other factors’ eigenvalues were below 1.0. Structure coefficients on
this factor ranged from between .64 and .79 (see Table 2).
CONVERGENT AND DISCRIMINANT VALIDITY
Migration history. Respondents born in Puerto Rico tended to have
lower PAS scores (M = 3.3) than did respondents born on the U.S.
mainland, M = 4.2, t(103) = -2.93,p * .01. Thus, individuals born in
Puerto Rico tended to be more Latino-oriented than bicultural.
Furthermore, psychological acculturation (as measured by the PAS) was
correlated positively with percentage of lifetime in the United
States, r(103) = .43, p * .01, such that greater time on the U.S.
mainland corresponded with a more Anglo/American orientation.
Language use. Respondents who chose to complete the questionnaire in
Spanish tended to have lower scores on the PAS (M = 3.1) than did
respondents who completed the questionnaire in English, M = 4.1,
t(104) = -4.22, p * .001. That is, respondents who chose the Spanish
version tended to be more Latino-oriented than were those who chose
the English version. Psychological acculturation also correlated
positively with use of English at home during the respondent’s
childhood, r(106) = .51, p * .01, indicating a greater Anglo/American
orientation with increased use of English in the home.
COMPARING MEASURES OF MIGRATION AND ACCULTURATION
Individuals’ migration histories traditionally have been used as
validation measures for acculturation scales. Although these measures
may be useful, it is also important to acknowledge a qualitative
difference between time spent in a culture and one’s sense of
belonging and attachment to that culture. Multiple regression analyses
were conducted to address this distinction using psychological
acculturation (i.e., respondents’ PAS scores) and percentage of
lifetime in the United States as predictors of the adapted cultural
behavior and preference items.
A separate correlational analysis indicated that the two predictor
variables bore a substantial positive correlation, r(103) = .43, p *
.01. Under such conditions, the standardized regression coefficients
that are obtained from standard regression analyses may be biased and
relatively unreliable, as compared to other indicators (Darlington,
1990). To promote the accurate interpretation of our findings,
semipartial correlations and structure coefficients instead will be
reported. Structure coefficients were computed by dividing the
correlation between each predictor variable and the criterion variable
by the multiple correlation (see Thompson & Borrello, 1985, for a more
detailed discussion of this procedure).
Results indicated that, together, psychological acculturation and
percentage of lifetime in the United States accounted for a
substantial portion of the variance in scores on most of the cultural
behavior and preference measures (R[sup 2] values ranging from. 14 to
.44). In particular, these variables were highly effective as
predictors for behaviors and preferences associated with language use,
although they were somewhat less effective as predictors for behaviors
and preferences associated with cultural foods (see Table 3).
Semipartial correlations and structure coefficients demonstrated high
levels of association between psychological acculturation (i.e.,
respondents’ PAS scores) and scores on all of the cultural behavior
and preference items (see Table 3). In contrast, semipartial
correlations and structure coefficients suggested that percentage of
lifetime in the United States is related fairly strongly to behaviors
and preferences associated with language use and holiday celebrations
yet has relatively weak relationships with respondents’ scores on the
other cultural behavior and preferences items (see Table 3). Thus, the
general pattern of results demonstrates that psychological
acculturation served as a stronger and more consistent correlate of
respondents’ cultural behaviors and preferences than did their
percentage of lifetime spent in the United States.
Study 3
Study 2 replicated findings of high internal consistency and validity
for respondents’ scores on the PAS with a large sample of Puerto Rican
respondents. A third study was conducted to gather further validity
evidence for PAS scores across two distinct age groups (adolescents
and adults) and with two methodological modifications. First, an
interview format was used rather than a self-administered
questionnaire to examine the robustness of the scale across modalities
of administration. Second, the response range was reduced to a 5-point
scale because most respondents from Study 2 used only a portion of the
response options from the 9-point scale.
Method
SAMPLES AND PROCEDURES
Puerto Rican adolescents and their parents were recruited through
door-to-door screening, media advertisements, and community networks
within the greater Boston area. Prospective participants who
identified themselves as Puerto Rican were contacted as part of a
larger study on Puerto Rican adolescent development. Respondents were
given $10 for their participation, which consisted of face-to-face
interviews in their homes. Respondents were interviewed in the
language of their choice (i.e., either Spanish or English) by trained
bilingual and bicultural interviewers. Informed consent was obtained
from respondents prior to the interviews.
Adolescent sample. A total of 247 Puerto Rican 13- and 14-year-old
adolescents participated in this study (118 males and 129 females). Of
the participants, 98 were born in Puerto Rico and 146 were born on the
U.S. mainland (3 were born in other places). Adolescents’ percentage
of lifetime in the United States ranged from less than 1% to 100% (M =
80%).
Parent sample. A total of 228 mothers of the adolescents also
participated in this study, ranging in age from 27 to 57 years (M = 39
years). Of these mothers, 201 were born in Puerto Rico and 21 were
born on the U.S. mainland (6 were born in other places). Parents’
percentage of lifetime in the United States ranged from 85% to 100% (M
= 92%).
MEASURES
For both adolescents and parents, interview protocols included the
same versions of the PAS and the items concerning migration history
and demographic factors, which were used in Studies 1 and 2. However,
we observed that 80% of the respondents from Study 2 did not use
Scores 8, 6, 4, and 2 on the 9-point scale and essentially worked with
a 5-point scale. Therefore, the original 9-point response scales were
collapsed to 5-point scales.
Cultural behaviors and preferences. The same versions of the cultural
behavior and preference items used in Studies 1 and 2 were included in
the interview protocols for this study. To match the format of the
other items, item responses were scored on Likert-type scales ranging
from 1 (only Hispanic/Latino) to 5 (only Anglo/American).
As in the previous studies, behavior and preference items pertaining
to language reading and speaking were combined to create composite
measures of language use (behavior items) and preferred language use
(preference items). Alpha coefficients were .87 and .86 for
adolescents’ and parents’ scores on the language use measure,
respectively. Alpha coefficients were .77 for both adolescents’ and
parents’ scores on the preferred language use measure.
In addition, items pertaining to cultural foods, music, holiday
celebrations, and family celebrations were combined to create
composite measures of cultural behaviors and cultural preferences.
Alpha coefficients of reliability were .72 for both adolescents’ and
parents’ scores on the cultural behaviors measure. Alpha coefficients
of reliability were .75 and .76 for adolescents’ and parents’ scores
on the cultural preferences measure, respectively.
Results
ADOLESCENT SAMPLE
Overall, the mean acculturation score for this sample was 1.57 on the
5-point scale (SD = .62). Scores on the PAS were shown to be
internally consistent, with an alpha coefficient of .91 and item total
correlations ranging from between .52 and .78. A principal components
analysis yielded a single primary factor of psychological
acculturation, which accounted for 55% of the variance. No additional
factors were extracted beyond this factor, considering that the
eigenvalues for all other factors were below 1.0. Structure
coefficients for items on this factor ranged from between .60 and .83
(see Table 2).
Migration history. Respondents born in Puerto Rico tended to have
lower PAS scores (M = 1.33) than did those born on the U.S. mainland,
M = 1.72, t(241) = 4.98, p * .001. Psychological acculturation also
was correlated positively with percentage of lifetime in the United
States, r(247) = .25, p * .01, indicating a stronger Anglo/American
orientation with more time on the U.S. mainland.
Language use. Respondents who chose the Spanish version of the
interview tended to have lower PAS scores (M = 1.52) than did those
who chose the English version, M = 1.99, t(243) = -3.75, p * .01.
Psychological acculturation also correlated positively with use of
English in the home during the respondents’ childhood, r(247) = .40, p
* .01, indicating a stronger Anglo/American orientation with increased
use of English in the home.
PARENT SAMPLE
The overall mean for mothers’ acculturation scores was 1.55 on the
5-point scale (SD = .61). Their scores on the PAS were shown to be
internally consistent, with an alpha coefficient of .91 and item total
correlations ranging from between .53 and .79. A principal components
analysis yielded a single primary factor of psychological
acculturation, which accounted for 56% of the variance. No additional
factors were extracted beyond this factor, and eigenvalues for all
other factors were less than 1.0. Structure coefficients for the items
on this factor ranged from between .61 and .84 (see Table 2).
Migration history. Paralleling the adolescent sample, respondents born
in Puerto Rico tended to have lower PAS scores (M = 1.46) than did
respondents born on the U.S. mainland, M = 2.40, t(219) = 7.53, p *
.001. Psychological acculturation also was correlated positively with
percentage of lifetime in the United States, r(221) = .45, p * .01,
indicating a stronger Anglo/American orientation with more time on the
U.S. mainland.
Language use. Respondents who chose the Spanish version of the
interview tended to have lower PAS scores (M = 1.50) than did those
who chose the English version, M = 2.01, t(224) = -4.00, p * .001.
Psychological acculturation also correlated positively with use of
English in the home during the respondent’s childhood, r(227) = .41, p
* .01, indicating an increased Anglo/American orientation with
increased use of English in the home.
COMPARING MEASURES OF MIGRATION AND ACCULTURATION
As in Study 2, analyses were conducted to address the distinction
between time spent in a given culture and one’s psychological
attachment to that culture. Multiple regression analyses were
performed using psychological acculturation (i.e., respondents’ PAS
scores) and percentage of lifetime in the United States as predictors
of the cultural behavior and preference measures.
Correlational analyses indicated that the two predictor variables were
correlated positively in the adolescent sample, r(246) = .25, p * .01,
and even more highly correlated in the parent sample, r(227) = .56, p
* .01. Semipartial correlations and structure coefficients, therefore,
will be reported to aid in the accurate interpretation of our findings
(see Table 4).
Adolescent sample. Overall, results from these analyses indicated that
psychological acculturation and percentage of lifetime in the Unite
331