus conformity scale is related to but not identical with achievement motivation.

In connection with sibsize influence on parental attention and interaction with children, an important effort at estimating the sibsize effect on time spent with children has been made by Peter Lindert (1978). Lindert used the basic data from the Cornell–Syracuse time-budget study of 1296 Syracuse families in 1967–1968. Lindert's analysis of these data is ingenious and described in detail in his book. Here we will simply note that a major finding is that the total amount of time parents spent on children was primarily a function of the ages and numbers of children, rather than of parental characteristics such as education, woman's labor force participation, mother's age, or father's socioeconomic status. These background variables turned out to be statistically insignificant when it came to time inputs. Hence, although parents of higher socioeconomic status and education spent more time per child, this was because such parents had fewer children.

Finally, in this section on parental "treatments" (parental interventions in the child's cognitive development) by sibsize we should mention that the intensive research by many psychologists on this topic (and on birth order) has been primarily concerned with the affective rather than the cognitive consequences of differential family size. Numerous psychologists have been interested in the changing emotional quality of interactions with children, shifting constellations of power, and also affection among families of different size. Studies of the process of interaction suggest that as additional children are born constellations do change. For some of the more recent studies of such processes see M. E. Lamb and B. Sutton-Smith (1982). However, these studies also suggest that it is difficult to characterize the changes in ways that psychiatrists might once have suggested—dethronement of the oldest child, hostility toward the new born, and so on. Rather, siblings have wide-ranging types of interactions and relationships—both outgoing and social as well as ambivalent and hostile. In effect, systematic studies of process do not lead us to think that the period of family-building is one so strongly characterized by particular types of sibling relationships that they inevitably lead to permanent personality changes or imprints on individuals depending on number of siblings, or ordinal position. We will see in our last section that there seem to be few, if any, major personality differences by sibsize.

Analysis of Family Size and Parents' Valuation of Intellectual Curiosity Versus Obedience

As has been suggested, one possible reason for the higher educational attainment of those in small families is that parents in small families may place a greater value on intellectual curiosity versus obedience than parents in large families. We should note here that given the data available, the discussion will be confined to expressed values, rather than behavior. Hence, the analysis will approach only marginally the issue of whether family size constrains and shapes behavior in spite of values.

Here we do not wish to engage the overall issue of self-direction versus conformity, but rather we will concentrate on those items in the Kohn scale relating to intellectual curiosity and obedience. These data are available from the General Social Surveys for selected years (1973, 1975, 1976, 1978, 1980, and 1983). Out of a range of 13 items characterizing a child of undesignated age and sex (items including "keeps neat and clean," "is considerate of others," "is a good student," "is honest," "has good manners") respondents were asked to pick the one most desirable, the three most desirable, the three least important, and the one least important. The results for each item were then coded as follows:

1 Mentioned as the one least important

2 Mentioned as one of three least important

3 Not mentioned

4 Mentioned as one of three most desirable

5 Mentioned as the one most desirable

The items, "Is interested in how and why things happen" and "Obeys parents," have been tabulated by family size of the parents in the GSS. This family-size variable was calculated from a combination of two items in the surveys—actual and expected number of children. (Respondents who had no children and expected none were excluded. For each respondent who had children and expected more, the expected number of additional children was added to the actual number already born. If a respondent expected no additional children over those already born, then the actual existing number was used. Finally, for childless respondents who expected children, the expected number was used.) Among 76 percent of respondents

no additional children were expected, and among 24 percent the total family size has been estimated by combining actual and expected fertility.

With regard to intellectual curiosity, the bivariate relation with parents' family size appears very weak (table 7.11). However, since Lenski (1961) and Alwin and Thornton (1984) have emphasized

an interaction with religion on the "thinks/obeys" polarity, it is worth examining the relationship between family size and intellectual curiosity for separate religious groupings. Accordingly, we have analyzed the data (not shown) by the following groupings: Catholics; Episcopalians, Jews, and None; Lutherans and Presbyterians; and Baptists and Methodists. This division serves not only to separate Catholics from non-Catholics, but to divide non-Catholics into those having similar levels of secularism and fundamentalism. We will summarize the analysis without showing the detailed findings.

Family size bears no bivariate relation to parents' valuation of intellectual curiosity among Catholics. However, within each non-Catholic group, those having smaller families are more likely to choose intellectual curiosity as desirable and less likely to choose it as unimportant. In addition, members of the grouping "Episcopalians, Jews, and None" are markedly less likely than any other group to regard this quality as unimportant, and markedly more likely to regard it as important, whereas Baptists and Methodists are at the other extreme, having the lowest proportion regarding intellectual curiosity as important and the highest proportion singling it out as unimportant.

As for obedience to parents we can see (table 7.11) that there is no overall relation with family size. There is also none among Catholics, but the item does show a relationship within the other religious groups. Those having large families are consistently more likely to pick this trait as desirable. There is also a noteworthy difference between Episcopalians, Jews, and None versus Baptists and Methodists in their evaluation of obedience as a desirable quality. Seventeen percent of the former but 35 percent of the latter single out this quality as desirable, while 10 percent of the former but only 5 percent of the latter see it as not important.

Parenthetically, we may now note that for one of the items from the Kohn scale, "Is a good student," tabulation (not shown) indicates no overall bivariate relation with family size and none among Catholics. There is however some association with family size among other religious groups, but it is in the opposite direction to our expectations—those having small families are the most likely to regard this trait as un important. This trait is selected as desirable by a small fraction of all religious groups, and specifically men-

tioned as not important by between 37 and 45 percent (depending on the group). The group having the largest percentage holding this quality not important is Episcopalians, Jews, and None. A possible interpretation of this anomalous finding is that, for developmentally oriented parents, emphasis on being a good student may be regarded as excessively directive and demanding—in effect, the opposite of what was intended by the item. It is also possible that, with few items in the scale innocuous enough to choose as unimportant, being a good student may seem like an acceptable or inoffensive candidate. This item clearly is not (or is no longer) measuring along the same dimension as the one on intellectual curiosity.

A final item in the Kohn scale worth mentioning is "Tries hard to succeed." This item is not associated with family size even at a bivariate level overall, or within any religious group.

Would the relationship between family size and intellectual curiosity versus obedience to parents be maintained if we controlled for possible antecedent influences on both family size and childrearing values—influences such as age, education, occupation, farm—nonfarm background? From table 7.12 we can see that the answer is negative. In this table, we show the R² (adjusted for degrees of freedom) for a multiple classification analysis within each religious group using intellectual curiosity and obedience to parents as dependent variables, together with a listing of the two most important predictor variables in order of relative importance. Quite clearly, the model explains only a modest amount of variation in the dependent variables and, in addition, the most important sources of explanation are education and age. Only among Episcopalians, Jews, and None does family size maintain a relation with obedience to parents after controls, but the relationship is modest. These results thus suggest that the apparent association of family size and different childrearing values emphasized by Bossard and Bell is really a result of the joint influence of education and generation on both family size and values.

It is also worth noting (see table 7.13) that our suspicion that a high score on the Kohn item "Is a good student" will be perceived as directive and nondevelopmental, rather than a companion to "intellectual curiosity," seems to be born out by a multiple classification analysis. As may be seen from the table, it is the less educated

and the older respondents who place a high value on being a good student, and the younger and more educated respondents value this quality the least.

Finally, this discussion of differential parental values has some methodological implications for our prior analysis of adult educational achievement in relation to sibsize. The discussion suggests that our controls for the socioeconomic status (father's education and SEI) of men's and women's parents, and the respondent's own

age (as an indicator of the period during which he or she was brought up), will serve to control to some extent for unmeasured

differences among respondents' parents in the kinds of childrearing values they espoused. As we have just seen, parents' childrearing values do not appear to be associated with the number of children they bear, once antecedent variables like education and age are controlled. This finding strengthens our interpretation of the independent effect of sibsize on education. We appear to have been measuring relatively pure sibsize effects, and not the influence of major unmeasured parental differences associated with the decision to have large versus small families.

Family Size and Parents' "Interventions"

Have we information on the amount of direct parental intervention in the lives of the children in the surveys under analysis? This type of information is not ideally collected from surveys and, hence, we do not have a wealth of data to present. We can, however, draw some insight from relatively indirect information concerning parental attention to youngsters' health and dental maintenance through visits to doctors and dentists.

Cycle II contains two questions relating to the last time the child had been to the dentist and the doctor. The results from each were coded as follows:

1 During past twelve months

2 One to two years ago

3 More than two years

4 Never

Table 7.14 shows the adjusted means by sibsize. The reader should note that, in addition to controls for background characteristics, visits to the doctor were controlled for an overall health status variable. In the case of doctor visits, sibsize was relatively the most important variable and, in the case of visits to the dentist, it ranked second to mother's education. There is a standard deviation difference between only children and those from families of seven or more in the unadjusted means; the difference remains substantial in the adjusted values for doctor visits, and approximately half a standard deviation for visits to the dentist. If we take these two

parental behaviors as indicative, parents of fewer children appear to intervene more actively with regard to their children's physical welfare (at the cost of both time and money) than do parents of large numbers of children. These data tell us nothing, of course, about whether parents of small versus large families also intervene more in matters relating to cognition, but the data do support some speculation that this might be the case.

Social and Personality Characteristics by Sibsize

As mentioned in the introduction to this chapter, our concern in this section will focus on those aspects of our subjects' personalities and temperaments that could prove important to educational attainment and about which systematic data are available. Specifically, we will concentrate on aspects of achievement motivation, on

over- and underperformance in school, and on self-confidence in ability. Although the relation of such variables to achievement outcomes may be in doubt (see Featherman 1972; O. D. Duncan, Featherman, and B. Duncan 1972; G. J. Duncan 1983; and Hill et al. 1983), many researchers are not convinced of the irrelevance of such variables to actual achievement.

Our focus on variables presumably related to educational attainment may leave some readers with a sense of dissatisfaction, since so much of what has been written in the past about number of siblings and personality has concentrated on how people from different sibsizes relate to and interact with others. In the section on "settings" we addressed the issue of whether only children (and those from small families generally) are more asocial, unpopular, lonely, and depressed. Our evidence has suggested just the opposite. Recent systematic research concerning the allegedly "spoiled," demanding, selfish, unsociable characteristics of only children, or of those from small versus large families, seems to provide virtually unequivocal answers. Either there are no differences by number of siblings, or the differences favor the only child and the small family (Claudy 1976; Falbo 1977; Blake 1981b ; Grotevant and Cooper 1982; Lamb and Sutton-Smith 1982; Ernst and Angst 1983).

The emphasis in this chapter on antecedents and correlates of cognitive development may be justified not only on the basis of the importance of this aspect of personality for educational advancement, but because there is widespread agreement in psychology that cognitive development is, in turn, an important predictor of future social development (see, for example, Kohlberg 1981a and b ; Loevinger and Knoll 1983; Mussen 1983; Parke and Asher 1983; Cairns and Valsiner 1984; Fischer and Silvern 1985).

Why are the results concerning the emotional and interactional aspects of personality by sibsize so unproductive of the effects that are most likely to be the focus of popular expectations? One reason, as suggested earlier, is that systematic studies of actual processes of interaction in families do not document the existence of the types of structured behavior (by sibsize) that "armchair" deduction from psychoanalytic theory presupposed. Moreover, in contrast to psychoanalytic theorizing, recent research in developmental psychology and sociology has abandoned the rigidity of "infant determinism" (see, Sewell 1952; Mischel 1969; Havighurst 1973; Brim

and Kagan 1980). It is becoming clear that personality is more plastic and changing over the life cycle than was previously believed. Varying childrearing practices, within a normally benign range, do not seem to be productive of major differences in personality development. It does seem clear, however, that deviantly "bad" parents (intense parental conflict, criminality, neglect, etc.) retard and disrupt development (for a summary, see Kohlberg 1981).

Here we may also point out structural reasons for a possible historical diminution in the personality effects of sibling number and birth order. First, children are increasingly exposed to a wide variety of emotionally engaging influences outside of the family, as well as to the influence of nonfamilial peers. Second, since occupations are ever more divorced from the family, both work activities and anticipations of these activities are more independent of siblings. Whereas in the past many people continued their sibling relationships into adult working life, the declining importance of family farms and family businesses means that this is rarely the case today. Long before young people leave home for college or work, they are thinking and dreaming about what they will do occupationally. Rarely does this fantasy include familial continuity in work relations. Finally, and as an extension of these points, under modern American conditions, such inheritance as there is appears either to be divided among children equally (Brittain 1978; Menchik 1980) or to be apportioned in a manner that is inversely related to the wealth of the sibling recipients (Becker and Tomes 1976; Griliches 1979; Tomes 1981), with no special advantage to a child of a particular birth order, such as the oldest. It is also no longer incumbent upon younger siblings to wait for their marriages until older siblings have been wed. Thus, as familial control over both occupation and marriage has declined, birth order has lost its saliency as a basis for role differentiation in the family (see, Schooler 1972, for a discussion of this point in relation to the oldest child).

Number of Siblings and Achievement Motivation

The issue of whether certain groupings in the population are characterized by more or less "achievement motivation" is probably most familiar to the reader as a consequence of the literature on

the "culture of poverty" or the "underclass" (see, for example, Banfield 1970; Rainwater 1970; Schiller 1976; Gecas 1979; Auletta 1982). One alleged component of the so-called culture of poverty is less achievement motivation, a component that is believed to be transmitted through the socialization process from parents to children, and which is said to perpetuate people as an underclass over generations.

We do not wish to engage the issue of the "underclass" here. Of relevance, however, is the analogous reasoning involved in the hypothesis that family size is of major motivational importance. This hypothesis is seen, for example, in the notion that small and large families are essentially different "systems" (see, Bossard 1953; Elder and Bowerman 1963). The small family system is said to be striving and achievement oriented, in contrast to the less compulsive and ambitious character of the large family system. If this hypothesis is correct, it could relate, of course, to differences in educational attainment by number of siblings.

Before we attempt to bring data to bear on this issue, a brief discussion of achievement motivation is in order. Following Atkinson and Birch (1978) we will distinguish the components of motivation into motives/incentives and expectancies. A motive is said to be a generalized disposition to approach or avoid a class of incentives. For example, the needs for achievement, power, or affiliation are classed as motives. An expectancy is the individual's assessment of the probability that his or her behavior will lead to a desired outcome. Both motives and expectancies are involved in overall motivation—people must need to achieve in order to take action (to try to bring about a result that would not occur without effort), but if the outcome does not seem remotely achievable to them, then action may not be initiated. Motives are believed to be strongly influenced by early socialization and relatively impervious to change during the life course. Expectancies, by contrast, are dynamic because they are influenced by a daily round of environmental cues, or more correctly, by people's perception and interpretation of environmental cues.

The notion of individuals being in situations of structured deprivation or opportunity involves, therefore, not simply whether such situations affect the development of particular motives, but whether the situations sensitize people to perceive and interpret environ-

mental cues in ways that contribute to achievement or failure, to action or inaction (Hyman 1953; Knupfer 1953). For this reason, the issue of deprived "cultures" (or, in essence, a structured situation of deprivation) cannot be dichotomized simply as leading to "flawed character" versus "restricted opportunity" (Schiller 1976). People who have been reared in situations of restricted opportunity may have strong needs to achieve ("unflawed characters"), but their rearing and the ordinary circumstances of their lives are such that "opportunities" are not perceived or interpreted correctly. There is an absence of the kind of "paradigm," to use Kuhn's terminology, that would enable them to sort environmental cues into a message that signals "opportunity." For purposes of simplicity, we will leave aside any discussion of the complexities involved when individuals transform blocked expectancies in the legitimate world into achievement through the use of negatively sanctioned means.

In chapter 6, we analyzed expectancies regarding educational levels. With regard to motives, our data are replete with the standard measures of achievement motives, both positive and negative (need for achievement and fear of failure). The variables are listed in figure 7.3. The results by number of siblings are easy to summarize. We find neither bivariate effects with sibsize nor suppressed effects that emerge when other components of the respondents' background are added. In an effort to go beyond the standard data relating to achievement motives, we have created a derived measure that relates to achievement motivation—"under-" or "overperformance" in school.

School Performance by Sibsize and "Under-" and "Overperformance."

The most common measure of school performance is grades. Our data for young people include self-reported grades in the prior year (for the Youth in Transition and High School and Beyond samples) and teacher evaluations of school performance (for Cycles II and III of the Health Examination Surveys). We have already seen that grades are negatively related to sibsize in all of the samples. This finding is hardly surprising since IQ is negatively related to sibsize, and IQ is an important determinant of grades.

More indicative of achievement motivation is whether students perform in school at a level that would be expected given their

Figure 7.3.
Indicators of Achievement Motivation (Youth in Transition,
High School and Beyond, Health Examination Survey—Cycle III).

ability, or whether they "underachieve" or "overachieve" in relation to ability, and how such behavior relates to sibsize. Following Lavin's methodological discussion of "overperformance" and "underperformance" in school (Lavin 1965), we have regressed grades on ability to obtain expected performance levels at each level of ability and, then, created a new variable that shows (in standard deviation units) respondents who perform below, at, and above what would be expected given their ability.

Table 7.15 shows predicted underperformance and overperformance in school for the youngsters in Cycles II and III of the Health Examination Surveys, the tenth-grade boys in Youth in Transition, and the tenth- and twelfth-grade boys and girls in High School and Beyond. It is worth noting first how similar are the overall proportions (the total lines) at predicted underperformance and overperformance levels among the surveys. Second, there seems to be a fairly consistent, but not invariant pattern of underperformance among those from large families. This pattern is more marked among girls than among boys. A comparison of seniors (High School and Beyond) versus sophomores (both Youth in Transition and High School and Beyond) suggests that school-leaving by underperforming boys from large families may contribute to the reversal of the sibsize/underperforming relation. Third, with regard to overperformance, it is rare for male only-children to exhibit this characteristic, but female singletons are somewhat more inclined to do so. Indeed, male only-children quite frequently underperform in relation to their abilities. Finally, children from sibsizes two to four are generally more likely to overachieve than those from sibsizes five or more. None of these differences are large, and many are not statistically significant. However, their appearance in so many different surveys generates some interest.

Are the tendencies to underperformance by those from large families due to the fact that they are more likely to come from educationally less-advantaged homes? Table 7.16 breaks down the data from table 7.15 into three educational groups for the youngsters' fathers. Since all of the surveys show a similar pattern, we present here only HES Cycle II and sophomores and seniors in HSB. With regard to underachievement, the dominant pattern among the surveys is for there to remain a higher-than-average underperformance value among children from sibsizes five or more.

As for overperformance, the differences by sibsize either disappear, or there is a tendency for those from small families to be only slightly more overperforming. Quite clear is a decline in underperformance with an increase in father's education (total lines), and an increase in overperformance as father's education goes up. Hence, the intuitive impression, such as by Bossard and Boll, of children from small families as overperforming (and/or not underperforming) is compounded by the association of father's education with small families, and by the association of both father's education and small families with less underperforming and somewhat more overperforming. Indeed, as we have suggested, some of the prejudice against children from small families (and especially only children) may actually be a reflection of the dual advantages such children have enjoyed (higher class and small family size) on the part of those who have suffered dual disadvantages (lower class and large family size). Insofar as being an only child seems particularly to lead to feminine overperformance, a further complication is added. However, there is a persisting tendency for singleton boys to underperform in relation to ability. A fairly obvious interpretation of underperformance in relation to grades by singleton boys and overperformance by singleton girls is that the boys feel totally without competition for their parents' approval and (being on average very bright) do not see fit to overextend themselves. Girls are also without actual competition, but many may feel symbolic competition—comparison with what they may believe to be their parents' vision of how well a boy would have done in their place. It is evident that there is much more to be learned about the effects of being an "only" on young people's achievement pressures. It is certainly clear that only boys, on average, are not feeling intense pressure to uphold extreme parental desires for achievement.

Confidence in Ability

As Ernst and Angst (1983) have noted, most psychoanalytic expectations of self-confidence and self-esteem by birth order have been that only children and firstborns would rank lower on these qualities. However, among the numerous studies cited by Ernst and Angst, most of which control neither for social class nor number of siblings, there is typically either no difference by birth order or firstborns score higher. The Project Talent data analyzed by Claudy, Farrell, and Dayton (1979) controlled for

parental background and showed no significant difference in self-confidence between only children and those from two-child families. A recent study by Falbo (1981), of undergraduates in college who responded to an advertisement, did control for number of siblings and social class. This research used an index of self-esteem (a short form of the Texas Social Behavior Inventory) that measures social self-confidence. Only one barely significant difference by birth order was found—between firstborns and last-borns, favoring firstborns.

Here we will focus our research question on an issue that would seem to be relevant to the educational process: Given that children from small families are known to have higher intellectual abilities than children from large ones, how does their confidence in their ability relate to their measured ability? To attempt to address this issue, we will have to confine our efforts to the Youth in Transition data set. Unfortunately, questions on confidence were not asked of sophomores in the High School and Beyond study and, among seniors in that study, only four items on the Rosenberg self-esteem scale were asked. The overall variability of response was slight, and there were no differences by sibsize even at the bivariate level. Equally, no questions on confidence were asked in either of the Health Examination Surveys.

The Self-Evaluation of School Ability index in the study Youth in Transition was composed of three questions to which the boy could respond on a six-point scale ranging from (1) "Far below average" to (6) "Far above average." The questions were:

1. How do you rate yourself in school ability compared with those in your grade in school?

2. How intelligent do you think you are compared with other boys your age?

3. How good a reader do you think you are compared with other boys your age?

Table 7.17 shows the adjusted means of the index of confidence in school ability by sibsize. These means result from a multiple classification analysis controlling for parents' education and socioeconomic status, family intactness, and size of the place where the youth was brought up. There is a marked and linear difference in

results by sibsize, with only children being over half a standard deviation (from the grand mean) higher than children in families of seven or more even after adjustments.

However, since boys from small families have higher IQs on average and higher grades than boys from large families, one would expect higher levels of self-confidence from those with few siblings. Hence, we may now ask whether boys from small families are more confident, and those from large families less so, than would be predicted on the basis of their IQs? The lowest tier (the total) of table 7.18 shows that this is indeed the case. Boys from small families are more confident than would be predicted, and those from large families less so. When the results are broken down by three levels of the father's education in table 7.18 (in an effort to control for the association of family background with both confidence and sibsize), we see that the relationship between sibsize and confidence disappears for those where the father has completed high school, but that among those with college-educated fathers overconfidence is a particularly marked characteristic of boys from

small families, and underconfidence particularly characterizes those from large families. Moreover, coming from a large family provides no more overconfidence if the father is college educated than it does if he has not completed high school. These data thus suggest that family size is not only negatively related to ability but also to confidence in such ability, and that this relationship is particularly evident among boys from the highest and lowest levels of parental education.

Conclusion

Do children from different-size families have home environments that vary in ways that may influence their cognitive development and academic achievement? Do extracurricular school activities diverge by sibsize in a manner that may have cognitive and academic implications? Do young people from different-size families have distinctive personality characteristics, such as achievement motivation and self-confidence, which may help to account for the educational deviations that we have seen to exist? And, finally, do parents of small versus large families have disparate childrearing values that may function as unmeasured antecedent variables explaining unequal educational achievement by sibsize? This chapter is devoted to providing some answers to these questions.

Our examination of the home environment makes a distinction between environments as "settings" and as "treatments." Following Spaeth, "settings" refer to the passive aspects of the environment—features that are available to be engaged but do not actively engage the child. "Treatments" are parental behaviors that are directed toward the child and that represent active interventions—encouragement, correction, goal setting and the like. We have regarded this distinction as heuristic and tentative, not ironclad, since many of the environmental features we discuss can be regarded as both passive and active. Using the Health Examination Survey (Cycle II) of young people, we have looked at quantitative data on the amount of time spent by young children (six to eleven year olds) in a variety of informal activities such as reading books, reading newspapers, TV watching, playing alone, and engaging in sports, Little League, and similar activities. We find that children from small families are more likely to engage in intellectual and cultural pursuits, and children from small families (particularly only children) spend much more time playing alone. However, data on popularity suggest that children from small families are more popular. Hence, the fact that such children spend a fair amount of time alone is not due to enforced isolation because of unpopularity. It does seem, however, that such children have the opportunity to develop the ability to be alone and to use such time profitably. Nothing from any of the surveys suggests that children from small families, especially only children, are lonely, depressed, or hopeless.

An interesting finding from the High School and Beyond data concerns whether the respondents were read to as children. Children from small families report themselves as having been more likely to have been read to as children than those from large families, in spite of the fact that, in many cases, children from large families presumably had siblings who could have performed this service. These data thus indicate that although siblings might theoretically provide help and instruction, children's chances of actually receiving such attention are enhanced if parents can provide it. The data do not support Zajonc's hypothesis concerning the teaching role of older siblings. In the same study, youngsters from small families were more likely to have experienced selected advantages such as having music or dance lessons or to have traveled outside the United States. In effect, our analysis suggests that children from small families and only children have more intellectually stimulating settings and a broader range of stimuli.

Among teenagers, our analysis by sibsize of the types of extracurricular activities in high school engaged in by seniors is also instructive. A factor analysis allowed us to divide nonsports activities into five types—intellectual (student government, school newspapers, etc.), cultural and artistic (drama, orchestra, etc.), community (youth groups, church groups, "Y," etc), vocational (farm clubs, carpentry, etc.), and hobby. Youngsters from small families are more likely to engage in intellectual and cultural activities (with or without sports) than are those from large families, and those from large families incline more toward community, vocational, and hobby activities. This is true even after we have removed youngsters from farm families (which are both larger than others and also have high rates of participation in community and vocational kinds of activities).

Turning to parental treatments, do parents of small versus large families conform to the speculation that small families are hard-driving and competitive, emphasizing external standards of excellence, whereas large families emphasize obedience, conformity and assimilation? Although there does seem to be some bivariate association between parental values for offspring and the number of children parents have, this association disappears when education, age, and religion are held constant. These results provide further backing to our contention that many of the personality characteris-

tics that have been associated with small versus large families are actually characteristics related to education and class rather than family size.

Turning to respondents' personality characteristics, we have been concerned with those aspects of personality that seem important for academic and intellectual advancement, and about which we have information. We have concentrated on over- and underperformance in school and on confidence in one's own intellectual ability.

With regard to over- and underperformance (defined as receiving school grades above or below what would be expected given the student's ability), differences by sibsize are not large, but children from sibsizes two to four are more likely to overperform than those from sibsizes five or more. Large families show a fairly consistent pattern of underperformance as well.

Information on confidence in one's own intellectual ability is available from Youth in Transition. Not only do youngsters from small families have more confidence in their ability than those from large families, but they have more confidence than would be expected on the basis of their ability scores. By contrast, boys from large families have less confidence than would be expected on the basis of such scores.

It thus appears that with regard to personality characteristics, young people from small families, including only children, seem to have none of the adverse traits often associated with an absence of many siblings. Moreover, children from small families are likely to overperform rather than underperform in school (although the differences on overperformance are not large), and to have a great deal of confidence in their own ability. Children from large families are more likely to underperform, and lack confidence in such objectively verified abilities as they have. We are thus led to believe that the prevalent notions about negative traits of only children and those from small families stem in part from the highly advantaged position that such children have, relative to children from large families—advantaged socioeconomically and in terms of the lack of dilution of such resources as exist. Since many more people have come from large families than from small ones, such people have "carried the day" in labeling those from small families as overindulged or spoiled by advantages.

Finally, this chapter has provided additional information relating to the causal interpretation advanced in this volume—that there are independent effects of sibsize on child achievement. As discussed in detail in our introduction, the alternative explanation of sibsize effects is that both sibsize and differential environment are the result of unmeasured antecedent parental characteristics that affect both the number of children parents have and the environments they provide. In effect, it is argued that sibsize effects are spurious. The results in this chapter supplement the data presented in our introduction concerning whether parents of small versus large families have disparate personality traits. Here we examine whether such parents have different childrearing values—differences that are not captured by socioeconomic controls. Our analysis indicates either no difference, or small differences in childrearing values by sibsize after controls for parental socioeconommic status are introduced. It thus seems that parental personality traits and childrearing values, at least as analyzed so far, are not among the unmeasured variables that might invalidate our interpretation of the sibsize/achievement relationship.