Académique Documents
Professionnel Documents
Culture Documents
DOI 10.1007/s10578-009-0146-9
ORIGINAL ARTICLE
Abstract This study examines how multiple indicators of adolescent and parent acculturation relate to longitudinal trajectories of Latino adolescent aggression. The hierarchical
linear modeling analysis is based on a final sample of 256 adolescents paired with one
parent. Of the adolescents, 66% were born outside of the United States and the remaining
34% were US-born. Families lived in two sites: 38% lived in North Carolina and 62% lived
in Arizona. The overall trajectory of Latino adolescent aggression displays a statistically
significant negative trend best characterized by a quadratic curve. We delineate significant
risk factors related to aggression levels, and show that gender, age, parent-reported
acculturation conflicts, and adolescent-reported parent-adolescent conflicts are associated
with higher levels of adolescent aggression. We discuss the study limitations, implications
of the findings, and fertile ground for future research.
Keywords Latinos Adolescents Aggression Externalizing conduct problems
Immigrants Acculturation Culture
Introduction
Nationally representative epidemiologic data suggest that aggressive behavior is a serious
concern for Latino youth. The 2005 Youth Risk Behavior Survey [1] reported prevalence
rates for physical fights (41%) and injured in a physical fight (5.3%) reported by Latino
high-school students, which were relatively equal to prevalence rates of African American
students (43.1 and 5.4%, respectively) but significantly higher prevalence than Caucasian
P. R. Smokowski (&) R. A. Rose
School of Social Work, University of North Carolina at Chapel Hill, 325 Pittsboro Street,
CB 3550, Chapel Hill, NC 27599-3550, USA
e-mail: smokowsk@email.unc.edu
M. Bacallao
Department of Social Work, University of North Carolina Greensboro, 257 Stone Building,
P.O. Box 26170, Greensboro, NC 27402-6170, USA
e-mail: m.bacallao@uncg.edu
123
590
students (33.1 and 2.4%, respectively). Furthermore, Latino students (9.8%) were significantly more likely than Caucasian students (7.2%) to report that within the last year they
had been threatened or injured with a weapon at school, and Hispanic students were
significantly more likely to not go to school because of safety concerns than were their
Caucasian peers (10.2 vs. 4.4%). Moreover, these patterns remain stable when gender is
considered. It may be that such experiences of threats and safety concerns are related to
exposure to gangs; in 1995, one-half of the Latino students surveyed by the Departments of
Education and Justice reported that street gangs were present in their schools during the
previous 6 months [2]. In 1999, 28% of Latino students reported gang exposure [2]. Based
on this national epidemiologic data, it is particularly important to examine risk factors that
increase the probability of aggressive behavior in Latino youth. Acculturation has surfaced
as one such risk factor salient for this minority group.
Acculturation, Adolescent Aggression, and Youth Violence
Defining Acculturation. Acculturation was first defined as phenomena which results when
groups of individuals having different cultures come into continuous first hand contact with
subsequent changes in the original culture patterns of either or both groups [3]. This
original definition stressed continuous, long-term change and allowed for the process to be
bidirectional, wherein both of the interacting cultures could make accommodations. During
the decades since acculturation was first defined, a number of alternative definitions have
been offered that often stress unidirectional, rather than bidirectional, change. For example,
Smith and Guerra [4] refer to acculturation as the differences and changes in values and
behaviors that individuals make as they gradually adopt the cultural values of the dominant
society. These unidirectional trends suggest that cultural change results from interactions
between dominant and nondominant groups, and such change is commonly characterized
by nondominant groups taking on the language, laws, religions, norms, and behaviors of
the dominant group [5, 6]. Many factors, such as differences in attitudes between generations and sociopolitical trends, have influenced the conceptualization of acculturation,
leaving no universally accepted definition of the term.
Adding further complexity, many other constructs in cultural research, such as assimilation, enculturation, acculturation stress, segmented assimilation, and biculturalism, have
been invoked under the umbrella of acculturation research. The term acculturation, which
denotes the bidirectional process of cultural change, is often erroneously used interchangeably with the term assimilation, which captures unidirectional adaptations made by
minority individuals to fit into the host society. Consequently, the original Redfield [3]
definition captures the bidirectional notion of acculturation whereas the description offered
by Smith and Guerra [4] denotes the unidirectional assimilation approach. These competing
unidirectional and bidirectional approaches dominate acculturation research, influencing
conceptualization, measurement, analytic strategies, and results of empirical studies in this
area [7]. In light of having no universally accepted definition for acculturation, for the
purposes of our background foundation for the current study, we discuss extant studies that
use both bidirectional and unidirectional approaches to assess acculturation. Accordingly,
these studies measured acculturation in a variety of ways (e.g., language use, generation
status, acculturation stress, ethnic identity, and with multidimensional psychometric scales).
Defining Aggression and Youth Violence. Violence is the intentional use of physical
force or power, threatened or actual, against oneself, against another person, or against a
group or community, that either results in or has a high likelihood of resulting in injury,
death, psychological harm, maldevelopment, or deprivation [8]. Adolescent interpersonal
123
591
youth violence includes violence between unrelated youth who may or may not know each
other in other contexts and environments. Similar to acculturation, adolescent interpersonal
youth violence has been assessed in multiple ways across studies. Youth violence
researchers use measures including gang membership, bullying, physical fighting, weapon
carrying, verbal threats, aggressive behavior, externalizing symptoms, and serious criminal
activity such as homicide or assaults. In the research literature linking acculturation to
youth violence, five studies used the Child Behavior Checklist or Youth Self-Report [9
13], two articles used the Diagnostic Interview Schedule for childrenIV disruptive
behavior disorders subscale [14, 15], and nine used various self-report items or measures of
delinquency, theft or vandalism, and deviancy. Only two studies included multiple
informants for youth violence [10, 13].
In our discussion below, we examine studies linking an array of acculturation measures
to these disparate measures of youth violence perpetration. After establishing the role of
acculturation in the wide area of youth violence research, we focus the current study on
aggressive behavior, which is arguably the most common subtype of youth violence,
characterized by arguing, destroying property, and fighting that often does not rise to the
level of serious criminal activity.
Acculturation and Youth Violence. The association between Latino adolescent acculturation and youth violence outcomes has been examined in 16 studies; 13 of these
investigations examined the perpetration of violence or aggressive behavior as the outcome, and three studies examined fear of being a victim of violence as the outcome [16].
Although the results of the reports that considered the perpetration of youth violence were
mixed, these results favored a significant positive association between acculturation and
youth violence. Of the 13 empirical investigations focused on adolescent acculturation and
youth violence perpetration, nine studies reported higher adolescent US cultural involvement (although defined in different ways and sometimes referred to as assimilation) was
associated with increased youth violence [9, 1113, 1721].
Buriel et al. [19] reported that third generation Latino adolescents had significantly
higher delinquency rates than first or second generation adolescents. First and second
generation adolescents did not significantly differ on delinquency. Similarly, Bui and
Thongniramol [18] examined a nationally representative subsample of 18,097 student data
obtained from the National Longitudinal Study of Adolescent Health. After running
logistic regression models, odds ratio statistics indicated that youth who were second- or
third-generation Hispanics were 60 and 88% (respectively) more likely to report violent
delinquency as compared to their first- generation counterparts. Several other studies show
differences for immigrant versus US-born youth, with risk of violence or delinquency
rising for the latter group [12, 13, 17, 21]. Although these studies establish a link between
generational status, nativity, and youth violence, these proxy measures of acculturation
described sociological cohort effects rather than person-centered psychological effects.
A second subgroup of studies considers individual, rather than cohort, effects by using
person or psychological measures of acculturation (e.g., language use, cultural involvement, cultural conflicts or acculturation stress). Among a predominantly Cuban sample of
2,360 adolescents living in Miami, Vega et al. [13] found that only language conflicts were
associated with total behavior problems as reported by the parents and teachers of
immigrant adolescents. However, among the US-born Cuban youth, language conflicts,
perceived discrimination, and perceptions of a closed society were associated with
behavior problems reported by teachers but not by parents. Furthermore, Dinh and her
colleagues [9], whose assessment of 330 Hispanic youth represents the only extant
longitudinal study, found higher acculturation significantly predicted higher levels of
123
592
123
593
123
594
involvement. Compared to simple markers or one-dimensional measures, this multidimensional, multiple informant approach allows for a much more refined analysis of
acculturation dynamics.
The current study contributes to the emerging body of literature on Latino adolescent
acculturation and youth violence in several ways. We examined the relationship of multiple indicators of adolescent and parent acculturation (e.g., Latino and US cultural
involvement) and acculturation stressors (e.g., perceived discrimination, acculturation
conflicts) with adolescent aggression. Our sample included parents and adolescents living
in rural, small town, and metropolitan areas located in two geographically separated states.
We extend the research knowledge of the cross-sectional studies reviewed above by
exploring longitudinal trajectories of acculturation and aggression. Based on extant
research, we hypothesized that (a) adolescent US cultural involvement would be positively
related to adolescent aggression; (b) adolescent aggressive behavior would rise over time;
and (c) acculturation stressors, such as parent-adolescent conflict, would be positively
associated with adolescent aggression.
Methods
Data collection procedures for the Latino Acculturation and Health Project have been
discussed in detail elsewhere [11]. This study used parent and adolescent perceptions of
the adolescents aggressive behavior as outcomes in a longitudinal rater effects hierarchical linear model (HLM) [41, 42]. The structure of the data can be summarized as time
(four waves) nested within rater, which is nested within participant. In the present
context, the adolescent is considered the study participant. With only one adolescent
from each family randomly selected for inclusion in the study, characteristics of the
adolescent, parent, or family as a whole were implied as characteristics of the adolescent,
and further levels (i.e., a parent or family level) would have been redundant. Each
participants level of aggressive behavior was assessed using two raters: the adolescent,
who provided data on self-perception of aggression; and the adolescents parent, who
provided data on his or her perception of the childs aggression. We assessed each raters
perceptions of the adolescents aggressive behavior at four time points with intervals of
approximately 6 months. Analyses were conducted on adolescent outcomes using a mix
of adolescent, parent, and family characteristics as predictors. All variables, except time,
were entered as participant-level (level 3) variables, whereas time was a level 1 variable.
Given that there was no intention to explain variation between raters, the analyses did
not include any rater-level variables.
Independent Variables
Time. The analysis used a variable occasions design [43], whereby time was measured
continuously rather than in discrete waves (W = 0, 1, 2, 3). We used time living in the
United States, measured in months, at each time the survey was administered. We identified the time in United States at the initial condition (W = 0) and then summed the
number of intervening months from the first to each subsequent wave (though separated by
approximate 6 month intervals, exact dates were recorded; the variable was recorded in
months with decimals representing increments of months). These two values were then
added together to obtain time living in United States at each assessment. For native
adolescents, the time at the first wave (Wave 0) was their age; for a non-native adolescent,
123
595
time at Wave 0 was recorded as time since his or her immigration. Time measured in this
format differs from time measured in discrete waves in two ways. First, initial condition is
not the same for every participant, whereas using W = 0 treats initial condition as being
the same. Second, the increments between each wave are not exactly the same for each
participant: at Wave 2 (6-month follow-up), the increment ranged from 3.6 to 12.5 months;
at Wave 3 (12-month follow-up) from 8.1 to 16 months, and at Wave 4 (18-month followup) from 13.6 to 22.4 months. Thus, we captured the variable for time with both a linear
component representing the average rate of change per month the adolescent lived in the
United States, and a quadratic component representing acceleration in this rate.
Demographic Variables. Demographic data for adolescents included gender
(female = 1, male = 0); age (in years); and nativity by site. Initially, nativity was a
dichotomous variable labeled 1 for Latin American nativity and 0 for US nativity; however, we found nativity was strongly related to site membership. All North Carolina (NC)
site participants were foreign-born immigrants, but Arizona (AZ) site participants varied in
nativity. Therefore, site and birthplace were recoded into two variables representing AZ
families with US-born adolescents and AZ families with foreign-born adolescents, with all
NC families as the common reference group. Also included as characteristics representing
family influences on adolescents were variables for parent education level (elementary or
no schooling coded 1 and high school or higher coded 0) and household annual income (in
thousands of dollars).
Culture-of-Origin Involvement. We defined this variable as the maintenance of ethnic
identity through language, media use, and enacting traditions from the persons native
culture, measured using the Bicultural Involvement Questionnaire (BIQ) [44]; The BIQs
culture-of-origin involvement subscale has 20 items measuring language, food, recreation,
and media use on a 5-point Likert scale with anchors labeled not at all to very much.
Examples of questions include How comfortable do you feel speaking Spanish (at home,
with friends, in general) and How much do you enjoy music, television programs,
books, and magazines from your native country? In this sample, internal consistency
reliability was .89 for adolescents and .90 for parents culture-of-origin involvement.
US Cultural Involvement. This variable, defined as assimilation of host culture language, media, norms, and traditions, was also measured using the BIQ [44]. Exactly
parallel to the culture-of-origin involvement scale, the BIQs US cultural involvement
subscale has 20 items measuring English language use, non-Latino US food preference,
recreation, and media use on a 5-point Likert scale with anchors labeled not at all to very
much. In this sample, internal consistency reliability was .90 for adolescents and .93 for
parents.
Average scores for culture-of-origin involvement and US cultural involvement were
calculated by adding the items and dividing by the number of items answered. This step
yielded final variables with the possible ranges identical to the original 5-point Likert scale.
Both parent and adolescent involvement variables of all types were included as adolescent
characteristics.
Parent-adolescent conflict was assessed using the Conflict Behavior Questionnaire-20
(CBQ-20) [45]. The CBQ scale provides an overall measure of negative communication
conflict within a parent-adolescent pair. The CBQ-20 uses 20 yes-no items to assess
positive and negative interactions that occur in both nonconflictual and argumentative
exchanges. Examples of questions include the following: (a) My parent(s) dont understand
me; (b) My parent(s) say I have no consideration for them; and (c) My parent(s) put me
down. Reliability for the CBQ-20 in this sample was .89 for both adolescents and parents.
The adolescent report of this variable was entered as a characteristic of the adolescent.
123
596
Acculturation conflicts were measured with a four-item scale used by Vega et al. [46].
Response options were measured on a 5-point Likert scale with anchors ranging from not
at all to frequently. The four items were (a) How often have you had problems with your
family because you prefer American customs; (b) How often do you think that you
would rather be more American if you had a chance; (c) How often do you get upset at
your parents because they dont know American ways [not in parent version]; and (d)
How often do you feel uncomfortable having to choose between non-Latin and Latin
ways of doing things? In this sample, internal consistency reliability was .76 for adolescents and .87 for parents. Both parent and adolescent versions of this variable were
entered as adolescent characteristics.
Dependent Variable: Aggression
Parents and adolescents reported adolescent aggression using the Child Behavior Check
List (CBCL/4-18 and Youth Self-Report [YSR]) [47]. The YSR aggression subscale
consists of 17 items measured using a 3-point Likert scale (not true, sometimes true, and
often true). The scale included the following items: (a) I argue a lot; (b) I destroy my
own things; and (c) I get in many fights. Parents completed the CBCL/4-18 with
similar questions concerning their child. The CBCL has been widely used with Latino
children [13]. The internal consistency reliability for this sample was .85 for adolescent
and .89 for parents.
Sample Characteristics
The details of sample characteristics and study variables by site and nativity are displayed
in Table 1. The analysis was based on a final sample of 256 adolescents paired with one
parent. Of the final sample, 66% of the adolescents were born outside of the United States
and the remaining 34% were US-born. More than a third of the families lived in North
Carolina (38%) and 62% of the sample lived in Arizona. Of the adolescents, 57% were
female. Most of the adolescents (88%) attended school, and the median grade was 10th
grade. The vast majority of parents were foreign-born (95%). Although most adolescents
lived with two parents (73%), 27% lived with a single parent. Of the participating parents,
91% were mothers, 72% of parents were married, and 68% were working at least one job.
In addition, 89% of parents had not graduated from high school, and 39% had less than a
9th-grade education.
Sampling Design: Nesting of Participants
Sampling adolescent-parent pairs resulted in a nesting of these two participants within a unit
commonly referred to as a cluster. The intraclass correlation (ICC), which measures the
proportion of variation attributed to raters and adolescents, was 34% at the rater level and
31% at the participant level when an unconditional linear growth model was used (only a
linear effect for time was entered; no demographics or survey responses). Given these ICC
levels, we rejected the notion that time and rater-level observations were independent, and
therefore controlled for rater-level differences and participant attributes that were invariant
to time or rater using HLM. However, some simplifications were necessary to deal with
other sources of dependence among the records. To prevent unaccounted-for dependence
between adolescent participants from the same family, only one adolescentparent dyad
123
597
Proportion of sample
0.57
0.34
0.28
0.32
Characteristic: continuous
Mean
SD
3.74
0.62
3.91
0.68
3.43
0.71
2.62
0.78
Adol biculturalism
7.17
0.85
Parent biculturalism
6.53
0.78
1.97
0.80
1.89
0.98
2.33
0.94
3.33
0.52
0.25
0.26
Adol age
16.02
1.65
10.13
5.51
26.66
18.37
0.40
0.32
Aggression (DV)
a
CO = country-of-origin
from each family were included in the final sample. For families with more than one
adolescent participating, the adolescentparent pair included in the sample was selected at
random. Finally, we discarded the records for those families with only an adolescent or a
parent participating (i.e., no pair within a family). These deletions resulted in the final
sample of 516 raters (258 participants paired with one parent), at the exclusion of 158
discarded records (148 due to randomization and 10 due to having only a parent or adolescent). We found significant differences between the included and excluded records on the
following characteristics: (a) time living in the United States (t = 3.34, p \ .01); (b)
familism (t = 3.95, p \ .01); (c) adolescent report on conflict behavior (t = 2.49, p \ .05);
(d) acculturation conflict (parent report t = 2.04, p \ .05, adolescent report t = 2.31,
p \ .05); and (e) adolescent cultural involvement (culture of origin, t = 2.13, p \ .05, US
cultural involvement, t = 2.11, p \ .05).
Missing Data. Several variablesincluding all cultural involvement variables, conflict
behavior, discrimination, familism and parent-adolescent conflicthad non-response rates
of 12%. Income had a 4% rate of non-response. The dependent variable had a response rate
that varied over the four waves: 12% non-response at Wave 1, 17% at Wave 2, 32% at Wave
3, and 40% at Wave 4. Given the potential for bias from non-response, analyses were
conducted on data subjected to multiple imputation [48]. Based on recent standards
123
598
recommended by Graham et al. [49] 50 data sets were simulated in the imputation process. To
handle missing data on the dependent variable, we used a procedure called multiple imputation, then deletion (MID) [50]. With MID, the imputation uses the dependent variable,
which ensures that the covariance structure of the analysis model is represented in the
simulated values. However, after imputation the simulated values of the dependent variable
are deleted before conducting the analysis because these simulated values provide no useful
information. These procedures were completed using SAS Proc MI and MIAnalyze. Collinearity was assessed using the variance inflation factor (VIF). A VIF of 10 would warrant
caution. The highest VIF observed was 2.9, and therefore we concluded that the data were not
collinear.
Analytic Strategy
The data consisted of three hierarchically nested levels. The first level consisted of each
raters assessment of adolescent aggression taken at four time points. The second level
consisted of random effects for each rater allowing for and capturing variability between
how adolescents and their parents perceive the adolescents aggression levels. No covariates were used at this level. The third level consisted of adolescent and family level
characteristics including demographics (adolescent gender, age, sample site by nativity);
socioeconomic status (parent education and income); adolescent perceptions of discrimination; both parent and adolescent perceptions of acculturation conflict and conflict
between the parent and adolescent; and cultural involvement for both US culture and
country-of-origin culture. A three-level hierarchical linear model was used to account for
the variability at all three levels [41, 42]. SAS version 9.1 Proc Mixed was used; restricted
maximum likelihood was used to estimate the reported models (a maximum likelihood
version of each model was run to facilitate a comparison of fit between Models 1 and 2,
which have different fixed effects [43]).
Model Form. The model used in the analysis was as follows:
2
rtij rtij N0; r2
Aggtij b0ij b1ij Ttij b2ij Ttij
2:1
b2ij p20j
2:2
3:0
p10j c100
3:1
p20j c200
3:2
In Eq. 1, aggression for adolescent j for rater i at time t was regressed on a constant (for
status at T = 0), rate of change (per month; Ttij) and a quadratic component for time
2
), which together capture variation in
representing acceleration in the growth rate (Ttij
perceived level of aggression over time.
In Eqs. 2 and 2.1, status at T = 0 and instantaneous rate of change coefficients (b0ij and
b1ij) were each then regressed on a constant p and random coefficient u representing raterlevel variation around this constant. The acceleration in change coefficient b2ij was
123
599
regressed on a rater-level constant only. No covariates were entered at the rater level to
explain variation between raters in perceived initial status of aggression or the instantaneous rate of change in aggression.
The constants in Eqs. 22.2 were then regressed on adolescent-level models. In Eq. 3.0,
the random intercept from the regression of initial status (p00j) was then regressed on a set
of adolescent- and family-level predictors, with a random component (e00j) representing
residual parameter variance. In Eq. 3.1, the rater-level means of rate of change and
acceleration were then regressed on participant level constants. The coefficients c are the
coefficients reported in the results; these represent participant-level means of all effects.
Two model variations are reported. In Model 1, only a linear time coefficient was included
(b1ij). The quadratic and random coefficients were not included (we assumed b2ij = 0 and
u1ij = 0). In Model 2, all effects were included. The average rate of change is properly
characterized as the instantaneous rate of change when the quadratic component is included.
Mean Centering. The use of time living in United States to define the time variable
facilitates an analysis that accounts for participants being at different points in their
developmental trajectories. However, the center (the point at which time = 0) of the time
variable must be chosen carefully. None of the participants had zero time in the United
States. If measured in its raw units, the intercept and instantaneous rate of change coefficients, which are interpreted where the quadratic time variable is at zero (at the point
where T = 0), would be interpreted at a point that could not occur for any study participant. To facilitate a meaningful interpretation of these coefficients, and to remove collinearity between the instantaneous rate of change and quadratic components, the time
variable, calculated as number of months living in the United States at each wave, was
centered at each adolescents mean across the four time points [42]. Thus, the point at
which T = 0 is not likely to be any of the points at which data were collected, but
importantly, the point does occur at some time between the first wave of data collection
and the last (i.e., within the study period) for all of the participants being studied. Having
T = 0 occur at some arbitrary point within the study period facilitates interpretation of the
random intercept and instantaneous rate of change coefficients at this value of T.
In addition to time, all continuous adolescent-level measures (e.g., age, income and all
scales except the dependent variable) were mean centered to facilitate interpretation of
random intercept at the rater and adolescent levels as the variation around the mean level of
the outcome at the mean time living in the United States [42].
Results
Model 1: Linear Nonrandom Time Living in the United States
This was the baseline model, having the simplest covariance structure for any model
accounting for nesting. Model characteristics are shown in Table 2. On average, aggression
levels fell during the study period (c = -.01, p \ .001). This model shows that higher
levels of adolescent US cultural involvement were significantly predictive of lower levels
of adolescent aggression (c = -.04, p \ .05). Alternatively, higher parent reported
acculturation conflict (c = .04, p \ .01) and adolescent reported parent-adolescent conflict
behavior (c = .44, p \ .001) were significantly predictive of higher levels of aggression.
Females were significantly more likely to be aggressive (c = .05, p \ .05), whereas older
children were significantly less likely to be aggressive (c = -.02, p \ .05). The two
random effects were both significant (v2 (rater level) = 488.32, p \ .001; v2 (adolescent
123
600
level) = 6.89, p \ .01). The fixed effect model explained 70% of the variance in the
participant-level model (level 3) (Table 2).
Model 2: Quadratic and Linear Random Slope for Time Living in the United States
This model contained two additional elements, a quadratic time variable and a random
slope for the linear (instantaneous) rate of change coefficient, producing a more complex
Table 2 Longitudinal hierarchical linear models predicting Latino adolescent aggression
Fixed effect
Model 1: linear, no
random time slope
SE
(b)
t Value
SE
(b)
t Value
0.00
-6.37***
-0.01 0.00
-7.40***
-0.01
0.0003 0.00
2.36*
0.33
0.00 0.02
0.24
0.01
0.02
0.01 0.02
0.38
0.01
0.02
0.32
-0.04 0.02
-1.98*
-0.04
0.02
-1.91
-1.11
-0.02 0.02
-1.12
-0.02
0.02
0.02 0.02
1.05
0.02
0.02
1.07
0.04 0.01
3.03**
0.04
0.01
2.85**
0.02 0.01
1.47
0.02
0.01
1.43
0.44 0.05
8.23***
0.45
0.05
8.15***
0.05 0.02
2.01*
0.05
0.02
1.97*
-0.02 0.01
-2.33*
-0.02
0.01
-2.45*
0.04 0.03
1.36
0.04
0.03
1.29
0.00 0.03
-0.16
0.00
0.03
-0.16
0.00 0.00
-0.85
0.00
0.00
-0.85
-0.03 0.03
-1.21
-0.03
0.03
-1.22
Variances/random effects
Est.
Chi-square
Est.
Residual
0.033
0.032
488.32***
0.0331
489.21***
0.000
27.06***
0.000
0.42
0.0088
0.009
7.37**
0.43
0.12
0.13
0.70
0.69
Chi-square
0.027
6.89**
0.48
CO = country-of-origin; (C) means the variable was centered before entering it in the model
123
601
fixed effects model and covariance structure than the baseline model. Plots of each raters
perception of adolescent aggression suggested that a nonlinear effect was present. An
analysis of maximum likelihood versions of Models 1 and 2 showed that Model 2 was a
better-fitting specification. The quadratic term capturing acceleration in the rate of change
in adolescent aggression had a coefficient of .0003 (p \ .05), while the instantaneous rate
of change coefficient was -.01 (p \ .001), which is the same magnitude and direction as in
the baseline model. However, the addition of a term for acceleration in change provides
further insight into the change in aggression reported by adolescents and their parents. The
coefficient of -.01 for instantaneous rate of change tells us that at the mean time in the US,
aggression was decreasing. Nevertheless, when the instantaneous rate of change and
quadratic coefficients are interpreted together, the results indicate that, on average,
aggression decreased until 16.67 months after the mean time in US, at which point
aggression started to increase. To provide a more intuitive meaning to this finding, for a
heuristic average adolescent, for whom the mean time in the US was 10.13 years
(121.56 months; refer to Table 1), aggression started to increase after 11.5 years
(138.23 months) in the United States.
In Model 2, relative to the baseline model, the only substantial change in the findings of
significance was that adolescent US cultural involvement was not significant. For both
Models 1 and 2, findings were not substantially different for parent reports of acculturation
conflict, and for adolescent reports of parent-adolescent conflict on behavior. Further, both
models were similar on characteristics for gender and age. A further investigation was
conducted to determine if the quadratic term or the random slope at the rater level was
responsible for the change in the significance of the effect of adolescent US cultural
involvement on reducing aggression when moving from the baseline to the more complex
model. (Given the mean centering of the time regressor, we expect the change to have a
substantive interpretation and not simply the result of a structural change from including a
nonlinear regressor.) The difference in significance findings was a result of both a drop in
the magnitude of the coefficient and an increase in the standard error, and the actual
difference in the t-value was minor (a change of .07). Individually, the addition of either
the quadratic term or the random slope for time caused the t-value for adolescent US
cultural involvement to decrease. This change indicates the presence of some confoundedness between adolescent US cultural involvement and the acceleration in change in
aggression and random linear slope. These findings may be due to a modest positive, nonsignificant association between time in the United States and adolescent US cultural
involvement that we observed.
Foreign- Versus US-Born Subgroup Analyses. We conducted a final subgroup analyses,
fitting model 2 separately to foreign-born and US born youth. On average, aggression levels
fell during the study period for both foreign and US-born youth (c = -.01, p \ .001).
However, the coefficient for acceleration in the rate of change was statistically significant
for foreign-born, but not for US-born youth. This suggests that US-born adolescents during
the study period displayed a linear decrease in their aggressive behavior; whereas, for
foreign-born youth the decline in their aggressive behavior flattened out over time. Further,
the coefficient for adolescent US cultural involvement approached standard values for
statistical significance (t-value = -1.93) for foreign-born youth, but was not relevant for
US-born adolescents (t-value = -0.14). Similarly, parent reports of acculturation conflicts
were a significant risk factor associated with more aggressive behavior for foreign-born
youth (t-value = 1.98, p \ .05), but not for US born youth (t-value = 1.82, ns). Parentadolescent conflict was the strongest, most significant risk factor related to aggressive
behavior in both foreign-born and US-born adolescents.
123
602
Discussion
This is the first investigation to go beyond the simple linear association between acculturation measures and adolescent aggression by examining the longitudinal trajectory of
aggressive behavior since time of immigration (or birth for native Latino adolescents). The
overall trajectory of Latino adolescent aggression displayed a statistically significant
negative trend that was best characterized by a quadratic curve, decreasing from baseline
(Wave 0) to Wave 2 follow-up 1 year later and slightly increasing at Wave 3 (18-month
follow-up; see Figs. 1, 2). However, when subgroups were considered, this trajectory,
while still decreasing, was linear for US-born adolescents and quadratic for foreign-born
youth.
It was surprising to find a significant negative trajectory for adolescent aggression; this
contradicted our hypotheses based on prior cross-sectional studies. Indeed, we hypothesized that adolescent US cultural involvement would be positively related to adolescent
aggression and that adolescent aggressive behavior would rise over time. The HLM models
showed that adolescent US cultural involvement was significantly and inversely related to
adolescent aggression, and that adolescent aggressive behavior decreased over time. These
findings contradict past cross-sectional research that reported a positive relationship
between assimilation measures and adolescent aggression [9, 12, 13, 1721]. It may still be
the case that subsequent generations of Latino youth report higher levels of aggressive
behavior and violence [18, 19]. Yet, the person-centered analyses adopted in this study
showed for the first time that individual trajectories of adolescent aggression are highest for
immigrant youth near the time of immigration and decrease thereafter. This finding adds
refinement to our understanding of the relationship between acculturation and aggression.
At the same time, the quadratic curve showed some increases near the end of our trajectories, indicating a complex longitudinal relationship. Notably, this decreasing trajectory
was characteristic of both foreign- and US-born adolescents.
Three theoretical explanations for this attenuating curve can be posited from cultural
psychology. The first explanation suggests that this curve signifies positive attributes of the
acculturation process. The common notion of assimilation or behavioral adaptation entails
0.6
Level of Aggression
0.5
0.4
0.3
0.2
Low Conflict Behavior
0.1
0
0
Wave
Fig. 1 Aggression by parent-child conflict behavioryouth
123
603
0.6
Level of Aggression
0.5
0.4
0.3
0.2
Low Acculturation Conflict
0.1
0
0
Wave
Fig. 2 Aggression by acculturation conflictparent
123
604
and familism were cultural assets associated with fewer internalizing problems and higher
self-esteem in their sample of 323 Latino adolescents living in North Carolina. Similarly,
Coatsworth et al. [54] compared the acculturation patterns of 315 Hispanic youth and found
that bicultural youth had the most adaptive pattern of functioning across a number of
different ecological domains. Considering these positive effects, both the decreasing trajectory for aggression and the positive effect found for adolescent US cultural involvement,
particularly for foreign-born adolescents, may provide further support for the benefits of
biculturalism.
Finally, decreased aggressive behavior may be a sign that, over time, Latino adolescents
learn strategies for coping with acculturation stress. Low-acculturated individuals experiencing high levels of stress display negative self-esteem, experience acculturation conflicts,
and are commonly cut-off from the benefits of their cultures of origin [13, 28]. These lowacculturated individuals often lack the resources and skills to successfully navigate within
their new environment and have greater difficulty with negative stereotypes and perceived
discrimination [25]. Gil et al. [27] found low-acculturated adolescents who were born in the
United States to have a particularly problematic profile of stressors and difficulties. Compared to foreign-born peers, these low-acculturated US-born Latino adolescents were much
more likely to perceive discrimination and internalize negative stereotypes. In reaction to
acculturation stress, Latino adolescents may feel the need to defend themselves or to
affiliate with gangs for safety [55]. However, over time, both native and immigrant Latino
adolescents may acquire coping mechanisms to peacefully handle acculturation conflicts,
decreasing the frequency and intensity of reactive aggressive behaviors.
The acculturation stress explanation for higher levels of aggression at baseline that
decrease over time has additional supporting evidence in our models. Latino adolescents
whose parents reported acculturation conflicts displayed significantly higher levels of
aggressive behavior at baseline as compared to adolescents whose parents reported fewer
acculturation conflicts. The trajectory marking decreasing adolescent aggression was still
evident for this group with higher parent-reported acculturation conflicts; however, the
actual levels of adolescent aggression were consistently higher in the context of acculturation conflicts (see Fig. 2).
This dynamic was even more pronounced for parent-adolescent conflict behavior (see
Fig. 1). Adolescents who reported conflict with their parents were significantly more
aggressive at baseline and, despite the decreasing trajectory of aggression, remained more
aggressive across all waves of data as compared to adolescents with fewer parent-adolescent conflicts. This positive relationship between parent-adolescent conflict and Latino
adolescent aggression extends previous cross-sectional research on the role of parentadolescent conflict as a risk factor [11, 40, 56]. In addition, this positive relationship
provides additional support for the notion that family processes mediate the relationship
between acculturation and Latino adolescent aggression [911, 20]. Parent-adolescent
conflict was a risk factor for aggressive behavior in both foreign-born and US-born adolescents. This conflict may gain additional fuel from acculturation experiences, but it
appears to be a salient risk factor regardless of level of acculturation and independent of
acculturation stress.
Implications for Practice
It is good news that Latino adolescent aggression appears to decrease naturally without
intervention. At the same time, however, practitioners and prevention scientists should
consider that there is an important opportunity for service provision shortly after
123
605
123
606
Summary
This study examined how multiple indicators of adolescent and parent acculturation relate
to longitudinal trajectories of Latino adolescent aggression. The hierarchical linear modeling analysis was based on a final sample of 256 adolescents paired with one parent. The
overall trajectory of Latino adolescent aggression displayed a statistically significant
negative trend that was best characterized by a quadratic curve, decreasing from baseline to
follow-up 1 year later and slightly increasing at 18-month follow-up. This quadratic curve
characterized foreign-born youth; their US-born counterparts displayed a linear decreasing
trajectory. This decreasing trend contradicts past reports of a positive association between
adolescent acculturation and aggression. Risk factors were also delineated, showing that
parent-reported acculturation conflicts, for foreign-born adolescents, and adolescentreported parent-adolescent conflicts, for all adolescents, were associated with higher levels
of adolescent aggression. Prevention practitioners are encouraged to target these risk
factors with family-focused programs that they strategically time to occur shortly after
immigration when acculturation stress and adolescent aggression are at their highest points.
Acknowledgments The authors wish to thank Dr. Flavio Marsiglia and Monica Parsai, MSW for their
work collecting data in Arizona and Melissa Chalot, MPH for project management. Special thanks go to the
Latino families who participated in this study. This study was supported by grants from the Center for
Disease Controls National Center for Injury Prevention and Control (R49/CCR42172-02) and from the
Centers for Disease Controls Office of the Director (1K01 CE000496-01).
References
1. Centers for Disease Control and Prevention (2006) Youth risk behavior surveillance-United States,
2005. Surveillance Summaries. MMWR 2006:55 (No. SS-5). Available from: http://www.cdc.gov/
mmwr/PDF/SS/SS5505.pdf
2. Kaufman P, Chen X, Choy SP, Peter K, Ruddy SA, Miller AK et al (2001) Indicators of school crime
and safety: 2001 (NCES 2002113 or NCJ-190075). US Department of Education and US Department
of Justice, Washington
3. Redfield R, Linton R, Herskovits M (1936) Memorandum for the study of acculturation. Am Anthropol
38:149152. doi:10.1525/aa.1936.38.1.02a00330
4. Smith EP, Guerra NG (2006) Introduction. In: Guerra NG, Smith EP (eds) Preventing youth violence in
a multicultural society. American Psychological Association, Washington, pp 314
5. Berry JW (1998) Acculturation stress. In: Balls Organista P, Chun KM, Marin G (eds) Readings in
ethnic psychology. Routledge, New York, pp 117122
6. Castro FG, Coe K, Gutierres S, Saenz D (1996) Designing health promotion programs for Latinos. In:
Kato PM, Mann T (eds) Handbook of diversity issues in health psychology. Plenum, New York, pp
319346
7. Cabassa LJ (2003) Measuring acculturation: where we are and where we need to go. Hisp J Behav Sci
25:127146. doi:10.1177/0739986303025002001
8. Dahlberg LL, Krug EG (2002) Violence: a global public health problem. In: Krug EG, Dahlberg LL,
Mercy JA, Zwi AB, Lozano R (eds) World report on violence and health. World Health Organization,
Geneva, pp 121
9. Dinh KT, Roosa MW, Tein JY, Lopez VA (2002) The relationship between acculturation and problem
behavior proneness in a Hispanic youth sample: a longitudinal mediation model. J Abnorm Child
Psychol 30:295309. doi:10.1023/A:1015111014775
10. Gonzales NA, Deardorff J, Formoso D, Barr A, Barrera M (2006) Family mediators of the relation
between acculturation and adolescent mental health. Fam Relat 55:318330. doi:10.1111/j.1741-3729.
2006.00405.x
11. Smokowski PR, Bacallao ML (2006) Acculturation and aggression in Latino adolescents: a structural
model focusing on cultural risk factors and assets. J Abnorm Child Psychol 34:657671. doi:10.1007/
s10802-006-9049-4
123
607
12. Vega WA, Gil AG, Warheit G, Zimmerman R, Apospori E (1993) Acculturation and delinquent
behavior among Cuban American adolescents: toward an empirical model. Am J Community Psychol
21:113125. doi:10.1007/BF00938210
13. Vega WA, Khoury E, Zimmerman R, Gil AG, Warheit G (1995) Cultural conflicts and problem
behaviors of Latino adolescents in home and school environments. J Community Psychol 23:167179.
doi:10.1002/1520-6629(199504)23:2\167::AID-JCOP2290230207[3.0.CO;2-O
14. Bird H, Canino G, Davies M, Duarte C, Febo V, Ramirez R et al (2006) A study of disruptive behavior
disorders in Puerto Rican youth: I. Background, design, and survey methods. J Am Acad Child Adolesc
Psychiatry 45:10321041. doi:10.1097/01.chi.0000227879.65651.cf
15. Bird HR, Davies M, Duarte CS (2006) A study of disruptive behavior disorders in Puerto Rican youth:
II. Baseline prevalence, comorbidity, and correlates in two sites. J Am Acad Child Adolesc Psychiatry
45:10421053. doi:10.1097/01.chi.0000227878.58027.3d
16. Smokowski PR, David-Ferdon C, Stroupe N (2009) Acculturation, youth violence, and suicidal
behavior in minority adolescents: a review of the empirical literature. J Prim Prev (forthcoming)
17. Brook JS, Whiteman M, Balka EB, Win T, Gursen MD (1998) African American and Puerto Rican drug
use: a longitudinal study. J Am Acad Child Adolesc Psychiatry 36:12601268
18. Bui H, Thongniramol O (2005) Immigration and self-reported delinquency: the interplay of immigration, generations, gender, race, and ethnicity. J Crim Justice 28:7180
19. Buriel R, Calzada S, Vasquez R (1982) The relationship of traditional Mexican American culture to
adjustment and delinquency among three generations of Mexican American male adolescents. Hisp J
Behav Sci 4:4155. doi:10.1177/07399863820041003
20. Samaniego RY, Gonzales NA (1999) Multiple mediators of the effects of acculturation status on
delinquency for Mexican American adolescents. Am J Community Psychol 27:189210. doi:10.1023/
A:1022883601126
21. Sommers I, Fagan J, Baskin D (1993) Sociocultural influences on the explanation of delinquency for
Puerto Rican youths. Hisp J Behav Sci 15:3662. doi:10.1177/07399863930151002
22. Carvajal SC, Hanson CE, Romero AJ, Coyle KK (2002) Behavioural risk factors and protective factors
in adolescents: a comparison of Latino and non-Latino whites. Ethn Health 7:181193. doi:10.1080/
1355785022000042015
23. Caetano R, Schafer J, Cunradi CB (2000) Intimate partner violence, acculturation and alcohol consumption among Hispanic couples in the United States. J Interpers Violence 15:3045. doi:10.1177/
088626000015001003
24. Markides K, Krause N, Mendes De Leon CF (1988) Acculturation and alcohol consumption among
Mexican Americans: a three-generation study. Am J Public Health 78:11781181. doi:10.2105/AJPH.
78.9.1178
25. Rogler LH, Cortes RS, Malgady RG (1991) Acculturation and mental health status among Hispanics.
Am Psychol 46:585597. doi:10.1037/0003-066X.46.6.585
26. Gil A, Wagner E, Vega W (2000) Acculturation, familism, and alcohol use among Latino adolescent
males: longitudinal relations. J Community Psychol 28:443458. doi:10.1002/1520-6629(200007)28:
4\443::AID-JCOP6[3.0.CO;2-A
27. Gil AG, Vega WA, Dimas JM (1994) Acculturative stress and personal adjustment among Hispanic
adolescent boys. J Community Psychol 22:4354. doi:10.1002/1520-6629(199401)22:1\43::AID-JCOP
2290220106[3.0.CO;2-T
28. Vega WA, Gil A, Warheit G, Apospori E, Zimmerman R (1993) The relationship of drug use to suicide
ideation and attempts among African American, Hispanic, and white non-Hispanic male adolescents.
Suicide Life Threat Behav 23:110119
29. Decker MR, Raj A, Silverman JG (2007) Sexual violence against adolescent girls: influences of
immigration and acculturation. Violence Against Women 13:498513. doi:10.1177/1077801207300654
30. Silverman JG, Decker MR, Raj A (2007) Immigration-based disparities in adolescent girls vulnerability to dating violence. Matern Child Health J 11:3743. doi:10.1007/s10995-006-0130-y
31. Hovey JD (1998) Acculturative stress, depression, and suicidal ideation among MexicanAmerican
adolescents: implications for the development of suicide prevention programs in schools. Psychol Rep
83:249250. doi:10.2466/PR0.83.5.249-250
32. Hovey JD, King CA (1996) Acculturative stress, depression, and suicidal ideation among immigrant and
second-generation Latino adolescents. J Am Acad Child Adolesc Psychiatry 35:11831192. doi:
10.1097/00004583-199609000-00016
33. Brown B, Benedict WR (2004) Bullets, blades, and being afraid in Hispanic high schools: an exploratory study of the presence of weapons and fear of weapons-related victimization among high school
students in a border town. Crime Delinq 50:372394. doi:10.1177/0011128703254916
123
608
34. Olvera RL (2001) Suicidal ideation in Hispanic and mixed ancestry adolescents. Suicide Life Threat
Behav 31:416427. doi:10.1521/suli.31.4.416.22049
35. Rasmussen K, Negy C, Carlson R, Burns J (1997) Suicide ideation and acculturation among low
socioeconomic status Mexican American adolescents. J Early Adolesc 17:390407. doi:10.1177/027243
1697017004003
36. Sanderson M, Coker AL, Roberts RE, Tortolero SR, Reininger BM (2004) Acculturation, ethnic
identity, and dating violence among Latino ninth-grade students. Prev Med 39:373383. doi:
10.1016/j.ypmed.2004.01.034
37. Swanson JW, Linksey AO, Quintero-Salinas R, Pumariega A, Holzer CE (1992) A binational school
survey of depressive symptoms, drug use, and suicidal ideation. J Am Acad Child Adolesc Psychiatry
31:669678. doi:10.1097/00004583-199207000-00014
38. Ng B (1996) Characteristics of 61 Mexican American adolescents who attempted suicide. Hisp J Behav
Sci 18:312. doi:10.1177/07399863960181001
39. Yu SM, Huang ZJ, Schwalberg RH, Overpeck M, Kogan MD (2003) Acculturation and the health and
well-being of US immigrant adolescents. J Adolesc Health 33:479488. doi:10.1016/S1054-139X(03)
00210-6
40. Smokowski PR, Rose R, Bacallao, ML (2008) Acculturation and Latino family processes: how parentadolescent acculturation gaps influence family dynamics. Fam Relat 57(3):295308
41. Guo S, Hussey D (1999) Analyzing longitudinal rating data: a three-level hierarchical linear model. Soc
Work Res 23:258269
42. Raudenbush SW, Bryk AS (2002) Hierarchical linear models: applications and data analysis methods,
2nd edn. Sage, Newbury Park
43. Snijders TAB, Bosker RJ (1999) Multilevel analysis: an introduction to basic and advanced multilevel
modeling. Sage, London
44. Szapocznik J, Kurtines W, Fernandez T (1980) Biculturalism involvement and adjustment in Hispanic
American youths. Int J Intercult Relat 4:353365. doi:10.1016/0147-1767(80)90010-3
45. Robin AL, Foster SL (1989) Negotiating parent-adolescent conflict: a behavioral-family systems
approach. Guilford, New York
46. Vega WA, Alderete E, Kolody B, Aguilar-Gaxiola S (1998) Illicit drug use among Mexicans and
Mexican Americans in California: the effects of gender and acculturation. Addiction 93:18391850.
doi:10.1046/j.1360-0443.1998.931218399.x
47. Achenbach TM, Rescorla LA (2001) Manual for ASEBA school-age forms and profiles. University of
Vermont, Research Center for Children, Youth & Families, Burlington
48. Schafer JL (1997) Analysis of incomplete multivariate data. Chapman Hall/CRC, Boca Raton
49. Graham JW, Olchowski AE, Gilreath TD (2007) How many imputations are really needed? Some
practical clarifications of multiple imputation theory. Prev Sci 8:206213. doi:10.1007/s11121007-0070-9
50. Von Hippel PT (2007) Regression with missing Ys: an improved strategy for analyzing multiply
imputed data. Sociol Methodol 37:83117. doi:10.1111/j.1467-9531.2007.00180.x
51. Bacallao ML, Smokowski PR (2005) Entre dos mundos (between two worlds) bicultural skills training
and Latino immigrant families. J Prim Prev 26:485509. doi:10.1007/s10935-005-0008-6
52. Gonzales NA, Knight GP, Morgan-Lopez A, Saenz D, Sirolli A (2002) Acculturation and the mental
health of Latino youths: an integration and critique of the literature. In: Contreras JM, Kerns KA, NealBarnett AM (eds) Latino children and families in the United States. Greenwood, Westport
53. LaFromboise T, Coleman HL, Gerton J (1993) Psychological impact of biculturalism: evidence and
theory. Psychol Bull 114:395412. doi:10.1037/0033-2909.114.3.395
54. Coatsworth JD, Maldonido-Molina M, Pantin H, Szapocznik J (2005) A person-centered and ecological
investigation of acculturation strategies in Hispanic immigrant youth. J Community Psychol 33:157174.
doi:10.1002/jcop.20046
55. Bacallao ML, Smokowski PR (2009) Obstacles to getting ahead: how assimilation mechanisms impact
Mexican immigrant families. J Health Soc Policy (forthcoming)
56. Smokowski PR, Bacallao ML (2007) Acculturation, internalizing mental health symptoms, and selfesteem: cultural experiences of Latino adolescents in North Carolina. Child Psychiatry Hum Dev
37:273292. doi:10.1007/s10578-006-0035-4
57. Coatsworth JD, Pantin H, Szapocznik J (2002) Familias unidas: a family-centered ecodevelopmental
intervention to reduce risk for problem behavior among Hispanic adolescents. Clin Child Fam Psychol
Rev 5:113132. doi:10.1023/A:1015420503275
58. Szapocznik J, Santisteban D, Kurtines W, Perez-Vidal A, Hervis O (1986) Bicultural effectiveness
training (BET): an experimental test of an intervention modality for families experiencing intergenerational/intercultural conflict. Hisp J Behav Sci 8:303330. doi:10.1177/07399863860084001
123