Rebel without a Cause or Effect: Birth Order and Social Attitudes

by Jeremy Freese, Lala Carr Steelman
Citation
Title:
Rebel without a Cause or Effect: Birth Order and Social Attitudes
Author:
Jeremy Freese, Lala Carr Steelman
Year: 
1999
Publication: 
American Sociological Review
Volume: 
64
Issue: 
2
Start Page: 
207
End Page: 
231
Publisher: 
Language: 
English
URL: 
Select license: 
Select License
DOI: 
PMID: 
ISSN: 
Abstract:

REBEL WITHOUT A CAUSE OR EFFECT:
BIRTH ORDER AND SOCIAL ATTITUDES
Jeremy Freese Brian Powell

Indiana University Indiana University
Lala Carr Steelman

Uiziversity of South Carolina

The enduring effects of an individual's birth order have been subject to a long and lively debate in sociology and other disciplines. Recently, in re- sponse to Sulloway's (1996) Born to Rebel: Birth Order, Family Dynamics, and Creative Lives, interest has increased in the possible effects of birth or- der on social attitudes. Using quantitative, historical data, Sulloway found that birth order is a better predictor of social attitudes than is gender, class, or race. His novel, evolutionary theory asserts the universal influence of birth order across eras and cultures. We use contemporary data to test Sulloway's contention that firstborn adults are more conservative, support- ive of authority, and "tough-minded" than laterborns. Examining 24 mea- sures of social attitudes from the General Social Survey (GSS), we find no support for these claims, either in terms of significant effects or even the direction of nonsignificant coefficients. An expanded inquiry using all (202) relevant attitudinal items on the GSS yields similar results. In our analysis, variables discounted by Sulloway-gender, race, social class, and family size-are all linked to social attitudes more strongly than is birth order: Our

findings suggest that birth-order theories may be better conceptualized in terms of modest effects in limited domains and in specific societies.

Sociologists of the family have long tried to direct more attention to how individu- als are affected by basic components of fam- ily structure and early childhood environ- ment. In recent years, these efforts have sparked considerable debate about the effects of family size (Downey 1995; Downey et al. 1999; Guo and VanWey 1999; Phillips 1999)

" Direct all correspondence to Jeremy Freese, Department of Sociology, Indiana University, 744 Ballantine Hall, Bloomington, IN 47405 Cjfreese@indiana.edu). We thank Robert Hauser and Rob Warren for their assistance regarding the Study of American Families data. We also thank Karen Lutfey, Robert Fulk, Clem Brooks, Erin Maher, Cher Jamison, Robert Carini, Carl Ek, L.

C. Steelman, Joseph Steelman, members of Indi- ana University's Seminar in Social Psychology and Political Economy Workshop, and, espe- cially, Frank Sulloway for their helpful comments on previous versions of this paper. The research was partially supported by a fellowship from the National Science Foundation for Freese, NSF grant SBR-9810435 for Powell, and NSF grant SBR-9810246 for Steelman.

and of being raised in single- versus two-par- ent homes (Cherlin, Chase-Lansdale, and McRae 1998; Furstenberg 1990; McLanahan and Sandefur 1994). Historically, however, the family structure variable that has perhaps received the most lively scrutiny both within and outside the realm of sociology has been birth order. Indeed, birth order has inspired some of the most striking dialogues among sociologists and between sociologists and other social scientists (Bayer 1967; Hauser and Sewell 1985; Retherford and Sewell 1991; Schachter 1963; Steelman and Mercy 1980; Zajonc and Markus 1975; Zajonc et al. 1991). Sociologists have examined the effects of birth order on achievement, educational performance, and personality, and have cri- tiqued some of the more ambitious claims that have been made about birth order. Until recently, scholars have shown little empirical interest in the relationship between birth or- der and social attitudes, despite the persistent (and often taken-for-granted) notion that adults who are firstborn children are more

American Sociological Review, 1999, Vol. 64 (April:207-231)

conservative than adults who are not. Instead, to comprehend the determinants of social at- titudes, sociologists have typically empha- sized the role of social structural position, cultural norms, and group identity, and they have relied on a relatively limited set of ex- planatory variables, most prominently gen- der, race, age, and social class.

Interest in birth order and social attitudes has been reignited by the recent publication of Sulloway's (1996) Born to Rebel: Birtlz Ordei; Family Dynamics, and Creative Lives. Using an impressive array of quantitative his- torical data, Sulloway finds evidence that firstborns are more politically conservative, supportive of existing authorities, and "tough-minded" (i.e., punitive) than are later- borns. Indeed, Sulloway's analyses suggest that, when appropriate controls are used, birth order is a better predictor of social attitudes- and, in turn, a better predictor of radical be- havior-than is gender or social class. These dramatic findings have received considerable public attention, and Sulloway's book ranks high among the decade's top-selling books in social science.

Yet despite its popularity, Sulloway's work cannot be easily dismissed as "pop social sci- ence." His methods of data collection and quantitative analyses are more thorough and sophisticated than virtually all previous re- search in the field, and he provides a plau- sible account of how flaws in previous stud- ies may have systematically obscured real birth-order effects.' Moreover, although Sulloway's basic premise is similar to long- held ideas about birth order (i.e., that first- borns are more conservative than laterborns) he offers an ingenious theoretical explana- tion of birth order's influence that draws upon recent work in sociobiology and evolu- tionary psychology. He is one of the first thinkers to elaborate an explicit theoretical link between birth order and social attitudes (and, in turn, behavior). Sulloway also offers detailed predictions about the effects of birth order on attitudes and how such effects vary

Borrz to Rebel is also difficult to dismiss as "pop" because of the reputation of its author: Sulloway is a former MacArthur fellow who has won numerous awards and plaudits for his earlier, critical work on Freud and psychoanalysis (Sulloway 1979).
AMERICAN SOCIOLOGICAL REVIEW

as the result of interactions with other inde- pendent variables. This combination of rig- orous analysis and creative theorizing has drawn accolades from scholars as diverse as Edward 0.Wilson and Robert K. Mert~n.~ The intellectual breadth of the book is evi- denced by reviews in prominent periodicals of the natural sciences (Science, Nature), so- cial sciences (Contemporary Sociology), and broader public (see Wolfe's [I9961 featured review in the New Republic).

Sulloway also makes a more cogent case than has anyone before for why sociologists, in particular, should care about birth order. First, he provides evidence that birth order has been a crucial yet overlooked source of social division, and that birth order has played an important role in shaping the ideo- logical differences among participants in some of modern history's most transforma- tive events. Second, Sulloway claims that his data explicitly challenge classic sociological analyses of social movements, including Marxian studies of the French Revolution (pp. 309-15) and Weberian claims about the Reformation (pp. 262-70). Third, by arguing that birth order is a more important determi- nant of social attitudes than other variables, he maintains that sociologists have overem- phasized the effects of social structural vari- ables and that greater attention should be de- voted to an individual's early family environ- ment. Finally, Sulloway's use of evolution- ary theory to explain birth-order effects prompts him to assert a broad generality to his findings, and he maintains that birth or- der is as critical to understanding ideological cleavages in contemporary society as it is in the historical disputes that comprise the bulk of his data.

We examine the relationship between birth order and social attitudes using data from a contemporary survey of adults in the United

On the book's jacket, Wilson calls Born to Rebel ". . . one of the most important treatises in the history of the social sciences," and Merton writes, "A quarter century in the making, this brilliant, searching, provocative, and readable treatise promises to remain definitive for at least as long." Other eminent scholars praising the book include zoologist Ernst Mays and anthro- pologist Sarah Blaffer Hrdy, who predicts that the book ". . . will have the same kind of long-term impact as Freud's and Darwin's."

States. We test whether firstborns are more conservative than laterborns, and we also test some of Sulloway's specific hypothesis about the effects of birth order on attitudes. Through this analysis, we seek not only to evaluate claims about the relationship be- tween birth order and social attitudes, but also to assess the influence of birth order relative to that of variables discounted by Sulloway's analysis: gender, race, social class, age, and number of siblings.

BACKGROUND
Birth Order

Birth order has had a long and cyclical his- tory in the social sciences. For the most part, modern birth-order theories have refashioned ideas already implicit in the centuries-old lore surrounding birth order; namely, that firstborns are smarter, more responsible, and more conservative than their younger sib- lings. Birth order first received scholarly at- tention in the work of Francis Galton (1874), who interpreted the disproportionate number of firstborns in his sample of English scien- tists as evidence of "the superiority of the el- dest." Although, from a contemporary per- spective, birth order may seem out-of-step with the hereditarian focus of Galton's schol- arship (he was a pioneer of the eugenics movement), at the time Galton's findings were consistent with the common view that the intrinsic quality of a woman's offspring degenerated with successive births-an idea that persisted long after Galton (Apperly

1939; Clark 1918; Gini 1915). Later, as psy- choanalytic theories of the self ascended, Adler (1928) proposed that firstborns dif- fered from laterborns because only firstborns experienced the trauma of having the undi- vided attention of parents taken away from them by the birth of a sibling. Adler (1956: 327) posited that this "dethronement" causes firstborns to try to imitate their parents, to identify more strongly with rules and author- ity, and to often become "power-hungry con- servatives." As a central part of Adler's "in- dividual psychology," birth order remains an active topic in developmental psychology, psychoanalysis, and family studies research (Berthoud 1996; Weinstein and Sackhoff 1987).

In the 1960s and earlv 1970s. a number of social psychologists shifted their emphasis from the dynamics of experimental small groups to dynamics within families. Some of these early models of family dynamics made strong declarations about the consequences of birth order. For example, Toman (1970) contended that one can use birth order and its interactions with other family structure variables to predict quite accurately "a person's major personality characteristics and those of his friends and lovers, the like- lihood of stability in his marriage, . . . what he is like at work, his politics and his prob- able philosophy" (p. 45; also see Toman 1993). Other researchers interested in intel- lectual development began to pay renewed attention to children's early home environ- ments. It was in this context that Zajonc and Markus (1975) first proposed that firstborns had an academic advantage over laterborns because home environments tended to be more intellectually stimulating for an eldest child than for her or his siblings. Their "confluence model" of birth order and aca- demic performance remains one of the most influential theories in social psychology.

Although bold claims have been made about the effects of birth order, work in so- ciological journals (including work pub- lished in American Sociological Review) has tended to remain skeptical. Schachter (1963) challenges previous research on birth order and eminence (also see Somit, Arwine, and Peterson 1996). Retherford and Sewell (1991) cap a series of sociological studies that have failed to find support for Zajonc and Markus's (1975) confluence model, de- spite its continuing popularity (Mercy and Steelman 1982; Steelman and Mercy 1980; also see reply to Rutherford and Sewell by Zajonc et al. 1991). Status attainment re- search by Blau and Duncan (1967), Hauser and Sewell (1985), and Blake (1989) chal- lenges previous assertions that firstborns have greater economic or educational attain- ment than laterborns (also see Olneck and Bills 1979; Steelman and Powell 1985; for contradictory findings see Bayer 1967). Re- views by Adams (1972) and Schooler (1972) catalogue some of the methodological flaws that mar early birth-order studies. steelman's (1985) overview of research suggests that the scholarly attention showered on birth order has come at the expense of other family con- figuration variables-most notably family size, for which better evidence of consistent effects exists (Blake 1989; but see Guo and VanWey 1999).

Nonetheless, sociologists have found evi- dence of birth-order effects in other domains. Birth order has been shown to influence pa- rental discipline patterns, funding for col- lege, leadership, and sociability (Felson and Russo 1988; Steelman and Powell 1985, 1991). Sociologists also have provided a number of reasons why birth order should be of theoretical interest to the discipline. Bonacich, Grusky, and Peyrot (1985) argue that structural position in the family plays a critical role in the formation of intrafamilial coalitions. Differential treatment of first- borns by parents has been shown to influence occupational choice and educational attain- ment in other societies (e.g., Greenhalgh 1985; Post and Pong 1998). In our own soci- ety, Steelman and Powell (1996) note that birth order still often acts as an important status characteristic within families; as such, they speculate firstborns and laterborns may reproduce many of the same dynamics that mark superordinate and subordinate relation- ships more generally.

When considering the relationship be- tween birth order and social attitudes, how- ever, the analogy between family structure and social structure should not be taken too far. To understand why individuals differ in their social attitudes, sociologists have typi- cally emphasized the role of social structural position, cultural norms, and group identity (Gurin, Miller, and Gurin 1980). Emphasis on these mechanisms has prompted sociolo- gists to stress a relatively limited number of explanatory variables, most prominently race, gender, age, and class. Yet the same theoretical mechanisms are not so easily ap- plied to birth order, because there is no such thing as a "firstborn subculture" and because firstborns do not saliently identify with one another as a group. As a result, any ideologi- cal differences between firstborns and later- borns imply a different set of underlying causes than those thought to create cleavages between the rich and poor, men and women, or blacks and whites. Indeed, birth-order theories (including Sulloway's, as we shall see) are often presented as opposing conven- tional sociological analyses, and the debate over birth order is often conceived as a dis- pute between those who emphasize the influ- ence of social structure and culture and those who champion the effects of early family en- vironment.
Born to Rebel

Using biographies and ratings from 94 his- torical experts, Sulloway's (1996) initial analyses examine data on more than 3,800 scientists who were pivotally involved in 28 different scientific controversies that took place over the past four centuries. Sulloway finds that while firstborn scientists tend to be more accomplished than laterborns, later- borns have been responsible for many of the most radical and important scientific break- throughs. Moreover, when radical new ideas are proposed, Sulloway finds that laterborn scientists have been more than twice as likely as firstborns to adopt them early, es- pecially when the ideas have liberal social implications. Firstborn scientists, by contrast, have only embraced radically innova- tive ideas when they have had clear conser- vative implications, such as scientific move- ments based largely on religious principles (for example, vitalism and idealistic taxono- mization).

From these initial, evocative observations, Sulloway moves from considering scientists to a broader examination of the influence of birth order on social movements. Using data extracted from biographies and expert rat- ings of thousands of historical figures, he finds that birth order and age are strong pre- dictors of liberal social attitudes, and that the influence of birth order on social attitudes has led laterborns disproportionately to sup- port liberal social movements and firstborns to resist them. According to Sulloway's data, birth order is the single best predictor of whether eminent individuals converted to Protestantism or remained Catholic during the Reformation. During the French Revolu- tion, a majority of firstborn deputies of the National Convention voted to execute Louis XVI, while most laterborn deputies voted to spare the King's life.3 In American political

Although at first glance the mere presence of firstborns among revolutionary deputies may

history, laterborns have been overrepresented among radical abolitionists, suffragettes, radical revolutionaries, social reformers, and liberal Supreme Court justices.

Although these instances appear to high- light behavioral differences between first- borns and laterborns, Sulloway argues that the behavioral differences are rooted at least in part in the effect of birth order on social attitudes. While Sulloway argues that birth order is crucial for understanding ideologi- cal differences, he claims his data also show that sociologists and other social scientists have exaggerated the explanatory role of gender, class, or race. He concludes that so- cial structural factors are not as important for understanding attitudinal differences as are family configuration and birth order.

Sulloway's thinking contradicts prevailing sociological wisdom about the effects of birth order relative to other variables, but the pos- sibility that he has found something others have missed gains support from both the so- phisticated quantitative methods he employs and the novel evolutionary theory that he of- fers to explain birth order's effects. Regard- ing the latter, Sulloway postulates that chil- dren have an innate tendency to develop the attitudes and personalities best-suited for maximizing the resources that they get from their parents. Because siblings must compete with one another for parental resources, Sulloway argues that the attitudes most likely to nab parents' attention are those which help children stake out a unique "family niche" relative to their brothers and sisters. In the early years, at least, firstborns tend to be larger, stronger, and more intellectually de- veloped than their siblings; as such, they al- ready occupy a dominant position in the sib- ling group and develop the conservative atti- tudes best suited for safeguarding this place.4 On the other hand, laterborn children are chronic underdogs who must seek out oppor-

seem to contradict Sulloway's theory, his treat- ment of the role of birth order in the French Revolution is much more nuanced than this and explains the seeming anomaly (see Sulloway 1996:306-26.)

"arents also are seen as having a sociobiologi- cal incentive to favor firstborns, because the pos- sibility of childhood mortality lowers laterborns' expected reproductive success relative to their older siblings (Daly and Wilson 1988).

tunities to differentiate themselves from their older siblings. In trying to secure their own family niche, laterborns develop attitudes that are more liberal, antagonistic to authority, and compassionate than do firstborns.

Even though Sulloway ties his theory to sociobiology and evolutionary psychology, he does not claim that there are any genetic differences between firstborns and laterborns. Rather, he contends that all children have the same evolved disposition to maximize paren- tal investment, but that the optimal strategy for doing so varies systematically by birth order. In this regard, Sulloway's theoretical arguments are exemplary of a recent move- ment in evolutionary approaches to human behavior that criticizes sociobiology's early genetic determinism and emphasizes the sen- sitivity of evolved adaptations to environ- mental conditions (Pinker 1997). Sulloway's theory has been hailed by many sociobiolo- gists and evolutionary psychologists because it provides a general means for incorporating environmental influences within an evolu- tionary framework, while still downplaying the explanatory potential of education, socio- economic background, and the other factors that some evolutionary psychologists have lumped together as the "incoherent environ- mentalism" of the "standard social science model" (Tooby and Cosmides 1992:37).

The theoretical language of evolutionary approaches allows Sulloway to envisage a broad generality to his finding^.^ Even though his sample is comprised mainly of eminent historical figures, most of whom were upper-crust white males, Sulloway claims that "the effects of birth order tran- scend gender, social class, race, nationality, and for the last five centuries-time" (p. 356). Sulloway also provides an appendix, in which he shows readers how to calculate

To support his claims about the contemporary relevance of his findings, Sulloway also conducts a meta-analysis of previous birth-order studies in psychology, in which he predicts and finds sig- nificant birth-order effects on all five basic per- sonality dimensions (also see Sulloway 1995). Unfortunately, this meta-analysis has a number of severe problems, which are both meticulously and persuasively documented by Harris (1998). Pre- liminary work using sibling data by Hauser, Kuo, and Cartmill (1997) finds no evidence of birth- order effects on any personality dimension.

their own "propensity to rebel" based on the logistic regression coefficients from his sample of historical elites (pp. 440-44). Such claims to generality are of course einpirical claims, and whether Born to Rebel ultimately measures up to its accolades and theoretical promise will depend on how well its ideas withstand empirical scrutiny using contem- porary data and representative samples. By testing both the general idea that firstborns are more conservative than laterborns and some of Sulloway's more specific claims about birth order and social attitudes, we seek to assess whether the renewal of inter- est in birth-order effects is justified and whether birth order deserves a place along- side gender, race, class, and age in studies of social attitudes.
DATA AND MEASURES
Data

We use data from the 1994 General Social Survey (GSS), conducted by the National Opinion Research Center (Davis and Smith 1994). The GSS is a full probability sample of noninstitutionalized, English-speaking adults in the United States. In 1994, as part of a special module on family mobility, GSS respondents were asked to provide back- ground information on each of their siblings, including their year of birth. To our knowl- edge, no other data set that contains infor- mation on a respondent's birth order com- bines the GSS's large, nationally representa- tive sample, high-quality data-collection techniques, and variety of questions on so- cial and political attitudes.

A difficulty in testing birth-order theories is that many individuals' early family lives do not lend themselves to easy classification as firstborns or laterborns. We use a sub-sample of GSS respondents that excludes only children, respondents with any step- or half-siblings, and respondents who report having a sibling born the same year as they were born. Only children differ from first- borns in that they do not compete with
AMERICAN SOCIOLOGICAL REVIEW

Step- and half-siblings imply varying rela- tions in a family between children and care- givers that may complicate the allocation of parental resources and may unfairly under- mine the expectations of birth-order theo- ries. We assume that most pairs of siblings born in the same year are twins, and twins, unfortunately, have received little consider- ation in the birth-order literature. In auxil- iary analyses, we retain only children and respondents with step- or half-siblings; retaining these individuals in our sample has little effect on the overall patterns we observe (these results are available from the authors on request).

After also excluding respondents who failed to report the birth year of any of their siblings, the sample used in our main analy- ses contains 1,945 of the original 2,992 re- spondents. The number of cases used in our analyses is sometimes considerably fewer than 1,945 because the 1994 GSS used a split-ballot design, and most of the questions eliciting social attitudes were administered only to a randomly selected subset (one-third to two-thirds) of all respondents. Preliminary analyses using items administered to the whole sample suggest no systematic differ- ences among respondents receiving different ballots.

Like the data used in Sulloway's and most other birth-order studies, the GSS allows comvarisons between firstborns and later- born; from different families. Some have ar- gued that birth-order research ideally should use data that allow for the direct comparison of siblings (Retherford and Sewell 1991), al- though Ernst and Angst (1983) find that stud- ies using inter- and intra-familial data yield similar results.' We are able to supplement our interfamilial analyses of the 1994 GSS data with intrafamilial comparisons that combine the GSS data with data from the

der or sibling rivalry." Nevertheless, he usually codes only children as firstborns in his analyses. Including only children as firstborns compro- mises the evolutionary theory that firstborns' social attitudes develop in response to rivalry from their younger sibling(s), which is why we exclude

younger siblings for parental attenti~n.~them here.

Ernst and Angst's (1983:170-71) discussion

Sulloway (1996:23) writes that only children of birth-order studies that use parent ratings of .'represent a kind of .controlled experiment'- siblings may be read as an exception to their what it is like to grow up unaffected by birth or- overall conclusions on this point.

Study of American Families (SAF) (Hauser and Mare 1997). The SAF attempted to in- terview one randomly selected sibling of GSS respondents; in all, 1,115 sibling inter- views were conducted. SAF includes a small subset of the social attitude items also used in the GSS. When examining sibling pairs from the SAFIGSS, we exclude step- and half-siblings and twins; we also exclude cases in which both siblings interviewed are later born^.^ While the SAF data enable us to compare siblings' attitudes, two caveats should be made: The SAF has a low response rate (43 percent) and the range of attitudinal items on the SAF is much narrower than in the GSS.
Measures of Birth Order

We measure birth order as a dichotomous variable indicating whether the respondent is the firstborn child in his or her family, as in- dicated by the year-of-birth information pro- vided by respondents. This tactic is the most common way of measuring birth order in pre- vious research. Sulloway's work has been criticized for using a measure of what he calls "functional" birth order, in which he makes case-by-case adjustments for instances in which early family environment is inconsis- tent with biological birth order-as, for ex- ample, when divorce and remarriage lead a firstborn child to be raised with older step- sibling^.^ As mentioned earlier, we dropped

The number of cases in our analysis of SAF data is considerably fewer than 1,115, partly be- cause we exclude these respondents, but also, as mentioned above, because the split-ballot format of the 1994 GSS led to some attitudinal items be- ing asked of only a randomly selected subset of respondents. The low response rate on the SAF resulted primarily not from refusals to participate but from GSS interviewers failing to get adequate contact information on the randomly selected sib- ling.

Sulloway's evolutionary explanation of birth- order effects depends crucially on this distinction between "functional" and "biological" birth order. However, the finding that biological birth order is unimportant once functional birth order is con- trolled is based on a test of only 29 biologically laterborn scientists who were raised as "func- tional" firstborns (Sulloway 1996:465). Given the high rates of infant mortality and other sources of childhood instability in the eras predominantly

all respondents with step- or half-siblings from our sample; by doing this, we attempt to maximize the correspondence between bio- logical birth order and functional birth or- der.1° In addition, we also tested other oper- ationalizations of birth order: the number of older siblings (birth rank); the number of older siblings divided by the total number of siblings (relative birth rank); the number of older brothers; the number of older children of the same sex; and a trichotomous variable differentiating firstborns, middleborns, and lastborns. None of these alternative measures yielded patterns substantively different from those presented here.
Measures of Social Attitudes

The GSS contains a large number of ques- tions on social attitudes. We sought to test a respondent's social attitudes in six domains:

(1) political identification; (2) opposition to liberal social movements; views on (3) race and (4) gender; (5) support for existing au- thority; and (6) "tough-mindedness." Initially, we chose 24 items and scales that rep-

represented in Sulloway's sample, it is almost cer- tain that more than 29 cases of laterborns raised as "functional" firstborns exist among these sci- entists. This raises the possibility that the "func- tional" birth status of some sample members was investigated more thoroughly than others, precisely because they otherwise would have been exceptions to the study's general findings. At least 3 of the 29 scientists-Louis Agassiz, Georges Cuvier, and Tycho Brahe-are probably among the most eminent 1 percent of all the scientists Sulloway examined, and they would have stood out as prominent exceptions to the overall find- ings had they remained coded as laterborns. If the probability of being recoded as a "functional" firstborn is correlated with the probability of oth- erwise being a (prominent) deviation from the overall pattern, then we should not be surprised that Sulloway was able to report strongly signifi- cant differences (p < .001) between these 29 sci- entists and other biological laterborns.

lo In Sulloway's historical data, one major cause of discrepancies between "biological" and "functional" birth order is infant mortality. We assume that measurement problems due to infant mortality are greatly diminished in a contempo- rary sample such as ours. Moreover, restricting the sample to only those respondents who reported having no deceased siblings does not af- fect the general pattern of results we report.

resented these broad headings and were also consistent with many of the specific asser- tions and historical examples presented in Sulloway's research." These dependent vari- ables are described in the text below, and their means, standard deviations, and metrics are presented in Appendix A. The analysis of these variables comprises the primary focus of this paper. We later expanded our inquiry to include all (202) attitudinal items that may distinguish respondents on the basis of con- servatism, support for existing authority, or "tough-mindedness." We later summarize the results of this broader inquiry.

Political self-identification. We measure a respondent's political self-identification with an item asking respondents to place them- selves on a seven-point scale ranging from "extremely liberal'' to "extremely conserva- tive," and a similar item with a scale ranging from "strong Democrat" to "strong Republi- can." We also use support for George Bush in the 1992 Presidential election as an indicator.

Opposition to liberal social movements.

Consistent with the long-standing stereotype about the "liberal" laterborn and "conserva- tive" firstborn, Sulloway provides historical evidence that laterborns have been both over- represented among prominent liberal social reformers and underrepresented among those who have resisted liberal social change. We test whether firstborns are less supportive than laterborns of a variety of traditionally liberal movements and causes: abortion rights, environmentalism, free speech, social

l1 Alphas for scales typically ranged between .70 and .90 (available from the authors on re- quest). The scales with the lowest alphas (mea- sures of individualistic views about racial equal- ity and support for English-only laws) are based on variables that were presented together as a set on the GSS. We tried to use existing GSS scales whenever possible to avoid the possible criticism that our scale construction decisions were (unin- tentionally or otherwise) biased against the hy- pothesis. Certainly, low alphas can lead to attenu- ated estimates of real effects. However, regard- ing the scales with the lowest alphas: (I) The es- timated effects of birth order on these scales are opposite the predicted direction; (2) as will be shown in Table 2, other independent variables did exert a significant effect on these scales; and (3) when we look at the individual items used to cre- ate the scales, we observe no significant effects for any item in the predicted direction.

welfare programs, the effort to decriminalize marijuana, the right-to-die movement, and animal rights.

Conservative views on race and gender.

Sulloway asserts that firstborns should be more resistant than laterborns to initiatives for racial and ethnic equality. Of all the so- cial reform movements studied by Sulloway, the most disproportionate number of later- borns is observed among participants in the abolitionist movement (Sulloway 1996: 152). Elsewhere, Sulloway describes first- borns as "particularly inclined toward rac- ism" (p. 286; also see Lieberman and Reynolds 1978; Sherwood and Nataupsky 1968). Because this latter conclusion is drawn from a sample that is almost exclu- sively white, we restrict our sample to whites when testing for birth-order differences in beliefs about racial equality. We measure at- titudes toward racial reforms both in terms of attitudes toward African Americans and attitudes toward immigration. For the former, we test whether the respondent believes the government is too generous to blacks, that whites should be able to segregate them- selves from blacks if they wish, and that the economic differences between blacks and whites are caused by racial differences in "in-born ability" and "motivation or will power." For attitudes toward immigration, we test a respondent's opposition to provid- ing benefits to immigrants and support for laws requiring government documents to be only in English.

To support his contention that firstborns have more traditional beliefs about gender than laterborns, Sulloway (1996:154-58) presents evidence from a sample of female American conservatives and reformers. He notes also that Anita Bryant and Phyllis Schafly are firstborn women who have been outspoken in their support of traditional gen- der roles, while Susan B. Anthony, Elizabeth Cady Stanton, and other suffragette leaders were predominantly laterborns (1996:154). We examine respondent support for a tradi- tional division of labor between spouses, in which the husband is the breadwinner and the wife stays home to look after their fam- ily. We also examine respondent beliefs about the appropriateness of mothers remain- ing in the labor force and the appropriateness of women seeking political offices.

Support for existing authority. As noted above, the idea that firstborns identify more strongly with authority than laterborns goes back at least to Adler (1928). Sulloway sup- ports this idea with examples from both the French and American revolutions in which laterborns rebelled against monarchies while their firstborn siblings or sons remained staunch royalists. We measure attitudes to- ward obedience with an item asking how im- portant the respondent believes it is to teach a child to obey, as compared to teaching the child to work hard, help others, be well- liked, or think for herlhimself. Support for existing authority is measured by a respon- dent's confidence in those running major so- cial institutions, including banks, the armed forces, organized religion, and Congress. We also use respondent patriotism ("How proud are you to be an American?") as an indicator of support for existing authority.

"Tough-mindedness."Following Eysenck's (1954) two-dimensional model of political attitudes, Sulloway argues that firstborns not only are more politically conservative but also are more "tough-minded" (i.e., less compassionate and more aggressive in their assessment of human affairs). As mentioned, Sulloway presents evidence that firstborns were disproportionately likely among French deputies to vote to execute Louis XVI, while laterborns voted to spare the King's life. Accordingly, we use a respondent's support for capital punishment to measure tough-mindedness. In addition, we measure tough-mindedness with an item asking whether the respondent believes that the justice system should be harsher in its sentencing of criminals, and an item asking whether the respondent feels that "it is sometimes necessary to discipline a child with a good, hard spanking."
Other Independent Variables

As noted above, our research seeks not only to examine the effect of birth order on social attitudes, but also to compare the influence of birth order on social attitudes with that of other variables that have received more sus- tained attention from sociologists. Conse- quently, after examining the bivariate rela- tionships between birth order and our mea- sures of social attitudes, we look at birth or- der in the context of multiple regression models that also include sex, age, race (coded as dummy variables for blacks and other nonwhites), parents' education, and sibship size. Because each of these variables has been posited to affect social attitudes di- rectly and because each may be correlated with birth order,I2 including these variables as controls also permits better estimates of actual birth-order effects.

In addition, subsequent models employ more stringent controls. Because birth-order theories typically place emphasis on child- hood environment, our next model adds con- trols for parents' occupational prestige (us- ing recent recodings by Hauser and Warren [1997]), parents' marital status, the loss of a parent to death before age 16, childhood re- ligion, and the region of the country in which the respondent was raised. To account for the possibility that observed birth-order effects may be caused by birth-order differences in achievement, our final model adds controls for the respondent's education and occupa- tional prestige.
RESULTS

Table 1 presents estimates of the effect of birth order on the different measures of so- cial attitudes. All measures are coded so that positive coefficients are consistent with the hypothesis that firstborns are more conserva- tive, supportive of authority, and "tough- minded" than laterborns. As described above, we provide results for four different models. Model 1 is the bivariate regression of the at- titudinal measures on birth order. Model 2 holds constant the respondent's age, sex, race, sibship size, and parents' education. Model 3 adds controls for parents' occupa-

l2 Ernst and Angst (1983) therefore advise that credible birth-order research must, at a minimum, control for the respondent's sibship size and so- cioeconomic background; Hare and Price (1969) counsel that birth-order studies also should con- trol for age. At the same time, we remind the reader that with cross-sectional data we cannot distinguish age effects from cohort effects. Sieff (1990) suggests that controlling for sex is impor- tant because laterborn children may be dispropor- tionately female. Similarly, controlling for race is important because fertility varies across racial and ethnic groups.

Table 1. Coefficients from the OLS and Logistic Regression of Selected hleasures of Social Attitudes on Birth Order: U.S. Adults with Full Siblings, GSS, 1994

Model 4
Model 1     Model 2     Model 3     Model 4     Results in Predicted
Social Attitude Measure     Coef.     S.E.     Coef.     S.E.     Coef.     S.E.     Coef.     S.E.     Direction?
--                
Political Identification                 
Identifies self as:                 
Conservat~ve     -.057     (.069)                 No
Republican     ,092 (.loo)                 Yes
Bush supporter ~n 1992     .220* (.104)                 Yes
Opposition to Liberal Movements                 
Opposes:                 
Abortion rights     ,154 (.155)                 Yes
Assisted suicide laws     ,039 (.139)                 Yes
Legalization of marijuana     -. 167 (.142)                 No
Animal rights     -.076     (.134)                 No
Environmental movement     ,059 (.056)                 Yes
Free speech     -.038     (.182)                 Yes
Social welfare programs     ,080 (.043)                 Yes
Resistance ro Racial Reforms                 
Thinks government is too generous to blacks "     ,083 (.059)                 Yes
Excluding blacks is OK h.003 (.062)                 Yes
Racial inequality due to     -.I42     (.076)                 No
blacks' lack of ability1 will power "                 
Against benefits for     -.008     (.149)                 
immigrants a,b                 
Supports English-only laws a -.079     (.074)                 
Belief in Trudlrional Gender Roles                 
Against mothers working     ,005 (.081)                 No
Supports traditional division     ,008 (.077)                 Yes
of labor between spouses                 
Against women in politics     -.019     (.152)                 Yes
Support for Existing Authority                 
Children should obey     -.095     (.110)                 Yes
Trusts social institutions     -.005     (.100)                 Yes
Patriotism     -.I56     (.136)                 No
"Tough-iMindednessn                 
Tough on crimeC     -.35OU (.155)                 
Supports:                 
Capital punishmentC     -.I72     (.124)                 No
Corporal punishmentb     -.001     (.112)                 Yes

Notes: Numbers in parentheses are standard errors. Model 1 is the bivariate regression. Model 2 controls for sibship size, age, sex, race, and parents' education. Model 3 adds controls for parents' marital status and occupational prestige, parental loss, respondents' religion, and region where respondent was raised. Model 4 adds controls for current income and education. Positive coefficients are consistent with the hypothesis that firstborns are more conservative, supportive of authority, and "tough-minded" than laterborns. Number of cases range from 595 to 1,894 (see Appendix A). a Racial equality items include only white respondents.

Ordered logistic regression used for ordinal dependent variables.

Binary logistic regression used for dichotomous dependent variables 'p < .05 "p < .O1 *"*p< ,001 (two-tailed tests)

tional prestige and marital status, parental loss prior to age 16, respondent's religion, and the region of the country where the re- spondent was raised. Model 4 adds the respondent's current family income and edu- cation.

If we look first at the results of the bivari- ate regression (Model I), the data do not ap- pear to support the hypothesis. The observed effect of birth order is indistinguishable from chance for 22 out of the 24 measures of so- cial attitudes, and only one of the two sig- nificant effects is in the predicted direction: firstborns were more likely than laterborns to have supported Bush in 1992.13 Meanwhile, contradicting Sulloway's findings about the "tough-mindedness" of laterborns, firstborn respondents are significantly more likely than laterborns to believe that the nation's courts are too harsh with criminals.

Introducing control variables in Models 2, 3, and 4 does not improve support for the hypothesis. In each of these models, signifi- cant effects are observed for only 3 of the 24 measures of social attitudes, and all sig- nificant effects are opposite the predicted direction. After controls are added the con- nection between birth order and support for Bush is no longer significant, while the rela- tionship between birth order and the ad- equacy of criminal sentencing remains. In addition, laterborns are more likely than firstborns to support capital punishment, and laterborns also are significantly more patriotic.

Many of the nonsignificant estimates also run opposite the predicted direction. In the bivariate regressions, fewer than half of the estimates are in the predicted direction (9 of 24), while in Model 4 slightly more than half of the estimates are in the predicted direc- tion (14 of 24). Indeed, when asked to place themselves on a liberal-conservative continuum, firstborns identified themselves as more liberal than laterborns, although this result was not significant. Moreover, while it is plausible that systematic birth-order ef- fects could exist in some domains of social attitudes but not in others, we observed in- consistent results within each of the six broad types of attitudes we defined.

l3 In our data, firstborns were neither over- nor underrepresented among Perot supporters.

Comparing Birth Order with Other Independent Variables

Above we noted that Sulloway's work poses a direct challenge to sociologists by assert- ing that birth order is a more important de- terminant of social attitudes than is gender or social class. In Table 2, we compare the effect on social attitudes of being firstborn with the effect of the control variables used in Model 2: sibship size, age, sex, race, and parents' education (which we use here as a proxy for social class). While birth order is significantly associated with only 3 out of the 24 measures of social attitudes (all oppo- site the predicted direction), Table 2 shows that each of the other independent variables is significantly associated with at least half

(12) of the dependent variables. When we compare standardized coefficients (not shown), we find that parents' education is a more powerful predictor of attitudes than birth order for 22 of 24 variables (92 per- cent), race for 16 of 19 (84 percent); sex for 19 of 24 (79 percent); age for 18 of 24 (75 percent); and sibship size for 17 of 24 (71 percent). Consequently, our data strongly suggest birth order is not as important for understanding attitudinal differences as are these other variables.

Several implications of these results are worth highlighting. First, by observing so many significant effects for other variables, we gain confidence that the null results we observe for birth order are not due to inad- equate or unreliable measures of social atti- tudes. Second, the linchpin of Sulloway's claim that his Darwinian approach supplants conventional sociological analyses is his finding that birth order is 14 times more powerful a predictor of attitudes than social class, but our data indicate that parents' edu- cation actually influences attitudes much more strongly than birth order. Third, birth order appears even less valuable for predict- ing social attitudes than sibship size, which suggests that, at least relative to historical data, Sulloway overstates the influence of birth order, not only compared to sociologi- cal mainstays like gender, race, and class, but also compared to other aspects of family configuration.

Finally, we should note that while the in- dependent variables other than birth order

Table 2. Coefficients from the OLS and Logistic Regression of Social Attitude hleasures on Birth
Order and Other Independent Variables: U.S. Adults with Full Siblings, GSS 1994

Social Attitude Measure

Political Identification

Identifies self as: Conservative Republican Bush supporter in 1992

Opposition to Liberal Movements

Opposes: Abortion Assisted suicide laws Legalization of marijuana Animal rightsC Environmental movement Free speech Social welfare programs

Resistance to Racial Reforms

Thinks government is too generous to blacks Excluding blacks is OK

Racial inequality due to blacks lack of abilitylwill power Against benefits for immigrants b.

Supports English-only lawsb

Belief in Traditional Gender Roles

Against mothers working

Supports traditional division of labor between spouses Against women in politics
Independent Variables "
Sibship         Parents'
Firstborn     Size     Age     Female     Black     Education

(Table 2 cor~tinlled or1 r7err page) (Table 2 continued frotn prevlous page)

Independent Variables a
Social Attitude Measure     Firstborn     Sibship Size     Ape     Female     Black     Parents' Education
Supportfor Existing Authorit? Children should obey                         

Trusts social institutions

Patriotism

"Tough-Mindedness"

Tough on crimed

Supports: Capltal punishment

Corporal punishment

Notes: Numbers in parentheses are standard errors. Model includes an additional control for other non- whites. Number of cases range from 595 to 1,894 (see Appendix A).

Tirstborn is measured with a dummy variable (1 = yes); sibship size is number of siblings; age is in years; female is a dummy variable (1 = yes); black is a dummy variable (1 = yes); and parents' education is the highest degree earned by the respondent's most educated parent, measured by a categorical variable (from 0 = no high school degree to 4 = graduate degree).

Racial equality items include only white respondents.

Ordered logistic regression used for ordinal dependent variables. *Binary logistic regression used for d~chotomous dependent variables 'p<.05 ""p<.Ol ***p < ,001 (two-tailed tests)

significantly affect a broad range of social which have tended to trade on broad charac- attitudes, only age does so in a seemingly terizations of firstborns as "conservative" consistent fashion (in terms of the direction and "identifying with authority" (Sulloway of coefficients). In contrast, respondents with 1996; also see Adler 1928; Toman 1993). Put well-educated parents tend to be more liberal simply, such labels may not be fine-grained on most attitudinal measures than do those enough to map a thoroughly consistent rela- with less-educated parents, yet they are also tionship between a single independent vari- more likely to identify themselves as Repub- able and opinions on the complex social is- licans. Females and blacks tend to be more sues of contemporary society. liberal than males and whites, but among other things, females are more likely than
Comparisons within Families

males to oppose the legalization of marijuana and blacks are more likely than whites to be- Like most other birth-order studies, Tables 1 lieve that spanking children is sometimes and 2 compare persons from different fami- necessary. Social scientists have known lies, while birth-order theories posit a pro- about such apparent inconsistencies for a cess of differentiation that takes place within long time, and considerable work has gone families. As described above, the SAF data into explicating the nuances of attitude for- allow for the comparison of GSS respondents mation across various social divisions with a randomly selected sibling. These in- (Brooks and Manza 1997; Inglehart 1990). terviews included 9 of the 24 measures of Yet these results may point to possible, social attitudes examined in Tables 1 and 2, deeper problems with birth-order theories, including at least one measure from each of

Table 3. Within-Family Comparisons of Firstborns and Laterborns on Selected Social Attitude Mea- sures: Scores on Items Included in both the GSS and the Study of American Families (SAF)

Social Attitude Measure

Identifies self as: Conservative Republican

Opposes free speech Excluding blacks is OK a Racial inequality due to blacks'

lack of abilitylwill power a Against women in politics Children should obey Trusts social institutions Supports corporal punishment
Mean Score     Mean Score     Difference     Difference     Number
for     for     in Predicted     Significant     of
Firstborns     Laterborns     Direction?     at p < .05?     Cases
3.19     3.21     No     No     492
3.22     3.22     No     No     483
2.00     1.81     Yes     No     274
.7 1     .70     Yes     No     249
1.60     1.78     No     No     382
1.23     1.26     No     No     286
1.56     1.41     Yes     No     345
4.21     4.25     No     No     315
1.78     1.82     No     No     319

Notes: Significance tests use two-tailed t-tests for paired observations. Variables are coded such that the hypothesis predicts the means for firstborns to be higher than laterborn means.

Weans are for white respondents only.

the six broad types we defined. Using the SAFIGSS data, we are able to test for birth- order differences within families; however, we remind readers of the low response rate and relatively narrow range of attitudinal items available in these data.

Table 3 compares the mean responses of firstborn SAFIGSS respondents with the mean responses of their laterborn siblings. The means of most of the variables difTer only slightly between firstborns and later- borns, and none of the observed differences is statistically significant.'%lthough the lack of significant results here may be par- tially attributed to the relatively small sample size, most of the observed differences are op- posite the predicted direction. Only the dif- ferences in respondent attitudes toward free speech, the exclusion of blacks, and the im- portance of obedience in childrearing are consistent with the hypothesis (3 of 9, or 33 percent). These results are consistent with Ernst and Angst's (1983) observation that in-

14The mean differences between firstborns and laterborns were also not significant when Model 2 controls were added. To test whether a few out- lying sibling pairs influenced our results, we compared the number of pairs in which firstborn siblings gave the more conservative response to the number of pairs in which the laterborn was more conservative, and this difference was not significant for any dependent variable.

ter- and intrafamilial data tend to yield simi- lar findings on the effects of birth order, and that when results differ studies using intra- familial data tend to be less likely to observe significant birth-order effects than compari- sons of children from different families.
Examining Additional GSS Items

Because the attitudinal variables discussed above represent only a portion of the attitu- dinal items available inthe 1994 GSS, it is possible that the results for the items we present are less favorable to hypothesized birth-order effects than results using other items we could have chosen. We tested whether our selection was inadvertently bi- ased by running regressions of all 202 GSS items that could be considered measures of conservatism, support for existing authority, and/or "tough-rnindednes~."'~ Table 4 sum- marizes the results of the bivariate regres-

-

l5 Even here, of course, researchers may dis- agree about which items they consider to be indi- cators of conservatism, support for authority, or "tough-mindedness." Of the 202 items we exam- ined, 197 (97 percent) were positively correlated with respondent self-identification as conserva- tive (albeit not all significantly). The 5 items that were not positively correlated with conservatice self-identification had been selected as indicators of supportive attitudes toward authority.

Table 4. Frequency Distribution Summarizing the Results from the OLS and Logistic Regressions of Relevant GSS Social Attitude Measures on Birth Order and Other Selected Independent
Variables

Number of GSS Items in Predicted

Total

Predicted Number of Direction Variable Direction GSS Items N Percent

Tests of tlze Mairt Effect of Birth Order on Social Attitude Measures

Firstborn (Model 1)" Conservative 202 100 (49.5) Firstborn (Model 2)b Conservative 202 117 (57.9) Firstborn (Model 4)C Conservative 202 104 (51.5)

Coinparing the Effects of Model 2 Variables on Social Attitude Measures

Firstborn Conservative 202 117 (57.9)

Age Conservative 202 154 (76.2) Parents' education Liberal 202 150 (74.3) Fernale Liberal 202 139 (68.8) Black Liberal 148 84 (56.8) Sibship size Conservative 202 115 (56.9)

Number of

Number of Significant Items Items Significant in Predicted at p < .05 Direction

N Percent N Percent
10     (5.0)     4     (2.0)
16     (7.9)     8     (4.0)
15     (7.4)     5     (2.5)

16 (7.9) 8 (4.0) 120 (59.4) 105 (52.0) 106 (52.5) 89 (44.1) 89 (44.1) 64 (31.7) 54 (36.5) 36 (24.3) 38 (18.9) 30 (14.9)

Note: Numbers in parentheses are percentages. Relevant GSS measures include all attitudinal items con- sidered to test respondents' conservatism, support for existing authority, and/or "tough-mindedness" (a com- plete list is available from the author on request).

a Model 1 is the bivariate regression model.

Model 2 controls for sibship size, age, sex, parents' education, and race.

Model 4, the full model, controls for age, sibship size, sex, race, parents' education, parents' occupa-

tional prestige, parental loss, childhood religion, region where respondent was raised, respondents' income,

and respondents' education.

sions of these items on birth order, as well as the results of regressions using the controls employed in Models 2 and 4 (from Table 1). Table 4 shows how many results were in the predicted direction, how many were signifi- cant, and how many of the significant results were as predicted.

Mirroring the results for the previous tables, we find little support for the theory that firstborns are more conservative than laterborns. The number of significant effects is exactly what chance predicts for the bivari- ate regressions, and only slightly more than half for both models that add controls. Im- portantly, in none of the models are more than half of the significant effects in the predicted direction, suggesting that even if there were some small number of real birth-order effects in the data that are not simply due to chance, theories such as Sulloway's offer no insight for predicting their direction. Because the measures of social attitudes are not indepen- dent of one another, we cannot use standard binomial confidence intervals to calculate the p-value of a given deviation away from the 5 percent of the items being significant or away from 50 percent of the coefficients being in the predicted direction. Even so, the number of variables in the predicted direction for Models 1 and 4 are both almost exactly 50 percent; that somewhat more than 50 percent of the items are in the correct direction for Model 2 is less convincing given the reversal in the direction of coefficients when controls are either added or dropped.

The possibility of strong birth-order effects in these data is thrown into even greater doubt when we look at the second panel of Table 4, which compares the effect of birth order on all GSS attitudinal items with the independent variables included as controls in Model 2. Although above we questioned the

adequacy of broad labels like "conservative" or "liberal" for understanding the relation- ship between an independent variable and at- titudes on complex social issues, here for purposes of comparison we ascribe a pre- dicted direction to each variable based on previous research (e.g., Huber and Form 1973; Hunt 1996; Schuman, Steeh, and Bobo 1985). Looking first at the results for respon- dents' age and parents' education, we see that approximately three-quarters of all the esti- mates are in the direction we assigned, over half the estimates are significant, and over 40 percent of the estimates are significant as predicted (as compared to only 4 percent for birth order). Sex is significantly associated with 44 percent of the GSS items, and sig- nificant as predicted for 32 percent. Even for the control variable that exerts the weakest apparent influence on social attitudes-sib- ship size-we still observe twice as many significant effects as for birth order and three times as many effects that are significant in the predicted direction.
Interaction Effects

Up to this point, we have examined main ef- fects of birth order and social attitudes; we find no evidence of the profound effects that Sulloway and others have predicted. Yet a key difference between Sulloway's and most earlier birth-order theories is that Sulloway argues that birth order is also influential through its interaction with other indepen- dent variables.16 Perhaps most plausibly, he proposes an interaction effect between birth order and age spacing between siblings (Sulloway 1996: 133-36). Drawing upon Hamilton's (1964) biological theory of inclu- sive fitness and kin selection, Sulloway claims that the most pronounced attitudinal differences occur when there is a moderate gap (2 to 5 years) between adjacent sib- lings.17 Consequently, firstborns who have a

l6 The 1994 GSS does not contain information that would have allowed us to test two other in- teraction effects discussed by Sulloway-those between birth order and shyness, and between birth order and childhood conflict between respondents and their parents.

l7 Using Hamilton's (1963) theory, Sulloway argues that siblings who are more closely or dis- tantly spaced are less costly (in an evolutionary

sibling close to their age, or who only have much younger siblings, are expected to be less conservative than firstborns with a mod- erate gap to their next oldest sibling.

Table 5 presents tests for each of these pos- sibilities. For firstborns with closely spaced younger siblings, the number of significant results is approximately consistent with what chance would predict (5.9 percent), and con- siderably less than half of the significant esti- mates are in the predicted direction (2 of 12). The results are similar for firstborns who have only distantly spaced younger siblings. Clearly, then, the data do not support the proposition that firstborns are more liberal when there is a small or large gap between them and their next oldest sibling.

In addition, Sulloway presents evidence of interaction effects between birth order and social class, which we also test in Table

5.18 His model of social attitudes reports a significant interaction, suggesting that up- per-class firstborns are more conservative than lower-class firstborns (p. 507).19Yet, in

sense) than those who are moderately spaced, and that as a result firstborns should feel a greater ri- valry with moderately spaced siblings than with other siblings. Sulloway also claims that laterborn children born after a long gap are much like first- borns and therefore tend toward conservativism. These claims also were not supported by tests us- ing the 202 measures.

Is In a footnote, Sulloway (1996:507) describes a model of social attitudes that includes an inter- action between birth rank and sibship size. He provides no justification for why such an interac- tion should be included other than that the result is significant using stepwise methods in his sample. Because birth rank and sibship size are highly correlated (r = .64 in the GSS data), using the product of birth rank and sibship size is very much like using the square of sibship size, which makes more theoretical sense because it raises the possibility that the marginal effect of siblings on attitudes changes with each additional sibling (cf. Downey 1995). We find no evidence of an inter- action effect for birth rank x sibship size in the GSS data once the square of sibship size is con- trolled.

l9 The text of Born to Rebel is ambiguous about the direction of this interaction effect; we thank Frank Sulloway (personal communication) for clarifying this matter. Sulloway alternatively de- scribes the interaction of birth order and social class on conservatism in terms of the birth-order effects being much more pronounced for upper-

Table 5. Frequency Distribution Summarizing the Results of Relevant GSS Social Attitude Measures on Interactions between Birth Order and Other Selected Independent Variables

Number of GSS Items Total in Predicted

Number of Direction Variable Direction GSS Items N Percent

Predicted

Tests of Effect of Birth Order x Sibling Spacing on Social Attitude Measures

Firstborn x Closely Less 202 86 (42.6) spaced younger sibling Conservative

Firstborn x Distantly Less 202 77 (38.1) spaced younger sibling Conservative

Tests of Effect of Birth Order x Other Variable on Social Attitude Measures.

Firstborn x Parents' education

Firstborn x Parents' occupational prestige

Firstborn x Parents' occupational prestige x Parental loss

More 202 106 (52.5) Conservative a

Less 202 102 (50.5) Conservative a

Less 202 116 (57.4) Conservative

Number of

Number of Significant Items Items Significant in Predicted at p < .05 Direction

N Percent N Percent

12 (5.9) 2 (1.0) 4 (2.0) 2 (1.0)

5 (2.5) 3 (1.5) 10 (5.0) 7 (3.5) 3 (1.5) 2 (1.0)

Nore: Numbers in parentheses are percentages. Each interaction in the table was tested separately. Mod- els also control for age, sex, race, sibship size, parents' education, parents' occupational prestige, and pa- rental loss. Relevant GSS measures include all attitudinal items considered to test respondents' conserva- tism, support for existing authority, and/or "tough-mindedness" (a complete list is available from the author on request).

While all other models in this study derive from predictions that birth order affects conservatism and tough-mindedness in the same direction, the prediction here is that upper-class firstborns are more conser- vative, but less tough-minded, than lower-class firstborns.

his model of the behavior of deputies during the French Revolution, he finds that upper- class firstborns were less tough-minded than lower-class firstborns (p. 323). Measuring social class both by parents' education and occupational prestige, we find no evidence of these patterns: The number of significant effects and the number of estimates in the predicted direction are fully in line with what chance would predict. Sulloway also argues that social class and birth order me- diate the effect of parental loss through a

and middle-class sample members than for lower- class members. When we restrict the sample to only those respondents whose parent's occupational prestige is above the sample median, we find no additional evidence for the theory: Ex- cluding measures of tough-mindedness, estimates for 106 of 193 (54.9 percent) items are in the pre- dicted direction, 17 (8.8 percent) are statistically significant at p < .05, and 7 (3.6 percent) are sig- nificant in the predicted direction.

three-way interaction effect (pp. 136-45). Lower-class firstborns, thrust into the role of surrogate parents, become even more doggedly conservative, while upper-class firstborns, who often do not have to bear this burden, become somewhat more liberal. This proposition is also not supported by our data: Significant three-way interaction effects are in the predicted direction for only 4 of 202 items when social class is measured by parents' education, and for only 2 of 202 measures when class is mea- sured by parents' occupational prestige.
CONCLUSION

Birth order has enjoyed one of the longest pedigrees of any variable in the social sci- ences, but most previous studies have looked at its effects on either personality or achieve- ment. Sulloway (1996) gives new reasons why sociologists should be interested in birth order, as he provides the first extensive theo- retical and empirical consideration of the re- lationship between it and social attitudes (and, in turn, behavior). Using quantitative historical data, Sulloway finds a pronounced tendency for firstborns to be more conserva- tive than laterborns, and he claims this dif- ference has led laterborns to be more likely than firstborns to embrace liberal social movements early. Moreover, because he maintains that these birth-order differences are ultimately rooted in evolved adaptations, he asserts that his findings apply not just to historical elites but throughout any society where children of different ages are raised together.

We have used contemporary data to test Sulloway's hypothesis that firstborns are more conservative, supportive of existing au- thority, and "tough-minded" than laterborns. We find no evidence that birth order affects social attitudes in the way Sulloway predicts. Instead, our data strongly contradict his as- sertion that birth order is a better predictor of social attitudes than are gender, class, or race.

How can we reconcile the discrepancy be- tween our findings and Sulloway's? One pos- sibility is that substantial birth-order effects on social attitudes may have existed in the historical eras that Sulloway studies but do not exist today. Primogeniture may have en- couraged conservatism among firstborns by binding them to their ancestral properties, while encouraging liberalism among later- borns by allowing them greater freedom to travel and have a broader range of experi- ences. The privileged position of firstborns has been long in decline, as evidenced, espe- cially over this century, by the steady re- placement of inheritance practices biased to- ward firstborns with practices that divide wealth equally among children (a process which has itself been the topic of socio- biological explanation [Hrdy and Judge 19931). In contemporary society, Steelman and Powell (1991) find that laterborn chil- dren may often have the upper hand in re- ceiving parental economic investments be- cause they are born at a time when their families are more likely to be economically secure.

Another possibility is that birth-order ef- fects may be confined to elites in both his-

AMERICAN SOCIOLOGICAL REVIEW

torical and contemporary times.20 Sulloway's sample includes members of lower social strata only when they rose from their condi- tions to make a place for themselves in his- tory. Practices such as primogeniture certainly affected wealthy families more than peasant families who had no lands or wealth to pass on, and birth-order differences in education and the opportunity to enter sci- ence or politics may also manifest them- selves most strongly among elites. We have no way of comparing the relative birth-order effects of elites and peasants in eighteenth- and nineteenth-century society, but we did test whether birth-order differences exist among contemporary economic or intellec- tual elites by restricting our sample first to respondents with upper-class parents and then to respondents who scored in the top 15 percent on the cognitive tests included on the 1994 GSS.21 In neither instance did we find any evidence that firstborns were more con- servative than laterborns.

Our failure to replicate Sulloway's find- ings may also point to problems in his analy- sis. Although reminiscent of some of Sorokin's (e.g., 1928) work, the strategy of using biographies and historians' ratings for testing general propositions about individual behavior is unusual, and the lack of estab- lished procedures for this method leaves open several potential problems. One prob- lem is that the ratings from historians were all obtained through in-person interviews conducted by Sulloway well after he began constructing his arguments about birth order, and although he used a standardized rating

20 Sulloway (1996:416-18) discounts the pos- sibility that his results are confined either to past eras or to elites. He finds that the birth-order ef- fects in his data do not significantly vary over time, and that the effects exist among 115 scien- tists and historical figures in his sample that were born after 1900. Concerning elites, Sulloway (1997:382) also argues that birth-order effects turn up in analyses of the "most obscure" scien- tists and historical figures in his data. Here, how- ever, it seems likely to us that the most obscure members of his sample are still more accurately characterized as "elite" than "typical" members of their respective societies.

21 We also found no significant birth-order dif- ferences among those who labeled themselves "extremely liberal" or "extremely conservative."

scale, the interview procedures themselves are not well-documented. As a result, the possibility of substantial interviewer effects cannot be ruled out. Another problem is that Sulloway's reliance on biographies results in a disturbing amount of missing data that many social scientists would not accept. For the eight variables in one of his key models, Sulloway has complete records on fewer than 200 of the 3,890 scientists in his sample. In addition, some procedures in Sulloway's quantitative analysis are questionable, espe- cially his use of stepwise regression-like pro- cedures to derive all of his main multivariate models.22 Because the inclusion or exclusion of independent variables is based on empiri- cal fit rather than on substantive justification, Sulloway's theoretical arguments often have the character of post hoc attempts to explain the observed results.

In any case, the true cause of the discrep- ancy between our findings and ~ulloway's can be revealed only through additional re- search. Surely, a study that focuses squarely on social behavior would be instructive. This said, it is important to remember that Sulloway's assertions about birth order and behavior are rooted in his claims about birth order's effect on social attitudes and on per- sonality. Regarding the latter, preliminary work by Hauser et al. (1997) compares sib- lings in the Wisconsin Longitudinal Study and finds no evidence for hypothesized ef- fects of birth order on personality. If further work supports our findings on social attitudes and Hauser et al.'s findings on personality, it would not be the first time that enthusiasm about birth order has given way to sobriety. Indeed, Somit et al. (1996: 105) liken the re- currence of birth-order theories to a "vam- pire" that cannot be extinguished by either rational argument or contravening evidence.

22 Specifically, Sulloway uses the All Possible Subsets Regression algorithm in the statistical package BMDP, which finds the set of variables that yields the best Mallows' C,,. For missing data, Sulloway uses imputation methods outlined by Rubin (1987), but these should not be con- fused with solvirzg such a pronounced problem. It is also not clear from Sulloway's text when these methods were applied, as he says only that he uses them "occasionally" (p. 453), even though his data would seem to require some accommo- dation for missing data in every model.

We do not contend, however, that birth or- der is unimportant for every question in which sociologists are interested. On the contrary, earlier we list some sociological studies that have documented an eclectic ar- ray of birth-order effects, and we note that, in other societies, whether one is a firstborn or laterborn can exert a profound effect on life outcomes. But the history of birth-order research suggests that those interested in the variable may be well advised to think in terms of modest effects in limited domains and in specific societies. To date, the sweep- ing and universalistic claims that periodi- cally have been offered about birth order have struggled to live up to their hubris in subsequent research. With regard to social at- titudes, our research indicates that broad claims, such as those offered by Sulloway, may be inherently problematic: None of the independent variables we examine is associ- ated with respondents being more liberal or more conservative across every social issue.

An additional benefit of the renewal of in- terest in birth order is that it forces us to think again about the possible influences of intrafamilial variables. At present, sociolo- gists still have only a limited understanding of why people from very similar back- grounds come often to hold very different beliefs about the social world, even when they are of the same sex and age. Without discounting the importance of structural po- sition and of culture, sociologists may profit from further considering the possible mecha- nisms of attitude formation that may operate within families. Birth order is only one of a number of basic family configuration vari- ables (others include sex composition of the sibship, the age spacing of siblings, single- or step-parenthood, and parental age at birth) that have been shown to exert significant ef- fects in some domains. Each merits greater consideration, both separately and as part of one's overall constellation of family traits, as potential determinants of social attitudes. In addition, although its effects were not as strong as that of other variables, our findings indicate that additional consideration of the effects of sibship size on social attitudes may prove lucrative.

The study of family dynamics and its ef- fects increasingly has become a key focal point in debates between advocates of tradi- tional sociological approaches and propo- nents of sociobiological or otherwise evolu- tionary ones. As we noted above, Sulloway's theory has been hailed as a cogent demon- stration of how evolutionary reasoning can be used to explain the influence of family environment. Although we find no evidence supporting Sulloway's theoretical claims, our results cannot be taken as an indictment of evolutionary perspectives generally, as they cannot speak to the many other Darwinian theories that have been offered (including those which purport to explain the effects of race, gender, or social class). Yet our study does illustrate how sociologists can empiri- cally engage evolutionary theories. All too often, sociologists have either ignored socio- biological and evolutionary psychological al- ternatives or critiqued these theories at a dis- tance from actual empirical investigation. By testing Sulloway's specific claims using na- tionally representative data, we hope to prompt additional tests of evolutionary hy- potheses by sociologists-and we underscore the importance of using empirical criteria to evaluate these ideas.

Jeremy Freese is a Ph.D. candidate in the De- partnzent of Sociology at Indiana University. His dissertation assesses the possible contributions of evolutionary psychology to our understanding of

AMERICAN SOCIOLOGICAL REVIEW

familial dynamics. A forthcoming article (with Brian Powell) in the American Journal of Sociol- ogy tests the applicability to contenzporary soci- eties of the Trivers-tVi1lard hypothesis, a widely accepted sociobiological theory on parental in- vestments. He is currently exanzining the relation- ship between patients' socioeconomic status and physicians' medical treatment designs (with Karen Lutfeyj, and racial differences in percep- tions of the etiology and treatment of mental ill- ness (with Jason Schnittker and Brian Powell). He is the 1998 co-recipient (with Jason

Schnittker) of the General Social Survey Gradu- ate Student Paper Award.

Brian Powell is Professor of Sociology at Indi- ana Universiv. His current research focuses on the connection between parental age and the flow of resources to children (with Lala Carr Steelman and Robert Carinij, the effect of teacher union- ization on state standardized test performance (with Lala Carr Steelman and Robert Carini), and racial variation in beliefs about the causes of mental illness (with Jason Schnittker and Jer- emy Freese).

Lala Carr Steelman is Professor of Sociology at the University of South Carolzna. She is collabo- rating with Brian Powell and Robert Carini in studying the relationship between parental age and social, intellectual, and economic invest- ments in children and the link between state teacher unionization rates and standardized test performance. In addition, she is studying racial variations in educational attainment and gender differences in psychological distress (with C. Andre' Mizell).
Appendix A. Means, Standard Deviations, and Descriptions of Variables in the Study
Mean
Variable     Metric     (S.D.)     N
Political Identification
Identifies self as:
Conservative     Placement on liberal-conservative
    scale.
Republican     Party identification.
Bush supporter     Voted for Bush (or would have voted
in 1992     for Bush) in 1992 Presidential election.

Opposition to Liberal Movements

Opposes:

Abortion rights Whether woman should be able to get a legal abortion if (I) her life is in danger, (2) baby may have serious defect, (3) she wants no more children,

(4)
    family can afford no more children,
(5)
    pregnancy is the result of rape,
(6)
    she does not wish to marry father, or 17) she wants abortion for anv reason.

0 = extremely liberal, 3.19 1,887 to 6 = extremely (1.39) conservative.

0 = strong Democrat, 2.90 1,894 to 6=strong Republican. (2.01)

0 = vote for Clinton .37 1,819 or Perot, to 1= vote for Bush. (.48)

0 = should be legal in all 2.29 1,153 these circumstances, to (2.44) 7 = should be illegal in all these circumstances.

(Continued on next page)

(Appendix A cor~tirzzredfroriz previous page)

Mean
Variable     Description     Metric     (S.D.)     N
Assisted suicide     Belief that doctors should not be     0 = disagrees, to 1 = agrees.     .26     1,214
laws     legally allowed to help patients with         (.44)     
    incurable diseases who want to end            
    their lives.            
Legalization of     Belief that marijuana use should     0 = should be legal, to     .76     1,245
marijuana     remain illegal.     I = should stay illegal.     (.43)     
Animal rights     Belief that animals do not have     0 = strongly disagrees, to     2.39     839
    same moral rights as humans.    4 = strongly agrees.     (1.20)     
Environmental     Whether respondent would be     0 = very willing to do all     1.54     843
movement     willing to (1) pay much higher     of these, to 3 = not at all     (.75)     
    prices, (2) pay much higher taxes,    willing to do any of these.         
    and (3) accept cuts in standard            
    of living to protect environment.            
Free speech     Belief that (I) atheists, (2) com-     0 = all of these should be     2.56     1,13 1
    munists, or (3) homosexuals should    permitted, to 9 = none of     (2.83)     
    not be able to (I) make a speech in    these should be permitted.         
    respondent's community, (2) teach            
    in college or university, and (3) have            
    their books shelved in public library.            
Social welfare     Belief that government has no     0 = disagrees strongly     1.46     1,301
programs     responsibility to (I) improve the     with all items, to 3 = agrees     (.71)     
    living standards of the poor or    strongly with all items.         
    (2) help poor people pay their            
    medical bills and that (3) the            
    government tries to do too much            
    to solve the country's problems.            

Thinks government Belief that the government gives 0 = disagrees strongly, to 1.73 773 is too generous too much attention to blacks and 3 = agrees strongly. (.76) to blacks" should spend less money trying to

improve the conditions of blacks.

Excluding blacks Belief that blacks should not push 0 = disagrees strongly, to .94 813

is OKa themselves where they are not 3 = agrees strongly. (.82) wanted and that whites have a right to keep blacks out of their neighbor- hood if desired.

Racial inequality Belief that blacks tend to have 0 = disagrees with all 1.82 1,102 due to blacks' worse jobs and incomes than whites items, to 4 = agrees with (1.20) lack of ability1 because most blacks (1) have less all items. will powera innate ability and (2) lack motivation,

but not because blacks (3) have less educational opportunities and (4) are discriminated against.

Against benefits Belief that illegal immigrants should 0 = believes immigrants 2.75 703 for immigrantsd (1) not be able to get work permits, are entitled to all these, to (1.19)

(2) not be able to attend public 4 = believes immigrants universities, and (3) not have their are entitled to none of children qualify as American citizens these. if born in the U.S., and (4) belief that legal immigrants should not be able to receive welfare benefits upon entering the country.

(Cot~tinuedon next page)

(Apperidix A continued fronz preisious page)

Variable Descrivtion

Supports English- Belief that English should be exclu- only lawsd sive language used (1) in schools,

(2) on election ballots, and (3) to conduct government business.

Belief in Traditional Gender Roles

Against mothers Belief that mothers should stay

working home (1) when they have child under school age and (2) after youngest child has started school, and that working mothers (3) have weaker relationships with children,

(4)
    cause children to suffer and
(5)
    cause family life to suffer.

Supports traditional Belief that (1) women should look division of labor after home and family, (2) it is more between spouses important for wife to help husband's

career than have one herself, (3) it is

bad if husband stays home and wife

works, and (4) it is better if husband

is achiever outside home and wife

takes care of family.

Against women in Belief that (1) men are emotionally politics better suited for politics than women,

(2) that women should leave running

U.S. up to men, and (3) that respon- dent would not vote for female Presidential candidate from own party.

Support for Existing Authority

Children should obey Evaluation of how important it is to teach a child to obey versus (1) to be well-liked, (2) to think for herlhimself,

(3) to work hard, (4) or to help others.

Trusts social Confidence in persons running

institutions (1) Congress, (2) the executive branch of the government, (3) the U.S. Supreme Court, (4) the military,

(5)
    major companies, (6) banks,
(7)
    education, (8) organized religion, and (9) medicine.

Patriotism How proud respondent is to be an
American.

Tough on crime Belief that courts are not harsh
enough with criminals.

Supports capital Attitude toward death penalty for
punishment those convicted of murder.

Supports corporal Belief that spanking is sometimes
punishment necessary as punishment for
children.

Metric

0= disagrees with all

items, to 3 = agrees with

all items.

0= disagrees with all

items, to 5 = agrees with

all items.

0= disagrees with all

items, to 4 = agrees with

all items.

0= disagrees with all

items, to 3 = agrees with

all items.

0= less important than

all these, to 4 = more

important than all these

0= hardly any confidence in all institutions, to 9 = great deal of confidence in all institutions.

0= not very proud, to 3 = extremely proud.

0= believes courts are too harsh or about right, to I =believes courts are not harsh enough.

0= opposes the death penalty, to 1 = favors the death penalty.

0= disagrees strongly, to 3 = agrees strongly.

Mean IS.D.) N

Note: Almost all variation in the number of available cases is due to the split ballot format of the GSS and the administration of many questionnaire items to a randomly selected subset (113 to 213) of the total sample. *Effects examined for white respondents only.
REFERENCES

Adams, Bert N. 1972. "Birth Order: A Critical Review." Socio~netry35:411-39. Adler, Alfred. 1928. "Characteristics of the First, Second, and Third Child." Children 3: 14-52.

. 1956. The Individual Psychology of Alfred Adler: A Systematic Presentation in Se- lections from His Writings. Edited by Heinz L. Ansbacher and Rowena R. Ansbacher. New York: Basic Books.

Apperly, Frank L. 1939. "A Study of American Rhodes Scholars." Journal of Heredity 30:493

95.

Bayer, Alan E. 1967. "Birth Order and the Attain- ment of the Doctorate: A Test of Economic Hypotheses." American Journal of Sociology 72:540-50.

Berthoud, Joy. 1996. Pecking Order: How Your Place irz the Family Affects Your Personality.

London, England: Gollancz.

Blake, Judith. 1989. Family Size and Achieve- ment. Berkeley, CA: University of California Press.

Blau, Peter M. and Otis Dudley Duncan. 1967. The Anterican Occupational Structure. New York: Wiley.

Bonacich, Phillip, Oscar Grusky, and Mark Peyrot. 1985. "Family Coalitions: A New Ap- proach and Method." Social Psychology Quar- terly 48:42-50.

Brooks, Clem and Jeff Manza. 1997. "Social Cleavages and Political Alignments in U.S. Presidential Elections, 1960 to 1992." American Sociological Review 62:937-46.

Cherlin, Andrew J., P. Lindsay Chase-Lansdale, and Christine McRae. 1998. "Effects of Paren- tal Divorce on Mental Health throughout the Life Course." American Sociological Review 63:239-49.

Clark, E. L. 1918. American Men of Letters, Their Nature and Nurture. New York: Columbia University Press.

Daly, Martin and Margo Wilson. 1988. Homicide. New York: Aldine de Gruyter.

Davis, James Allan and Tom W. Smith. 1994. General Social Surveys 1972-1994: Cunzulatzve Codebook. Chicago, IL: National Opinion Research Center.

Downey, Douglas B. 1995. "When Bigger Is Not Better: Family Size, Parental Resources, and Children's Educational Performance." American Sociological Review 60:746-61.

Downey, Douglas B., Brian Powell, Lala Carr Steelman, and Shana Pribesh. 1999. "Much Ado about Siblings: Change Models, Sibship Size, and Intellectual Development." Anzerican Sociological Review 64: 193-98

Ernst, CCcile and Jules Angst. 1983. Birth Order: Its Influence on Personality. New York: Springer-Verlag.

Eysenck, Hans J. 1954. The Psychology of Poli- tics. London, England: Routledge and Kegan Paul.

Felson, Richard. B. and Natalie Russo. 1988. "Pa- rental Punishment and Sibling Aggression." Social Psychology Quarterly 5 1: 11-18.

Furstenberg, Frank F., Jr. 1990. "Divorce and the American Family." Annual Review of Sociol- ogy 16:379-403.

Galton, Francis. 1874. English Men of Science: Their Nature and Nurture. London, England: Macmillan.

Gini, Corrado. 1915. "Superiority of the Eldest." Journal of Heredity 6:37-39.

Guo, Guang and Leah K. VanWey. 1999. "Sib- ship Size and Intellectual Development: Is the Relationship Causal?" American Sociological Review 64: 169-87.

Greenhalgh, Susan. 1985. "Sexual Stratification: The Other Side of 'Growth with Equity' in East Asia." Population and Developnzent Review 11:265-314.

Gurin, Patricia, Arthur H. Miller, and Gerald Gurin. 1980. "Stratum Identification and Con- sciousness." Social Psychology Quarterly 43: 30-47.

Hare, E. H. and J. S. Price. 1969. "Birth Order and Family Size: Bias Caused by Changes in Birth Rate." British Journal of Psychiatry 115:647-57.

Hamilton, William D. 1964. "The Genetical Evo- lution of Social Behavior: 11." Journal of Theo- retical Biology 7: 17-52.

Harris, Judith Rich. 1998. The Nurture Assump- tion. New York: The Free Press.

Hauser, Robert M., Hsiang-Hui Daphne Kuo, and Randi S. Cartmill. 1997. "Birth Order and Per- sonality among Adult Siblings: Are There Any Effects?" Paper presented at the annual meet- ings of the Population Association of America, March, Washington, DC.

Hauser, Robert M. and Robert D. Mare. 1997. Study of American Families 1994 [MRDF]. Madison, WI: Data and Program Library Ser- vice [distributor].

Hauser, Robert M. and William H. Sewell. 1985. "Birth Order and Educational Attainment in Full Sibships." American Educational Research Journal 22: 1-23.

Hauser, Robert M. and John Robert Warren. 1997. "Socioeconomic Indexes for Occupa- tions: A Review, Update, and Critique." Sociological Methodology 27:177-298.

Hrdy, Sarah Blaffer and Debra S. Judge. 1993. "Darwin and the Puzzle of Primogeniture."

Human Nature 4: 1-45.

Huber, Joan and William H. Form. 1973. Inconze and Ideology: An Analysis of the American Po- litical Fornzula. New York: Free Press.

Hunt, Matthew 0. 1996. "The Individual, Soci- ety, or Both? A Comparison of Black, Latino, and White Beliefs about the Causes of Pov- erty." Social Forces 75:293-322.

Inglehart, Ronald. 1990. Culture Shift in Ad- vanced Industrial Societies. Princeton, NJ: Princeton University Press.

Lieberman, Leonard and Larry T. Reynolds. 1978. "The Debate over Race Revisited: An Empirical Investigation." Phylon 39:333-43.

McLanahan, Sara and Gary Sandefur. 1994.

Growing Up with a Single Parent: What Hurts, What Helps. Cambridge, MA: Harvard Univer- sity Press.

Mercy, James A. and Lala Carr Steelman. 1982. "Familial Influence on the Intellectual Attain- ment of Children." American Sociological Re- view 47:532-42.

Olneck, Michael R. and David B. Bills. 1979. "Family Configuration and Achievement: Ef- fects of Birth Order and Family Size in a Sample of Brothers." Social Psychology Quar- terly 42: 135-48.

Phillips, Meredith. 1999. "Sibship Size and Aca- demic Achievement: What We Now Know and What We Still Need to Know." American So- ciological Review 64: 188-92.

Pinker, Steven. 1997. How the Mind tt'orks. New York: Norton.

Post, David and Suet-Ling Pong. 1998. "The Waning Effect of Sibship Competition on School Attainment in Hong Kong." Conzparative Education Review 42:99-116.

Retherford, Robert D. and William H. Sewell. 1991. "Birth Order and Intelligence: Further Tests of the Confluence Model." American So- ciological Review 56: 14 1-58.

Rubin, Donald B. 1987. Multiple Imputation for Nonresponse in Surveys. New York: John Wiley and Sons.

Schachter, Stanley. 1963. "Birth Order. Emi- nence, and Higher Education." American So- ciological Review 28:757-68.

Schooler, Carmi. 1972. "Birth Order Effects: Not Here, Not Now!" Psychological Bulletin 78:161-175.

Schuman, Howard, Charlotte Steeh, and Lawrence Bobo. 1985. Racial Attitudes in America: Trends and Interpretations. Cambridge, MA: Harvard University Press.

Sherwood, John and Mark Natpusky. 1968. "Pre- dicting the Conclusions of Negro-White Intel- ligence Research from the Biographical Char- acteristics of the Investigator." Journal of Per-

sonality and Social Psychology 8:53-58.

Sieff, Daniela F. 1990. "Explaining Biased Sex Ratios in Human Populations." Current An- thropology 3 1 :25-37.

Somit, Albert, Alan Arwine, and Steven A. Peterson. 1996. Birth Order and Political Be- havior. Lanham, MD: University Press of America.

Sorokin, Pitirim. 1928. Contemporary Sociologi- cal Theories. New York: Harper and Brothers.

Steelman, Lala Carr. 1985. "A Tale of Two Vari- ables: A Review of the Intellectual Conse- quences of Sibship Size and Birth Order." Review of Educational Research 55:353-86.

Steelman, Lala Carr and James A. Mercy. 1980. "Unconfounding the Confluence Model: A Test of Sibship Size and Birth Order Effects on In- telligence." Anzerican Sociological Review 45:571-82.

Steelman, Lala Carr and Brian Powell. 1985. "The Social and Academic Consequences of Birth Order: Real, Artifactual, or Both?Vour- nal of Marriage and the Fatnily 47: 117-24.

. 1991. "Sponsoring the Next Generation: Parental Willingness to Pay for Higher Educa- tion." American Journal of Sociology 96: 1505-

29.

. 1996. "The Family Devalued: The Treat- ment of the Family in Small Groups Litera- ture." Pp. 21 1-36 in Advances in Group Pro- cesses, vol 13, edited by B. Markovsky, M. J. Lovaglia, R. Simon, and E. J. Lawler. Green- wich, CT: JAI Press.

Sulloway, Frank J. 1979. Freud, Biologist of the Mind: Beyond the Psychoanalytic Legend. New York: Basic Books.

, 1995. "Birth Order and Evolutionary Psy- chology: A Meta-Analytic Overview." Psychological Inquiry 6:75-80.

. 1996. Born to Rebel: Birth Order, Fam- ily Dynamics, and Creative Lives. New York: Pantheon.

. 1997. Born to Rebel: Birth Order, Fam- ily Dynamics, and Creative Lives. Paperback ed. New York: Pantheon

Toman, Walter. 1970. "Never Mind Your Horo- scope: Birth Order Rules All.'' Psychology To- day 3:45-48, 68-69. , 1993. Fan~ily Constellation: Its Effects on Personality and Social Behavior. 4th ed. New York: Springer.

Tooby, John and Leda Cosmides. 1992. "The Psy- chological Foundations of Culture." Pp. 19-136 in The Adapted Mind: Evolutionary Ps)>chology and the Generation of Culture, edited by J. H. Barkow, L. Cosmides, and J. Tooby. Oxford, England: Oxford University Press.

Weinstein, Lawrence and Janis Sackhoff. 1987.

"Adler is Right." Bulletin of the Psyclzonomic Society 25:201.

Wolfe, Alan. 1996. "Up from Scientism." New Republic, December 23, pp. 29-35.

Zajonc, R. B. and Gregory B. Markus. 1975. "Birth Order and Intellectual Development." Psyclzological Review 82:74-88.

Zajonc, R. B.. Gregory B. Markus, Michael J. Berbaum, John A. Bargh, and Richard L. Moreland. 1991. "One Justified Criticism Plus Three Flawed Analyses Equals Two Unwar- ranted Conclusions: A Reply to Retherford and Sewell." American Sociological Review 56: 159-65.

Comments
  • Recommend Us