You are on page 1of 14

See discussions, stats, and author profiles for this publication at: https://www.researchgate.

net/publication/264972250

Behavioral and Emotional Rating Scale–2 Parent Report: Exploring a Spanish


Version With At-Risk Students

Article  in  Behavorial Disorders · August 2009


DOI: 10.1177/019874290903500104

CITATIONS READS

4 2,009

4 authors:

Jill D. Sharkey Sukkyung You


University of California, Santa Barbara Hankuk University of Foreign Studies
105 PUBLICATIONS   1,824 CITATIONS    94 PUBLICATIONS   2,205 CITATIONS   

SEE PROFILE SEE PROFILE

Gale M. Morrison Amy-Jane Griffiths


University of California, Santa Barbara Chapman University
67 PUBLICATIONS   2,144 CITATIONS    43 PUBLICATIONS   180 CITATIONS   

SEE PROFILE SEE PROFILE

Some of the authors of this publication are also working on these related projects:

“I Don’t Wanna Live No More”: Differences in Prevalence Rates of Chronic Hopelessness, Suicidal Ideation, and Suicide Attempt among Adolescents by Gang
Membership and Latina/o Ethnicity View project

Adolescents' Resilience to Cyberbullying Victimization: the Role of School Climate View project

All content following this page was uploaded by Jill D. Sharkey on 23 August 2014.

The user has requested enhancement of the downloaded file.


Behavioral and Emotional Rating Scale–2 Parent
Report: Exploring a Spanish Version
With At-Risk Students

Jill Sharkey, Sukkyung You, Gale Morrison, and Amy Griffiths


University of California, Santa Barbara

ABSTRACT: Youth strengths are as important to consider as deficits in understanding developmental


outcomes and thus are an important component of psychoeducational assessment. Latino/as are
understudied with regard to strength-based constructs, although cultural and socioeconomic factors
may be related to differences in Latino/a parents’ views of their children’s strengths and school
experiences. The Behavioral and Emotional Rating Scale–2 (BERS-2) Parent Report is one of the most
widely used measures to examine school-based strengths of students. With two sets of data collected
with Spanish-speaking Latino/a parents of students aged 9 to 14 years, the authors examined the
BERS-2 to explore and then preliminarily confirm latent constructs present in reports of their
children’s strengths. A three-factor model, labeled Self-Control, School Participation, and Emotional
Health, emerged as an alternative fit to the data. Results indicate that the BERS-2 measures culturally
sensitive constructs, yet amended procedures may enhance the assessment of student strengths from
the parent perspective, particularly when working with Spanish-speaking Latino/a families.

& The ability to identify strengths is a crucial System of Empirically Based Assessment across
component of psychoeducational assessment. more than 70 cultures and found somewhat
Longitudinal research has found that youth different relevance of constructs among groups
strengths are as important to consider as their but relatively consistent findings across cul-
deficits for understanding developmental tra- tures. They also found that scores varied more
jectories (Garmezy, 1993). Strengths help within than between groups. In a summary of
students and their families avoid negative their findings, the researchers concluded that
outcomes and achieve healthy development. understanding the differential functioning of
Moreover, understanding strengths allows for various assessments within and across cultural
more accurate and effective intervention and groups is important (a) to diversify normative
treatment plans (Jimerson, Sharkey, Nyborg, & data, (b) to compare recent immigrants to
Furlong, 2004). Thus, strength-based assess- nonimmigrant functioning from the same
ment has the potential to identify important country of origin, and (c) to further enhance
individual and environmental factors to con- the accuracy of diagnostic criteria for mental
sider when understanding students’ treatment disorders. Validated through multicultural
needs and may also promote intervention norming, strength-based assessment has the
success through relationship building. potential to inform more accurate diagnosis
In response to the now well-established and treatment of multicultural youth with
link between strengths and outcome, psychol- behavioral disorders.
ogists have begun to incorporate a strengths Although very little multicultural research
approach with the traditional deficit-focused has been applied to strength-based assessment,
model to promote understanding the internal there is debate among scholars regarding the
and external factors that influence a child’s consistency of positive psychology traits across
abilities and disabilities. With strength-based cultures. Peterson and Seligman (2004) have
assessment in its infancy, it is unknown conducted the most extensive multicultural
whether strength-based concepts apply with research in positive psychology to date. They
diverse populations. Precedent in deficit-based examined character strengths across numerous
assessment indicates the multicultural robust- cultural groups and found 24 individual virtues
ness of assessments such as the Achenbach present in all cultures, which led them to
scales (Rescorla et al., 2007). Achenbach and conclude that positive traits transcend culture.
colleagues (2008) have studied the Achenbach Yet other scholars argue that a cultural context

Behavioral Disorders, 35 (1), 53–65 November 2009 / 53


is always required to understand the role of was designed to be completed by any adult
strengths for diverse people, and the same (e.g., parent, teacher, school psychologist)
strengths are not equally relevant across with knowledge about a student. After devel-
cultures (Snyder & Lopez, 2007). A reasonable opment procedures were completed, the BERS
hypothesis is that a variety of virtues or positive was published and implemented in numerous
traits are likely to be strengths across cultures, projects across the country to help develop
but the relative value or use of these traits in treatment plans based on strengths.
diverse cultures and their relation to success In the BERS-2 revision, Epstein (2004)
may vary within and between cultures (Ped- made subtle changes to the original BERS to
rotti & Edwards, 2010). The only way to create teacher, parent, and self-report versions.
understand the influence of various positive For the parent report, the original 52-item
traits across cultures is to measure them with BERS was maintained, and five career strength
instruments that have multicultural validity. items were added. The test manual does not
mention if alternative trial items were tested,
The Behavioral and Emotional Rating and it appears that the original 52 items were
Scale–Second Edition retained for separate parent, teacher, and child
versions without scrutiny. That the same items
The Behavioral and Emotional Rating would load on the same five factors for
Scale–Second Edition (BERS-2) Parent Rating parents, teachers, and youths is unlikely given
Scale (PRS) is one of the most widely used developmental and cognitive differences
measures to examine school-based strengths of among groups (Furlong et al., 2007). None-
students (Furlong, Sharkey, Boman, & Cald- theless, new normative samples were collected
well, 2007). The BERS-2 was developed for the with 927 parents of students from 34 states
Children’s Mental Health Services System of with 14% of Hispanic ethnic background,
Care Project (Center for Mental Health Servic- which was nationally representative in com-
es, 2001), which involves wraparound services parison with national census data. The BERS-2
for children with emotional and behavioral was subsequently published and implemented
disorders, and was subsequently implemented nationwide.
for the national evaluation of the Comprehen- Although the BERS-2 is a widespread
sive Community Mental Health Services for measure that includes research-based items
Children and Families Program (Center for and some psychometric examination, review-
Mental Health Services, 2001). The BERS-2 ers of the original BERS have raised concerns
assesses multiple types of strengths (e.g., social that have yet to be addressed (Furlong et al.,
skills, emotional control) in different domains 2007). For example, Doll (2001) noted that the
(i.e., school and family), and its factor structure underlying factor analysis needs to be verified
has been validated for typical as well as at-risk and a conceptual model should be offered.
students. The measure has been used most However, to validate the parent and youth
frequently to assess the strengths of students versions, authors (Buckley, Ryser, Reid, &
with significant emotional and behavioral Epstein, 2006) conducted a confirmatory factor
concerns. analysis with the five-factor BERS-2 and tested
Epstein and Sharma (1998) created the only the original five-factor solution. They
original BERS using empirical procedures concluded that the scale was an acceptable
along with professional judgment. According fit with samples of parents and youths despite
to the test manual, they created a list of 1200 unacceptable root mean square error of
items based on literature in areas of develop- approximation (RMSEA) values of .148 (parent)
mental psychology, resilience, protective fac- and .120 (youth). Thus, the original five-factor
tors, and strength-based assessment. After structure of the BERS-2, parent version, was
consultation with experts and preliminary not substantiated, rendering its use for clinical,
analyses, authors winnowed items down to research, and evaluation purposes question-
ones that distinguished between children with able. The authors appeared to base their
and without emotional disturbance. Explorato- conclusion of a replicated factor structure on
ry factor analysis completed scale develop- levels of alternative measurement indices (i.e.,
ment with 52 items in five domains (i.e., comparative fit index [CFI], Tucker-Lewis Index
Interpersonal Strength, Family Involvement, [TLI], Bentler-Bonett normed fit index [NFI]);
Intrapersonal Strength, School Functioning, however, the RMSEA was the only index
and Affective Strength). The original BERS reported that takes into consideration sample

54 / November 2009 Behavioral Disorders, 35 (1), 53–65


size and model complexity. Given that to date, a sharp delineation between the role of the
no independent assessment of reliability and school and their role as parents: The role of
validity has been undertaken with the BERS-2 parents is to provide nurturance and to teach
parent report, studies need to further explore the morals, respect, and good behavior, whereas
factor structure of the BERS-2 with various the role of the school is to instill knowledge in
parent groups that represent diverse popula- students (Tinkler, 2002). There is clearly a
tions. Rigorous psychometric evaluation by need for educators to develop a stronger
external evaluators is important to promoting understanding of the cultural values that
the most efficient, reliable, and valid measures children bring with them from their home
for assessing and monitoring student strengths. environment to school (Tinkler, 2002) and the
In the case of the BERS-2, the corpus of implications of those cultural differences on
information needed to validate its use with parental understanding of their children’s
parents, as described in the test manual, is still educational experience.
developing.
Purpose
The Importance of Understanding the
Latino/a1 Parent Perspective In this study, we explore the factor structure
of the BERS-2 with Spanish-speaking parents of
Latino/as’ origins are rooted in South and at-risk youths. Because it is a measure used to
Central American countries including Mexico, assess strength characteristics for children with
Cuba, Puerto Rica, El Salvador, Guatemala, significant mental health concerns including
Colombia, and Costa Rica (Miville, 2010). attention deficit/hyperactivity disorder, opposi-
Although they are the largest and most rapidly tional defiant disorder, and mood disorders, it is
growing ethnic minority group in the United crucial to identify the psychometric properties
States, Latino/as are understudied with regard to for this population. Thus, as was done in the
psychological processes in general and strength- original BERS-2 norming process, we purpose-
based constructs in particular, which has led to fully selected samples of youths with emotional
the perception of cultural exceptions as weak- and behavioral concerns.
nesses rather than potential strengths (Pedrotti &
Edwards, 2010). Given their population growth Research Questions and Hypotheses
(U.S. Census Bureau, 2001), overrepresentation
in families with incomes below the poverty level This study explores a Spanish version of
(Smokowski, Reynolds, & Bezruczko, 1999), the BERS-2 PRS. First, we examined the ques-
higher frequency in families at risk for poor tion, What are latent constructs present in
behavioral and mental health outcomes (Vega et Latino/a parents’ reports regarding their chil-
al., 1998), and relatively high risk for school dren’s strengths? Second, we assessed the
dropout and academic failure (Tinkler, 2002), it hypothesis that the sample-specific factor struc-
is important to understand the perspectives of ture would be a better fit than the original norm-
Latino/a parents about their children’s strengths. based factor structure to a second set of data.
Although Latino/as are not a homogenous
group, there are general characteristics that Methods
differentiate the Latino/a identity from Europe-
an Americans values, particularly for recent Participants
immigrants who are more likely to be less
acculturated to majority group values. Latino/a Sample 1. Participants were parents of
parents distinguish themselves from parents of students in fourth through seventh grade from
other backgrounds by expressing concern low-socioeconomic neighborhoods in two
about showing respect to authority figures, school districts in the Central Coast region of
and they may not become involved in educa- California. Participating schools enrolled 94%
tional matters for fear of showing disrespect to to 99% of students who were classified as
teachers (Drummond & Stipek, 2004). Latino/a socioeconomically disadvantaged and 66% to
parents may often feel intimidated by teachers 72% of students who were classified as
and the school system in general (Chrispeels & English-language learners. Parents who identi-
Rivero, 2001) and may feel anxious, unwel- fied as Latino/a and responded in Spanish,
come, and misinformed when entering the representing 52% of the overall sample, were
school environment. Latino/a parents often see selected for analysis. Data were collected at

Behavioral Disorders, 35 (1), 53–65 November 2009 / 55


two time points. At the first data collection performs competently in classroom tasks and
period, parents represented 97 students (69% in school in general (e.g., ‘‘Reads at or above
male) aged 9 to 14 years (M 5 10.87 years). At grade level’’). The Intrapersonal Strength sub-
the second data collection period, the parent scale consists of 11 items to assess the child’s
group represented 112 students (76.8% male) perceptions of his or her accomplishments and
aged 10 to 15 years (M 5 12.09 years). competence in various areas (e.g., ‘‘Identifies
Sample 2. Participants were parents of personal strengths’’). The Family Involvement
youths enrolled in a community program that subscale has 10 items to explore how often the
provided comprehensive services to criminally child participates in family activities and his or
involved families with identified substance her relationship with the family unit (e.g.,
abuse problems. Parents who identified as ‘‘Participates in family activities’’). Finally, the
Latino/a and responded in Spanish were Interpersonal Strength subscale includes 15
selected for analysis, representing 48% of the items and assesses the child’s ability to control
overall sample. Data were collected at two emotions and behaviors in social situations
time points. At the first data collection period, (e.g., ‘‘Is kind towards others’’). Mooney,
participants represented 83 students (58% Epstein, Ryser, and Pierce (2005) conducted
male) aged 13 to 17 years (M 5 15.21 years). a series of three studies to test the reliability
At the second data collection period, parents and validity of the BERS-2 PRS. With 78
represent 73 students (62% male) aged 14 to parents of elementary and middle school youth
17 years (M 5 15.18 years). in Nebraska (91% white), they found test-retest
Although immigration history was not correlations at or above .80 for all subscales
obtained from these samples, a 2005 demo- and the strength index. With 85 and 55 parents
graphic survey in the area revealed that of of elementary and middle school youth in
individuals who were classified as Hispanic or Nebraska (93% and 100% white, respectively),
the authors found 90% significant correlations
Latino/a, 92% identified as Mexican, 0.5% as
between the BERS-2 PRS and the Social Skills
Puerto Rican, 0.3% as Cuban, and 7% as
Rating System and the Child Behavior Check-
Other Hispanic or Latino/a. Thus, the majority
list, respectively. Mooney and colleagues
of Spanish-speaking parents of children in both
(2005) noted the need for additional studies
studies were likely recent immigrants from
using larger and more diverse samples to
Mexico, with fewer from other Central and
establish other types of reliability and validity.
South American countries.
Procedures
Measures
Translation procedures were implemented
Behavioral and Emotional Rating Scale–2 to create a Spanish version of the BERS-2 PRS.
Parent Rating Scale. The BERS-2 PRS, exclud- First, the BERS-2 PRS was translated into Spanish
ing the career strengths subscale, was used to by a bilingual, bicultural Latino/a American
measure youth strengths from a parent per- graduate student trained in social-emotional
spective. The BERS-2 PRS consists of 52 items assessment and the purpose of the BERS-2. This
completed by parents to assess behavioral and preliminary translation was subsequently back-
emotional strengths of children between the translated into English by a second trained
ages of 5 and 18 years within a school setting. bilingual, bicultural Latino/a American graduate
Based on factor analyses studies with the student who provided additional feedback
original BERS, the BERS-2 consists of five regarding how well items preserved intent rather
subscales: Affective Strength, School Function- than merely a literal translation of the measure.
ing, Intrapersonal Strength, Family Involve- The measure was piloted with Spanish-speaking
ment, and Interpersonal Strength (Epstein, families, and wording was adjusted based on
2004). The items are rated using a Likert-type feedback. The Spanish version was subsequently
scale ranging from 0 to 3 (i.e., 0 5 not at all requested and obtained by the original BERS-2
like, 1 5 not much like, 2 5 like, 3 5 very authors (M. Furlong, personal communication,
much like). The Affective Strength subscale May 22, 2008). Although the BERS-2 Parent
includes 7 items that reflect the students’ Spanish has not been published, it has been used
ability to give and receive affection from in evaluation projects, further necessitating its
others (e.g., ‘‘Shows concerns for the feelings empirical examination. This study represents a
of others’’). The School Functioning subscale first step in establishing reliability and validity of
includes 9 items and assesses if the child this Spanish version of the BERS-2.

56 / November 2009 Behavioral Disorders, 35 (1), 53–65


Participants in the first sample were cutoff or minimum factor loading, can be used
recruited for a research study because they to identify the ideal number of factors.
were referred for discipline to school princi- However, an overreliance on the statistics of
pals or had been suspended or expelled from a particular EFA can lead to results that fit for
school. Parents in the second sample were one sample but are neither theoretically sound
recruited for an intervention project through nor true for the population estimated. Thus, we
truancy, juvenile probation, or adult parole considered empirical data within a theoretical
systems. Upon completion of recruitment framework that explains item groupings.
procedures including informed consent, the Confirmatory factor analysis. Confirmatory
BERS-2 PRS was administered to parents in factor analysis allows researchers to test a
both studies by a graduate student researcher priori models by plotting the proposed factor
during a one-on-one, face-to-face visit as part structure with measured variables loading onto
of a larger set of study procedures. Parents proposed, or latent, variables (Kline, 1998).
were given the choice to complete the The fit of the proposed model was evaluated
measure in English or Spanish, and those based on CFI (Bentler, 1990), nonnormed fit
who completed the measure in Spanish were index (NNFI; also known as the Tucker Lewis
retained in this study. Directions for comple- Index; Bentler & Bonett, 1980) and RMSEA
tion were explained verbally in the parents’ (Steiger & Lind, 1980). The CFI provides a
primary language, and any questions were measure of fit, which assesses the improve-
answered prior to or during the survey. Parents ment in fit of the hypothesized model relative
were asked to complete a second set of to a null model. Although it is generally
measures, including the BERS-2, 1 to 2 years accepted that a CFI value equal to or greater
(first sample) and 6 months (second sample) than .90 represents a well-fitting model
after their initial response. The first set of (McDonald & Ho, 2002), a revised cutoff
responses was used to conduct the first stage of value close to .95 has been recommended
analysis, and the second set of responses was (Hu & Bentler, 1999). The NNFI and RMSEA
used to conduct the second stage of analysis. were also included. We chose these indices
Because this was a secondary data analysis, because they are relatively independent of
combining two samples allowed us to achieve sample size and take into account model
the sample size necessary for the required complexity, which is an important property
analysis. Using the second set of data acquired for comparing several alternative models with
from the same parents allowed us to confirm different degrees of complexity. Values of .95
the exploratory factor structure on a unique, or above for NNFI (Hu & Bentler, 1999) and
although not independent, set of data. RMSEA values of about .08 indicate fair fit,
whereas .05 or less indicates good fit (Browne
& Cudek, 1993).
Overview of the Statistical Analyses

Analyses were conducted in two stages of


Results
factor analyses. First, exploratory factor anal-
ysis (EFA) was employed to explore the Stage 1: EFA
underlying structure of a relatively large set
of variables included in the BERS-2 Parent Before conducting EFA, a content analysis
Report. After conducting EFA, confirmatory of highly correlated items was carried out to
factor analysis (CFA) was conducted to test the avoid multicollinearity. When intercorrela-
fit of the proposed factor structure to a second tions among some variables are high (..85),
sample. Because the BERS items are ordinal one should be dropped as redundant. Examin-
variables that violate the assumption of multi- ing the correlation matrix, the absolute values
variate normality, analysis was based on a of correlations were not too high (ranged from
robust weighted least squares (WLS) estimation .28 to .81); thus, all items were retained. A
using Mplus 4.0 (Muthen & Muthen, 2006). robust WLS estimation was used with a
Exploratory factor analysis. Exploratory promax rotation with study sample 1 (N 5
factor analysis is a method of examining how 180). Oblique rotation was used given the
scale items may optimally group together into correlation between BERS dimensions. Empir-
distinct subsets to measure an overall construct ical approaches, such as a scree test and
in the most parsimonious fashion. Empirical pattern of factor loading, were considered
techniques, such as creating an eigenvalue within a theoretical framework considering

Behavioral Disorders, 35 (1), 53–65 November 2009 / 57


TABLE 1
Rotated Pattern Matrix From Exploratory Factor Analysis: Study 1 Sample
BERS-2 Item I II III
B1 Family belongingness 0.673 0.065 0.234
B2 Trusts a significant person 0.630 0.147 0.206
B3 Accepts a hug 0.714 0.104 0.059
B4 Participates in community activities 0.600 0.267 20.137
B5 Self-confidenta 0.386 0.142 0.407
B6 Acknowledges painful feelings 0.636 0.016 0.262
B7 Maintains positive family relationshipsa 0.562 20.016 0.476
B8 Demonstrates a sense of humor 0.746 0.064 0.090
B9 Asks for helpa 0.577 0.012 0.323
B10 Uses anger management skills 20.081 0.149 0.721
B11 Communicates with parents about behavior 0.253 0.173 0.520
a
B12 Remorseful of behavior that hurt others 0.340 20.002 0.538
B13 Show concern for feelings of othersa 0.672 20.108 0.419
a
B14 Completes a task on first request 20.067 0.367 0.481
B15 Interacts positively with parents 0.194 20.002 0.795

B16 Reacts to disappointment in a calm manner 0.093 0.009 0.766


B17 Considers consequences of own behavior 0.205 0.151 0.571
B18 Accepts criticism 0.453 0.138 0.144
B19 Participates in church activities 0.528 0.284 20.049
B20 Demonstrates age-appropriate hygiene skills 0.646 0.193 0.053
B21 Requests support from peers and friends 0.605 0.216 0.017
B22 Enjoys a hobby 0.575 0.268 0.032
B23 Discuss problems with others 0.555 0.006 0.140
B24 Completes school tasks on time 20.049 0.853 0.153
B25 Accepts the closeness and intimacy of others 0.784 20.042 20.039
B26 Identifies own feelings 0.611 0.118 0.231
B27 Identifies personal strengths 0.607 0.174 0.177
B28 Accepts responsibility for own actions 0.177 0.114 0.603
B29 Interacts positively with siblingsa 0.302 0.090 0.486
B30 Loses a game gracefully 0.145 20.042 0.637
B31 Completes homework regularly 0.115 0.819 0.078
B32 Is popular with peers 0.721 0.017 0.032
B33 Listens to others 0.722 20.095 0.243
B34 Express concern for others 0.827 20.102 0.173
B35 Admits mistakesa 0.324 20.065 0.674
B36 Participates in family activitiesa 0.529 0.138 0.336
B37 Accepts ‘‘no’’ for an answer 0.072 0.294 0.517
B38 Smiles often 0.845 0.058 0.008
B39 Pays attention in class 0.053 0.829 0.024
B40 Computes math problems at above grade level 0.080 0.639 0.125
B41 Reads at or above grade level 0.131 0.569 0.094
B42 Enthusiastic about lifea 0.625 0.054 0.311

58 / November 2009 Behavioral Disorders, 35 (1), 53–65


TABLE 1 (Continued )
Rotated Pattern Matrix From Exploratory Factor Analysis: Study 1 Sample
BERS-2 Item I II III
a
B43 Respects the right of others 0.304 0.182 0.459
a
B44 Shares with others 0.622 20.011 0.343
B45 Complies with rules at home 20.001 0.285 0.632
B46 Apologizes to others when wrong 0.175 0.182 0.529
B47 Studies for tests 0.029 0.667 0.259
B48 Talks about the positive aspects of life 0.494 0.228 0.237
B49 Is kind toward others 0.753 0.011 0.208
B50 Uses appropriate languagea 0.480 0.091 0.361
B51 Attends school regularlya 0.529 0.430 0.036
B52 Use note-taking and listening skills in school 0.183 0.766 0.038

Note. Numbers in bold type signify items that load on that specific factor.
aDouble-loaded item.

extant literature to confirm that the final factor EFA three-factor model and original five-factor
selection was interpretable and substantively model using study sample 2 (N 5 185). We
plausible. Through this process, a three-factor also evaluated a one-factor model. This
model emerged as the most meaningful and examination was conducted on an exploratory
parsimonious model. The fit of the three-factor basis because EFA and CFA are not typically
solution in RMSEA and standardized root performed on the same participants. Although
mean square residual (SRMR) was acceptable. our data were from two distinct data collection
The value of RMSEA was .07, and the value of periods, the same participants were surveyed
SRMR was .04. Table 1 displays which items twice.
related to each of three obtained factors and The results of the three CFA models are
reports pattern coefficients. Fourteen double- shown in Table 3. Fit indices indicated that the
loaded items were eliminated to have a three-factor model proposed from EFA yielded
simpler structure; thus, 38 items of the 52 a slightly better fit than the other two models.
original items were finally selected for the Improvements were seen for all of the model fit
three-factor solution. Cronbach’s alphas and indices. The one-factor model was not accept-
intercorrelations for these three factors are able in terms of CFI, NNFI, and RMSEA. The
presented in Table 2. The intercorrelations NNFI values for three-factor and five-factor
ranged from .65 and .66, and they were not models were good (both greater than .95),
high enough to suggest that the scales sub- whereas the RMSEA value for the three-factor
stantially overlapped. model (.08) was acceptable but not for the
five-factor model (.09) based on accepted
criteria (Browne & Cudek, 1993). The CFI
Stage 2: CFA
value was not ideal yet was improved for the
The second stage sought to explore the three-factor model compared with the five-
factor structure of BERS based on the proposed
TABLE 3
Fit Indexes for the Confirmatory Factor
TABLE 2 Models
Descriptive and Intercorrelation Statistics of
the Three-Factor Model From Exploratory Group S-B x2 df CFI NNFI RMSEA
Factor Analysis: Study 1 Sample One factor 264.20 63 .81 .93 .13

M SD a 1 2 3 Three factor 153.54 65 .92 .97 .08

1. Emotional Health 2.79 0.84 .96 — Five factor 193.68 75 .89 .96 .09

2. School Participation 2.07 0.85 .92 .65 —


Note. CFI 5 comparative fit index; NNFI 5 nonnormed fit
3. Self-Control 2.61 0.87 .93 .66 .65 — index; RMSEA 5 robust root mean square error of
approximation.

Behavioral Disorders, 35 (1), 53–65 November 2009 / 59


TABLE 4
Descriptive and Intercorrelation Statistics of the Five-Factor Model From Confirmatory Factor
Analysis: Study 2 Sample
M SD a 1 2 3 4 5
1. Interpersonal Strength 2.23 0.60 .91 —
2. Family Involvement 2.25 0.57 .82 .94 —
3. Intrapersonal Strength 2.49 0.54 .89 .87 .85 —
4. School Functioning 2.09 0.68 .89 .80 .77 .70 —
5. Affective Strength 2.49 0.57 .80 .88 .90 .93 .65 —

factor model, and the other two indices (NNFI ture of a Spanish version of the BERS-2 to gain
and RMSEA) were acceptable. further insight into its utility as a culturally
The correlation patterns of the three-factor sensitive measure of youth strengths. Cultural
and five-factor BERS scales were also exam- and socioeconomic factors may influence
ined to check validity (see Tables 4 and 5). parents’ views of their children’s strengths
Intercorrelations among three factors were not and school experiences. Culture influences
too high (range 5 .68 to .81), which suggests the ways students behave, and for ethnic
discriminant validity of BERS scales. However, minority students, it may also contribute to a
intercorrelations among the five factors were mismatch between the behavioral expecta-
significantly higher (range 5 .65 to .94), which tions of their home and school environments
suggests the five-factor BERS scales are redun- (Aaroe & Nelson, 2000).
dant. Therefore, the three-factor model Results suggest that a three-factor BERS
emerged as the most meaningful and parsimo- model is a better fit to the data than the
nious model using both substantive and original five-factor BERS model given statisti-
statistical criteria. Standardized factor loadings cal, theoretical, and practical considerations.
for the three-factor model are provided in Using descriptive labels, Emotional Health is
Table 6. Results demonstrated that the stan- composed of 21 items (e.g., trusts a significant
dardized factor loadings of each construct are person), School Participation is composed of 7
substantively large (..58), suggesting that all items (e.g., completes school tasks on time),
factors are well determined with valid indica- and Self-Control is composed of 10 items (e.g.,
tors. reacts to disappointment in a calm manner). As
we will describe in additional detail, these
factors appear to be consistent with extant
Discussion
literature related to Latino/a identity generally
As one of the most widely used measures and for Mexican Americans specifically.
to examine school-based strengths of students Although within-group differences are
of children with emotional and behavioral substantial, common cultural characteristics
disorders involved with wraparound services, related to recent immigrants to the United
it is crucial to evaluate the reliability and States from Mexico include a unique mix of
validity of the BERS-2 as implemented in Native American and European cultures. For
practice. This study explores the factor struc- example, the importance of daughters’ virtue
to family honor originates in Spanish culture,
TABLE 5 whereas the prestigious role of the eldest son
Descriptives and Intercorrelation of the in the family originates in Mayan tradition
Three-Factor Model From Confirmatory (McNeill et al., 2001). Given their unique
Factor Analysis: Study 2 Sample cultural influences, there are certain cultural
norms that tend to describe traditional Mexi-
M SD a 1 2 3 can Americans and even extend to define an
1. Emotional Health 2.46 0.51 .91 — overall Latino/a culture. Familismo is one such
2. School Participation 2.05 0.75 .90 .68 — value that refers to extended family members’
3. Self-Control 2.13 0.65 .88 .81 .78 — contribution to emotional and financial sup-
port of everyone in the family and that the

60 / November 2009 Behavioral Disorders, 35 (1), 53–65


TABLE 6
Standardized Factor Loadings of Confirmatory Factor Analysis
BERS-2 Item BERS-2 Subscale I II III
Factor 1: Emotional Health
1. Family belongingness FAM .783
2. Trusts a significant person FAM .589
3. Accepts a hug AS .718
4. Participates in community activities FAM .793
6. Acknowledges painful feelings AS .786
8. Demonstrates a sense of humor INTRA .668
18. Accepts criticism INTER .583
19. Participates in church activities FAM .609
20. Demonstrates age-appropriate hygiene skills INTRA .717
21. Requests support from peers and friends INTRA .629
22. Enjoys a hobby INTRA .655
23. Discuss problems with others AS .583
25. Accepts the closeness and intimacy of others AS .610
26. Identifies own feelings INTRA .877
27. Identifies personal strengths INTRA .795
32. Is popular with peers INTRA .576
33. Listens to others INTER .735
34. Express concern for others AS .771
38. Smiles often INTRA .681
48. Talks about the positive aspects of life INTRA .808
49. Is kind toward others INTER .861
Factor 2: School Participation
24. Completes school tasks on time SCHL .917
31. Completes homework regularly SCHL .890
39. Pays attention in class SCHL .867
40. Computes math problems at above grade level SCHL .632
41. Reads at or above grade level SCHL .621
47. Studies for tests SCHL .817
52. Use note-taking and listening skills in school SCHL .913
Factor 3: Self-Control
10. Uses anger management skills INTER .687
11. Communicates with parents about behavior FAM .702
15. Interacts positively with parents FAM .777
16. Reacts to disappointment in a calm manner INTER .709
17. Considers consequences of own behavior INTER .836
28. Accepts responsibility for own actions INTER .744
30. Loses a game gracefully INTER .657
37. Accepts ‘‘no’’ for an answer INTER .611
45. Complies with rules at home FAM .785
46. Apologizes to others when wrong INTRA .727

Note. INTRA 5 Intrapersonal Strength; INTER 5 Interpersonal Strength; SCHL 5 School Functioning; FAM 5 Family
Involvement; AS 5 Affective Strength.

Behavioral Disorders, 35 (1), 53–65 November 2009 / 61


needs of the family transcend the needs of the cent immigrants, were more traditional, in-
individual (McNeill et al., 2001). Each family cluding the belief that school holds the
member is expected to cooperate and contrib- responsibility to educate children, parents
ute to the benefit of the family as a whole. should not question teachers’ methods, and
Because Mexican American cultural values children should obey educators (Rodriguez &
include strong family loyalty and allegiance, Olswang, 2003). Despite potentially different
which may differ from the behavioral styles of attitudes toward school, and potentially differ-
majority-culture youth who may strive for ent relations between the school factor and
more self-expression and individuality (Griggs educational outcomes, parents of different
& Dunn, 1996), Mexican American parents backgrounds appear to rate the school items
may be less likely to identify family strengths cohesively.
as distinct from personal strengths. Results
indicate that the original BERS Family Func- Limitations and Future Directions
tioning variables primarily converged with
Interpersonal Strength items, which is under- Sample sizes used for factor analyses were
standable because parents’ basis for under- relatively small given traditional guidelines, as
standing their child’s interpersonal functioning larger samples tend to provide more stable
is within the family system. results. However, a simulation study by
The Behavioral Self-Control factor is con- MacCallum, Widaman, Zhang, and Hong
sistent with Latino/a personal identity and (1999) demonstrated that a common rule of
expectations for behavior. Respeto dictates thumb regarding sample size in factor analysis
respectful behavior toward others due to age, is not valid. On the basis of simulation, their
gender, socioeconomic level, and authority findings showed that the level of communality
status (McNeill et al., 2001). Children are is more important than the role of sample size.
expected to obey their parents and demon- According to their guidelines, our study
strate appropriate behavior at school and sample, near 200 (study 1 sample 5 180 and
particularly toward authority figures such as study 2 sample 5 185), is not problematic
teachers. In Spanish terms, controlarse is a because the mean level of communality is
value of controlling oneself (Dana, 1993). consistently high (.7) and factors are well
Related items in this factor include ‘‘uses determined in our study. It is ideal to conduct
anger management skills,’’ ‘‘reacts to disap- EFA and a subsequent CFA with distinct
pointment in a calm manner,’’ and ‘‘considers samples of participants. Although our data
consequences of own behavior.’’ were from two distinct data collection periods,
The Emotional Health factor is also con- the same participants were surveyed twice.
sistent with Latino/a personal identity. A Conducting CFA with an independent sample
behavioral strength prioritized in Latino/a would strengthen our findings.
culture is represented by the concept of Within-group differences are often just as
simpatı́a, which refers to a priority for affec- meaningful as between-group differences. Ac-
tional social behaviors. Mexican Americans culturation level is likely to be related to
value maintaining pleasantries and avoiding perception of strengths, and the more parents
conflict in relationships. In research, having become acculturated to majority-group values,
close relationships with parents is related to the more likely their report of strengths is to
important positive outcomes for Mexican match parents of majority-group students. In
American youth (Love & Buriel, 2007). Items this study, we were not able to determine the
in the Emotional Health scale include ‘‘family level of acculturation, age of arrival in the
belongingness,’’ ‘‘accepts a hug,’’ and ‘‘ac- United States, or education level. However, it
cepts the closeness and intimacy of others’’ is likely that most participants were immigrants
and seem consistent with the simpatı́a identity. with low levels of acculturation given their
The School scale included school items preference for Spanish materials. Future stud-
from the original BERS and thus appears to ies should examine the relationship between
have a consistent latent structure across acculturation level and perception of strengths.
normative samples and the current sample. The primary goal of the BERS-2 is to
Research in cross-cultural differences between identify strengths of students with emotional
Mexican American and Anglo-American and behavioral difficulties who are referred for
mothering in the school context found that wraparound services. Thus, it is important to
Mexican American mothers, particularly re- explore the underlying factor structure of the

62 / November 2009 Behavioral Disorders, 35 (1), 53–65


BERS-2 with high-risk samples of youth. and question the relevance of results before
Although previous psychometric evaluation making recommendations.
of the BERS-2 suggests that the factor structure The BERS-2 has been published and used
is valid for typical students as well as students with at-risk families across the country as part
with emotional and behavioral disorders, to be of a national system of care evaluation and for
robust for use in schools with all students, psychological assessment. Results are com-
future research should continue to explore and monly used to inform intervention strategies.
confirm the factor structure of the BERS-2 with Given the widespread use of the BERS-2, it is
a typical school population. crucial for independent research to examine its
It is unknown whether a more parsimoni- psychometric properties. Results of this study
ous three-factor model is unique to the suggest that the BERS-2 may measure cultur-
Spanish-speaking participants assessed in this ally sensitive constructs. Thus, working with
study or to a broader tendency of parents families to design interventions that improve
reporting on their child’s strengths because the on these strengths is likely to be acceptable,
factor structure of the BERS-2 has not been although this hypothesis needs to be tested.
thoroughly explored with regard to its fit to Future studies should explore the factor
parent respondents. As a model of cross- structure of the BERS-2 with larger samples of
cultural research in the field of deficit-focused parent and youth groups that represent
assessment, the Child Behavior Checklist diverse populations in the United States.
(Achenbach, 1991) has been studied in nu- Ultimately, the most robust measure will be
merous societies across the world. This re- validated across cultures and in different
search has demonstrated that the factor struc- languages to encourage comparison between
ture of the eight U.S.-derived syndrome scales societies (Rescorla et al., 2007). Such re-
fit data from 30 other societies (Ivanova et al., search should include an examination of how
2007). Thus, it is possible that a consistent acculturation affects parent perspectives of
factor structure exists for strength-based scales strengths. By considering strengths from di-
such as the BERS-2, with separate norms to verse parent perspectives, it is more likely that
reflect group differences in strength identifica- parents of minority students will be more
tion. Further research is necessary to determine involved in school-based interventions. As a
if the factor structure identified in the current second step, it is important to relate assess-
study is unique to Spanish-speaking Latino/a ment results to treatment recommendations.
parents or consistent across parents of diverse Interventions designed to enhance strengths
cultures. should be developed and tested to determine
Given the substantial limitations to this if they are effective at promoting healthy
study, it is not meant to provide a comprehen- outcomes. Without these types of research
sive psychometric analysis of the BERS-2 efforts, strength-based assessment will contin-
Spanish version. Rather, it is presented as an ue to be advocated for and conducted but
exploration to generate further hypotheses without direction as how to proceed with
regarding strength constructs and multicultural intervention.
perspectives. Future studies should capitalize
on the limitations present in this study to
advance the study of strength-based assess- NOTE
ment. The BERS-2 is a promising measure that 1. The term Latino/a was selected to represent the
would benefit from additional reliability and participants in this study because all participants
validity analysis. self-identified as Latino/a on the demographic
portion of the questionnaire administered to them.
Although using a single label to refer to a large and
Conclusion diverse group does not accurately reflect the identity
of individual participants, we felt it was more
Using the BERS-2 with multicultural fam- accurate to use a general term to avoid overgener-
alization of the results to a specific ethnic group
ilies necessitates an understanding of the
(e.g., Mexican Americans) that was not represented
cultural factors that guide healthy functioning
by all participants. Latino/a is a more inclusive term
for a particular group. As every family is than Hispanic because it represents the diversity of
unique in their cultural values, acculturation, languages (e.g., Spanish, Portuguese, French), an-
and relationship with the school, a sensitive cestry (e.g., Europe, Africa, and the Americas), and
test administrator should be aware of potential values of immigrants to the United States from Latin
cultural effects on a youth’s school experience America (Miville, 2010). Readers should keep in

Behavioral Disorders, 35 (1), 53–65 November 2009 / 63


mind that results summarize group data that reflect Epstein, M. H. (2004). Behavioral and Emotional
general trends among participants. Rating Scale: A strengths-based approach to
assessment (2nd ed.). Austin, TX: PRO-ED.
Epstein, M. H., & Sharma, J. (1998). Behavioral and
REFERENCES Emotional Rating Scale: A strengths-based ap-
Aaroe, L., & Nelson, J. R. (2000). A comparative proach to assessment. Austin, TX: PRO-ED.
analysis of teachers’, Caucasian parents’, and Furlong, M. J., Sharkey, J. D., Boman, P., & Caldwell,
Hispanic parents’ views of problematic school R. (2007). Cross-validation of the Behavioral and
survival behaviors. Education and Treatment of Emotional Rating Scale-2 Youth Version: An
Children, 23, 314–324. exploration of strength-based latent traits. Jour-
Achenbach, T. M. (1991). Manual for the Child nal of Child and Family Studies, 16, 696–
Behavior Checklist/4-18 and 1991 profile. Bur- 711.
lington, VT: University of Vermont, Department Garmezy, N. (1993). Children in poverty: Resilience
of Psychiatry. despite risk. Psychiatry, 56, 127–136.
Achenbach, T. M., Becker, A., Döpfner, M., Heier- Griggs, S., & Dunn, R. (1996). Hispanic-American
vang, E., Roessner, V., & Steinhausen, H., et al. students and learning style. Internet site: http://
(2008). Multicultural assessment of child and www.eric.ed.gov/ERICDocs/data/ericdocs2/
adolescent psychopathology with ASEBA and content_storage_01/0000000b/80/2a/25/b8.pdf
SDQ instruments: Research findings, applica- (Accessed June 6, 2006).
tions, and future directions. Child Psychology Hu, L., & Bentler, P. M. (1999). Cutoff criteria for fit
and Psychiatry, 49, 251–275. indexes in covariance structure analysis: Con-
Bentler, P. M. (1990). Comparative fit indices in ventional criteria versus new alternatives. Struc-
structural models. Psychological Bulletin, 107, tural Equation Modeling: A Multidisciplinary
238–246. Journal, 6, 1–55.
Bentler, P. M., & Bonett, D. G. (1980). Significance Ivanova, M. Y., Achenbach, T. M., Dumenci, L.,
tests and goodness of fit in the analysis of Rescorla, L. A., Almqvist, F., & Bilenberg, N., et
covariance structures. Psychological Bulletin, al. (2007). Testing the 8-syndrome structure of
88, 588–606. the Child Behavior Checklist in 30 societies.
Browne, M. W., & Cudeck, R. (1993). Alternative Journal of Clinical Child and Adolescent Psy-
ways of assessing model fit. In K. A. Bollen & J. chology, 36, 405–417.
S. Long (Eds.), Testing structural equation Jimerson, S. R., Sharkey, J. D., Nyborg, V. M., &
models (pp. 136–162). Newbury Park, CA: Furlong, M. J. (2004). Strength-based assessment
Sage. and school psychology: A summary and synthe-
Buckley, J. A., Ryser, G., Reid, R., & Epstein, M. H. sis. California School Psychologist, 9, 9–19.
(2006). Confirmatory factor analysis of the Kline, R. B. (1998). Principles and practice of structural
Behavioral and Emotional Rating Scale–2 equation modeling. New York: Guilford.
(BERS-2) Parent and Youth Rating Scales. Love, J. A., & Buriel, R. (2007). Language brokering,
Journal of Child and Family Studies, 15, 27– autonomy, parent-child bonding, biculturalism,
37. and depression: A study of Mexican American
Center for Mental Health Services. (2001). Annual adolescents from immigrant families. Hispanic
report to Congress on the evaluation of the Journal of Behavioral Sciences, 29, 472–491.
Comprehensive Community Mental Health Ser- MacCallum, R. C., Widaman, K. F., Zhang, S., &
vices for Children and their Families Program, Hong, S. (1999). Sample size in factor analysis.
2001. Atlanta, GA: ORC Macro. Psychological methods, 4, 84–99.
Chrispeels, J. H., & Rivero, E. (2001). Engaging McDonald, R. P., & Ho, M.-H. (2002). Principles and
Latino/a families for student success: How practice in reporting structural equation analy-
parent education can reshape parents’ sense of ses. Psychological Methods, 7, 64–82.
place in the education of their children. Pea- McNeill, B. W., Prieto, L. R., Niemann, Y. F.,
body Journal of Education, 76, 119–169. Pizarro, M., Vera, E. M., & Gómez, S. P. (2001).
Dana, R. H. (1993). Multicultural assessment per- Current directions in Chicana/o psychology. The
spectives for professional psychology. Boston, Counseling Psychologist, 29, 5–17.
MA: Allyn & Bacon. Miville, M. (2010). Latina/o identity development:
Doll, B. (2001). Review of the Behavioral and Updates on theory, measurement, and counsel-
Emotional Rating Scale: A strength-based ap- ing implications. In L. A. Suzuki, M. Casas, J. G.
proach to assessment. In B. S. Plake, & J. C. Ponterotto, & C. M. Alexander (Eds.), Handbook
Impara (Eds.), The fourteenth mental measure- of multicultural counseling (3rd ed.). (pp. 241–
ments yearbook (pp. 142–144). Lincoln, NE: 252). Thousand Oaks, CA: SAGE Publications.
University of Nebraska–Lincoln. Mooney, P., Epstein, M. H., Ryser, G., & Pierce, C.
Drummond, K. V., & Stipek, D. (2004). Low-income D. (2005). Reliability and validity of the
parents’ beliefs about their role in children’s Behavioral and Emotional Rating Scale–Second
academic learning. Elementary School Journal, Edition: Parent Rating Scale. Children &
104, 197–213. Schools, 27, 147–155.

64 / November 2009 Behavioral Disorders, 35 (1), 53–65


Muthen, L. K., & Muthen, B. O. (2006). Mplus user’s Vega, W. A., Kolody, B., Aguilar-Gaxiola, S.,
guide. Los Angeles, CA: Author. Alderete, E., Catalano, R., & Caraveo-Anduaga,
Pedrotti, J. T., & Edwards, L. M. (2010). The J. (1998). Lifetime prevalence of DSM-III-R
intersection of positive psychology and multi- psychiatric disorders among urban and rural
culturalism in counseling. In J. G. Ponterotto, J. Mexican Americans in California. Archives of
M. Casas, L. A. Suzuki, & C. M. Alexander (Eds.), General Psychiatry, 55, 771–778.
Handbook of multicultural counseling (3rd ed.,
pp. 165–174). Thousand Oaks, CA: Sage.
Peterson, C., & Seligman, M. E. P. (2004). Character
strengths and virtues: A handbook and classifi- AUTHORS’ NOTE
cation. Washington, DC: American Psycholog-
ical Association. Authors wish to thank the principal investiga-
Rescorla, L., Achenbach, T., Ivanova, M. Y., tors of studies used in this secondary data
Dumenci, L., Almquist, F., Bilenberg, N., et al. analysis. Data were originally collected
(2007). Behavioral and emotional problems through support from a field-initiated research
reported by parents of children ages 6 to 16 in grant to Gale Morrison (P.I.) at the University
31 societies. Journal of Emotional and Behav- of California, Santa Barbara from the US
ioral Disorders, 15, 130–142.
Department of Education, Special Education
Rodriguez, B. L., & Olswang, L. B. (2003). Mexican-
Programs, grant # H324C000072 and a
American and Anglo-American mothers’ beliefs
and values about child rearing, education, and
Challenge 2 grant evaluation subcontract to
language impairment. American Journal of Shane R. Jimerson, Michael J. Furlong, and
Speech-Language Pathology, 12, 452–462. Manuel Casas (P.I.s) at the University of
Smokowski, P., Reynolds, A., & Bezruczko, N. California, Santa Barbara from the State of
(1999). Resilience and protective factors in California Board of Corrections.
adolescence: An autobiographical perspective
from disadvantaged youth. Journal of School Address correspondence to Jill Sharkey, Uni-
Psychology, 37, 425–448. versity of California, Santa Barbara, Counsel-
Snyder, C. R., & Lopez, S. J. (2007). Positive psychol- ing, Clinical, and School Psychology, Gevirtz
ogy: The scientific and practical explorations of Graduate School of Education, University of
human strengths. Thousand Oaks, CA: Sage.
California, Santa Barbara, CA 93101. Phone:
Steiger, J. H., & Lind, J. M. (1980, June). Statistically
(805) 893-3441; Fax: (805) 893-3375; E-mail:
based tests for the number of common factors.
Paper presented at the annual meeting of the
jsharkey@education.ucsb.edu.
Psychometric Society, Iowa City, IA.
Tinkler, B. (2002). A review of literature on Hispanic/
Latino/a parent involvement in K-12 education.
Internet site: http://www.eric.ed.gov/ERICDocs/
data/ericdocs2/content_storage_01/0000000b/80/ MANUSCRIPT
27/ce/30.pdf (Accessed March 3, 2006).
U.S. Census Bureau. (2001). The Hispanic population: Initial Acceptance: 6/09/09
Census 2000 brief. Washington, DC: Author. Final Acceptance: 10/20/09

Behavioral Disorders, 35 (1), 53–65 November 2009 / 65

View publication stats

You might also like