Birth Order Thesis

The Pennsylvania State University
The Graduate School
College of the Liberal Arts
ORDERED DELINQUENCY: THE EFFECTS OF BIRTH ORDER ON
DELINQUENCY
A Thesis in
Crime, Law and Justice
by
Patrick R. Cundiff
Copyright 2010 Patrick R. Cundiff
Submitted in Partial Fulfillment

of the Requirements
for the Degree of
Master of Arts
August 2010
The thesis of Patrick R. Cundiff was reviewed and approved* by the following:
Jeremy Staff
Associate Professor of Crime, Law, and Justice and Sociology
Thesis Advisor
Derek Kreager
Assistant Professor of Crime, Law, and Justice
Lori Burrington
Assistant Professor of Crime, Law, and Justice
John McCarthy
Professor of Sociology
Head of the Department of Sociology and Crime, Law, and Justice
*Signatures are on file in the Graduate School
ii
Abstracti
Juvenile delinquency has long been associated with birth order (i.e., the ordinal position
in the birth sequence of a given family) in popular culture, though little research has attempted to
explain why. Drawing from Adlerian birth order theory, Sulloway’s born to rebel hypothesis,
and Blake’s resource dilution model, I examine the relationship between birth order and a variety
of delinquent outcomes in adolescence. Following some recent research on birth order and
intelligence, I use methods that allow for examination of both between-individual (i.e., cross-
sectional comparisons) and within-family (i.e., sibling comparisons) differences to better address
potential spurious relationships. I also include measures of school achievement, sibship size, and
both parental supervision and attachment to help explain why birth order is related to
delinquency. I use data on approximately 16,500 adolescents from the National Longitudinal
Study of Adolescent Health (Add Health). My findings suggest that the relationship between
birth order and delinquency is spurious; birth order effects on delinquency that were observed
between individuals become negligible and statistically non-significant when the effects are
examined within families (i.e., among siblings). The implications of this finding are discussed.
iii
TABLE OF CONTENTS
LIST OF FIGURES: ................................................................................................................................... v
LIST OF TABLES: .....................................................................................................................................vi
SECTION 1: INTRODUCTION................................................................................................................. 1
SECTION 2: THEORETICAL PERSPECTIVES ...................................................................................... 2
Birth Order Theory ................................................................................................................................. 2
Resource Dilution Model ........................................................................................................................ 7
Spurious Arguments ............................................................................................................................... 9
SECTION 3: CURRENT STUDY ............................................................................................................ 13
SECTION 4: METHODS ......................................................................................................................... 14
Data ...................................................................................................................................................... 14
Measures ............................................................................................................................................... 15
Outcome Variables............................................................................................................................ 15
Predictor Variables ........................................................................................................................... 18
Analytic Strategy .................................................................................................................................. 23
SECTION 5: RESULTS ........................................................................................................................... 24
Comparing Single Children and Firstborns ........................................................................................... 24
Examining the Effects of Birth Order on Potential Mediating Variables .............................................. 25
Examining the Effects of Birth Order on Delinquency: Between-Individual Analyses ......................... 27
Examining the Effects of Birth Order on Delinquency: Within-Family Analyses................................. 45
SECTION 6: DISCUSSION AND CONCLUSION ................................................................................. 50
REFERENCES ......................................................................................................................................... 55
iv
LIST OF FIGURES:
Figure A: Graphic Representations of Predicted Likelihood of Delinquency
v
LIST OF TABLES:
Table 1: Descriptive Statistics
Table 2: Results from Logistic Regressions Examining the Negligible Differences Between
Firstborns and Single Children
Table 3: OLS Regressions Predicting Potential Mediating Variables
Table 4: Logistic Regression Predicting Seatbelt Non-Use
Table 5: Logistic Regression Predicting Marijuana Use
Table 6: Logistic Regression Predicting Violence
Table 7: Logistic Regression Predicting Non-Violent Crime
Table 8: Logistic Regression Predicting Binge Drinking
Table 9: Logistic Regression Predicting Sexual Activity
Table 10: Logistic Regression Predicting Violation of Parents' Rules
Table 11: Fixed Effects Logistic Regression Examining Within Family Differences in Seatbelt Non-
Use
Table 12: Fixed Effects Logistic Regression Examining Within Family Differences in Marijuana Use
Table 13: Fixed Effects Logistic Regression Examining Within Family Differences in Non-Violent
Crime
Table 14: Fixed Effects Logistic Regression Examining Within Family Differences in Binge Drinking
Table 15: Fixed Effects Logistic Regression Examining Within Family Differences in Sexual Activity
vi
SECTION 1: INTRODUCTION
Social scientists have long shown the effects of birth order and family size on outcomes
such as intelligence and educational achievement, but little attention has been paid to the effects
of birth order and family size on deviance and delinquency. My thesis will attempt to test
competing hypotheses of the effects of birth order on delinquency and other problem behaviors
in adolescence derived from birth order theory and the resource dilution model. As I review in
more detail below, birth order theory, as proposed by Adler (1928), suggests a curvilinear
relationshipii in which firstborns or single childreniii would score the lowest on a delinquency
scale, middleborns would score the highest, and the youngest children would score between
firstborns and middleborns. More recently, Sulloway (1996), building upon Adler’s work, has
argued that laterbornsivare more likely to rebel than firstborns.
In contrast, the resource dilution model proposed by Blake (1981) assumes that parental
resources are finite and become diluted with the addition of more children. The assumption of
the model is that firstborns begin life with all of the parental resources but lose half upon the
addition of another sibling. The second child begins life with only a proportion of the resources
the firstborn once held, the third child would have yet a smaller proportion of the total resources
available, and this division of resources continues with the birth of each additional sibling. Thus,
birth order would have a linear relationshipv with delinquency in which the youngest child would
score the highest, the oldest child would score the lowest, and the middle child would score
somewhere between the youngest and the oldest. While every birth position may be affected by
the dilution of resources, it stands to reason that the youngest child will begin life with fewer
resources than children of the first or middle ordinal positions.
1
In this thesis, I use data from the National Longitudinal Study of Adolescent Health (Add
Health) to examine the effects of birth order on delinquency. Unlike previous research, I use a
large nationally representative dataset to better identify and parse out the effects, if any, of birth
order on delinquency, as well as the underlying mechanisms for these associations. For instance,
birth order theory suggests that differences in measures of school achievement may explain the
curvilinear relationship between birth order and delinquency. The resource dilution model, on
the other hand, suggests that the mechanisms explaining the linear effect of birth order on
delinquency may be parental supervision and attachment. Furthermore, I examine both between-
individual and within-family differences in the effects of birth order on delinquency to better
address potential spurious relationships. I begin with a review of these theoretical perspectives.
SECTION 2: THEORETICAL PERSPECTIVES
Birth Order Theory

Adler’s (1928) seminal work ―Characteristics of the First, Second, Third Child‖ provided
the basis of birth order theory. Adler argued that the first child has one of two paths: the ―path of
the single child‖ or the ―path to dethronement.‖ For example, if a first born child remains a
single child, he or she will likely remain the center of attention. The constant attention from
parents expedites development and makes the child more self-confident. However, the child is
also prone to developing a ―spoiled‖ or entitled attitude. On the other hand, the arrival of a
sibling can be a traumatic experience for a first born as the child suffers dethronement, so that
the child is no longer the center of attention in their parents’ eyes. The path to dethronement
causes the first child to determinedly attempt to reclaim the center of attention, which results in
conformity to conventional goals and values. Moreover, the younger child thrives in his or her
ordinal position as desires and ambitions are nurtured and cultivated by the parents. This
2
nourishment leads the second child to become ―more charming and likeable than their older
sibling‖ (Adler, 1928: 24). Nevertheless, the younger child is not without trials as he or she is
constantly in competition with a larger and more intelligent rival.
According to Adler, the arrival of the third sibling, however, does not dramatically shift
the balance of power. The third child receives similar treatment to the second child, but the
youngest child is generally more apt to overcome the competition between siblings and to
establish his or her importance in the family. If the third child is unable to compete, he or she
will seek the center of attention through maintaining a prince or princess mentality. Adler
argued that this mentality will lead to laziness, shirking of responsibilities, and generating
elaborate excuses. Since sibling rivalries are often fueled by the competition over parental
resources, this competition can become strain inducing and lead to jealousy among the siblings.
Since the inception of Adler’s birth order theory, a sizable body of research has found
varying and contradictory evidence regarding the effects of birth order on adolescent adjustment.
Most of this research centers on educational achievement (Altus 1966, Adams and Phillips 1972,
Adams 1972, Forer 1976), intelligence (Black et al. 2007), creativity (Runco and Bahleda 1987),
conformity (Becker and Carroll 1962, Sampson 1962, Bragg and Allen 1970), and risk taking
(Nisbett 1968, Eisenman 1987). In terms of educational achievement, studies have shown that
firstborns have a greater likelihood of educational attainment and intellectual professional
development than other ordinal positions (Altus 1966, Adams and Phillips 1972, Adams 1972,
Forer 1976). Research on the relationship between birth order and creativity has found a
curvilinear relationship in which firstborns are the most creative, middleborns are the least
creative, and laterborns are nearly as creative as firstborns (Runco and Bahleda 1987).
Furthermore, the results of research on birth order and conformity (Becker and Carroll 1962,
3
Sampson 1962, Bragg and Allen 1970) show that firstborns are more likely to conform (whether
through influence by an authority figure or fulfilling a specific role) than laterborns; however, an
interaction between sex and birth order reveals that firstborn males are more likely to conform
than laterborn males, while firstborn females are less likely to conform than laterborn females.
Research on the subject of risk taking and birth order has also found somewhat mixed results.
For instance, Nisbett (1968) finds that firstborns are less likely to participate in dangerous sports
for fear of injury than laterborns. On the other hand, Eisenman (1987) finds that firstborns are
more likely to engage in risk taking than laterborns.
Although empirical tests of birth order theory have generally produced mixed results,
several consistent personality profiles have been built around each ordinal position. Forer (1969,
1976), for instance, provides personality profiles and tendencies for a variety of ordinal positions
and sibshipvi sex distributions.vii Overall, Forer’s (1976) descriptions focus primarily on single
children, firstborns, middleborns, and youngest children.viii Single children are described as
pampered and spoiled, but also confident and optimistic. According to Forer, holding the center
of attention can lead to a false sense of entitlement as the individual matures and has greater
amounts of interactions outside the immediate family. Polit and Falbo (1987), through a meta-
analysis, found that single children scored higher in achievement and adjustment. Falbo’s work
(1984) on single children has found that single children typically have smaller social networks
than their firstborn counterparts, as well as the tendency to adopt the social characteristics of the
mother. Moreover, firstborns have a tendency to be more conservative. Firstborns are also
likely to be concerned with dealing with the pressures of upholding the family mores and
attitudes. Baskett (1984) found that firstborns were more likely than other ordinal positions to
receive negative responses from parents and other siblings following misconduct or failure.
4
When a sibling is added, the firstborn may react with jealousy and anxiety following
dethronement. As earlier noted, firstborn children crave and need parental approval, thus they
are prone to succumbing to parental pressure. Because the middleborn position stimulates
maximum sibling competition; the stress and strain from the middle ordinal position can either
lead to great success in life or constant failure. Moreover, due to the divided parental attention,
middle children may have difficulties in adjustment and role identification. Thus, middleborns
would be theorized to have the highest likelihood of having social problems with peers and
figures of authority. Forer’s personality profiles tend to indicate that middleborn children are the
most prone to difficulties in conventional achievement. Research conducted by Blau and
Duncan (1967) found that children occupying polar ordinal positions tended to have more
successful careers than middleborn children. Research by Tygart (1991) proposed that a
curvilinear relationship between birth order and delinquency could be a result of middle children
lacking parental attention or supervision, thus forcing middle children to seek attention or
approval from their peer group. Through a process of social learning within the peer group,
middle children utilize delinquency to gain attention and approval. The youngest child typically
is recognized as the baby of the family regardless of maturation, and this role is both
advantageous and disadvantageous. While the youngest child is characterized as more
extroverted than older siblings, youngest children generally tend to have low acceptance of
responsibility.
More recently, Sulloway (1996) found a significant birth order effect upon the propensity
to rebel. In Born to Rebel, he found that laterborns were more likely than firstborns to engage in
rebellious activities, based upon his analysis of thousands of historical figures. Sulloway’s
findings have been rigorously tested (Salmon and Daly 1998, Paulhus et al. 1999, Zweigenhaft
5
and Von Ammon 2000, Zweigenhaft 2002, Healey and Ellis 2007) and each study has found
support for Sulloway’s born to rebel hypothesis. Sulloway’s theory begins with the assumption
that siblings who are raised together typically have markedly different personalities, and these
differences are comparable to differences observed between individuals not from the same
family. Sulloway contends that these differences are shaped by competition over family
resources, such as parental affection and parental resources. This competition creates rivalries
among the siblings, as each child vies for the family’s resources. The born to rebel hypothesis
emphasizes the impact birth order plays upon family niches. Sulloway’s findings point to
rebelliousness as a possible mechanism behind the effects of birth order. Rebellion as Sulloway
constructed it more resembles the propensity to upset the status quo. Laterborns are seen as
more open to radical change because the change favors them in the competition for parental
resources as it upsets the tradition associated with investment in the firstborn. While Sulloway
does not speak directly about deviance, it would follow that those open to radical change may
also be open to deviance or risk taking as a method of radical change. Related research
conducted by Heaven (1994) found that ―venturesomeness‖ or risk taking was associated with
higher delinquency scores. Venturesomeness and risk taking can be considered proximal
measures of rebelliousness.
The rebirth of interest into the effects of birth order has generated a considerable amount
of literature in the fields of psychology and sociology, yet few studies have looked at the effects
of birth order on delinquency. Using a sample of youth from Israel, Rahav (1980) found a
curvilinear relationship between ordinal position and delinquency rates in which middle children
scored the highest and oldest and youngest children scored lower. However, like other research
on birth order effects, Rahav did not adequately address potential spurious relationships between
6
birth order and delinquency. Similarly, Argys et al. (2006) found evidence that middleborns and
lastborns were more likely to use substances and engage in risky adolescent behavior (sexual
activity. While Argys et al.’s findings were more robust than Rahav’s (1980), they neglect issues
plaguing much of birth order research. As I review in more detail below, the ―effects‖ of birth
order may also be linear, as suggested by research based upon the longstanding resource dilution
model.
Resource Dilution Model

The history of the resource dilution model can be traced back to Dumont’s (1890) work,
which focused on the ―law of capillary action‖ which states that the presence of siblings dilutes
resources that are crucial for social mobility; however, Blake (1981) was the first to coin the
term resource dilution model and has since been the prominent proponent. The resource dilution
model assumes that given a finite number of parental resources, the number of children will be
inversely related to the amount of resources available for each child. For example, single
children from birth hold one hundred percent of available parental resources and never lose
shares of any resources. The assumption of the model is that firstborns begin with one hundred
percent of parental resources but lose half upon the arrival of another sibling. The second child
begins life with a division of the resources the firstborn once held, the third child begins with yet
another division of the total resources available and this division of resources continues with the
birth of each additional sibling. As older siblings move away from parents, the division of
resources swings in the youngest child’s favor; however, this process does not typically occur
prior to the youngest child beginning the socialization process. This process marks a pivotal
point in a child’s development. Lack of resources at this time in a child’s development can have
harmful and long lasting effects. Previous research (Hirschi 1969, Sampson and Groves 1989,
7
Sampson and Laub 1990) has noted links to deficiencies in parental resources (social bonds,
socio-economic resources, supervision) and propensities for delinquency. Using theories of
strain and control, one can see how the resource dilution model would predict a linear
relationship between birth order and delinquency. Larger family sizes, and thus increasing birth
orders, face greater economic strain as resources must be spread over a greater number of
children. Additionally, with more children, a parent’s attention is more divided. This division
leads to less immediate supervision; with supervision lacking among larger families (and higher
ordinal positions), the opportunity for deviance necessarily increases. Under the resource
dilution model, children born last would begin with fewer parental resources and fail to gain total
parental resources equal to those of older siblings prior to beginning the socialization process.
Blake (1989a) theorizes that birth order differences in intelligence primarily result from
biological capabilities of parents and dilution of parents’ resources by other children.
Additionally, Blake contends that previous studies on the effects of birth order have had
difficulty in controlling for the following confounding factors: the prevalence fallacy (or the bias
introduced by looking at the proportionate prevalence of a particular birth order in a selected
group), absence of sibship size controls, period effects, parental background differences, child
spacing, and selection biases. Blake (1989a) notes ―systematic and patterned birth-order
differences in cognitive ability are quite possibly nonexistent‖ (p. 182). Tests of the resource
dilution model have found some mixed results (Blake 1981, Blake 1989a, Blake 1989b, Downey
1995, Guo and VanWey 1999, Downey 2001, Rodgers 2001, Steelman et al. 2002). Several
studies (Blake 1981, Blake 1989a, Blake 1989b, Downey 1995, Downey 2001) found evidence
that increased sibship size results in lower educational attainment. Studies (Guo and VanWey
8
1999, Rodgers 2001, Steelman et al. 2002) have also found the link between increased sibship
size and educational attainment to be non-existent or weak.
Prior to Blake’s research, the confluence model (Zajonc 1976) offered an explanation for
differences in intellectual development by ordinal position. According to the confluence model,
intellectual development is dependent upon the average intellectual level of the family, and with
each additional child, this level decreases. The average intellectual level of the family decreases
with the addition of children because the low education levels of new children drive down the
average intellectual level. Importantly, empirical support for this model is limited (Brackbill and
Nichols 1982, Galbraith 1982a, Galbraith 1982b). Attempts have been made to answer
criticisms and to revive the confluence model by controlling for potentially confounding factors,
(Berbum et al. 1982, Zajonc and Mullaly 1997) but these attempts have done little to repair the
damage caused by earlier criticisms. Within the scope of this study, support for the resource
dilution model would be found if the results suggest a linear relationship between birth order and
delinquency in which firstborns score the lowest in terms of delinquency and each subsequent
ordinal position scores higher than its predecessor.
Spurious Arguments
Through the years there has been a strong body of literature (Capra and Dittes 1962,
Kammeyer 1967, Schooler 1972, Edwards and Klemmack 1973, Schooler 1973, Ernst and Angst
1983, Hauser and Sewell 1985, Steelman and Powell 1985, Freese et al. 1999, Seff et al. 2005)
that suggests that the effects of birth order are negligible. This area of research has often focused
on attempting to prove that observed effects of birth order are in fact spurious. Spuriousness
results when the observed effects of a predictor variable are found to be the product of some
other variable. Several studies (Edwards and Klemmack 1973, Hauser and Sewell 1985, Seff et
9
al. 2005) found contradictory results to previous research. Edwards and Klemmack (2005) found
weak and limited results for the hypothesis that firstborns are the conservators of tradition.
Hauser and Sewell (1985) found no link between birth order and intellectual development or
social success. Seff et al. (2005) found evidence contrary to Nisbett’s (1968) study on
participation in dangerous sports. Additional studies (Capra and Dittes 1962, Kammeyer 1968)
have identified potential pitfalls when using birth order as a variable. Capra and Dittes (1962)
identified severe sample biases associated with prior research on birth order. Findings from
Kammeyer’s (1968) study suggest that previous research attempted to utilize ordinal position
beyond its capabilities and future research should attempt to extract the true effects of birth order
by better controlling for possible confounding variables.
Birth order theory’s strongest criticisms (Schooler 1972, Ernst and Angst 1983, Steelman
1985), however, have focused on methodology flaws and whether the effects of birth order are in
fact causal. For instance, Schooler (1972) conducted a meta-analysis on birth order effects from
previously published and unpublished studiesix and found that results from prior research reveal
no reliable evidence of birth order effects. Schooler makes note of several variables that should
be accounted for: density of sibling spacing, sex of siblings, family size, and social class trends.
In response to Schooler’s (1972) meta-analysis, Breland (1973) proposed that not all birth order
effects are completely erased by controlling for population biases or socioeconomic-status
differences. In response to Breland (1973), Schooler (1973) made the contention that ―even if
birth order differences in intellectual functioning exist in childhood, they are very slight and
exert at most minimal effects‖ (p. 213). Steelman and Powell (1985) found no significant
relationship between birth order and academic performance and the conclusions lead them to
believe that any observed birth order effects may be artificial.
10
Ernst and Angst (1983) delivered perhaps the most extensive analysis of birth order.
Ernst and Angst tested birth order effects on all available fields (including: intelligence, school
achievement, occupational status, personality, socialization, and mental illness) and their
subsequent findings diminished the credibility of previous research. Their analysis found that in
regards to school achievement and substance use (cigarettes and alcohol), birth order had
negligible effects. Furthermore, birth order had no effects on intelligence or personality. Ernst
and Angst further tested relationships between birth order and variables related to personality
while controlling for structural and individual variables and found that birth order effects
disappeared. Additionally, Ernst and Angst identified that general differences in scores were
between subjects, not between siblings. They also found little support for the influence of
sibship size upon socialization, which, of course, contradicts hypotheses derived from the
resource dilution model. While these criticisms focused on early work with birth order theory,
few criticisms have focused on Sulloway’s (1996) ―born to rebel‖ hypothesis. Recent analyses
(Freese et al. 1999, Zweigenhaft and Von Ammon 2000, Zweigenhaft 2002) have yielded mixed
results. Both studies conducted by Zweigenhaft focused on rebellious activities (such as civil
disobedience, participating in protests or demonstrations, and marijuana use) and found support
for Sulloway’s work, while Freese et al. failed to find support. Freese et al.’s analysis focused
on examining Sulloway’s contention that firstborns display more conservative attitudes, tough
mindedness, and supportive attitudes towards authority. The analysis found no significant
effects for Sulloway’s contention, even stating that the nonsignificant coefficients were in the
wrong direction.
In addition to tests of Sulloway’s born to rebel hypothesis, many researchers (Retherford
and Sewell 1991, Rodgers et al. 2000, Rodgers 2001, Wichman et al. 2006, Wichman et al.
11
2007) have criticized previous literature’s failure to examine within-family, or between siblings,
differences when examining the effects of birth order and family size on intelligence. Retherford
and Sewell’s work (1991) demonstrated that the use of within-family analyses provide a
strikingly different result from works that focus on between-individual differences. In line with
the work of Retherford and Sewell (1991), Rodgers (2001) and Rodgers et al.’s (2000) work
provide evidence suggesting that any birth order effects are a product of differences between
families, not within. More recently, Wichman et al. (2006) utilized a multilevel approach in their
examination of the effect of birth order on intelligence. They found that when controlling for
maternal age, the effects of birth order on intelligence become negligible and statistically non-
significant. Wichman et al. (2006) noted that ―the fundamental cause of supposed birth order
effects lies between, not within, families‖(p. 125). In defense of birth order, Zajonc and
Sulloway (2007) questioned the Wichman et al.’s study design and analyses. Zajonc and
Sulloway provided analyses demonstrating that the effects of birth order are both a between-
individual and within-family phenomenon. In reply to Zajonc and Sulloway’s (2007) work,
Wichman et al. (2007) contend that the research design by Zajonc and Sulloway failed to
appropriately control for influences on intelligence that vary across families.
In summary, the evidence is mixed on whether the effects of birth order observed in
previous studies are indeed causal. The greatest issue in evaluating the effects of birth order is
the level of evaluation. Much of the recent criticisms have debated whether birth order should be
evaluated between individuals or within families, in line with this debate I test the effects of birth
order at both levels.
12
SECTION 3: CURRENT STUDY
This study will examine the relationship between birth order and delinquency. Research
examining the link between birth order and delinquency has become dated and there exists a
growing body of literature focused on testing hypotheses generated by Sulloway’s born to rebel
hypothesis. The primary goal of this study is to sort out the influence, if any, that birth order has
upon delinquency. If a relationship between birth order and delinquency exists, this study will
then seek to clarify the mechanism at work.
The following hypotheses are derived from: birth order theory, Sulloway’s born to rebel
hypothesis, and the resource dilution model.
Hypothesis 1: The relationship between birth order and delinquency will be curvilinear,
with firstborn and lastborn children scoring significantly lower on the
delinquency scale than middleborn children.
If hypothesis 1 is confirmed, the results will lend support to birth order theory; which suggests
that low achievement explains why middleborn children are more delinquent than firstborn and
laterborn children.
Hypothesis 2: The relationship between birth order and delinquency will be linear, with
firstborns scoring significantly lower on delinquency scales and each ordinal
position after scoring higher than the previous ordinal position.
If hypothesis 2 is confirmed, the results will lend support to the resource dilution model, which
suggests that lack of parental supervision explains the high rates of delinquency among the
youngest children.
Hypothesis 3: Firstborns will score lower on scales of deviance and delinquency than
laterborns.
13
If hypothesis 3 is confirmed, the results will lend support to Sulloway’s born to rebel hypothesis.
Recall that Sulloway argues that laterborns are most delinquent because they have a tendency
toward rebellion.
Hypothesis 4: The relationship between birth order and delinquency will be spurious.
Finally, if hypothesis 4 is confirmed, the results would lend support for studies that have
questioned whether the effects of birth order are causal.
Figure A: Graphic Representations of Predicted Likelihood of Delinquency
High
Resource Dilution
Delinquency
Model
Birth Order Theory
Born To Rebel
Null Hypothesis
Low
Firstborn Middleborn Lastborn
Ordinal Position
SECTION 4: METHODS
Data
These hypotheses will be tested using data from the National Longitudinal Study of
Adolescent Health (Add Health). Add Health is a longitudinal study of a nationally
representative sample of adolescents in grades 7 to 12. From 1994 to 2001, the study collected
14
four waves of student data, with additional surveys administered to parents, siblings, and school
administrators. Since the effects of birth order are theorized to be established prior to
adolescence, I do not use the follow-up data in this thesis. Thus, this analysis focuses on only
the wave 1 in-home survey data. Additionally, wave 1 has the largest sample size giving the
analyses greater statistical power to detect even small effects of birth order on delinquency.
Harris et al. (2008) sampled 80 high schools and 52 middle schools from the US with unequal
probability of selection. Incorporating systematic sampling methods and implicit stratification
into the Add Health study design ensured that this sample is representative of US schools with
respect to region of country, urbanicity, school size, school type, and ethnicity. An in-school
survey was administered during one 45-60 minute class period to all available students in each of
the sampled schools; in-home interviews were then conducted approximately six months after
the in-school survey to approximately 200 adolescents randomly selected from each school.
Interviews lasted 90 minutes and were completed confidentially through the use of laptop
computers and headphones. The respondent’s parent(s) were also interviewed. Add Health
yielded a response rate of 79 percent for wave 1. The sample size following the coding and
exclusion of missing cases ranged from approximately 13,000 to 16,500 adolescents with sample
size dependent upon outcome variables.
Measures
Outcome Variables
For the purposes of this study, I use a variety of measures of delinquency and problem
behaviors including: violence, non-violent crime, marijuana use, binge drinking, sexual activity,
and measures of risk taking. These measures are explained in detail below.
15
Violence
The violence measure consists of the summed score of three binary-coded violent events:
involvement in a serious physical fight, seriously injuring someone, and participation in a group
fight. For instance, respondents were asked questions pertaining to how often each event took
place such as: ―In the past 12 months, how often did you take part in a fight where a group of
your friends was against another group?‖, ―How often did you get into a serious physical fight?‖,
and ―How often did you hurt someone badly enough to need bandages or care from a doctor or
nurse?‖. If a respondent reported committing any of these offenses in the past year, they were
coded as ―1‖ and coded ―0‖ if they had not engaged in at least one of these acts during the past
12 months. The Cronbach’s alpha coefficient for this measure was .69.
Non-Violent Crime
The non-violent crime measure was created similarly to the violence measure, consisting
of eight items: shoplifting, burglary, theft of less than $50, theft of more than $50, graffiti,
damage of another’s property, selling drugs, and automobile theft. Respondents were asked
questions regarding how many times the offense had occurred; specifically, respondents were
asked the following: ―In the past 12 months, how often did you take something from a store
without paying for it?‖, ―How often did you drive a car without the owner’s permission?‖, ―In
the past 12 months, how often did you steal something worth more the $50?‖, ―How often did
you go into a house or building to steal something?‖, and ―How often did you steal something
worth less than $50?‖. Similar to the measure of violence, if a respondent reported committing
any of these non-violent offenses in the past year, they were coded as ―1‖ (and coded ―0‖ if they
16
had not engaged in at least one of these acts during the past 12 months). The Cronbach’s alpha
coefficient for this measure was .75.
Marijuana Use
The marijuana use variable incorporates responses indicating whether the adolescent had
used marijuana during the past 30 days. Respondents were asked questions regarding the
amount of substance use such as: ―During the past 30 days, how many times did you use
marijuana?‖ Responses were then coded into a dichotomous outcome variable, which indicates
whether the adolescent had engaged in marijuana use during the previous 30 days (coded ―1‖),
versus no use in the past 30 days (coded ―0‖). Thus, marijuana use is a 30 day prevalence
measure.
Binge Drinking
The binge drinking variable is based upon adolescent responses to the following question:
―Over the past 12 months, on how many days did you drink five or more drinks in a row?‖ The
responses were then coded into a dichotomous outcome variable which indicates whether the
adolescent had ever engaged in binge drinking during the past 12 months (coded ―1‖), versus no
binge drinking in the past year (coded ―0‖).
Sexual Activity
The outcome variable ―sexual activity‖ was created based upon the adolescents’ response
to a question asking whether they had ever engaged in sexual intercourse at the time of the
17
survey. If the adolescent had previously engaged in sexual intercourse, they were coded as ―1‖
(―0‖ if they had not previously engaged in sexual intercourse).
Risk Taking
I created two measures of risk taking: (1) seatbelt non-use, and (2) violation of parents’
rules. Two measures of risk taking were created based upon the low Cronbach’s alpha (.18) for a
combined scale. Additionally, the combination of the two items made little substantive sense as
they represent different approximations of risk taking. The outcome variable of seatbelt use
asked adolescents to estimate the overall frequency they used a seatbelt when riding in or driving
in an automobile. Respondents were able to select from the following: never, rarely, sometimes,
most of the time, and always. These responses were then coded dichotomously, where
respondents who selected never or rarely were coded as ―1‖ and respondents who selected any of
the other categories (sometimes, most of the time, and always) were coded as ―0.‖ The outcome
variable of violation of parents’ rules is based upon adolescents’ responses to a question asking
about whether they had ever spent the night away from home without their parents’ permission.
The responses were dichotomously coded, where respondents who had spent the night away
from home without permission were coded as ―1‖ and respondents who had not were coded as
―0.‖
Predictor Variables
Birth order is the key predictor variable utilized in this study. Birth order information
was obtained through the use of the following questions ―Which child are you—the first, the
second, or what?‖ and ―How many children have your biological parents had together?‖
Responses to both questions ranged from 1-15. Based upon the responses to both questions,
18
respondents were then coded into one of four categories: single child, firstborn, middleborn, or
lastborn. From this variable two additional variables were also created, both of which excluded
single children. Note that this exclusion is based upon the contention that single children and
firstborn children generally have negligible differences (Falbo 1984, Polit and Falbo 1987). The
exclusion of single children also allows for comparisons between all three theoretical positions:
birth order theory, Sulloway’s born to rebel hypothesis, and the resource dilution model. The
first of the two created variables will place adolescents into one of the three following categories:
firstborn, middleborn, or lastborn. The second created variables will place adolescents into
either the firstborn category or the laterborn category, which is in line with the work of Sulloway
(1996).
To address the hypothesized underlying mechanisms of birth order and delinquency,
predictor variables indicating academic achievement, intellectual ability, parental monitoring,
and parental attachment will be included in the analyses. The academic achievement measure is
based upon grade point averages (GPA) constructed from self-reported grades in the following
subjects: English, Mathematics, History or Social Sciences, and Science. The construction of the
GPA variable is done through calculating the GPA for all courses in which each participant was
enrolled. Thus, if a survey participant was enrolled in only two of the four subjects, his/her GPA
score would not be missing. My calculation of GPA allows for GPA to range from 1.0 (D) to 4.0
(A). The picture vocabulary test (PVT) administered during the survey will serve as a measure
of intellectual ability. Scores on the PVT ranged from 13 to 146.
The parental monitoring measure is based upon adolescent responses regarding the
general presence of parents before they leave for school, when they return from school in the
afternoon, when they eat dinner and when they go to bed in the evening. The responses ranged
19
for presence before school, when they return from school, and when they go to bed ranged from
always (coded as ―5‖) to never (coded as ―1‖). The max score between parents for each time
point (morning, afternoon, and evening) was used in the calculation of parental presence. The
responses for the frequency with which the adolescents ate dinner with at least one parent over
the past seven days ranged from 0 to 7. The final parental presence measure was calculated by
summing the max parental presence scores for morning, afternoon, and evening presences with
the score from the frequency of eating dinner together in the past seven days; creating a
distribution that ranged from 0 (indicating no parental presence) to 22 (indicating near constant
or constant presence). The parental attachment measure is based upon responses to questions
regarding how close a respondent feels to each parent, as well as activities the individual has
recently (within the past four weeks) done with a parent (e.g., shopping, sport, religious event,
talked about problem, talked about social life, went to entertainment, talked about school,
worked on school project, and talked about other school things). The responses to the closeness
and sum of activities done were averaged; the max parental attachment score (mother or father)
is used for each adolescent.
The following variables will be used to control for possible confounding influences:
sibship size, twin status, sex, race, family stability, parental education, and age. Sibship size was
obtained using the following: ―How many children have your biological parents had together?‖
Controlling for sibship size will help to reduce the effect of variance between subjects by
limiting the effect that coming from a large or small family will have upon the outcome
variables. While one could expect sibship size and birth order to be strongly correlated as
adolescents from larger families necessarily have a greater chance of being classified as a
middleborn, the correlation between the two variables is minute (r =0.07). The variable twin
20
status was obtained through asking ―Are you a twin?‖ It is necessary to control for the effects of
twins due to the oversampling of twins conducted by Add Health. Intact family is determined
by responses to questions regarding residences of parents; adolescents were coded for intact
family as ―1‖ if both biological parents were still residing with child, and ―0‖ if either the
biological mother or father were not residing with the child. Parental education was generated
by using the maximum score of educational attainment score between the two parents; the scores
ranged from 0 (no education) to 5 (professional degree). Demographic variables of sex (coded as
a dummy variable where a code of ―1‖ indicates that the individual is female), race/ethnicity
(coded as a series of dummy variables), and age will also be utilized to control for age effects
upon delinquency. Table 1 provides descriptive statistics for both outcome and predictor
variables used in this study (with the exclusion of single children).
21
Table 1: Descriptive Statistics-Excluding Missing Cases and Single Children (N=12,787)
Variable N Mean (%N) SD Min. Max.
Birth Order Theory
Firstborn 4,795 (37.50%) ~ 0 1
Middleborn 3,134 (24.51%) ~ 0 1
Lastborn 4,858 (37.99%) ~ 0 1
Born to Rebel Hypothesis

Firstborn 4,795 (37.50%) ~ 0 1
Laterborn 7,992 (62.50%) ~ 0 1
Outcome Variables
Non-Use of Seatbelt 12,787 0.11 0.31 0 1
Marijuana Use 12,787 0.13 0.34 0 1
Violence 12,787 0.39 0.49 0 1
Non-Violent Crime 12,787 0.41 0.49 0 1
Binge Drinking 12,787 0.26 0.44 0 1
Sexual Activity 12,787 0.36 0.48 0 1
Violation of Parents' Rules 12,787 0.13 0.33 0 1
Controls
Age 12,787 15.63 1.70 11 21
Female 6,547 (51.20%) ~ 0 1
Male 6,240 (48.80%) ~ 0 1
White 7,035 (55.02%) ~ 0 1
Black 2,371 (18.54%) ~ 0 1
Hispanic 2,170 (16.97%) ~ 0 1
Asian 987 (7.72%) ~ 0 1
Other Race 224 (1.75%) ~ 0 1
Sibship Size 12,787 3.00 1.33 2 14
Family Education 12,787 2.97 1.25 0 5
Intact Family 12,787 0.62 0.49 0 1
Twin Status 12,787 0.08 0.28 0 1
Mediating Variables
Parental Presence 12,787 17.31 3.64 3 22
Parental Attachment 12,787 3.86 1.06 0.5 7
GPA 12,787 2.80 0.77 1 4
PVT 12,787 100.64 14.61 13 138
Total 12,787
22
Analytic Strategy
The analytic strategy consists of examining the effects of birth order utilizing both
between-individual and within-family analyses to assess genuine impacts of birth order on
delinquency as well as the underlying mechanisms. First, ordinary least-squares (OLS)
regressions were estimated to examine the relationship between birth order and the explanatory
variables (GPA, PVT, parental presence, and parental attachment), which were hypothesized to
mediate the relationship between birth order and delinquency. Mediators significantly related to
birth order net of controls will be included in the full models of both the between-individual and
within-family analyses. Following these OLS regressions, I then estimated a series of logistic
regressions examining the effects of birth order on the risky behaviors of interest in this thesis
(seatbelt non-use, marijuana use, violence, non-violent crime, binge drinking, sexual activity,
and violation of parents’ rules). These cross-sectional analyses capture any between-individual
differences in the effects of birth order on delinquency. The cross-sectional analyses are
consistent with much of the previous literature that found significant birth order effects. Finally,
I used fixed-effects models to further analyze any significant relationships between birth order
and the outcome variables previously mentioned. As birth order is conceptualized to be a within-
family process, it is necessary to examine the effects of birth order using a methodology that is
able to control for unobserved family characteristics. Using a fixed-effects analytic framework
allows for the control of unobserved family-stable characteristics that may be correlated with
delinquency. Utilizing family-specific fixed-effects to control for unobserved family
characteristics increases confidence that the effects of birth order on delinquency are causal.
The logistic fixed effects equation can be written as:
23
𝐾 𝑁−1
log(𝑦𝑖𝑧 ) = 𝛽0 + 𝛽𝑘 𝑥𝑖𝑧𝑘 + 𝑑𝑖𝑛 + 𝜀𝑢 , 𝜀𝑢 ~𝑁(0, 𝜎 2 )

𝑘=1 𝑛=1
in which K is the number of family varying covariates, N is the total number of individuals, β0 is
the constant, yiz is a problem behavior for person i in family z, βkxizk is an observed value on the
kth family-varying covariate, such as age or parental attachment, and εu is the unobserved error.
The logistic fixed effects models were estimated in STATA using the xtlogit command. To
insure the appropriateness of the use of fixed effects over random effects, Hausman tests were
conducted (Hausman 1978). Note that, for all outcomes examined in this thesis, the Hausman
tests were found to be statistically significant (results not shown but available from the author).
This indicates that the fixed effects models were more appropriate.
SECTION 5: RESULTS
Comparing Single Children and Firstborns

Do single children differ significantly from firstborns in their early problem behaviors?
Table 2 provides the coefficients from models estimated with birth order coded as four dummy
variables (e.g., single children, firstborn, middleborn, and lastborn). As shown in Table 2, no
statistically significant differences were observed between single children and firstborns
regarding seatbelt non-use, marijuana use, violence, property crime, and violation of parents’
rules. In the analyses of the outcome variables of binge drinking and sexual activity, statistically
significant differences were observed; however, these differences were negligible. Based upon
the results presented in Table 2, single children and firstborns do not show differences in
problem behaviors. Thus, in subsequent models I excluded single children and focused only on
adolescents who reported having at least one sibling.
24
Table 2: Results From Logistic Regressions Examining the Negligible Differences Between Firstborns and Single Children-(SE)
Seatbelt Marijuana Violence Property Binge Sex Violation
Single Children 0.09 (.10) 0.20 (.11) 0.12 (.08) 0.02 (.08) 0.19 (.08)* 0.20 (.08)** 0.13 (.11)
Middleborn 0.30 (.12)* 0.57 (.12)*** 0.14 (.08)† 0.27 (.08)** 0.37 (.09)*** 0.14 (.09) 0.24 (.13)†
Lastborn 0.26 (.08)** 0.36 (.09)*** 0.07 (.06) 0.14 (.06)* 0.40 (.06)*** 0.26 (.07)*** 0.21 (.09)*
N 16424 16180 16312 16276 16387 16300 16420

† p<.10 * p<.05 **p<.01 ***p<.001
Note: Numbers in parentheses are standard errors
Note: Reference group equals Firstborn
Examining the Effects of Birth Order on Potential Mediating Variables

Tables 3 displays the results of analyses testing the relationship between birth order
variables and potential mediating variables (parental presence, parental attachment, GPA, and
PVT). Table 3a provides OLS regression results based upon the birth order theory (BOT)
categorization of birth order (firstborn, middleborn, and lastborn), with firstborns serving as the
reference category. The results indicate that, net of controls, firstborns score significantly higher
on all outcome variables. Specifically, middleborns yielded negative coefficients for parental
presence, parental attachment, GPA, and PVT (-0.33, -0.15, -0.10, and -2.25 respectively) when
compared to firstborns. Lastborns also yielded negative coefficients for all outcome variables (-
0.41, -0.11, -0.08, and -1.93 respectively) when compared to firstborns. These results lend
support for the use of these outcome variables as mediating variables in later analyses.
In regards to the control variables utilized in these analyses, family education, age, and
race dummy variables (black and Asian compared with white) were found to significantly
decrease parental presence; specifically, as both family education and age increased parental
presence decreased. The measures of intact family (in relation to non-intact families) and sibship
size were found to significantly increase parental presence. Examination of the parental
attachment outcome results indicates that intact family and age significantly decrease parental
attachment, whereas family education significantly increases parental attachment. Family
25
education, intact family, and gender were found to significantly increase an individual’s GPA,
while agex and race dummy variables (Hispanic, black, and other race) significantly decreased
GPA. The intelligence measure of PVT was found to be significantly increased by family
education and intact family while gender and race dummy variables (Hispanic, black, Asian, and
other race) significantly decreased PVT. Taken together, these results suggest that parental
presence and parental attachment are differentially affected by variables related to the
adolescent’s family, and the measures of academic achievement (GPA) and intellectual ability
(PVT) are similarly influenced by the control variables.
Table 3b provides the OLS regression results based upon the born to rebel hypothesis
(BRH) categorization of birth order (firstborn, laterborn), with firstborns serving as the reference
category. Similar to the results of the analysis of BOT birth order, the analysis of BRH birth
order reveal that firstborns score higher on all outcome variables than laterborns. Specifically,
laterborns yielded statistically negative coefficients on parental presence, parental attachment,
GPA, and PVT (-0.38, -0.12, -0.09, and -2.02 respectively). The effects of the control variables
were substantively similar to those of the BOT birth order analyses. As with Table 3a, Table 3b
provides evidence that parental presence, parental attachment, GPA, and PVT could potentially
serve as mediating variables.
26
Table 3: OLS Regressions Predicting Potential Mediating Variables-(SE)
Parental Presence Parental Attachment GPA PVT
A: Birth Order Theory
Middleborn -0.33 (.12)** -0.15 (.04)*** -0.10 (.03)** -2.25 (.38)***
Lastborn -0.41 (.08)*** -0.11 (.03)*** -0.08 (.02)*** -1.93 (.35)***
Sibship Size 0.17 (.03)*** 0.00 (.01) 0.00 (.01) -0.83 (.17)***
Family Education -0.24 (.05)*** 0.12 (.01)*** 0.15 (.01)*** 3.57 (.22)***
Intact Family 1.64 (.10)*** -0.16 (.03)*** 0.17 (.02)*** 2.10 (.37)***
Age -0.54 (.03)*** -0.06 (.01)*** -0.02 (.01)** 0.23 (.11)
Female -0.13 (.08) 0.04 (.03) 0.21 (.02)*** -1.02 (.28)***
Hispanic -0.12 (.17) 0.01 (.04) -0.12 (.05)* -7.69 (.83)***
Black -0.74 (.16)*** 0.04 (.04) -0.21 (.04)*** -9.68 (.71)***
Asian -0.58 (.26)* -0.16 (.06)* 0.18 (.06)*** -8.10 (1.22)***
Other Race -0.43 (.38) -0.15 (.10) -0.19 (.07)* -4.75 (1.84)**
Twins -0.04 (.31) 0.08 (.06) 0.01 (.05) -1.97 (.93)
Constant 25.43 (.49)*** 4.62 (.16)*** 2.57 (.12)*** 93.28 (2.03)***
N 14,233 14,279 13,829 13,602
B: Born to Rebel
Laterborn -0.38 (.08)*** -0.12 (.03)*** -0.09 (.02)*** -2.02 (.32)***
Sibship Size 0.18 (.03)*** -0.01 (.01) -0.01 (.01) -0.87 (.17)***
Family Education -0.24 (.05)*** 0.12 (.01)*** 0.15 (.01)*** 3.56 (.22)***
Intact Family 1.64 (.10)*** -0.16 (.03)*** 0.17 (.02)*** 2.09 (.37)***
Age -0.54 (.03)*** -0.06 (.01)*** -0.02 (.01)** 0.23 (.11)
Female -0.13 (.08) 0.04 (.03) 0.21 (.02)*** -1.02 (.28)***
Hispanic -0.12 (.16) 0.00 (.04) -0.12 (.05)* -7.71 (.83)***
Black -0.74 (.16)*** 0.04 (.04) -0.21 (.04)*** -9.68 (.71)***
Asian -0.58 (.26)* -0.16 (.06)* 0.18 (.06)*** -8.11 (1.22)***
Other Race -0.44 (.38) -0.15 (.10) -0.19 (.07)* -4.74 (1.84)**
Twins -0.03 (.31) 0.07 (.06) 0.01 (.05) -2.02 (.93)
Constant 25.40 (.49)*** 4.64 (.16)*** 2.58 (.12)*** 93.41 (2.00)***
N 14,233 14,279 13,829 13,602
† p<.10 * p<.05 **p<.01 ***p<.001
Note: PVT stands for picture vocabulary test, which serves as a measure of intellectual ability
Note: In the analyses of Birth Order Theory and Born to Rebel, Firstborns serve as the reference category
Examining the Effects of Birth Order on Delinquency: Between-Individual Analyses

Turning to the outcome variables, Table 4a provides the results of the logistic regression
predicting seatbelt non-use. For the analyses presented in Table 4a BOT birth order was
examined, with firstborns serving as the reference category in Models 1 and 2, and lastborns
27
serving as the reference category in Model 3xi. Model 1 provides the results of the analysis of
the effect of birth order, net of controls, on the likelihood of never or rarely wearing a seatbelt.
The results indicate that both middleborns and lastborns are significantly more likely than
firstborns to never or rarely wear a seatbelt (yielding odds ratios of 1.31 and 1.29, respectively).
Model 2 provides the results of the analysis testing for mediation. The mediation variables
proved to have an impact upon differences between firstborns and middleborns, as well as
firstborns and lastborns. The significant difference between firstborns and middleborns observed
in Model 1 became non-significant with the addition of the mediating variables in Model 2;
specifically, this difference decreased by 35 percent when the potentially mediating variables
were added. While the difference between firstborns and lastborns remained statistically
significant, the magnitude of this difference decreased (a 24 percent decrease was observed)
Examination of the control and mediating variables in Model 2 indicates that family
education, intact family, gender, PVT, GPA, and parental attachment significantly increase an
individual’s likelihood of wearing a seatbelt. Blacks are more likely to never wear or rarely wear
a seatbelt in a car than whites and Asians are more likely to wear a seatbelt than whites. Model 3
provides results of the analysis testing for non-linear associations. The results yield no evidence
of a significant curvilinear relationship; however, while the difference between lastborns and
middleborns is not statistically significant, examination of the coefficients provides some support
for birth order theory over the resource dilution model as middleborns were found to have the
highest likelihood of seatbelt non-use.
Table 4b provides the results of the logistic regression testing the effect BRH birth order
categorization on the likelihood of seatbelt non-use. Model 1 provides the results of the birth
order’s effect on seatbelt non-use net of controls. The results indicate that laterborns are
28
significantly more likely than firstborns to never wear or rarely a seatbelt (odds ratio of 1.30).
Model 2 provides the results of the mediation analysis. Similar to the relationship between
firstborns and lastborns presented in Table 4a, Model 2 finds that the difference between
firstborns and laterborns remained significantly different, with a reduction in the magnitude of
the difference (a reduction in magnitude of 27 percent). As anticipated, the relationships
between the control variables and the outcome variable (seatbelt non-use) were not significantly
different from the relationships observed in the BOT birth order analyses previously discussed.
29
Table 4: Logistic Regression Predicting Seatbelt Non-Use (N=12,787)-Odds Ratio (SE)
Non-Use of Seatbelt
Model 1 Model 2 Model 3
A: Birth Order Theory b OR (SE) b OR (SE) b OR (SE)
Firstborn ~ ~ ~ ~ ~ ~ -0.20 0.82 (.09)*
Middleborn 0.27 1.31 (.13)* 0.18 1.19 (.14) -0.02 0.98 (.13)
Lastborn 0.26 1.29 (.09)** 0.20 1.22 (.09)* ~ ~ ~
Sibship Size -0.02 0.98 (.03) -0.03 0.97 (.03) -0.03 0.97 (.03)
Family Education -0.30 0.74 (.04)*** -0.19 0.83 (.04)*** -0.19 0.83 (.04)***
Intact Family -0.29 0.75 (.08)*** -0.23 0.79 (.09)** -0.23 0.79 (.09)**
Age 0.04 1.04 (.03) 0.01 1.01 (.03) 0.01 1.01 (.03)
Female -0.58 0.56 (.09)*** -0.51 0.60 (.09)*** -0.51 0.60 (.09)***
Hispanic -0.16 0.85 (.16) -0.26 0.77 (.16) -0.26 0.77 (.16)
Black 0.41 1.51 (.14)** 0.29 1.33 (.14)* 0.29 1.33 (.14)*
Asian -0.70 0.50 (.24)** -0.74 0.48 (.26)* -0.74 0.48 (.26)*
Other Race -0.03 0.97 (.29) -0.20 0.82 (.29) -0.20 0.82 (.29)
Twins 0.22 1.25 (.26) 0.25 1.29 (.25) 0.25 1.29 (.25)
PVT -0.01 0.99 (.00)* -0.01 0.99 (.00)**
GPA -0.40 0.67 (.06)*** -0.40 0.67 (.06)***
Parental Presence -0.01 0.99 (.01) -0.01 0.99 (.01)
Parental Attachment -0.27 0.77 (.03)*** -0.27 0.77 (.03)***
Constant -1.41 (.51)** 1.84 (.73)* 2.04 (.74)**
Model 1 Model 2
B: Born to Rebel b OR (SE) b OR (SE)
Laterborn 0.26 1.30 (.08)** 0.19 1.21 (.09)*
Sibship Size -0.02 0.98 (.03) -0.03 0.97 (.03)
Family Education -0.30 0.74 (.04)*** -0.19 0.83 (.04)***
Intact Family -0.29 0.75 (.08)*** -0.23 0.79 (.09)*
Age 0.04 1.04 (.03) 0.01 1.01 (.03)
Female -0.58 0.56 (.09)*** -0.51 0.60 (.09)***
Hispanic -0.16 0.85 (.16) -0.26 0.77 (.16)
Black 0.41 1.51 (.14)** 0.29 1.33 (.14)*
Asian -0.70 0.50 (.24)** -0.74 0.48 (.26)**
Other Race -0.03 0.97 (.29) -0.20 0.82 (.30)
Twins 0.22 1.25 (.26) 0.25 1.29 (.25)
PVT -0.01 0.99 (.00)**
GPA -0.40 0.67 (.07)***
Parental Presence -0.01 0.99 (.01)
Parental Attachment -0.27 0.77 (.03)***
Constant -1.41 (.52)** 1.85 (.73)**
† p<.10 * p<.05 **p<.01 ***p<.001
30
Table 5a provides the results of the logistic regression testing the effect of BOT birth
order categorization on the likelihood of marijuana use within the past 30 days. Model 1
indicates that both middleborns and lastborns are significantly more likely to have engaged in
marijuana use within the past 30 days than firstborns net of controls (odds ratios of 1.82 and
1.42, respectively). Model 2 provides the results of the mediation analysis. The mediating
variables did not explain much of the observed differences between firstborns and middleborns
or firstborns and lastborns; however, a reduction in the magnitude of each difference was
observed (reductions of 15 percent and 21 percent respectively), though the effects of birth order
remained statistically significant. Intact family, race dummy variables (black, and Asian), GPA,
parental presence, and parental attachment were found to decrease the likelihood of marijuana
use, while family education, age, and PVT were found to significantly increase the likelihood of
marijuana use. Model 3 provides the results of the curvilinearity analysis. Similar to seatbelt
non-use, the results in Model 3 indicate that there is not a statistically significant curvilinear
relationship between birth order and likelihood of marijuana use. While the curvilinear
relationship is not statistically significant, examination of the coefficients lends support to birth
order theory’s position over that of the resource dilution model as middleborns displayed the
highest likelihood of marijuana use.
Table 5b provides the results of the logistic regression examining the relationship
between BRH birth order categorization and the likelihood of marijuana use within the past 30
days. The results from Model 1 provide evidence that laterborns are significantly more likely
than firstborns to have used marijuana within the past 30 days, net of controls (odds ratio of
1.53). Model 2 provides the results of the mediation analysis, in which laterborns were found to
be significantly more likely than firstborns to have engaged in marijuana use within the past 30
31
days, with the introduction of the potentially mediating variables yielding a 19 percent reduction
in magnitude. No significant differences in the coefficients of the control variables were
observed when compared to results presented in Table 5a.
32
Table 5: Logistic Regression Predicting Marijuana Use (N=12,787)-Odds Ratio (SE)
Marijuana Use
Firstborn ~ ~ ~ ~ ~ ~ 0.76 0.76 (.08)**
Middleborn 0.60 1.82 (.12)*** 0.51 1.66 (.13)*** 1.26 1.26 (.11)†
Lastborn 0.35 1.42 (.09)*** 0.28 1.32 (.10)** ~ ~ ~
Sibship Size -0.14 0.87 (.04)*** -0.12 0.88 (.04)** 0.88 0.88 (.04)**
Family Education 0.03 1.03 (.04) 0.10 1.11 (.04)** 1.11 1.11 (.04)**
Intact Family -0.62 0.54 (.07)*** -0.40 0.67 (.08)*** 0.67 0.67 (.08)***
Age 0.26 1.30 (.03)*** 0.21 1.23 (.03)*** 1.23 1.23 (.03)***
Female -0.20 0.82 (.09)* -0.06 0.94 (.09) 0.94 0.94 (.09)
Hispanic -0.17 0.84 (.16) -0.23 0.79 (.15) 0.79 0.79 (.15)
Black -0.14 0.87 (.13) -0.26 0.77 (.13)* 0.77 0.77 (.13)*
Asian -0.57 0.57 (.25)* -0.51 0.60 (.24)* 0.60 0.60 (.24)*
Other Race 0.45 1.57 (.29) 0.31 1.36 (.28) 1.36 1.36 (.28)
Twins -0.03 0.97 (.18)† -0.02 0.98 (.19) 0.98 0.98 (.19)
PVT 0.01 1.01 (.00)* 1.01 1.01 (.00)*
GPA -0.67 0.51 (.05)*** 0.51 0.51 (.05)***
Parental Presence -0.09 0.91 (.01)*** 0.91 0.91 (.01)***
Parental Attachment -0.15 0.86 (.04)*** 0.86 0.86 (.04)***
Constant -5.49 (.46)*** -1.87 (.65)** -1.59 (.65)*
Model 1 Model 2
Laterborn 0.42 1.52 (.09)*** 0.34 1.40 (.09)***
Sibship Size -0.10 0.90 (.03)** -0.09 0.92 (.03)*
Family Education 0.04 1.04 (.04) 0.11 1.11 (.04)
Intact Family -0.61 0.54 (.07)*** -0.39 0.68 (.07)***
Age 0.26 1.30 (.03)*** 0.21 1.23 (.03)***
Female -0.19 0.82 (.09)* -0.06 0.95 (.09)
Hispanic -0.16 0.85 (.16) -0.22 0.80 (.16)
Black -0.13 0.87 (.12) -0.25 0.78 (.12)
Asian -0.56 0.57 (.25)* -0.51 0.60 (.24)*
Other Race 0.44 1.56 (.29) 0.30 1.35 (.28)
Twins 0.00 1.00 (.19) 0.01 1.01 (.20)
PVT 0.01 1.01 (.00)*
GPA -0.68 0.51 (.05)***
Parental Presence -0.09 0.91 (.01)***
Constant -5.61 (.47)*** -1.97 (.66)**
† p<.10 * p<.05 **p<.01 ***p<.001
33
Table 6a provides the results of the logistic regression examining the relationship
between BOT birth order categorization and the likelihood of violence. The results presented in
Model 1 indicate no significant relationship between birth order and violence net of controls.
The results of the mediation analysis presented in Model 2 decrease the magnitude of the already
non-significant relationship between birth order and violence (reductions of 51 percent and 87
percent respectively). Family education, intact family, age, gender, twin status, GPA, parental
presence, and parental attachment all significantly decreased the likelihood of violence. Race
dummy variables of Hispanic and black significantly increased the likelihood of violence.
Model 3 presents the results of the curvilinearity analysis, and similar to Model 1, finds no
significant relationship between birth order and violence.
between BRH birth order categorization and the likelihood of violence. As with the examination
of BOT birth order categorization, no significant relationship is found between BRH birth order
categorization and likelihood of violence. The mediation analysis presented in Model 2 shows a
71 percent reduction in the magnitude of the already non-significant relationship. Again,
coefficients of the control variables do not significantly differ from one birth order categorization
to the other.
34
Table 6: Logistic Regression Predicting Violence (N=12,787)-Odds Ratio (SE)
Violence
Firstborn ~ ~ ~ ~ ~ ~ -0.01 0.99 (.06)
Middleborn 0.12 1.13 (.08) 0.06 1.06 (.09) 0.05 1.05 (.08)
Lastborn 0.07 1.07 (.06) 0.01 1.01 (.06) ~ ~ ~
Sibship Size 0.01 1.01 (.02) 0.01 1.01 (.02) 0.01 1.01 (.02)
Intact Family -0.25 0.78 (.05)*** -0.13 0.88 (.06)* -0.13 0.88 (.06)*
Age -0.12 0.88 (.02)*** -0.15 0.86 (.02)*** -0.15 0.86 (.02)***
Female -0.91 0.40 (.05)*** -0.85 0.43 (.05)*** -0.85 0.43 (.05)***
Hispanic 0.30 1.35 (.09)*** 0.21 1.24 (.09)* 0.21 1.24 (.09)*
Black 0.66 1.94 (.08)*** 0.52 1.68 (.08)*** 0.52 1.68 (.08)***
Asian 0.14 1.15 (.12) 0.17 1.18 (.12) 0.17 1.18 (.12)
Other Race 0.25 1.29 (.18) 0.13 1.14 (.18) 0.13 1.14 (.18)
Twins -0.27 0.76 (.11)* -0.28 0.76 (.11)** -0.28 0.76 (.11)**
PVT 0.00 1.00 (.00)* 0.00 1.00 (.00)*
GPA -0.45 0.64 (.04)*** -0.45 0.64 (.04)***
Parental Presence -0.03 0.97 (.01)*** -0.03 0.97 (.01)***
Parental Attachment -0.03 0.97 (.02) -0.03 0.97 (.02)
Constant 2.46 (.26)*** 4.96 (.40)*** 4.97 (.40)***
Model 1 Model 2
Laterborn 0.08 1.09 (.06) 0.02 1.02 (.06)
Sibship Size 0.01 1.02 (.02) 0.01 1.01 (.02)
Intact Family -0.24 0.78 (.05)*** -0.13 0.88 (.06)*
Age -0.12 0.89 (.02)*** -0.15 0.86 (.02)***
Female -0.91 0.40 (.05)*** -0.85 0.43 (.05)***
Hispanic 0.30 1.35 (.09)*** 0.22 1.24 (.09)*
Black 0.66 1.94 (.08)*** 0.52 1.68 (.08)***
Asian 0.14 1.15 (.12) 0.17 1.19 (.12)
Other Race 0.25 1.29 (.18) 0.13 1.14 (.18)
Twins -0.26 0.77 (.11)* -0.27 0.76 (.11)**
PVT 0.00 1 (.00)*
GPA -0.45 0.64 (.04)***
Parental Attachment -0.03 0.97 (.02)
Constant 2.44 (.26)*** 4.94 (.39)***
† p<.10 * p<.05 **p<.01 ***p<.001
35
between BOT birth order categorization and the likelihood of the commission of a non-violent
crime. Model 1 indicates that both middleborns and lastborns are significantly more likely to
engage in non-violent crime than firstborns (odds ratios of 1.35 and 1.15, respectively). Model 2
provides the results of the mediation analysis. As indicated by the results in Model 2, the
significant difference between firstborns and lastborns is explained away by the four
hypothesized mediating variables. The relationship between middleborns and firstborns
remained significant; however, a decrease in the magnitude of the difference was observed.
Specifically, the magnitude of the relationship between firstborns and lastborns was reduced by
40 percent, while the relationship between middleborns and firstborns was only reduced by 17
percent with the introduction of the mediating variables. Intact family, age, gender, GPA,
parental presence, and parental attachment significantly reduced the likelihood of engaging in
non-violent crime, while family education and race dummy variables (Hispanic, and other race)
significantly increased the likelihood. Model 3 provides the results of the curvilinearity analysis.
The results indicate that there is no significant curvilinear relationship between birth order and
non-violent crime; however, examination of the coefficients reveals, that while non-significant,
there exists a curvilinear relationship that would fall more in line with the predictions of birth
order theorists over resource dilution model theorists.
between BRH birth order categorization and the likelihood of the commission of a non-violent
crime. Model 1 indicates that, net of controls, laterborns are more likely than firstborns to
commit a non-violent crime (odds ratio of 1.20). Model 2 provides the results of the mediation
analysis. The results indicate that the relationship between birth order and non-violent crime
36
cannot be fully explained by controls or mediating variables. As with previously discussed
results, the relationship remains significant; however, a decrease in the magnitude of the
coefficient is observed (30 percent reduction). Control variables yield coefficients similar to the
coefficients presented in Table 7a.
37
Table 7: Logistic Regression Predicting Non-Violent Crime (N=12,787)-Odds Ratio (SE)
Non-Violent Crime
Firstborn ~ ~ ~ ~ ~ ~ -0.08 0.92 (.06)
Middleborn 0.30 1.35 (.09)** 0.25 1.28 (.09)** 0.17 1.18 (.09)
Lastborn 0.14 1.15 (.06)* 0.08 1.08 (.06) ~ ~ ~
Sibship Size -0.05 0.95 (.02)* -0.03 0.97 (.02) -0.03 0.97 (.02)
Family Education 0.04 1.04 (.02) 0.08 1.08 (.03)** 0.08 1.08 (.03)**
Intact Family -0.35 0.71 (.06)*** -0.21 0.81 (.06)** -0.21 0.81 (.06)**
Age -0.02 0.98 (.02) -0.08 0.93 (.02)*** -0.08 0.93 (.02)***
Female -0.57 0.56 (.05)*** -0.50 0.61 (.05)*** -0.50 0.61 (.05)***
Hispanic 0.32 1.37 (.10)** 0.34 1.41 (.09)*** 0.34 1.41 (.09)***
Black -0.12 0.88 (.09) -0.18 0.84 (.08)* -0.18 0.84 (.08)*
Asian 0.15 1.16 (.15) 0.26 1.30 (.16) 0.26 1.30 (.16)
Other Race 0.46 1.58 (.19)* 0.39 1.48 (.20)* 0.39 1.48 (.20)*
Twins -0.06 0.94 (.14) -0.03 0.97 (.15) -0.03 0.97 (.15)
PVT 0.01 1.01 (.00)*** 0.01 1.01 (.00)***
GPA -0.47 0.62 (.04)*** -0.47 0.62 (.04)***
Constant 0.24 (.31) 2.91 (.42)*** 2.99 (.41)***
Model 1 Model 2
Laterborn 0.18 1.20 (.06)** 0.13 1.14 (.06)*
Sibship Size -0.03 0.97 (.02) -0.01 0.99 (.02)
Family Education 0.04 1.04 (.02) 0.08 1.08 (.03)**
Intact Family -0.34 0.71 (.06)*** -0.20 0.82 (.06)**
Age -0.02 0.98 (.02) -0.08 0.93 (.02)***
Female -0.57 0.56 (.05)*** -0.50 0.61 (.05)***
Hispanic 0.32 1.38 (.10)** 0.35 1.42 (.09)***
Black -0.12 0.89 (.09) -0.18 0.84 (.08)*
Asian 0.15 1.17 (.15) 0.26 1.30 (.16)
Other Race 0.45 1.57 (.19)* 0.38 1.47 (.20)*
Twins -0.04 0.96 (.14) -0.01 0.99 (.15)
PVT 0.01 1.01 (.00)***
GPA -0.48 0.62 (.04)***
Constant 0.17 (.30) 2.84 (.41)***
† p<.10 * p<.05 **p<.01 ***p<.001
38
between BOT birth order categorization and the likelihood of binge drinking. The results
presented in Model 1 indicate that, net of controls, both middleborns and lastborns are
significantly more likely to engage in binge drinking than firstborns (odds ratios of 1.42 and 1.51
respectively). Model 2 provides the results of the mediation analysis. The mediation variables
are unable to render the relationship between birth order and binge drinking non-significant. A
reduction in the magnitude of both coefficients is observed, 19 percent and 12 percent
respectively. Race, twin status, GPA, parental presence, and parental attachment significantly
decreased the likelihood of binge drinking; age is the only control variable to significantly
increase the likelihood of binge drinking. Model 3 presents the results of the curvilinearity
analysis. The results indicate that no significant curvilinear relationship between birth order and
binge drinking exists; however, examination of the coefficients tends to lend support the
predictions based upon the resource dilution model, as lastborns have the highest likelihood of
binge drinking.
Table 8b provides the results of the logistic regression analyzing the relationship between
BRH birth order categorization and the likelihood of binge drinking. Model 1 indicates that net
of controls, laterborns are significantly more likely to engage in binge drinking than firstborns
(odds ratio of 1.48). Model 2 provides the results of the mediation analysis. The results indicate
that while a reduction (14 percent) in the magnitude of the effect of birth order is observed, the
significant relationship remains. Again, no significant differences between coefficients of the
control variables presented in Table 8a and Table 8b were observed.
39
Table 8: Logistic Regression Predicting Binge Drinking (N=12,787)-Odds Ratio (SE)
Binge Drinking
Firstborn ~ ~ ~ ~ ~ ~ -0.36 0.70 (.07)***
Middleborn 0.35 1.42 (.09)*** 0.28 1.33 (.09)** -0.08 0.93 (.08)
Lastborn 0.41 1.51 (.06)*** 0.36 1.43 (.07)*** ~ ~ ~
Sibship Size -0.07 0.93 (.03)** -0.05 0.95 (.03) -0.05 0.95 (.03)
Family Education -0.04 0.96 (.03) 0.00 1.00 (.03) 0.00 1.00 (.03)
Intact Family -0.29 0.75 (.07)*** -0.09 0.91 (.07) -0.09 0.91 (.07)
Age 0.40 1.49 (.02)*** 0.36 1.43 (.02)*** 0.36 1.43 (.02)***
Female -0.23 0.79 (.07)** -0.15 0.86 (.08) -0.15 0.86 (.08)
Hispanic -0.21 0.81 (.11)* -0.24 0.78 (.11)* -0.24 0.78 (.11)*
Black -1.06 0.35 (.12)*** -1.20 0.30 (.13)*** -1.20 0.30 (.13)***
Asian -0.80 0.45 (.19)*** -0.76 0.47 (.18)*** -0.76 0.47 (.18)***
Other Race -0.36 0.70 (.18) -0.49 0.61 (.18)* -0.49 0.61 (.18)*
Twins -0.43 0.65 (.16)* -0.45 0.64 (.16)** -0.45 0.64 (.16)**
PVT 0.00 1.00 (.00)† 0.00 1.00 (.00)†
GPA -0.47 0.62 (.05)*** -0.47 0.62 (.05)***
Parental Attachment -0.05 0.95 (.03)† -0.05 0.95 (.03)†
Constant -6.75 (.41)*** -3.95 (.50)*** -3.60 (.50)***
Model 1 Model 2
Laterborn 0.39 1.48 (.06)*** 0.34 1.40 (.07)***
Sibship Size -0.08 0.92 (.03)** -0.06 0.94 (.03)
Family Education -0.04 0.96 (.03) 0.00 1.00 (.03)
Intact Family -0.30 0.74 (.07)*** -0.09 0.91 (.07)
Age 0.40 1.49 (.02)*** 0.36 1.43 (.02)***
Female -0.23 0.79 (.07)** -0.15 0.86 (.08)
Hispanic -0.21 0.81 (.11)* -0.25 0.78 (.11)*
Black -1.06 0.35 (.12)*** -1.20 0.30 (.13)***
Asian -0.80 0.45 (.19)*** -0.77 0.47 (.18)***
Other Race -0.35 0.70 (.18) -0.49 0.61 (.18)*
Twins -0.44 0.64 (.16)* -0.46 0.63 (.16)**
PVT 0.00 1.00 (.00)†
GPA -0.47 0.62 (.05)***
Parental Attachment -0.05 0.95 (.03)†
Constant -6.73 (.41)*** -3.92 (.51)***
† p<.10 * p<.05 **p<.01 ***p<.001
40
Table 9a provides the results of the logistic regression analyzing the relationship between
BOT birth order categorization and the likelihood of sexual activity. Model 1 indicates that, net
of controls, there is a significant difference between firstborns and lastborns (odds ratio of 1.29)
and no significant difference between firstborns and middleborns. Model 2 provides the results
of the mediation analysis. After the introduction of the mediating variables into the model, the
difference between firstborns and lastborns remains significant, however, the mediating variables
caused a 25 percent reduction in the coefficient’s magnitude. Family education, intact family,
the race dummy variable of Asian, twin status, GPA, and parental presence significantly reduce
the likelihood of sexual activity, while age and the race dummy variable of black significantly
increased the likelihood of sexual activity. Model 3 presents the results of the curvilinearity
analysis. The results indicate no significant curvilinear relationship between birth order and
sexual activity. Examination of the non-significant coefficients provides some support for
predictions based upon the resource dilution model.
Table 9b provides the results of the logistic regression analyzing the relationship between
BRH birth order categorization and the likelihood of sexual activity. Model 1 indicates that net
of controls, laterborns are more likely to engage in sexual activity than firstborns (odds ratio of
1.24). Model 2 provides the results of the mediation analysis. The addition of the hypothesized
mediating variables fails to render the relationship between birth order and sexual activity non-
significant; only a reduction (32 percent) in the magnitude of the effect was observed. Control
variables in Table 9b display similar coefficients to control variables presented in Table 9a.
41
Table 9: Logistic Regression Predicting Sexual Activity (N=12,787)-Odds Ratio (SE)
Sexual Activity
Firstborn ~ ~ ~ ~ ~ ~ -0.19 0.83 (.08)*
Middleborn 0.13 1.14 (.10) 0.04 1.04 (.09) -0.15 0.86 (.08)†
Lastborn 0.25 1.29 (.08)*** 0.19 1.21 (.08)* ~ ~ ~
Sibship Size -0.11 0.90 (.03)*** -0.10 0.91 (.03)*** -0.10 0.91 (.03)***
Intact Family -0.53 0.59 (.05)*** -0.35 0.70 (.06)*** -0.35 0.70 (.06)***
Age 0.58 1.79 (.02)*** 0.56 1.76 (.03)*** 0.56 1.76 (.03)***
Female -0.13 0.88 (.06)* -0.02 0.98 (.06) -0.02 0.98 (.06)
Hispanic -0.11 0.89 (.11) -0.19 0.83 (.11) -0.19 0.83 (.11)
Black 0.92 2.51 (.10)*** 0.80 2.22 (.10)*** 0.80 2.22 (.10)***
Asian -0.75 0.47 (.22)*** -0.72 0.49 (.20)*** -0.72 0.49 (.20)***
Other Race 0.14 1.15 (.25) 0.01 1.01 (.25) 0.01 1.01 (.25)
Twins -0.37 0.69 (.14)** -0.39 0.68 (.15)** -0.39 0.68 (.15)**
PVT 0.00 1.00 (.00) 0.00 1.00 (.00)
GPA -0.54 0.58 (.04)*** -0.54 0.58 (.04)***
Parental Attachment -0.05 0.95 (.03)† -0.05 0.95 (.03)†
Constant -8.67 (.41)*** -5.86 (.52)*** -5.67 (.54)***
Model 1 Model 2
Laterborn 0.22 1.24 (.08)** 0.15 1.16 (.07)*
Sibship Size -0.12 0.88 (.03)*** -0.12 0.89 (.03)***
Intact Family -0.53 0.59 (.05)*** -0.36 0.70 (.06)***
Age 0.58 1.79 (.02)*** 0.56 1.76 (.03)***
Female -0.13 0.88 (.06)* -0.02 0.98 (.06)
Hispanic -0.12 0.89 (.11) -0.20 0.82 (.11)
Black 0.92 2.50 (.10)*** 0.80 2.22 (.10)***
Asian -0.75 0.47 (.22)*** -0.73 0.48 (.20)***
Other Race 0.14 1.15 (.25) 0.01 1.01 (.25)
Twins -0.39 0.68 (.15)** -0.41 0.67 (.15)**
PVT 0.00 1.00 (.00)
GPA -0.54 0.58 (.04)***
Parental Attachment -0.05 0.96 (.03)†
Constant -8.61 (.40)*** -5.80 (.51)***
† p<.10 * p<.05 **p<.01 ***p<.001
42
Table 10a provides the results of the logistic regression analyzing the relationship
between BOT birth order categorization and the likelihood of the violation of parents’ rules.
Model 1 indicates that neither lastborns nor middleborns are significantly more likely to violate
parents’ rules than firstborns net of controls. The results in Model 2 indicate that the non-
significant differences observed in Model 1 between firstborns and lastborns as well as firstborns
and middleborns are further reduced (differences were reduced in magnitude by 58 percent and
65 percent respectively). While GPA had the largest effect, I would contend that parental
attachment is of the greatest importance as it should be related to the violation of parents’ rules
and could be driving the relationship between GPA and violation of parents’ rules. Intact family,
gender, the race dummy variable black, GPA, parental presence, and parental attachment
significantly reduce the likelihood of the violation of parents’ rules, while age significantly
increases the likelihood. Model 3 presents the results of the curvilinearity analysis. The results
find no significant curvilinear relationship. Examination of the non-significant coefficients
provides some support for predictions based upon birth order theory.
Finally, Table 10b provides the results of the logistic regression analyzing the
relationship between BRH birth order categorization and the likelihood of the violation of
parents’ rules. Model 1 indicates that net of controls, laterborns are more likely than firstborns to
violate parents’ rules (odds ratio of 1.20). Model 2 presents the results of the mediation analysis.
As with the results of the mediation analysis using BOT birth order categorization, the results of
the mediation analysis using BRH birth order categorization prove to render the relationship
between birth order and the likelihood of the violation of parents’ rules non-significant (a
reduction in magnitude of 60 percent was observed). Control variables yield results similar to
the results of the BOT birth order categorization.
43
Table 10: Logistic Regression Predicting Violation of Parents' Rules (N=12,787)-Odds Ratio (SE)
Violation of Parents' Rules
Firstborn ~ ~ ~ ~ ~ ~ -0.07 0.93 (.10)
Middleborn 0.21 1.24 (.13) 0.08 1.08 (.13) 0.00 1.00 (.11)
Lastborn 0.17 1.19 (.09)† 0.07 1.07 (.10) ~ ~ ~
Sibship Size 0.05 1.05 (.03) 0.07 1.08 (.03)* 0.07 1.08 (.03)*
Family Education -0.05 0.95 (.03) 0.02 1.02 (.03) 0.02 1.02 (.03)
Intact Family -0.76 0.47 (.08)*** -0.55 0.58 (.08)*** -0.55 0.58 (.08)***
Age 0.31 1.36 (.02)*** 0.25 1.28 (.03)*** 0.25 1.28 (.03)***
Female -0.35 0.70 (.08)*** -0.27 0.76 (.08)*** -0.27 0.76 (.08)***
Hispanic -0.15 0.86 (.13) -0.25 0.78 (.13) -0.25 0.78 (.13)
Black -0.24 0.79 (.14) -0.43 0.65 (.14)*** -0.43 0.65 (.14)***
Asian -0.09 0.91 (.22) -0.15 0.86 (.21) -0.15 0.86 (.21)
Other Race 0.39 1.47 (.25) 0.21 1.23 (.27) 0.21 1.23 (.27)
Twins -0.31 0.73 (.19) -0.35 0.70 (.19)† -0.35 0.70 (.19)†
PVT 0.00 1.00 (.00) 0.00 1.00 (.00)
GPA -0.49 0.61 (.06)*** -0.49 0.61 (.06)***
Constant -6.25 (.39)*** -1.62 (.61)** -1.55 (.63)*
Model 1 Model 2
Laterborn 0.18 1.20 (.09)* 0.07 1.08 (.10)
Sibship Size 0.05 1.06 (.03)† 0.07 1.08 (.03)*
Family Education -0.05 0.95 (.03) 0.02 1.02 (.03)
Intact Family -0.76 0.47 (.08)*** -0.55 0.58 (.08)***
Age 0.31 1.36 (.02)*** 0.25 1.28 (.03)***
Female -0.35 0.70 (.08)*** -0.27 0.76 (.08)***
Hispanic -0.15 0.86 (.12) -0.25 0.78 (.13)
Black -0.24 0.79 (.14)† -0.43 0.65 (.14)***
Asian -0.09 0.91 (.22) -0.15 0.86 (.21)
Other Race 0.39 1.47 (.25) 0.21 1.23 (.27)
Twins -0.30 0.74 (.19) -0.35 0.70 (.19)†
PVT 0.00 1.00 (.00)
GPA -0.49 0.61 (.06)***
Constant -6.26 (.39)*** -1.62 (.62)**
† p<.10 * p<.05 **p<.01 ***p<.001
44
Examining the Effects of Birth Order on Delinquency: Within-Family Analyses
Despite the findings shown above of significant relationships between birth order and
delinquency, the question remains whether these associations are causal or spurious. To address
these concerns, Table 11 provides the results of the fixed-effects model examining within-family
effects of birth order on seatbelt non-use. The sample for the within-family analyses comes from
the sibling pairs data in AddHealth. Families were constructed by matching the family
identification numbers given to each survey participant. Note that it is important to insure that
the two samples (between-individual and within-family) are not markedly different from one
another. To test for sample differences, I performed independent samples t-testsxii for all
variables of interest. The results of the independent samples t-tests indicate that of the 16
predictor variables, seven predictor variables were significantly different between the samples.
However, these differences were trivial, even though some of the differences were statistically
significant. Overall, the within-family sample included slightly fewer laterborns, slightly smaller
sibships, and was slightly more racially diverse. Based upon the results of the independent
samples t-tests, the within-family sample appears to be generalizable to the full sample. Model 1
provides the results using BOT birth order categorization while Model 2 provides the results of
BRH birth order categorization. In Model 1, the results suggest that within-families, seatbelt
non-use is not significantly affected by an individual’s birth order. Like Model 1, Model 2 again
indicates that within-families, seatbelt non-use is not affected by an individual’s birth order.
Analysis of the control variables indicates that within families, gender, GPA, and parental
presence significantly reduce an individual’s likelihood of seatbelt non-use.
45
Table 11: Fixed Effects Logistic Regression Examining Within Family Differences
in Seatbelt Non-Use (N=526, 225 Families)-Odds Ratio (SE)
Seatbelt Non-Use
Model 1 Model 2
b OR (SE) b OR (SE)
Middleborn 0.11 1.11 (0.26) ~ ~ ~
Lastborn -0.04 0.96 (0.30) ~ ~ ~
Laterborn ~ ~ ~ 0.06 1.06 (0.25)
Age -0.02 0.98 (0.09) 0.01 1.01 (0.08)
Female -0.37 0.69 (0.22)† -0.37 0.69 (0.22)†
Twins 0.09 1.09 (0.47) 0.13 1.14 (0.47)
GPA -0.53 0.59 (0.17)*** -0.52 0.59 (0.16)***
PVT 0.00 1.00 (0.01) -0.01 0.99 (0.01)
Parental Attachment -0.18 0.84 (0.12) -0.18 0.84 (0.12)
Parental Presence -0.07 0.93 (0.04)† -0.07 0.93 (0.04)†
† p<.10 * p<.05 **p<.01 ***p<.001
Note: Model 1 tests Birth Order Theory, Model 2 tests Born to Rebel
Note: 1,457 families were dropped from the analyses due to lack
of variation among siblings
Table 12 provides the results of the fixed effects model examining within-family effects
of birth order on marijuana use. Model 1 indicates that within families marijuana use is not
affected by birth order. Model 2 indicates that within families, marijuana use is not affected by
birth order as categorized by BRH. In both models, differences in age significantly increased the
likelihood of marijuana use, while GPA was found to significantly decrease the likelihood of
marijuana use. The significance of age may be a result of opportunity as younger children may
not have yet had the opportunity to use marijuana.
46
in Marijuana Use (N=497, 213 Families)-Odds Ratio (SE)
Marijuana Use
Model 1 Model 2
b OR (SE) b OR (SE)
Middleborn 0.40 1.49 (0.28) ~ ~ ~
Lastborn -0.08 0.93 (0.28) ~ ~ ~
Laterborn ~ ~ ~ 0.16 1.17 (0.25)
Age 0.35 1.42 (0.12)** 0.41 1.51 (0.11)***
Female -0.26 0.77 (0.23) -0.22 0.80 (0.22)
Twins -0.20 0.81 (0.45) -0.25 0.78 (0.44)
GPA -0.62 0.54 (0.17)*** -0.62 0.54 (0.17)***
PVT 0.00 1.00 (0.01) 0.00 1.00 (0.01)
Parental Attachment -0.03 0.97 (0.13) -0.06 0.94 (0.13)
Parental Presence -0.03 0.97 (0.04) -0.03 0.97 (0.04)
† p<.10 * p<.05 **p<.01 ***p<.001
of birth order on non-violent crime. Within Model 1, the results indicate that birth order as
categorized by BOT has no effect upon non-violent crime when examining variation within-
families. Model 2, similar to Model 1, finds no significant effect for birth order as categorized
by BRH on non-violent crime. Age, gender, GPA, parental attachment, and parental presence
significantly reduced the likelihood of non-violent crime, while interestingly; twin status
significantly increased an individual’s likelihood.
47
in Non-Violent Crime (N=1,156, 493 Families)-Odds Ratio (SE)
Non-Violent Crime
Model 1 Model 2
b OR (SE) b OR (SE)
Middleborn -0.02 0.98 (0.16) ~ ~ ~
Lastborn -0.13 0.88 (0.19) ~ ~ ~
Laterborn ~ ~ ~ -0.05 0.95 (0.15)
Age -0.18 0.84 (0.06)** -0.16 0.85 (0.06)**
Female -0.72 0.49 (0.17)*** -0.71 0.49 (0.16)***
Twins 0.71 2.03 (0.33)* 0.73 2.08 (0.33)*
GPA -0.40 0.67 (0.12)*** -0.40 0.67 (0.12)***
PVT 0.01 1.01 (0.01) 0.01 1.01 (0.01)
Parental Attachment -0.17 0.85 (0.08)* -0.16 0.85 (0.08)*
Parental Presence -0.08 0.93 (0.03)** -0.08 0.92 (0.03)**
† p<.10 * p<.05 **p<.01 ***p<.001
Table 14 provides the results of the fixed effects models examining within-family effects
of birth order on binge drinking. Both Model 1 and Model 2 provide results suggesting that birth
order has no significant effect upon an individual’s likelihood of engaging in binge drinking
when conducting a within-family analysis. GPA and parental attachment were found to
significantly reduce the likelihood of binge drinking, while age significantly increased the
likelihood.
48
in Binge Drinking (N=900, 385 Families)-Odds Ratio (SE)
Binge Drinking
Model 1 Model 2
b OR (SE) b OR (SE)
Middleborn -0.02 0.98 (0.20) ~ ~ ~
Lastborn -0.02 0.98 (0.22) ~ ~ ~
Laterborn ~ ~ ~ -0.02 0.98 (0.19)
Age 0.38 1.46 (0.08)*** 0.38 1.46 (0.07)***
Female -0.26 0.77 (0.18) -0.26 0.77 (0.18)
Twins 0.16 1.17 (0.36) 0.16 1.17 (0.36)
GPA -0.51 0.60 (0.14)*** -0.51 0.60 (0.14)***
PVT -0.01 0.99 (0.01) -0.01 0.99 (0.01)
Parental Attachment -0.27 0.76 (0.10)** -0.27 0.76 (0.10)**
Parental Presence -0.03 0.97 (0.03) -0.03 0.97 (0.03)
† p<.10 * p<.05 **p<.01 ***p<.001
of birth order on sexual activity. As with the previous fixed effects analyses, no significant birth
order effects were observed in either Model 1 or Model 2. GPA and parental presence
significantly reduced the likelihood of sexual activity, while age and PVT significantly increased
the likelihood.
49
in Sexual Activity (N=1022, 444 Families)-Odds Ratio (SE)
Sexual Activity
Model 1 Model 2
b OR (SE) b OR (SE)
Middleborn -0.18 0.83 (0.20) ~ ~ ~
Lastborn 0.05 1.05 (0.21) ~ ~ ~
Laterborn ~ ~ ~ -0.08 0.92 (0.18)
Age 0.67 1.96 (0.09)*** 0.64 1.90 (0.08)***
Female -0.20 0.82 (0.18) -0.20 0.82 (0.18)
Twins -0.56 0.57 (0.35) -0.54 0.58 (0.35)
GPA -0.30 0.74 (0.14)* -0.29 0.75 (0.13)*
PVT 0.01 1.01 (0.01) 0.01 1.01 (0.01)
Parental Attachment 0.03 1.03 (0.09) 0.02 1.02 (0.09)
Parental Presence -0.10 0.90 (0.03)*** -0.10 0.91 (0.03)***
† p<.10 * p<.05 **p<.01 ***p<.001
In summary, I find significant effects for birth order while controlling for individual and
family characteristics when conducting analyses at the between-individual level. When
analyzing the effects of birth order at the within-family level, however, the effects of birth order
become statistically non-significant. As I discuss more fully below, this non-significance at the
within-family level leads me to question the impact of birth order on delinquency.
SECTION 6: DISCUSSION AND CONCLUSION
Research concerning the effects of birth order has a long and storied past, complete with
peaks of popularity (1950’s, 1960’s, and 1990’s) and valleys of disparagement (1970’s and
1980’s). A recent re-imagination of birth order theory by Sulloway (1996) has served to reignite
50
debate surrounding the true effects of birth order. Given this rebirth of interest into the effects of
birth order (especially concerning intelligence), criticisms have re-emerged questioning previous
studies’ use of between-individual designs (Retherford and Sewell 1991, Rodgers et al. 2000,
Rodgers 2001, and Wichman 2006). Birth order theory, Sulloway’s born to rebel hypothesis,
and the resource dilution model all hypothesize that the effects of birth order lie within-families;
yet, the majority of studies designed to test this very hypothesis rely on a between-individual
comparisons. This thesis sought to build upon this body of work, especially the research of
Rahav (1980) and Argys et al. (2006), in determining the effects of birth order upon delinquency
utilizing both between-individual and within-family designs.
The results from the between-individual analyses provide mixed support for Hypothesis 1
and Hypothesis 2. The delinquency outcomes of seatbelt non-use, marijuana use, and non-
violent crime provide support for Adlerian birth order theory, which predicts a curvilinear
relationship in which middleborn children will have the highest likelihood of delinquency,
followed by lastborn children and firstborn children respectively (see Figure A). The
delinquency outcomes of binge drinking and sexual activity provide support for the resource
dilution model, which predicts a linear relationship in which lastborn children will have the
highest likelihood of delinquency followed by middleborn children and firstborn children
respectively (see Figure A). In the between-individual analyses, Sulloway’s born to rebel
hypothesis was supported across the following outcomes: seatbelt non-use, marijuana use, non-
violent crime, binge drinking, and sexual activity. The findings regarding the born to rebel
hypothesis provide consistent support for Hypothesis 3 across a range of outcomes.
Nonetheless, given the recent criticisms regarding between-individual study designs, I
utilized a fixed effects methodology and re-analyzed all statistically significant outcomes. The
51
results of the within-family analyses proved strong and consistent support for Hypothesis 4.
Across all outcomes that were significant in the between-individual analyses (e.g., seatbelt non-
use, marijuana use, non-violent crime, binge drinking, and sexual activity), non-significant
relationships between the delinquent outcomes and birth order are observed in the within-family
analyses. It is important to note that due to data restrictions, the within-family models utilize a
smaller sample than the between-individual models. Given the lack of significance within-
families, the effects of birth order observed between-individuals seem to indicate that some
unobserved family characteristic is driving the relationship between birth order and delinquency.
This finding adds to a growing literature criticizing the continued study of birth order without
adequately considering possible unobserved or spurious influences.
This research improves upon previous research with its use of sophisticated statistical
procedures, a large nationally representative sample, and examination of both between-
individuals and within-family differences. The combination of the methods and data yield more
reliable and convincing results, especially when compared to previous analyses that have had
difficulty in controlling for potentially confounding variables and selection effects. Nonetheless,
this research has some limitations. First, the explanatory variables I utilized in this study might
not reflect difficult to measure concepts such as rebellion. Though I used a variety of outcomes
to assess the robustness of my results, still I might not be able to capture the ―rebelliousness‖ (or
openness to radical change) that Sulloway found in his historical analysis. Second, the within-
family analyses utilize a far smaller sample than the between-individual analyses. The use of the
smaller sample could potentially be underestimating the significance of the effects of birth order.
Nonetheless, the sample size ranges from 233 to 551 cases, and to my knowledge, no other
dataset exists that would have a greater number of sibling pairs. While my findings suggest an
52
unobserved family factor driving the relationship between birth order and delinquency, the
analyses do not necessarily provide a clear picture of what the unobserved factor could be.
Based upon research examining the link between families and crime, as well as siblings and
crime, I can speculate as to the possible unobserved factors. First, while my analyses include
measures of parental attachment and parental supervision, my analyses do not control for the
quality of parenting that fluctuates from family to family. Research by Larzelere and Patterson
(1990) provides evidence that family management practices (specifically discipline and
supervision) are strong predictors of delinquency. The impact of parental discipline in the
relationship between birth order and delinquency would in theory restrict juveniles’ opportunity
structures to commit delinquent acts, as well as serve as a deterrent to repeated offenses.
Second, based upon research by Haynie (2001, 2002) and Haynie and Osgood (2005), the
significant differences observed in the between-individual analyses may be a product of
delinquent peers or delinquent peer networks. My analyses are focused specifically on variables
related to families, thus neglecting other potentially influential factors such as peers. Finally,
research by Widmer (1997), Conger (1999), Slomkowski et al. (2001), and Conger et al. (2003)
suggests that older siblings are influential in shaping the behaviors and norms adopted by
younger siblings. Therefore, adolescents with older siblings who exhibit deviant behavior will
be more likely to mimic this behavior at an earlier age. Given differences in sibling spacing or
density, one could explain the between-individual results as an example of younger siblings
mimicking the behavior of older siblings who are not necessarily included in this study.
In conclusion, the results of the analyses provide evidence that birth order does not play a
significant role in shaping adolescents’ deviant behavior. This thesis indicates that birth order
need not be cause for concern for criminologists. While birth order theories, as evidenced by the
53
results, are not necessary in criminological research, the analyses point to other family
mechanisms (intact family, family education, parental supervision, and parental attachment) that
offer consistent predictions in regards to delinquent outcomes. Examination of the direct effect
of birth order may be misplacing the focus of what matters more, i.e. family-level social
processes. While direct effects were not observed, it may be in haste to neglect the study of the
indirect effects of birth order that may condition reporting of certain family mechanisms (e.g.,
parental attachment, and parental supervision). Further research is necessary in criminology to
test for the effects of the family mechanisms discussed above, as well as analyze the influence of
the sibling relationship in regards to delinquent or deviant outcomes.
54
REFERENCES
Adams, Bert N. 1972. ―Birth Order: A Critical Review.‖ Sociometry 35(3): 411-439
Adams, Russell L. and Beeman N. Phillips. 1972. ―Motivational and Achievement Differences
Among Children of Various Ordinal Birth Positions.‖ Child Development 43: 155-164
Adler, Alfred. 1928. ―Characteristics of the First, Second, Third Child.‖ Children 3: 24 and 52
Altus, William D. 1966. ―Birth Order and Its Sequelae.‖ Science 151: 44-49
Argys, Laura M., Daniel I. Rees, Susan L. Averett, and Benjama Witoonchart. 2006. ―Birth
Order and Risky Adolescent Behavior.‖ Economic Inquiry 44(2): 215-233
Baskett, Linda Musun. 1984. ―Ordinal Position Differences in Children’s Family Interactions.‖
Developmental Psychology 20(6): 1026-1031
Becker, Selwyn W. and Jean Carroll. 1962. ―Ordinal Position and Conformity.‖ Journal of
Abnormal and Social Psychology 65(2): 129-131
Berbaum, Michael L., Gregory B. Markus and R. B. Zajonc. 1982. ―A Closer Look at
Galbraith’s ―Closer Look.‖ Developmental Psychology 18(2): 174-180
Black, Sandra E., Paul J. Devereux and Kjell G. Salvanes. 2007. ―Older and Wiser? Birth Order
and IQ of Young Men.‖ NBER Working Paper Series 13237: 1-25
Blake, Judith. 1981. ―Family Size and the Quality of Children.‖ Demography 18(4): 421-442
Blake, Judith. 1989a. Family Size and Achievement. London: University of California Press
Blake, Judith. 1989b. ―Number of Siblings and Educational Attainment.‖ Science 245: 32-36
Blau, Peter M. and Otis Dudley Duncan. 1967. The American Occupational Structure. New
York: Wiley
Bragg, Barry W. and Vernon L. Allen. 1970. ―Ordinal Position and Conformity: A Role Theory
Analysis.‖ Sociometry 33(4): 371-381
Brackbill, Yvonne and Paul L. Nichols. 1982. ―A Test of the Confluence Model of Intellectual
Development.‖ Developmental Psychology 18(2): 192-198
Breland, Hunter M. 1973. ―Birth Order Effects: A Reply to Schooleer.‖ Psychological Bulletin
80(3): 210-212
Capra, Paul C. and James E. Dittes. 1962. ―Birth Order as a Selective Factor among Volunteer
Subjects.‖ Journal of Abnormal and Social Psychology 64(4): 302
Clark, Roger D. and Glenn A. Rice. 1982. ―Family Constellations and Eminence: The Birth
Orders of Nobel Prize Winners.‖ Journal of Psychology 110(2): 281-287
Conger, K.J. 1999. Older siblings’ influence on girls’ antisocial behavior: Does gender matter?
Paper presented at biennial meeting of Sociology for Research in Child Development,
Albuquerque, NM
Conger, K.J., E. Heylen, C. Slomkowski and R. Rende. 2003. Adolescent girls’ problem
behaviors: Evaluating the influence of older siblings. Manuscript in preparation
Downey, Douglas B. 1995. ―When Bigger is not Better: Family Size, Parental Resources, and
Children’s Educational Performance.‖ Family Size and Educational Performance 60:746-
761
Downey, Douglas B. 2001. ―Number of Siblings and Intellectual Development: The Resource
Dilution Explanation.‖ American Psychologist 56: 497-504
55
Downey, Douglas B., Brian Powell, Lala Carr Steelman and Shana Pribesh. 1999. ―Much Ado
About Siblings: Change Models, Sibship Size, and Intellectual Development.‖ American
Sociological Review 64: 193-198
Dumont A. 1890. Civilization et Depopulation. Paris: LeCronsier et Babe
Edwards, John N. and David L. Klemmack. 1973. ―Birth Order and the Conservators of
Tradition Hypothesis.‖ Journal of Marriage and the Family November: 619-626
Ernst, Cecile and Jules Angst. 1983. Birth Order: Its Influence on Personality. Berlin: Springer-
Verlag
Eisenman, Russell. 1987. ―Creativity, Birth Order, and Risk Taking.‖ Bulletin of the
Psychonomic Society 25(2): 87-88
Falaris, Evangelos M. and H. Elizabeth Peters. 1998. ―Survey Attrition and Schooling Choices.‖
The Journal of Human Resources 33: 531-554
Falbo, Toni. 1984. The Single-Child Family. New York: Guilford Press
Forer, Lucille K. 1969. Birth Order and Life Roles. Springfield: Charles C Thomas
Forer, Lucille K. and Henry Still. 1976. Birth Order Factor: How Your Personality is Influenced
by Your Place in the Family. New York: David McKay Company, Inc.
Freese, Jeremy, Brian Powell and Lala Carr Steelman. 1999. ―Rebel Without a Cause or Effect:
Birth Order and Social Attitudes.‖ American Sociological Review 64(2): 207-231
Galbraith, Richard C. 1982a. ―Just One Look was All it Took: Reply to Berbaum, Markus, and
Zajonc.‖ Developmental Psychology 18(2): 181-191
Galbraith, Richard C. 1982b. ―Sibling Spacing and Intellectual Development: A Closer Look at
the Confluence Models.‖ Developmental Psychology 18(2): 151-173
Guo, Guang and Leah K. VanWey. 1999. ―Sibship Size and Intellectual Development: Is the
Relationship Causal?‖ American Sociological Review 64: 169-187
Harris, Kathleen Mullan. 2008. The National Longitudinal Study of Adolescent Health (Add
Health), Waves I & II, 1994–1996; Wave III, 2001–2002 [machine-readable data file and
documentation]. Chapel Hill, NC: Carolina Population Center, University of North
Carolina at Chapel Hill.
Harris, Kathleen Mullan, Carolyn Tucker Halpern, Pamela Entzel, Joyce Tabor, Peter S.
Bearman, and J. Richard Udry. 2008. The National Longitudinal Study of Adolescent
Health: Research Design [WWW document]. URL:
http://www.cpc.unc.edu/projects/addhealth/design.
Hauser, Robert M. and William H. Sewell. 1985. ―Birth Order and Educational Attainment in
Full Sibships.‖ American Educational Research Journal 22: 1-23
Hausman, J. A. 1978. ―Specification Tests in Econometrics.‖ Econometrica, 46(6): 1251-1271
Haynie, Dana L. 2001. ―Delinquent Peers Revisited: Does Network Structure Matter.‖ American
Journal of Sociology 106(4): 1013-1057
Haynie, Dana L. 2002. ―Friendship Networks and Delinquency: The Relative Nature of Peer
Delinquency.‖ Journal of Quantitative Criminology 18(2): 99-134
Haynie, Dana L. and D. Wayne Osgood. 2005. ―Reconsidering Peers and Delinquency: How do
Peers Matter?‖ Social Forces 84(2): 1109-1130
Healey, Matthew D. and Bruce J. Ellis. 2007. ―Birth Order, Conscientiousness, and Openness to
Experience: Tests of the Family-Niche Model of Personality Using a Within-Family
Methodology.‖ Evolution and Human Behavior 28: 55-59
Heaven, Patrick C. L. 1994. ―Family of Origin, Personality, and Self-Reported Delinquency.‖
Journal of Adolescence 14: 445-459
56
Hirschi, Travis C. 1969. Causes of Delinquency. Berkeley: University of California Press.
Kammeyer, Kenneth. 1967. ―Birth Order as a Research Variable.‖ Social Forces 46: 71-80
Larzelere, Robert E. and Gerald R. Patterson. 1990. ―Parental Management: Mediator of the
Effect of Socioeconomic Status on Early Delinquency.‖ Criminology 28(2): 301-324
Leman, Kevin. 2002. The New Birth Order Book: Why You Are the Way You Are. Grand Rapids:
Fleming H. Revell
Nisbett, Richard E. 1968. ―Birth Order and Participation in Dangerous Sports.‖ Journal of
Personality and Social Psychology 8(4): 351-353
Paulhus, Delroy L., Paul D. Trapnell and David Chen. 1999. ―Birth Order Effects on Personality
and Achievement Within Families.‖ Psychological Science 10(6): 482-488
Polit, Denise F. and Toni Falbo. 1987. ―Only Children and Personality Development: A
Quantitative Review.‖ Journal of Marriage and the Family 49: 309-325
Rahav, Giora. 1980. ―Birth Order and Delinquency.‖ British Journal of Criminology 20(4): 385-
395
Retherford, R. D. and W. H. Sewell. 1991. ―Birth Order and Intelligence: Further Tests of the
Confluence Model.‖ American Sociological Review 56: 141-158
Rodgers, J. L., H. H. Cleveland, E. van den Oord, and D. C. Rowe. 2000. ―Resolving the Debate
Over Birth Order, Family Size, and Intelligence.‖ American Psychologist 55: 599-612
Rodgers, Joseph Lee. 2001. ―What Causes Birth Order-Intelligence Patterns? The Admixture
Hypothesis, Revived.‖ American Psychologist 56: 505-510
Runco, Mark A. and Michael D. Bahleda. 1987. ―Birth-Order and Divergent Thinking.‖ Journal
of Genetic Psychology 148(1): 119-125
Salmon, Catherine A. and Martin Daly. 1998. ―Birth Order and Familial Sentiment: Middleborns
are Different.‖ Evolution and Human Behavior 19: 299-312
Sampson, Edward E. 1962. ―Birth Order, Need Achievement, and Conformity.‖ Journal of
Abnormal and Social Psychology 64(2): 155-159
Sampson, Robert J. and W. Byron Groves. 1989. ―Community Structure and Crime: Testing
Social-Disorganization Theory.‖ American Journal of Sociology 94(4):774-802
Sampson, Robert J. and John H. Laub. 1990. ―Crime and Deviance Over the Life Course: The
Salience of Adult Social Bonds.‖ American Sociological Review 55:609-627
Schooler, Carmi. 1972. ―Birth Order Effects: Not Here, Not Now!‖ Psychological Bulletin 78(3):
161-175
Schooler, Carmi. 1973. ―Birth Order Effects: A Reply to Breland.‖ Psychological Bulletin 80(3):
213-214
Seff, Monica A., Viktor Gecas and James H. Frey. 1993. ―Birth Order, Self-Concept and
Participation in Dangerous Sports.‖ The Journal of Psychology 127(2): 221-232
Slomkowski, C., R. Rende, K.J. Conger, R.L. Simons and R.D. Conger. 2001. ―Sisters, brothers,
and delinquency: Evaluating social influence during early and middle adolescence.‖
Child Development 72: 271-283
Steelman, Lala Carr and Brian Powell. 1985. ―The Social and Academic Consequences of Birth
Order: Real, Artifactual, or Both?‖ Journal of Marriage and the Family February: 117-
124
Steelman, Lala Carr, Brian Powell, Regina Werum and Scott Carter. 2002. ―Reconsidering the
Effects of Sibling Configuration: Recent Advances and Challenges.‖ Annual Review of
Sociology 28: 243-269
57
Sulloway, Frank J. 1996. Born to Rebel: Birth Order, Family Dynamics, and Creative Lives.
New York: Pantheon Books
Toman, Walter. 1993. Family Constellation: Its Effects on Personality and Social Behavior. New
York: Springer Publishing Company
Tygart, C.E. 1991. ―Juvenile Delinquency and Number of Children in a Family: Some Empirical
and Theoretical Updates.‖ Youth and Society 22(4): 525-536
Wichman, A. L., J. L. Rodgers, and R. C. MacCallum. 2006. ―A Multilevel Approach to the
Relationship Between Birth Order and Intelligence.‖ Personality and Social Psychology
Bulletin 32(1): 117-127
Wichman, A. L., J. L. Rodgers, and R. C. MacCallum. 2007. ―Birth Order Has No Effect on
Intelligence: A Reply and Extension of Previous Findings.‖ Personality and Social
Psychology Bulletin 33: 1195-1200
Widmer, Eric D. 1997. ―Influence of Older Siblings on Initiation of Sexual Intercourse.‖ Journal
of Marriage and the Family 59: 928-938
Wilson, Bradford and George Edington. 1981. First Child, Second Child…: Your Birth Order
Profile. New York: McGraw-Hill Book Company
Zajonc, R.B. 1976. ―Family Configuration and Intelligence: Variations in Scholastic Aptitude
Scores Parallel Trends in Family Size and the Spacing of Children.‖ Science 192: 227-
236
Zajonc, R. B. and P.R. Mullaly. 1997. ―Birth Order: Reconciling Conflicting Effects.‖ American
Psychologist 52: 685-699
Zajonc, R. B. and F. J. Sulloway. 2007. ―The Confluence Model: Birth Order as a Within-Family
or Between-Family Dynamic?‖ Personality and Social Psychology Bulletin 33: 1187-
1194
Zweigenhaft, Richard L. 2002. ―Birth Order Effects and Rebelliousness: Political Activism and
Involvement with Marijuana.‖ Political Psychology 23(2): 219-233
Zweigenhaft, Richard L. and Jessica Von Ammon. 2000. ―Birth Order and Civil Disobedience:
A Test of Sulloway’s ―Born to Rebel‖ Hypothesis.‖ The Journal of Social Psychology
140(5): 624-627
i
This research uses data from Add Health, a program project designed by J. Richard Udry, Peter
S. Bearman, and Kathleen Mullan Harris, and funded by a grant P01-HD31921 from the Eunice
Kennedy Shriver National Institute of Child Health and Human Development, with cooperative
funding from 17 other agencies. Special acknowledgment is due Ronald R. Rindfuss and Barbara
Entwisle for assistance in the original design. Persons interested in obtaining data files from Add
Health should contact Add Health, Carolina Population Center, 123 W. Franklin Street, Chapel
Hill, NC 27516-2524 (addhealth@unc.edu). No direct support was received from grant P01-HD31921 for this
analysis.
58
ii
For graphic representation, see Figure A.
iii
In an effort to avoid confusion, the term “single child” will be used when discussing children with no siblings,
otherwise known as only children.
iv
Laterborns is a term used to describe children in the family that were not firstborns.
v
For graphic representation, see Figure A.
vi
Sibship is a term that refers to all siblings within one family.
vii
This paper focuses on examining the link between biological birth order and delinquency. Thus, personality
profiles of step children and adopted children will not be discussed at length. However, it is theorized (Toman
1993) that step children will typically keep the role they filled in their biological family with the possibility of
playing multiple roles; adopted children will typically accept the role associated with their ordinal position in terms
of the rest of the siblings.
viii
Twins are not mentioned at exceptional length within this study as it has been theorized (Wilson 1981, Toman
1993) that twins will typically adopt the same birth order role aside from the role of an only child; thus the
possibility exists of multiple firstborns, middleborns, and youngest children.
ix
By analyzing both published and unpublished work, Schooler avoided potential biases associated with the “file
drawer effect.”
x
Supplemental analyses included and age squared term, though this term was found to be non-significant and was
dropped from the analyses presented.
xi
For all analyses using BOT birth order categorization, firstborns served as the reference category in Model 1 and
Model 2. In Model 3, lastborns served as the reference category.
xii
Results of the independent samples t-tests are available from author upon request.
59

Birth Order Thesis

Uploaded by

Document Information

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Birth Order Thesis

Uploaded by

Copyright:

Available Formats

The Pennsylvania State University

The Graduate School

College of the Liberal Arts

ORDERED DELINQUENCY: THE EFFECTS OF BIRTH ORDER ON

Crime, Law and Justice

Copyright 2010 Patrick R. Cundiff

Submitted in Partial Fulfillment

*Signatures are on file in the Graduate School

LIST OF FIGURES: ................................................................................................................................... v

LIST OF TABLES: .....................................................................................................................................vi

SECTION 2: THEORETICAL PERSPECTIVES ...................................................................................... 2

Birth Order Theory ................................................................................................................................. 2

Resource Dilution Model ........................................................................................................................ 7

Spurious Arguments ............................................................................................................................... 9

SECTION 3: CURRENT STUDY ............................................................................................................ 13

SECTION 4: METHODS ......................................................................................................................... 14

Predictor Variables ........................................................................................................................... 18

Analytic Strategy .................................................................................................................................. 23

SECTION 5: RESULTS ........................................................................................................................... 24

Comparing Single Children and Firstborns ........................................................................................... 24

Examining the Effects of Birth Order on Potential Mediating Variables .............................................. 25

Examining the Effects of Birth Order on Delinquency: Between-Individual Analyses ......................... 27

Examining the Effects of Birth Order on Delinquency: Within-Family Analyses................................. 45

SECTION 6: DISCUSSION AND CONCLUSION ................................................................................. 50

Table 3: OLS Regressions Predicting Potential Mediating Variables

Table 4: Logistic Regression Predicting Seatbelt Non-Use

Table 5: Logistic Regression Predicting Marijuana Use

Table 6: Logistic Regression Predicting Violence

Table 7: Logistic Regression Predicting Non-Violent Crime

Table 8: Logistic Regression Predicting Binge Drinking

Table 9: Logistic Regression Predicting Sexual Activity

Table 10: Logistic Regression Predicting Violation of Parents' Rules

argued that laterbornsivare more likely to rebel than firstborns.

resources than children of the first or middle ordinal positions.

SECTION 2: THEORETICAL PERSPECTIVES

Birth Order Theory

constantly in competition with a larger and more intelligent rival.

firstborns have a greater likelihood of educational attainment and intellectual professional

more likely to engage in risk taking than laterborns.

most prone to difficulties in conventional achievement. Research conducted by Blau and

advantageous and disadvantageous. While the youngest child is characterized as more

Resource Dilution Model

socio-economic resources, supervision) and propensities for delinquency. Using theories of

biological capabilities of parents and dilution of parents’ resources by other children.

introduced by looking at the proportionate prevalence of a particular birth order in a selected

size and educational attainment to be non-existent or weak.

differences in intellectual development by ordinal position. According to the confluence model,

ordinal position scores higher than its predecessor.

by better controlling for possible confounding variables.

effects are completely erased by controlling for population biases or socioeconomic-status

believe that any observed birth order effects may be artificial.

In addition to tests of Sulloway’s born to rebel hypothesis, many researchers (Retherford

appropriately control for influences on intelligence that vary across families.

order at both levels.

then seek to clarify the mechanism at work.

hypothesis, and the resource dilution model.

questioned whether the effects of birth order are causal.

Figure A: Graphic Representations of Predicted Likelihood of Delinquency

Firstborn Middleborn Lastborn

Adolescent Health (Add Health). Add Health is a longitudinal study of a nationally

probability of selection. Incorporating systematic sampling methods and implicit stratification

size dependent upon outcome variables.

coefficient for this measure was .75.