Don’t trust anyone over 30: Parental legitimacy as a mediator between
parenting style and changes in delinquent behavior over time
, Ellen S. Cohn, Cesar J. Rebellon, Karen Van Gundy
University of New Hampshire, Durham NH 03824, USA
Both law and society scholars and developmental psychologists have focused on the
legitimacy of authority ﬁgures, although in different domains (police versus parents). The
purpose of the current research is to bridge these two ﬁelds by examining the relations
among parenting style (i.e., authoritarian, authoritative, permissive), the perception of
parental legitimacy, and changes in delinquency over time. It is hypothesized that parental
legitimacy mediates the relation between parenting style and future delinquent behavior.
Middle school and high school students completed questionnaires three times over
a period of 18 months. Parenting style and delinquent behavior were measured at time 1,
parental legitimacy at time 2, and delinquency again at time 3. The results show that
authoritative parenting was positively related to parental legitimacy, while authoritarian
parenting was negatively associated with parental legitimacy. Furthermore, parental
legitimacy was negatively associated with future delinquency. Structural equation
modeling indicated that parental legitimacy mediated the relation between parenting
styles and changes in delinquency over the 18-month time period. The implications for
parenting style and parental legitimacy affecting delinquent behavior are discussed.
Ó2011 The Foundation for Professionals in Services for Adolescents. Published by Elsevier
Ltd. All rights reserved.
As they grow older, adolescents begin to exert their autonomy by deﬁning more areas of their lives as outside of
parental control and authority (Smetana, 2002). Consequently, parents may feel that they are sometimes losing the battle
in exerting their inﬂuence over their children and enforcing rules. One factor that may aid parents in this endeavor is the
extent to which their children perceive them as legitimate authority ﬁgures. Indeed, research by law and society scholars
has shown that adolescents’perception of the legitimacy of the legal system has a direct effect on their delinquent
behavior (Fagan & Tyler, 20 05). However, legal scholars have only examined the legitimacy of legal and government
authorities (e.g., police), making it unclear if their ﬁndings generalize to authorities outside of the legal system, such as
parents. Developmental psychologists, on the other hand, have examined the effects of legitimacy of parental authorities
on compliance, but have largely ignored research by legal scholars examining the relation between the perceptions of
legitimacy and delinquency (e.g., Darling, Cumsille, & Martínez, 2007). No researcher has examined the relation between
parental legitimacy and delinquent behavior.
Additionally, although developmental researchers highlight individual variation in when and where adolescents view
their parents as legitimate authorities, relatively little is known about what factors create this individual variation (Darling
*Corresponding author. Department of Psychology, University of New Hampshire, 10 Library Way, Durham NH 03824, USA. Tel.: þ1 603 862 1619;
fax: þ1 603 862 4986.
E-mail addresses: firstname.lastname@example.org (R. Trinkner), Ellen.Cohn@unh.edu (E.S. Cohn), Cesar.Rebellon@unh.edu (C.J. Rebellon), Karen.VanGundy@unh.
edu (K.V. Gundy).
Contents lists available at ScienceDirect
Journal of Adolescence
journal homepage: www.elsevier.com/locate/jado
0140-1971/$ –see front matter Ó2011 The Foundation for Professionals in Services for Adolescents. Published by Elsevier Ltd. All rights reserved.
Journal of Adolescence 35 (2012) 119–132
et al., 2007). The styles parents use to rear their children (Baumrind, 1967, 1971, 1991) may potentially have powerful
inﬂuences on adolescents’perceptions of parental authority, although these inﬂuences have not been examined empirically.
Moreover, while researchers have found some evidence that parenting styles inﬂuence adolescents’engagement in delin-
quent behavior (e.g., Simons, Simons, Burt, Brody, & Cutrona, 2005), little is known about the mechanisms that mediate this
relation. Perceptions of parental legitimacy may not only be inﬂuenced by parenting styles, but also may be a mechanism by
which parenting styles affect adolescent delinquency. However, this has not yet been investigated.
This is the ﬁrst study to address speciﬁcally if parenting styles inﬂuence adolescents’beliefs about the legitimacy of
parental authority and if those perceptions affect adolescent reports of delinquent behavior. Our primarygoal was to examine
if adolescents’perceptions of parental legitimacy mediated the relation between parenting style and future delinquent
behavior. To assess this relation, we developed a measure of parental legitimacy based on a measure of legal legitimacy
developed by legal scholars (Sunshine & Tyler, 2003). We then tested if parental legitimacy mediated the relation between
parenting styles and changes in delinquent behavior over an 18-month time period in a sample of middle school and high
Deﬁnitions and conceptions of legitimacy abound within past research (Jost & Major, 2001). For the purposes of the
present paper, the concept of legitimacy is deﬁned as a psychological property of an individual that leads others to perceive
his or her authority as appropriate, proper, and just (Tyler, 2006a, 2006b). Within this framework, legitimacy perceptions
have two primary components: the extent to which an individual trusts an authority and the extent to which he or she feels
an obligation to obey the directives of that authority (Sunshine & Tyler, 2003).
Legitimacy is an important property for any authority ﬁgure to possess. Unlike an illegitimate authority, those that are
perceived to be legitimate do not have to rely on instrumental control of rewards and punishments to control behavior
(Tyler, 2006a). While such a strategy can be effective, it is costly and inefﬁcient as a relatively large amount of resources are
needed to ensure that people are obeying the rules and being rewarded (or punished) appropriately. Instead, authorities can
exert their control by fostering legitimacy among those connected to them. Ultimately, authorities become legitimate when
society has judged that they have a right or are entitled to obedience through some type of value system. Once established,
legitimacy is maintained to a large extent through social norms and socialization pressures. Thus, individuals are more likely
to follow the rules of a legitimate authority, even when that authority is not physically present (Sunshine & Tyler, 2003). This
allows an authority to direct his or her resources elsewhere, rather than using them to ensure that rules are not being
Research within the legal system has shown the effectiveness of authorities perceived as legitimate. For example, when
individuals view the police as legitimate, they are more likely to show support for the institution of policing (Sunshine & Tyler,
2003), cooperate with police (Tyler, 2006b), notify them when laws are broken (Tyler, 2006b), and not violate the law (Sunshine
& Tyler, 2003). Adolescents that believe the police are legitimate report less engagement in crime during the previous year
(Fagan & Tyler, 2005). However, legal scholars who have focused on adolescents’perceptions of legitimacy (e.g., Fagan &
Piquero, 2007) have ignored the legitimacy of parents, important authority ﬁgures in the life of adolescents. Because of this
lack of research, it is also not entirely clear what parenting factors will promote parental legitimacy. Furthermore, there has
been relatively little research examining speciﬁcally how adolescents’perceptions of the legitimacy of any type of authority
affect their engagement in delinquent behavior (see Fagan & Piquero, 2007; Fagan & Tyler, 2005 for important exceptions).
Legitimacy of parental authority
Much of the developmental psychology research on parental authority has tended to focus on where and when adoles-
cents will view parental authority as legitimate. Most of this research has shown that adolescents’perception of parental
legitimacy is domain speciﬁc(Milnitsky-Sapiro, Turiel, & Nucci, 2006; Smetana & Daddis, 2002). The decision to give parents
legitimate authority is dependent upon the particular issue in question. For example, Tisak (1986) found that adolescents
were more likely to perceive parental rules pertaining to stealing as more legitimate than rules concerning household chores
or friendships. More recently, Darling, Cumsille, and Peña-Alampay (2005) found that most adolescents perceived parents as
legitimate within moral domains (e.g., doing physical or psychological harm); however, there was greater variability in
legitimacy perceptions within personal domains (e.g., choice of friends or use of free time). Similar to past research on legal
authorities, when adolescents do view their parents as legitimate within a given domain and feel an obligation to obey
parental rules within that domain, they are less likely to break those rules (Darling et al., 2007).
Although developmental psychologists have examined when adolescents will view parental authority as legitimate, there
has been relatively few examinations of how these perceptions impact adolescents’engagement in rule-violating behavior
within a legal context. For example, Darling et al. (2007) primarily focused on behavior violating rules at home rather than
behavior violating public laws (i.e., delinquency). Moreover, no researchers have examined if the perceptions of parental
legitimacy impact changes in delinquent behavior over time. Finally, although it is well established that there are individual
differences in adolescent perceptions of parental legitimacy, relatively little is known about how these differences develop or
what factors inﬂuence adolescent perceptions (Darling et al., 2007; Darling & Steinberg, 1993).
R. Trinkner et al. / Journal of Adolescence 35 (2012) 119–132120
Parenting style may help to explain individual differences in the extent to which adolescents view their parents as
legitimate authority ﬁgures. In her now classic typology, Baumrind (1967, 1971) identiﬁed three general styles of parenting.
Authoritative parents are both demanding and controlling, but they are also warm and receptive to their children’s needs.
They are receptive to bidirectional communication in that they explain to their children why they have established rules and
also listen to their children’s opinions about those rules. Children of authoritative parents tend to be self-reliant, self-
controlled, and content. On the other hand, authoritarian parents are demanding and highly controlling, but detached and
unreceptive to their children’s needs. These parents support unilateral communication where they establish rules without
explanation and expect them to be obeyed without complaint orquestion. Authoritarian parenting produces childrenwho are
discontent, withdrawn, and distrustful. Finally, in contrast to authoritarian parenting, permissive parents are nondemanding
and noncontrolling. They tend to be warm and receptive to their children’s needs, but place few boundaries on their children.
If they do establish rules, they rarely enforce them to any great extent. These parents tend to produce children who are the
least self-reliant, explorative, and self-controlled out of all the parenting styles.
There has been a tremendous amount of research showing that parenting styles impact many aspects of child develop-
ment (for a review see Newman, Harrison, Dashiff, & Davies, 2008). Although there is general consensus that authoritarian
and permissive parenting result in unfavorable outcomes and authoritative parenting produces beneﬁcial outcomes,
surprisingly few researchers have examined the relation between the three typologies and adolescents’delinquent behavior
speciﬁcally (Hoeve et al., 2009). These few exceptions have found somewhat mixed results. In a large cross-section of
a socioeconomically and ethnically diverse sample, Steinberg, Mounts, Lamborn, and Dornbusch (1991) found that adoles-
cents within authoritative households, regardless of personal background, reported less delinquent behavior. In a longitu-
dinal examination, Simons et al. (2005) found that authoritative parenting deterred adolescents from not only engaging in
delinquent behavior over a two year time span, but also from associating with deviant peers. However, in a relatively small
sample of Dutch adolescents, Hoeve et al. (2007) found that neither authoritative nor authoritarian parenting predicted
delinquent behavior ten years later. In a later meta-analysis, Hoeve et al. (2009) did ﬁnd that authoritative parenting was
negatively related to delinquency, although deﬁnitive conclusions on the relation between parenting style and delinquency
were limited because of the small number of studies available.
It is important to note that parenting stylesreﬂect a conﬁguration of attitudes and goals parents direct toward theirchildren,
rather than a conglomeration of speciﬁc parenting behaviors. As Darling and Steinberg (1993) noted in their integrative model
of parenting styles, manyof the same speciﬁc parenting behaviors are used by all threetypes of parenting. Speciﬁc behaviors are
less important than how adolescents interpret those behaviors. Parenting styles provide a context or climate for children to
interpret parenting behaviors accordingly, thereby transforming parent–child interactions. Of particular importance to the
present study, their model suggests that parenting styles affect adolescent outcomes by changing the degree to which
adolescents accept their parents’attempts to socialize them. In later research, Darling et al. (2007) suggested that parental
legitimacy may be a marker for adolescents’willingness to be socialized. In other words, the more adolescents believe their
parents are legitimate authority ﬁgures, the more they will also report being open to the socialization efforts of their parents.
Coupled with Darling and Steinberg’s (1993) integrative model of parenting described above,this suggests that (1) the style that
parents use to rear their children should inﬂuence the degree to which their children perceive them as legitimate authority
ﬁgures and (2) the perception of parental legitimacy is an important mediator between parenting style and adolescent
outcomes (e.g., delinquent behavior). To our knowledge, researchers have not examined these suggestions empirically.
In the current study, our goal was to examine the relations among parenting style, perceptions of parental legitimacy, and
adolescent engagement in delinquent behavior over an eighteen month time period. We tested a number of hypotheses. First,
we predicted that authoritative parenting would be associated with less delinquent behavior over time, while authoritarian
and permissive parenting would be associated with more delinquent behavior over time. Second, we predicted that
adolescents who perceived their parents as legitimate authority ﬁgures would report engaging in less delinquent behavior
over time. Third, we hypothesized that all three parenting styles would inﬂuence adolescents’perceptions of parental
legitimacy. More speciﬁcally, we predicted that authoritative parenting would foster the belief that parents are legitimate
authorities, while both authoritarian and permissive parenting would attenuate such beliefs. Finally, we hypothesized that
adolescent perceptions of parental authority would mediate fully the relation between parenting styles and future delinquent
In order to test these hypotheses we used data generated from the New Hampshire Youth Study (NHYS; see Cohn, Bucolo,
Rebellon, & Van Gundy, 2010), an ongoing, longitudinal survey of middle school and high school students examining the
psychological, sociological,developmental, andlegal factorsthat inﬂuenceadolescentdelinquency. Analyses reportedare based on
data collectedduring the fall of2007, fallof 2008, and the springof 2009. These collectionperiods correspond tothe third, ﬁfth,and
sixthphases of the NHYS. However,to ease presentation,we will refer to themas T1 (fall of 2007), T2 (fallof 2008), and T3 (spring of
2009). Because the NHYS employs a rotated schedule of measurements that varies by phase, we were unable to use phase four
(spring of 2008) as there was no parental legitimacy measure included in the questionnaire during this phase. Measures of
parenting style and delinquent behavior were included at T1, parental legitimacy at T2, and delinquent behavior again at T3.
R. Trinkner et al. / Journal of Adolescence 35 (2012) 119–132 121
Four urban communities were chosen to provide a diverse sample of adolescents in New Hampshire. To recruit students,
we entered schools (eight middle schools and ﬁve high schools) and distributed parental informed consent forms outlining
the purpose and procedures of the study. Students were told that they would receive a ten dollar gift certiﬁcate to a national
bookstore for every phase of data collection they completed with each session separated by six months. Only students who
returned the parental consent forms and completed their own informed consent forms were included in the NHYS.
At the start of the third phase (T1; fall, 2007) of the NHYS, 1054 students had returned completed parental consent forms.
Nine hundred forty one students participated in the NHYS at T1 (fall, 2007), 867 at T2 (fall, 2008), and 840 at T3 (spring, 2009).
There were a number of reasons forour inability to collect data from all participating students at each time point. For example,
some of the schools would only allow a relatively small number of data collection sessions (one or two), making it difﬁcult to
collect from students that were absent or could not attend for academic reasons. Some students also decided not to
participate at particular phases because they were taking classes they did not want to miss. Some of the largest schools
participating in the NHYS have curricula in which students travel to other schools in the city throughout the day to take
classes. Scheduling data collection sessions for these schools at times when all of the participating students were available
was difﬁcult if not impossible. We also had a particularly difﬁcult time collecting from all students in phase 6 (T3) as the high
school students in the sample were graduating, making data collection during late spring/early summer difﬁcult. Out of the
1054 students that returned parental consent forms, only 27 did not complete a survey at any of the phases used in the
present paper (1027 participated in at least one phase).
After eliminating students with missing or incomplete data, a total of 596 students completed the surveys at all three
collection sessions. Of these 596 students, 366 (61.4%) were middle school students and 230 (38.6%) were high school
students. The racial composition was similar for both middle school and high school samples with 482 (80.9%) students from
the total sample reporting they were Caucasian, 32 (5.4%) Hispanic American, 30 (5.0%) “Other,”19 (3.2%) African American,17
(2.9%) Asian American, 8 (.8%) multi-racial, and 3 (.5%) reporting they were Native American. The middle school sample had
a slightly lower percentage of girls (n¼221, 60.4%) than the high school sample (n¼151, 65.7%). Finally, middle school
ranged from 11.0 to 14.0 (M¼12.27, SD ¼.49) at T1 and 12.0 to 16.0 (M¼13.75, SD ¼.54) at T3. High school
student ages ranged from 14.0 to 16.0 (M¼15.33, SD ¼.50) at T1 and 15.0 to 18.0 (M¼16.78, SD ¼.54) at T3.
All demographics were taken from T1. Participants reported their sex, race/ethnicity, grade/year in school, each parent’s
highest level of education completed (1: Less than High School;6:Graduate Degree), and average grades (1: Mostly A’s;9:
Mostly F’s). Average grades were recoded so that high scores reﬂected higher grades (M¼7.6 8 , SD ¼1.33). Additionally,
a composite score of mother’s education and father’s education was created as a proxy for socio-economic status (SES) with
higher scores indicating higher status (range 2–12, M¼7.30, SD ¼2.51).
Amodiﬁed version of the Parental Authority Questionnaire (PAQ; Buri, 1991) was used to measure parenting style as
conceptualizedby Baumrind (1971, 1991) at T1. The original PAQ consists of 30 items (10 for each parentingstyle) for each parent.
Because of time and space constraints, we were unable to include the full scale in the NHYS. Past research has shown that
adolescentsreport their parents are morelikely to use the same parenting style than anyparticular combinationof styles (Simons
&Conger,2007). As such, each item was ﬁrst reworded to reﬂect parents in general, rather than mother and father separately.
However, the resulting30-item scale was still too longfor our purposes. Because Buri (1991) did notconduct a factor analysis, we
were unable to shorten the scale further based on factor loadings in the original development of the scale. Thus, following
a similar procedure used by Buri to establish content validity, weshortened the scale by selecting4 items for each parenting style
that all the authors agreed were the best representations of each parenting style as conceptualized by Baumrind (1971, 1991).
The resulting 12 item scale included items measuring authoritative parenting style (e.g., “My parents tell me how I should
act and explain the reasons why”), authoritarian parenting style (e.g., “My parents feel that parents must use force to get
children to act the way they are supposed to”), and permissive parenting style (e.g., “My parents feel that children can do
whatever they like”). For all twelve items, participants reported their agreement on a four point Likert Scale (1: Disagree
Strongly;4:Agree Strongly). Items were averaged for each parenting style with higher scores indicating greater use of each
Although age appears to have a wide range, our data had exceedingly few people at the minimum and maximum of our age ranges. Rather, ages were
extremely clustered around two modes, reﬂecting a low variance in age in reality.
R. Trinkner et al. / Journal of Adolescence 35 (2012) 119–132122
parenting style. Each student had a score reﬂecting the extent to which their parents used authoritative (M¼2.79, SD ¼.69,
¼.74), authoritarian (M¼2.15, SD ¼.67,
¼.73), and permissive (M¼1.86, SD ¼.58,
¼.66) parenting styles.
To make certain that the modiﬁed PAQ had a similar dimensionality in comparison to the original PAQ, factor analysis was
performed using principal components analysis with varimax rotation. Only components with eigenvalues greater than 1 and
items with rotated factor loadings of .50 or higher were retained. This analysis provided a three component solution. The ﬁrst
component contained all four authoritative items and accounted for 20.38% of the variance. The second component contained
all four authoritarian items and accounted for 19.90% of the variance, while the third component contained all four permissive
items and accounted for 17.33% of the variance. These results suggest that the modiﬁed PAQ used in the present study is
functioning in a similar manner to the original PAQ. Past research has shown that authoritative parenting is associated with
beneﬁcial outcomes, while authoritarian and permissive parenting are associated with unfavorable outcomes (Hoeve et al.,
2009). We found a similar pattern in our data in that authoritative parenting was associated with less delinquency and more
parental legitimacy, while authoritarian and permissive parenting were associated with more delinquency and less parental
legitimacy (see results). In light of these ﬁndings, we are conﬁdent that our modiﬁed version of the PAQ is tapping the same
parenting style constructs as the original version.
Sunshine and Tyler’s (2003) police legitimacy scale was used as a basis to create a measure of parental legitimacy at T2. Ten
items were selected and reworded to reﬂect participants’trust in their parents (e.g., “My parents can be trusted to make
decisions that are right for me”) and their obligation to obey their parents’rules (e.g., “I should do what my parents tell me to
do even when I do not like the way they treat me”). Respondents reported their agreement with each item on a four point
Likert Scale (1: Strongly Disagree;4:Strongly Agree). All items were averaged with higher scores indicating greater parental
legitimacy (M¼3.05, SD ¼.57,
The Delinquency Component of the National Longitudinal Youth Survey (Wolpin, 1983) was used as a measure of
delinquency at T1 and T3. This 25 item measure asked students to report how many times they engaged in 25 speciﬁc
behaviors in the past six months from three areas: property offenses (e.g., “taken something from a store without paying for
it”), violent offenses (e.g., “attacked someone with the idea of seriously hurting or killing them”), and substance use (e.g., “had
an alcoholic drink”). Due to the non-normal and highly skewed distribution of the frequency reports, participant responses
were recoded into yes (1) or no (0) for each behavior and then summed. Although this did not eliminate the skewness, it did
alleviate it to some extent. Higher scores indicated more engagement in a greater variety of delinquent behavior (T1:
M¼2.63, SD ¼3.40,
¼.87; T3: M¼3.14, SD ¼3.97,
Although higher scores on this measure reﬂect engagement in a greater variety of delinquent behaviors, we used this scale
as an overall indicator of frequency of delinquency for several reasons. First, using a variety measure as a proxy of overall
engagement in delinquent behavior is a fairlycommon procedure, especially with delinquency measures taken from or based
upon the National Longitudinal Youth Survey (e.g., Apel, Bushway, Paternoster, Brame, & Sweeten, 2008; Cohn et al., 2010;
Han, Miller, & Waldfogel, 2010; Hoffman, 2010; Sweeten, Bushway, & Paternoster, 2009). Second, Monahan and Piquero
(2009) recently showed that variety measures and frequency measures of delinquent behavior were strongly correlated
with each other over late adolescence and early adulthood, indicating that adolescents who engaged in a greater variety of
delinquent behaviors were also engaging in a greateramount of delinquent behaviors. Our data support such a notion as there
was a high correlation between the frequency measure of delinquent behavior and the variety measure of delinquent
behavior at both T1 (r
¼.94, p<.001) and T3 (r
¼.88, p<.001). Third, Bendixen, Endresen, and Olweus (2003) showed that
utilizing variety measures of delinquency compared to frequency measures produced greater internal consistency, higher
stability over time, higher group differences, and stronger associations with conceptually related variables. They concluded
that variety measures actually provided a more sensitive assessment of overall offending than frequency measures. Similar to
their ﬁndings, our data also showed that the frequency measure had little internal consistency, especially compared to the
variety measure. Based on the above research, we believe that our variety measure is an acceptable proxy of overall
engagement in delinquent behavior.
All three phases of the current study had similar procedures. All measures and procedures were reviewed and approved by
the University Institutional Review Board. The researchers scheduled times to come to each school to administer the ques-
tionnaire in mass testing sessions in various locations around the school (e.g., study halls, auditoriums, school libraries, and
cafeterias). Students were placed one seat apart from each other to ensure conﬁdentiality. They were then given verbal
instructions on how to complete the questionnaire and were told to raise their hand if they had any questions. Students were
assured that their responses would be kept conﬁdential and instructed to not place any identifying information on their
questionnaire. The questionnaire took approximately 35 min to complete and was divided into two sections. To reduce
fatigue, students were given a short break and a fruit snack after completing Section 1. After completing the entire ques-
tionnaire, students signed in with the researchers. During the ﬁrst collection session, arbitrary ID numbers were placed on the
completed surveys to allow researchers to match students’responses in all three phases. At T2 and T3 students were asked for
R. Trinkner et al. / Journal of Adolescence 35 (2012) 119–132 123
their names as they turned in their surveys. A member of the research team thenwrote the corresponding ID number from T1
on their surveys. Finally, respondents received a $10 gift certiﬁcate to a national bookstore, were thanked for their time, and
dismissed back to their regularly scheduled class.
During the last phase of the NHYS (T3), the questionnaire was also offered in an on-line version to participants who could
not complete the questionnaire in person at their respective school. The on-line version was identical to the paper-and-pencil
version in all respects. Those students who could not complete the survey at their school were contacted by mail, telephone,
or e-mail, and given their ID number and the URL to the on-line questionnaire. After completing the questionnaire, they were
directed to a second unrelated website where they provided their mailing address. Upon completion, the $10 gift certiﬁcate
was mailed to them. Twenty-two participants in the ﬁnal matched sample completed this on-line version. The on-line
participants did not differ signiﬁcantly from paper and pencil participants in the reporting of delinquency at T3
(t(594) ¼1.10, p>.05).
We began our assessment of the degree to which parental legitimacy mediated the relation between parenting style and
delinquent behavior by using bivariate correlations and OLS regression to examine whether each parenting style was
signiﬁcantly associated with parental legitimacy and delinquency. Next, building on the foundation provided by Baron and
Kenny (1986), we followed prior research (e.g., Rebellon, 2002) to assess mediation effects via a series of negative-
binomial regression models
employing delinquency as our dependent variable. Our ﬁrst model included authoritarian,
authoritative, and permissive parenting as predictors of delinquency after statistically controlling for sex, average grades, SES,
and cohort. Our second model added parental legitimacy as a predictor, thus allowing a comparison of parenting coefﬁcients
before and after adjusting for legitimacy. To the degree that parental legitimacy was a signiﬁcant predictor of delinquency in
Model 2 and the absolute values of parenting coefﬁcients from Model 1 were attenuated in Model 2, results would suggest
support for the notion that perceptions of parental legitimacy mediated the relation between parenting style and respon-
dents’respective levels of delinquent behavior. Models 3 and 4 were congruent with Models 1 and 2, but added a lagged-
measure of delinquency to assess the relation between parenting styles, parental legitimacy, and changes in delinquency
over the 18 month time period.
Following our negative-binomial analyses, we assessed the robustness of our results by employing an alternative method
for examining the degree to which parental legitimacy mediated the relation between parenting style and delinquency. In
particular, we used Lisrel 8 (Jöreskog & Sörbom, 1993) to estimate two structural-equation models that each included both
direct paths linking parenting style with delinquency and indirect paths linking parenting style with delinquency via the
intervening inﬂuence of parental legitimacy.
The ﬁrst structural equation model paralleled the comparison of Model 1 with
Model 2 from our negative-binomial analyses to examine whether parental legitimacy mediated the relation between
parenting styles and respondents’respective levels of delinquent behavior. The second structural equation model (paralleling
Models 3 and 4) repeated the same general procedure, but included a lagged-measure of delinquency to examine the
mediating inﬂuence of legitimacy on the relation between parenting styles and changes in delinquent behavior over time.
Finally, we used Sobel tests (Sobel, 1982; MacKinnon, Warsi, & Dwyer, 1995) to examine the signiﬁcance of the indirect paths
between parenting style and delinquency via parental legitimacy.
Given the skewed nature of our dependent variable, typical maximum-likelihood estimation was not appropriate as our
structural equation estimation method. We therefore created a categorical variable that collapsed the approximately ﬁve
percent of respondents who reported 10 or more delinquent behaviors into one category (i.e., engagement in 10 or more
delinquent behaviors) and estimated our models using Lisrel’s weighted-least-squares (WLS) algorithm. We began each
analysis by estimating the polychoric correlation matrix among our indicator variables, as well as each one’s associated
asymptotic covariance matrix. This distribution-free method of estimation did not assume multivariate normality among
indicator items. Rather, it assumed normality of the latent structure underlying each indicator and bivariate normality
underlying the estimated polychoric correlations (see Browne, 1984).
Long (1997) describes the problems associated with using OLS regression to analyze count, or non-negative integer, dependent variables and suggests
using Poisson regression as an alternative. However, Poisson regression is problematic when the dependent variable exhibits overdispersion, resulting in
underestimated standard errors which subsequently tend to over-estimate the statistical signiﬁcance of regression coefﬁcients. In these cases, he suggests
using negative-binominal regression. Our ﬁrst set of analyses used both Poisson and negative-binomial regression. Inspection of the alpha coefﬁcients from
our negative-binomial regressions revealed evidence of overdispersion. We therefore opted to omit presentation of Poisson regression results, which
consistently appear to yield biased tests of statistical signiﬁcance (available upon request) and only present negative-binomial results. It is important to
note that negative-binomial coefﬁcients cannot be interpreted in the same way that OLS or logistic coefﬁcients are interpreted. For a detailed discussion of
interpreting negative-binomial coefﬁcients, please see Long (1997).
These models serve several purposes. First, they provide a means of assessing the robustness of negative-binomial results insofar as they employ an
alternative estimation algorithm (e.g., weighted-least-squares). Second, they allow for a more straight-forward and direct assessment of mediation than do
negative-binomial models. Speciﬁcally, rather than combining the results from two models to assess mediation via changes in regression coefﬁcients from
one model to another, path models simply provide simultaneous estimates of the direct and indirect links between parenting style and delinquency. Third,
whereas the magnitude of negative-binomial coefﬁcients are difﬁcult to compare directly for variables whose ranges differ, our structural equation results
yield fully-standardized coefﬁcients that can more easily be compared. Finally, structural equation results yield ﬁt statistics with arguably more agreed-
upon interpretations, thus allowing us to assess whether the models presented provide good ﬁt to the empirical data.
R. Trinkner et al. / Journal of Adolescence 35 (2012) 119–132124
By using a listwise deletion procedure, we eliminated a large portion of our sample from the analyses. To examine whether
listwise deletion compromised the validity and representativeness of the analyses, we conducted a missing values analysis
using the SPSS 18.0 missing values module (SPSS, 2010). This analysis revealed that between 0 (cohort and sex) and 18%
(Delinquency(T3)) of the values were missing for each variable in our analysis with the majority of variables missing
approximately 10% of the values. Little’s (1988) MCAR test was signiﬁcant indicating that values were not missing completely
at random (
(95) ¼186.91, p<.001). As such, following the SPSS (2010) manual guidelines, we used the expectation-
maximization (EM) multiple imputation of missing data procedure to create ﬁve datasets with imputed values. To retain
the same scale characteristics present in our original data, we speciﬁed the minimum and maximum imputed values for all
scales to avoid values outside the possible range allowed by each scale. Additionally, we speciﬁed that imputed values for the
delinquency measure be rounded to the nearest integer to maintain the count nature of this scale.
A visual inspection of the descriptive statistics for each variable in the original (listwise deleted) and the imputed datasets
showed that the mean and standard deviations of each variable (available upon request) were nearly identical across the
datasets. To further examine possible bias due to missing values, we conducted our OLS, negative binomial and structural
equation analyses for each of the imputed datasets. Next, following the procedures outlined by Schafer and Graham (2002)
and Sinharay, Stern, and Russell (2001), we aggregated the parameter estimates and standard errors across the imputed
datasets for each model. After conducting signiﬁcance tests, we found results substantively identical as those from the
original listwise deleted data (available upon request). In all cases, the direction and magnitude of our results were similar.
However, in a few instances, paths were signiﬁcant that were not in the original analyses. Upon further inspection, we found
that these particular paths approached signiﬁcance in the original analyses. We suspect that they were signiﬁcant in the
aggregate data due to the large increase in power afforded by not deleting participants with missing data. The analyses
presented below utilized the aggregated data from the multiple imputation procedure.
The ﬁrst step in our analyses was to examine the relation between each parenting style, parental legitimacy, and delin-
quency (see Table 1). Bivariate correlations revealed that authoritarian parenting was associated with less parental legitimacy
(r¼.20, p<.01) and that authoritative parenting was associated with more parental legitimacy (r¼.32, p<.01). Permissive
parenting was not signiﬁcantly associated with parental legitimacy. Authoritarian parenting was positively associated with
adolescent reports of delinquency at T1 (r¼.22, p<.01) and T3 (r¼.10, p<.05). In contrast, authoritative parenting was
negatively associated with delinquency at T1 (r¼.20, p<.01) and T3 (r¼.22, p<.01). Permissive parenting was positively
correlated with delinquency at T1 (r¼.12 , p<.01), but not at T3. Finally, adolescent perceptions of parental legitimacy were
associated with less delinquent behavior at T1 (r¼.36, p<.01) and T3 (r¼.38, p<.01).
To assess whether the relations between parenting style and parental legitimacy remained while controlling for sex (coded
0¼women; 1 ¼men), SES, cohort (coded 0 ¼middle school; 1 ¼high school), and average grades, we conducted an OLS
regression analysis estimating the independent association of each parenting style with parentallegitimacy net of controls (see
Table 2). This regressionwas signiﬁcant (F(7,1019) ¼34.15, p<.05) and accounted for 19% of the variance (r
¼.19) in parental
legitimacy. Similar to the correlation analysis, authoritarian parenting negatively predicted parental legitimacy (
p<.05), while authoritative parenting positively predicted legitimacy (
¼.33, p<.05) after accounting for all control variables.
In light of preliminary evidence that parenting styles predicted parental legitimacy (a standard step in Baron and Kenny’s
(1986) method of assessing mediation), we next examined the degree to which parenting was associated with delinquency
and to which the inclusion of parental legitimacy in multivariate analyses attenuated any such relation. Results from Model 1
in Table 3 suggested evidence that authoritarian parenting was associated with involvement in a greater array of delinquent
Estimated bivariate correlations among parenting styles (T1), parental legitimacy (T2), and delinquency (T1 & T3).
Authoritarian Authoritative Permissive Parental legitimacy Delinquency (T1) Delinquency (T3)
Authoritative .04 –
Permissive .31** .10* –
Parental legitimacy .20** .32** .07 –
Delinquency (T1) .22** .20** .12** .36** –
Delinquency (T3) .10* .22** .07 .38** .61** –
R. Trinkner et al. / Journal of Adolescence 35 (2012) 119–132 125
behaviors 18 months later (
¼.15, p<.05). Conversely, authoritative parenting (
¼.30, p<.05) was associated with
involvement in less delinquency. Both results remain net of controls.
Model 2 expanded Model 1 by including parental legitimacy as a predictor (see Table 3). Parental legitimacy was strongly
associated negatively with delinquency overtime (
¼.53, p<.05) while controlling for cohort, sex, average grades, and SES.
Additionally, after controlling for parental legitimacy, authoritarian parenting was no longer associated with delinquency,
suggesting that parental legitimacy may fully mediate the relation between authoritarian parenting and delinquent behavior.
Conversely, the coefﬁcient linking authoritative parenting and delinquency in model 2 (
¼.15 , p<.05) was substantially
lower in magnitude than the coefﬁcient in Model 1 (
¼.30), suggesting that parental legitimacy might only partially
mediate the relation between authoritative parenting and future delinquent behavior.
Model 3 examined the relations among parenting styles and change in delinquency over the 18 month time period by
adding a lagged-measure of delinquency in addition to the same set of predictors from Model 1 (see Table 3). First, results
revealed that prior engagement in delinquent behavior predicted future delinquency (
¼.12, p<.05), indicating a signiﬁcant
stability in delinquency over time. Second, authoritative parenting was associated negatively with changes in delinquency
across time (
¼.24, p<.05) net of controls, while the other two styles of parenting were not associated with changes in
delinquency across time.
Paralleling the difference between Models 1 and 2, Model 4 included the same setof predictors used in Model 3, but added
parental legitimacy as a predictor of change in delinquency. Similar to Model 2, parental legitimacy (
¼.36, p<.05) was
strongly negatively associated with changes in delinquent behavior across time after controlling for the demographic vari-
ables discussed previously. Additionally, the coefﬁcient linking authoritative parenting and changes in delinquency in Model
¼.24) was reduced in magnitude in Model 4 (
¼.15, p<.05), again suggesting that parental legitimacy may only
partially mediate the relation between authoritative parenting and future delinquency.
Structural equation models
As an alternative method of examining the degree to which legitimacy mediated the relation between parenting style and
delinquency, Figs. 1 and 2 depict the results of two structural equation models estimated via Lisrel 8.Fig. 1 corresponds
conceptually to the comparison of Models 1 and 2 in Table 3, but examines mediation by simultaneously estimating paths
linking parenting style directly to delinquency and indirectly to delinquency via the intervening mechanism of parental
legitimacy. We estimated correlations (available upon request) among predictor variables, but omitted those estimates for
ease of presentation. Authoritarian (
¼.05, n.s.) and permissive parenting (
¼.03, n.s.) were not directly associated with
future delinquency after accounting for controls. However, authoritative parenting (
¼.10, p<.05) maintained a weak
direct association with future delinquent behavior. All three types of parenting were signiﬁcantly associated with parental
legitimacy. Of note, authoritative parenting (
¼.37, p<.05) was associated positively with parental legitimacy, while
authoritarian parenting (
¼.19, p<.05) and permissive parenting (
¼.05, p<.05) were associated negatively with
parental legitimacy. Subsequently, parental legitimacy was associated negatively with delinquency (
¼.23, p<.05). Thus,
structural equation results parallel the general ﬁndings from regression analyses and suggest that the relation of parenting
style to delinquent behavior is mediated by the intervening inﬂuence of parental legitimacy. Further, ﬁt statistics suggested an
adequate ﬁt of the conceptual model to the empirical data (
(4) ¼58.67, p<.05, AGFI ¼.94, CFI ¼.96, NFI ¼.96,
OLS regression estimates predicting parental legitimacy
as a function of parenting styles and controls.
Average Grades .13*
F(7, 1019) ¼34.15, p<.05; r
Note: Cohort (dummy coded 0 ¼middle school, 1 ¼high
school), Sex (dummy coded 0 ¼women, 1 ¼men),
SES ¼socio-economic status.
Because a large proportion of our participants reported not engaging in any delinquent behavior (indicating that our dependent variable may have an
excess of zeros), we ran an additional set of zero-inﬂated negative-binomial analyses. These analyses substantively replicated our results. While parenting
style and parental legitimacy were not related to whether an individual did or did not engage in delinquency, they were signiﬁcant predictors of the
frequency of delinquent behavior for those individuals that reported engaging in some type of delinquent behavior.
R. Trinkner et al. / Journal of Adolescence 35 (2012) 119–132126
Finally, following MacKinnon et al. (1995), Sobel tests indicated that the indirect paths linking authoritarian
parenting (z¼3.03, p<.05), authoritative parenting (z¼3.14, p<.05), and permissive parenting (z¼2.91, p<.05) with
delinquency via parental legitimate were signiﬁcant.
Fig. 2 repeated the general strategy outlined in Fig. 1, but added a lagged measure of delinquency as a predictor to provide
a structural equation model of change in delinquent behavior rather than level of delinquent behavior (see Fig. 2). As found in
the negative-binomial analyses, delinquency (
¼.57, p<.05) exhibited signiﬁcant stability over time. In this model, none of
the parenting styles were signiﬁcantly associated with changes in delinquent behavior after accounting for the controls.
However, all three parenting styles were associated with parental legitimacy in the expected directions (authoritarian:
¼.18 , p<.05, authoritative:
¼.35, p<.05, permissive:
¼.05, p<.05). Parental legitimacy, in turn, was associated
negatively with changes in delinquency over time (
¼.14, p<.05). Thus, Fig. 2 suggested that parental legitimacy fully
mediated the relation between parenting style and changes in delinquency over time.
Finally, Sobel tests indicated that the
indirect paths linking authoritarian parenting (z¼2.52, p<.05), authoritative parenting (z¼2.59, p<.05), and permissive
parenting (z¼2.46, p<.05) with changes in delinquency via parental legitimate were signiﬁcant.
We were interested in examining the relations among the styles parents use to rear their children, adolescents’percep-
tions of the legitimacy of parental authority, and their delinquent behavior over an 18-month time period. Overall, the results
supported our hypotheses. First, as predicted, we found that authoritative parenting was associated negatively with delin-
quency, while authoritarian parenting was associated positively with delinquency; however, contrary to expectations, we
found little evidence that permissive parenting was associated with delinquency. Secondly, adolescents’perceptions of
parental legitimacy were associated negatively with delinquent behavior. Third, authoritative parenting was associated
positively with perceptions of parental legitimacy and authoritarian parenting negatively with perceptions of legitimacy, as
predicted. Once again contrary to expectations, permissive parenting was not related to parental legitimacy. Finally, structural
equation models indicated that parental legitimacy mediated the relation between parenting styles and future delinquency.
Authoritative parenting led to increased parental legitimacy in the future, while authoritarian and permissiveparenting led to
Negative binomial estimates predicting delinquency (T3) as a function of parenting style, parental legitimacy, past delinquent behavior and controls.
Model 1 Model 2 Model 3 Model 4
Cohort .13* .06 .15* .17*
Sex .30* .30* .18* .20*
Average Grades .21* .18* .11* .10*
SES .01 .01 .01 .01
Authoritarian .15* .04 .05 .01
Authoritative .30* .15* .24* .15*
Permissive .07 .06 .05 .05
Delinquency(T1) ––.12* .10*
Parental legitimacy –.53* –.36*
Constant 2.88* 4.20* 2.21* 3.15*
Alpha .75* .64* .49* .44*
N1027 1027 1027 1027
Note: Cohort (dummy coded 0 ¼middle school, 1 ¼high school), Sex (dummy coded 0 ¼women, 1 ¼men), SES ¼socio-economic status, T1 ¼Fall 2007,
T3 ¼Spring 2009.
The “constant”in negative-binomial regression is similar to the traditional constant in OLS regression. The “alpha”value corresponds to whether there is
overdispersion in our dependent variable. As can be seen above, all of our models had signiﬁcant alphas indicating that we had overdispersion, leading to our
use of negative-binomial regression instead of Poisson regression (see footnote
While some may expect a non-signiﬁcant chi-square statistic in order to show that our model ﬁts the data, chi-square statistics automatically increase
with larger sample sizes (as is the case here) making it difﬁcult to obtain non-signiﬁcant chi-square values regardless of model ﬁt. As such, we have chosen
to present alternative indices to assess the model ﬁt. Of these alternative ﬁt indices, the RMSEA ﬁt index may be slightly high. However, the 90% conﬁdence
interval for the RMSEA in both of the non-lagged and lagged models included the value of .05, which generally indicates a good ﬁt to the data.
To provide a stronger testof the structural paths in our presented models (Figs. 1 and 2), we also tested the models using latent variable analysis. Based
on prior research (Rebellon & Waldman, 2003), we constructed separate count measures for violent offenses, property offenses, and drug offenses. We used
each of these three counts as indicator variables for our new latent dependent measure of delinquency. Adding the latent variable analysis, however,
increased the complexity of models that were already mathematically complex on account of the asymptotic covariance matrices that we estimated in our
WLS estimation procedure with polychoric correlations. As a result, after adding latent variable estimation, the lagged model would not converge, although
the non-lagged model did converge. The non-lagged model (which omitted the three lagged indicators of delinquency) yielded results substantively
identical to those presented for both direct and indirect paths. Because the lagged results did not converge using our WLS estimation procedure, we further
tested this model using more traditional ML estimation that does not account for the skew of indicator variables and does not invoke an asymptotic
covariance matrix. This model yielded results substantively identical for the direct and indirect paths of authoritarian and authoritative parenting; however,
it failed to yield a statistically signiﬁcant direct or indirect path linking permissive parenting with delinquent behavior. However, in light of the repeated
models that we have run using imputed data, different estimation models (WLS & ML), and both latent and non-latent measurement strategies, we are
conﬁdent that the results presented are highly robust.
R. Trinkner et al. / Journal of Adolescence 35 (2012) 119–132 127
less parental legitimacy. Parental legitimacy then predicted less delinquency six months later. This relation remained for both
future levels of delinquency and changes in delinquency over time.
The current study addressed a number of gaps within past research. First, according to Darling and Steinberg’s (1993)
integrative model of parenting styles, adolescent outcomes are impacted by parenting styles through the intervening
mechanism of adolescents’openness to be socialized by their parents. The extent to which adolescents view their parents as
legitimate authority ﬁgures is considered an important marker of such willingness to be socialized (Darling et al., 2007). The
present results support the integrative model of parenting styles (Darling & Steinberg, 1993). Structural equation models
showed that all three parenting styles predicted adolescents’perceptions of parental legitimacy, which, in turn, negatively
predicted delinquency after controlling for the effects of sex, age, average grades, and socio-economic status. In addition,
parental legitimacy mediated the relation between parenting styles and reports of future delinquency, as well as changes in
delinquency over time.
Second, there are relatively few studies examining the inﬂuence of parenting styles on delinquent behavior (Hoeve et al.,
2009). Although past research on this topic has been inconclusive (Hoeve et al., 2009), there is general consensus that
authoritative parenting produces favorable outcomes for children and that unfavorable outcomes are products of authori-
tarian and permissive parents (Baumrind, 1971, 1991). The current results provide further evidence supporting this consensus.
The more parents utilized an authoritative parenting style, the less likely their children were to report they engaged in
delinquent behavior over time. Additionally, although not as consistent, adolescents were more likely to report engaging in
delinquent behavior the more they reported their parents used an authoritarian parenting style. Thepresent results, however,
provided little support that permissive parenting is related to delinquency.
Third, while it is generally agreed that authoritative parenting is more effective than authoritarian and permissive styles
(Baumrind, 1971, 1991), little is known about why some parenting styles are more efﬁcient than others (Darling & Steinberg,
1993). Our results showed that parental legitimacy was an important mechanism by which parenting styles affected
adolescent behavior. The style that parents used to rear their children had a direct inﬂuence on whether those children
perceived their parents as legitimate authority ﬁgures. Adolescents who perceived parents as legitimate were then less likely
to engage in delinquent behavior. Thus, authoritative parenting may be more effective than the other stylesbecause this style
makes adolescents more willing to accept their parents’attempts to socialize them and subsequently follow their rules.
Conversely, authoritarian parents have the opposite effect in that they actually reduce the likelihood of their children
perceiving their authority as legitimate. Adolescents from authoritarian parents are more likely to resist their parents’
attempts at socialization.
Fig. 1. Structural equation model of parenting styles, parental legitimacy, controls and level of delinquency. (Predictor correlations estimated but omitted for ease
of presentation. Available upon request.)
R. Trinkner et al. / Journal of Adolescence 35 (2012) 119–132128
Finally, the current study is the ﬁrst to combine theorizing on the legitimacy of authority from both legal scholars and
developmental psychologists. Although both legal (Tyler, 2006a) and developmental psychology researchers (Darling &
Steinberg, 1993) have studied authority ﬁgures extensively, they have generally ignored ﬁndings from the other ﬁeld. In
the present study we combined a theoretical framework of legitimacy common within legal scholarship (Levi, Sacks, & Tyler,
2009; Tyler, 2006a) with an integrative model of parental authority (Darling & Steinberg, 1993) well established within
developmental psychology. The present study suggests that this interdisciplinary approach is beneﬁcial. In particular, while
legal research on the legitimacy of authorities has largely ignored non-legal entities (Tyler, 2006a), the present results provide
preliminary evidence that past research within the legal domain can be generalized to non-legal entities. The present
research extends ﬁndings showing that adolescents’perceptions of police legitimacy are strong predictors of engagement in
delinquency by demonstrating that perceptions of legitimate parental authority predict delinquency as well. Conversely,
developmental psychologists emphasize that adolescents’perceptions of parental legitimacy are largely tied to the speciﬁc
issue at hand (Cumsille, Flaherty, Darling, & Martinez, 2006; Darling et al., 2005). However, following past research by legal
scholars (Sunshine & Tyler, 2003), we developed a scale that measured adolescents’perceptions of parental legitimacy more
generally. Thus, the present results extend past research by developmental psychologists by showing that adolescents’global
perceptions of the legitimacy of parental authority are important predictors of behavior in addition to issue-speciﬁc
perceptions of legitimacy.
The current study was limited in several respects. First, we relied on adolescent reports of parenting style, making it
difﬁcult to assess if parents truly utilized each style to the extent that adolescents reported they did. Thus, we cannot say for
certain if parents’actual behaviors inﬂuence legitimacy or if adolescents’perceptions of parents’behavior are the key
components. Future research can address this limitation by including both adolescent perceptions of parenting behaviors, as
well as parents’reports of their own behavior. Similarly, becausewe used adolescents’self-reports of delinquency, we have no
way of knowing whether these reports accurately represented adolescents’true delinquent behavior or were inﬂuenced by
other factors (e.g., social desirability). To address this limitation, future research should utilize a more objective measure of
delinquency than what was used here.
Fig. 2. Structural equation model of parenting styles, parental legitimacy, controls and change in delinquency. (Predictor correlations estimated but omitted for
each of presentation. Available upon request)
R. Trinkner et al. / Journal of Adolescence 35 (2012) 119–132 129
Second, we did not measure parental legitimacy on an issue-by-issue basis as is traditionally done by developmental
psychologists. Instead, we assessed adolescents’perceptions of parental legitimacy more generally, as legal scholars do when
measuring perceptions of police legitimacy. We do not believe this type of measurement was problematic because we were
concerned solely with delinquent behavior. Such behavior would be categorized in the moral or conventional domains in
which adolescents are more likely to give parents authority than other domains (Darling et al., 2007). Moreover, Darling et al.
did admit that it was unclear if adolescents were basing their responses on more traditional measures of parental legitimacy
on an issue-by-issue basis or using a more abstract idea of parental legitimacy in general. Future research can address this
limitation by utilizing multi-level modeling to examine simultaneously the effects of parental legitimacy for speciﬁc issues
and global perceptions of parental legitimacy on delinquency.
Third, none of our parenting measures differentiate between mothers and fathers, but rather asked adolescents to report
about their parents. Thus, we are unable to assess if the model presented here differs between maternal and paternal
authority or whether such differences (if any) are dependent upon the sexof the child. Past research has shown that it is more
common for parents to use the same parenting style than different styles (Simons & Conger, 2007). Furthermore, if both
parents are using the same parenting style, it seems likely that they would have similar inﬂuences on adolescents’perceptions
of parental legitimacy. However, such an assumption is problematic without empirical validation. In addition, to our
knowledge, researchers have not examined if adolescents’perceptions of parental legitimacy differ between mothers and
fathers. Future research can address this concern by differentiating between mothers and fathers to examine if the model
presented here applies to both parents.
Finally, we used Cronbach’s alpha to assess the reliability of our delinquency measure. There has been growing debate over
whether such a procedure is sufﬁcient in scales that use dichotomous items. While some argue that Cronbach’s alpha is
acceptable regardless of the number of response options (see Cortina, 1993), others contend that Rasch Dichotomous Models
should be used in such instances (Linacre, 1997). While there is evidence supporting both sides of the argument, we followed
past research utilizing this delinquency measure (e.g., Apel et al., 2008; Cohn et al., 2010; Han et al., 2010; Hoffman, 2010;
Sweeten et al., 2009) and used Cronbach’s alpha to assess reliability. However, variations of the present delinquency
measure have been used extensively within delinquency research going back to the original National Youth Survey (Elliott,
Huizinga, & Morse, 1985). Because of its widespread use, it would be beneﬁcial for future research, speciﬁcally focused on
this issue, to examine the reliability of this measure by comparing Cronbach’s alpha to Rasch models. Although such an
examination would be beneﬁcial, it was beyond the scope of the present paper.
Future research can extend the present ﬁndings in a number of ways. First, past research has identiﬁed a number of
different mediators between parenting and delinquent behavior, such as the extent to which adolescents can control their
behavior (Patock-Peckham & Morgan-Lopez, 2006), the degree to which they approve of deviant norms (Simon, Simons,
Chen, Brody, & Lin, 2007), whether they engaged in aggressive behavior as small children (Narusyte, Andershed,
Neiderhiser, & Lichtenstein, 2007), and whether they have an internal versus external locus of control (Shaw & Scott,
1991 ). The present study found that parental legitimacy is an important mediator between parenting style and delinquent
behavior. However, as indicated by our analyses with the imputed datasets, parental legitimacy did not fully mediate this
relation consistently in regards to authoritative parenting. Such ﬁndings suggest that there may be other important mediators
as indicated by the studies cited above, particularly in the case of authoritative parenting. Unfortunately this could not be
investigated here because parental legitimacy was examined in isolation. Future research should examine models that
include multiple mediators, including parental legitimacy.
Second, legal scholars can greatly expand their conception of legitimate authorities by incorporating past research by
developmental psychologists. For example, the present ﬁndings indicate strongly that the ﬁndings of legal scholars can be
expanded to a non-legal entity such as one’s parents. However, it is unclear if these ﬁndings apply to other authority ﬁgures
(e.g., teachers). Past research by developmental psychologists has shown that adolescents’perceptions of teacher legitimacy
inﬂuence their engagement in rule-violating behavior within the school context (Gregory & Ripski, 2008; Laupa & Turiel,
1993). To date, legal scholars have not examined the legitimacy of teacher authority. Future legal research should examine
adolescents’perceptions of a number of different authorities to assess the extent towhich past research can be generalized to
authorities outside the law. In addition, developmental psychologists have shown that adolescents’perceptions of the
legitimacy of parental authority are domain-speciﬁc (e.g., Cumsille et al., 2006). In other words, whether adolescents’view
their parents as legitimate is dependent upon the speciﬁc issue at hand. Legal researchers can expand their theoretical
conception of legitimacy by examining perceptions of legal authorities’legitimacy on an issue-by-issue basis rather than
general legitimacy perceptions as is traditionally done.
Alternatively, developmental psychologists can expand their conception of parental legitimacy byassimilating research by
legal scholars. As stated previously, past research has shown that parental legitimacy varies across situations and issues (e.g.,
Cumsille et al., 2006), but relatively little is known about what creates these individual variations (Darling et al., 2007). We
have shown that parenting style is an important factor that may explain this variation. Developmental psychologists can
examine this variation further by including factors that inﬂuence the perception of legitimacy identiﬁed by past legal
research, such as procedural justice. Procedural justice, the extent to which authorities use fair procedures when interacting
with the public, is a major precursor to perceiving legal authorities as legitimate (Levi et al., 2009; Tyler, 2000, 2006a, 2006b).
R. Trinkner et al. / Journal of Adolescence 35 (2012) 119–132130
Thus, another potential factor that may inﬂuence issue-speciﬁc parental legitimacy may be the extent to which adolescents’
believe their parents are being fair when enforcing their rules about a particular issue.
We were interested in examining the relation among parenting styles, adolescents’perceptions of parental legitimacy, and
engagement in delinquent behavior. We found that the style that parents use to rear their children inﬂuences the way those
children view parental authority in the future, which in turn affects adolescents’delinquent behavior over time. The present
results show that parental legitimacy is one technique for parents to exert control over their children. Additionally, parents
are more likely to be viewed as legitimateauthorities if they utilize authoritative parenting practices rather than authoritarian
or permissive practices, which tend to undermine parental authority. Our data offer further evidence that authoritative
parenting is an effective way for parents to successfully socialize their children and that its inﬂuence works largely through its
effect on youth perceptions of parental legitimacy.
This research was supported by grant SES-0550145 from the National Science Foundation Law and Social Sciences
Program. We would like to thank the following people for assistance in data collection: Stephanie Bartone, Jessica Bean,
Rebekah Blanchard, Michelle Braley, Karinne Brobst, Donald Bucolo, Jess Carig, Jessica Eddy, Steven Falk, Jason Frazier, Jessie
French, Jaime Gallagher, Marissa Hill, Jessica Lucier, Edward MacDonald, & Devon Mahler,
Apel, R., Bushway, S., Paternoster, R., Brame, R., & Sweeten, G. (2008). Using state child labor laws to identify the causal effect of youth employment on
deviant behavior and academic achievement. Journal of Quantitative Criminology, 24(4), 337–362. doi:10.1007/s10940-008-9055-5.
Baron, R. M., & Kenny, D. A. (1986). The moderator–mediator variable distinction in social psychological research: conceptual, strategic, and statistical
considerations. Journal of Personality and Social Psychology, 51(6), 1173–1182. doi:10.1037/0022-3518.104.22.1683.
Baumrind, D. (1967). Child care practices anteceding three patterns of preschool behavior. Genetic Psychology Monographs, 75(1), 43–88, Retrieved from
Baumrind, D. (1971). Current patterns of parental authority. Developmental Psychology, 4(1, Pt.2), 1–103. doi:10.1037/h0030372.
Baumrind, D. (1991). The inﬂuence of parenting style on adolescent competence and substance use. The Journal of Early Adolescence, 11(1), 56–95. doi:10.
1177/027243169111100 4 .
Bendixen, M., Endresen, I. M., & Olweus, D. (2003). Variety and frequency scales of antisocial involvement: which one is better? Legal & Criminological
Psychology, 8(2), 135–150, Retrieved from EBSCOhost.
Browne, M. W. (1984). Asymptotically distribution-free methods for the analysis of covariance structures. British Journal of Mathematical and Statistical
Psychology, 37(1), 62–83, Retrieved from EBSCOhost.
Buri, J. R. (1991). Parental authority questionnaire. Journal of Personality Assessment, 57(1), 110, Retrieved from EBSCOhost.
Cohn, E. S., Bucolo, D. O., Rebellon, C. J., & Van Gundy, K. (2010). An integrated model of legal and moral reasoning and rule-violating behavior: the role of
legal attitudes. Law and Human Behavior, 34(4), 295–309. doi:10.1007/s10979-009-9185-9.
Cortina, J. M. (1993). What is coefﬁcient alpha? An examination of theory and applications. Journal of Applied Psychology, 78(1), 98–104. doi:10.1037/0021-
Cumsille, P., Flaherty, B. P., Darling, N., & Martinez, M. (2006). Chilean adolescents’beliefs about the legitimacy of parental authority: individual and age-
related differences. International Journal of Behavioral Development, 30(2), 97–106. doi:10.1177/0165025406063554.
Darling, N., Cumsille, P., & Martínez, M. (2007). Adolescents’as active agents in the socialization process: legitimacy of parental authority and obligation to
obey as predictors of obedience. Journal of Adolescence, 30(2), 297–311. doi:10.1016/j.adolescence.2006.03.003.
Darling, C., Cumsille, P., & Peña-Alampay, L. (2005). Rules, obligation to obey, and obedience in three cultures: age-related differences and the development
of autonomy. New Directions for Child Development, 108,47–60. doi:10.1002/cd.127.
Darling, N., & Steinberg, L. (1993). Parenting style as context: an integrative model. Psychological Bulletin, 113(3), 487–496. doi:10.1037/0033-2909.113.3.487.
Elliott, D., Huizinga, D., & Morse, B. (1985). The dynamics of deviant behavior: A national survey (Project Report). Boulder, CO: Behavioral Research Institute.
Fagan, J., & Piquero, A. R. (2007). Rational choice and developmental inﬂuences on recidivism among adolescent felony offenders. Journal of Empirical Legal
Studies, 4(4), 715–748. doi:10.1111/j.1740-1461.20 07.00105.x.
Fagan, J., & Tyler, T. R. (2005). Legal socialization of children and adolescents. Social Justice Research, 18(3), 217–242. doi:10.1007/s11211-005-6823-3.
Gregory, A., & Ripski, M. B. (2008). Adolescent trust in teachers: implications for behavior in the high school classroom. School Psychology Review, 37(3),
337–353, Retrieved from EBSCOhost.
Han, W., Miller, D. P., & Waldfogel, J. (2010). Parental work schedules and adolescent risky behaviors. Developmental Psychology, 46(5), 1245–1267. doi:10.
Hoeve, M., Dubas, J., Eichelsheim, V. I., van der Laan, P. H., Smeenk, W., & Gerris, J. M. (2009). The relationship between parenting and delinquency: a meta-
analysis. Journal of Abnormal Child Psychology, 37(6), 749–775. doi:10.1007/s10802-009-9310-8.
Hoeve, M., Smeenk, W., Loeber, R., Stouthamer-Loeber, M., van der Laan, P. H., Gerris, J. R. M., et al. (2007). Long-term effects of parenting and family
characteristics on delinquency of male young adults. European Journal of Criminology, 4(2), 161–194. doi:10.1177/1477370807074854.
Hoffman, J. P. (2010). A life-course perspective on stress, delinquency, and young adult crime. American Journal of Criminal Justice, 35(3), 105–120. doi:10.
Jöreskog, K., & Sörbom, D. (1993). LISREL 8 user’s reference guide. Chicago, IL: Scientiﬁc Software International.
Jost, J. T., & Major, B. (2001). The psychology of legitimacy: Emerging perspectives on ideology, justice, and intergroup relations. New York, NY: Cambridge
Laupa, M., & Turiel, E. (1993). Children’s concepts of authority and social contexts. Journal of Educational Psychology, 85(1), 191–197. doi:10.1037/0022-0663.
85 .1.19 1.
Levi, M., Sacks, A., & Tyler, T. (2009). Conceptualizing legitimacy, measuring legitimating beliefs. American Behavioral Scientist, 53(3), 354–375. doi:10.1177/
Linacre, J. M. (1997). Cronbach alpha or Rasch reliability: which tells the “truth”.Rasch Measurement Transactions, 11, 580–581.
Little, J. A. (1988). A test of missing completely at random for multivariate data with missing values. Journal of the American Statistical Association, 83(404),
R. Trinkner et al. / Journal of Adolescence 35 (2012) 119–132 131
Long, J. S. (1997). Regression models for categorical and limited dependent variables. Thousand Oaks, CA: Sage Publications.
MacKinnon, D. P., Warsi, G., & Dwyer, J. H. (1995). A simulation study of mediated effect measures. Multivariate Behavioral Research, 30(1), 41–62. doi:10.
Milnitsky-Sapiro, C., Turiel, E., & Nucci, L. (2006). Brazilian adolescents’conceptions of autonomy and parental authority. Cognitive Development, 21(3),
Monahan, K. C., & Piquero, A. R. (2009). Investigating the longitudinal relation between offending frequency and offending variety. Criminal Justice and
Behavior, 36(7), 653–673. doi:10.1177/0093854809335527.
Narusyte, J., Andershed, A., Neiderhiser, J. M., & Lichtenstein, P. (2007). Aggression as a mediator of genetic contributions to the association between
negative parent-child relationships and adolescent antisocial behavior. European Child & Adolescent Psychiatry,16(2), 128–137. doi:10.1007/s00787-006-
Newman, K., Harrison, L., Dashiff, C., & Davies, S. (2008). Relationships between parenting styles and risk behaviors in adolescent health: an integrative
literature review. Revista Latino-Americana de Enfermagem, 16(1), 142–150. doi:10.1590/S0104-1169200800 0100022.
Patock-Peckham, J. A., & Morgan-Lopez, A. A. (2006). College drinking behaviors: mediational links between parenting styles, parental bonds, depression,
and alcohol problems. Psychology of Addictive Behaviors, 21(3), 297–306. doi:10.1037/0893-164X.21.3.297.
Rebellon, C. J. (2002). Reconsidering the broken homes/delinquency relationship and exploring its mediating mechanism(s). Criminology, 40(1), 103–136.
Rebellon, C. J., & Waldman, I. (2003). Deconstructing ‘Force and Fraud’: an empirical assessment of the Generality of crime. Journal of Quantitative Crim-
inology, 19(3), 303–331. doi:10.1023/A:1024961026078.
Schafer, J. L., & Graham, J. W. (2002). Missing data: our view of the state of the art. Psychological Methods, 7(2), 147–177. doi:10.1037/1082-989X.7.2.147.
Shaw, J. M., & Scott, W. A. (1991). Inﬂuence of parent discipline style on delinquent behaviour:the mediating role of control orientation. Australian Journal of
Psychology, 43(2), 61–67. doi:10.1080/00049539108259101.
Simons, L., & Conger, R. D. (2007). Linking mother–father differences in parenting to a typology of family parenting styles and adolescent outcomes. Journal
of Family Issues, 28(2), 212–241. doi:10.1177/0192513X06294593.
Simons, R. L., Simons, L., Burt, C., Brody, G. H., & Cutrona, C. (2005). Collective efﬁcacy, authoritative parenting and delinquency: a longitudinal test of
a model integrating community - and family-level processes. Criminology: An Interdisciplinary Journal, 43(4), 989–1029. doi:10.1111/j.1745-9125.2005.
Simon, R. L., Simons, L., Chen, Y., Brody, G. H., & Lin, K. (2007). Identifying the psychological factors that mediate the association betweenparenting practices
and delinquency. Criminology: An Interdisciplinary Journal, 45(3), 481–517. doi:10.1111/j.1745-9125.2007.00086.x.
Sinharay, S., Stern, H. S., & Russell, D. (2001). The use of multiple imputation for the analysis of missing data. Psychological Methods, 6(4), 317–329. doi:10.
Smetana, J. G. (2002). Culture, autonomy, and personal jurisdiction in adolescent-parent relationships. In R. V. Kail, H. W. Reese, R. V. Kail, & H. W. Reese
(Eds.), Advances in child development and behavior, Vol. 29 (pp. 51–87). San Diego, CA US: Academic Press, Retrieved from EBSCOhost.
Smetana, J. G., & Daddis, C. (2002). Domain-speciﬁc antecedents of parental psychological control and monitoring: the role of parenting beliefs and
practices. Child Development, 73(2), 563–580. doi:10.1111/1467-8624.00424.
Sobel, M. E. (1982). Asymptotic conﬁdence intervals for indirect effects in structural equation models. In S. Leinhart (Ed.), Sociological methodology 1982
(pp. 290–312). San Francisco, CA: Jossey-Bass.
SPSS. (2010). PASW missing values 18. Retrieved from. http://support.spss.com/ProductsExt/Statistics/Documentation/18/client/User%20Manuals/English/
Steinberg, L., Mounts, N. S., Lamborn, S. D., & Dornbusch, S. M. (1991). Authoritative parenting and adolescent adjustment across varied ecological niches.
Journal of Research on Adolescence, 1(1), 19–36, Retrieved from EBSCOhost.
Sunshine, J., & Tyler, T. R. (2003). The role of procedural justice and legitimacy in shaping public support for policing. Law & Society Review, 37(3), 513–548.
Sweeten, G., Bushway, S. D., & Paternoster, R. (2009). Does dropping out of school mean dropping into delinquency? Criminology: An Interdisciplinary Journal,
47(1), 47–91. doi:10.1111/j.1745-9125.2009.00139.x.
Tisak, M. S. (1986). Children’s conceptions of parental authority. Child Development, 57(1), 166–176. doi:10.2307/1130648.
Tyler, T. R. (2000). Social justice: outcome and procedure. International Journal of Psychology, 35(2), 117–125. doi:10.1080/002075900399411.
Tyler, T. R. (2006a). Psychological perspectives on legitimacy and legitimation. In S. T. Fiske, A. E. Kazdin, & D. L. Schacter (Eds.), Annual review of psychology,
Vol. 57 (pp. 375–400), Palo Alto, CA, US: Annual Reviews. Retrieved from EBSCOhost.
Tyler, T. R. (2006b). Why people obey the law (2nd ed.). New Haven, CT: Yale University Press.
Wolpin, K. (1983). The national longitudinal handbook: 1983–1984. Columbus, Ohio: Center for Human Resources Research, Ohio State University.
R. Trinkner et al. / Journal of Adolescence 35 (2012) 119–132132