Acculturation, mental health, and help-seeking

The dimensionality of acculturation constructs has received significant research attention [7,8]. Historical conceptualizations of acculturation have often characterized acculturation as a unidimensional process, with individuals being on a continuum ranging from unacculturated to assimilated [6,9,10]. Within this perspective, higher acculturation to the host culture results in the loss of an individual’s sense of orientation and affiliation with their culture of origin [10]. Alternatively, scholars have conceptualized acculturation as a bidimensional phenomenon whereby there are parallel processes—the maintenance of one’s culture as well as the adoption of the host culture—resulting in four distinct acculturation categorizations, namely integration, assimilation, separation, and marginalization [11]. Bidimensional frameworks view cultural maintenance and acculturation to the host culture as conceptually independent [9]. Research has also conceptualized acculturation as a multidimensional process that occurs across various life domains (e.g., cultural behaviors, values, attachment, identity, belonging) [7]. Further, researchers have drawn attention to the need to consider limitations associated with conceptualizing the acculturation process in host-heritage dichotomies within pluralistic societies, as culture extends beyond traditional racial/ethnic orientations (e.g., sexual orientation, gender identity, nationality, disability), which has resulted in the development of integrative models that address various cultural orientations [12].

Approaches to conceptualizing acculturation may focus on cultural specificity or cultural fairness. The cultural specificity approach emphasizes group-specific experiences (e.g., identity, values, pride, identification) for a target population (e.g., Latinxs, Asians) [4,9,10]. A cultural fairness approach attempts to promote cross-group comparisons by using generalizable indicators of the acculturation process but often emphasizing behavioral indicators that may be more distal influences on mental health outcomes relative to psychological aspects of acculturation [4,9,10].

The process by which acculturation impacts mental health outcomes varies as a result of acculturative stress [13]. Acculturative stress refers to discomfort emerging from an individuals’ attempt to reconcile differences in language, norms, family expectations, and social roles between their origin and host country [13]. These challenges may result in individuals experiencing identity conflict, perceived cultural marginalization/discrimination, and/or pressure to assimilate, which can contribute to the development of mental health problems [1,3,14].

Acculturation may also shape help-seeking behaviors. For example, Latinx caregivers experiencing elevated acculturative stress may need mental health supports but may be less likely to seek support due to stigma, mistrust of institutions, limited culturally responsive care, and/or differing cultural beliefs about causes and treatment of mental health problems [1,3,14]. Alternatively, higher levels of bicultural integration may facilitate help-seeking by increasing familiarity with available mental health services, congruence between available treatments and perceived causes of mental health problems, and comfort navigating healthcare systems [15,16].

Research linking acculturation to mental health outcomes has largely emphasized behavioral measures, limiting our understanding of associations between psychological acculturation, acculturative stress, and related outcomes [7,17]. The underutilization of psychological acculturation measurements, like the PAS, may be due to the absence of formal psychometric evaluations of this measure, which may influence decisions regarding its inclusion in psychological research. Addressing this methodological limitation is crucial as measures of psychological acculturation may capture an internal cultural orientation process that is more proximally related to psychological functioning and outcomes relative to external behavioral indicators of acculturation [5].

Measures of acculturation

Several unidimensional, bidimensional, and multidimensional measures of acculturation have been developed and/or psychometrically evaluated but largely emphasize behavioral indicators of acculturation and focus on specific target populations [4,10]. Unidimensional measures often assess acculturation as a continuum between heritage and host culture [4,10]. These measures frequently assess language use/preferences, cultural heritage or pride, media use, ethnic identity, classification, behaviors, interactions, values, and/or attitudes among other factors (e.g., Acculturation Rating Scale for Mexican Americans, Caetano Scale, Anderson Scale) [4,10]. Bidimensional measures often include separate subscales representing acculturation and enculturation formed by indicators representing language use, language proficiency, and electronic media use among other factors (e.g., Bidimensional Acculturation Scale; Acculturation Attitude Scale-Revised) [9,10]. Multidimensional measures commonly assess acculturation across various subscales including language use/preferences, ethnic identity, classification, cultural heritage, behaviors, interactions, values, attitudes, and/or interactions (e.g., Acculturation Rating Scale for Mexican Americans II, Hazuda Scale) [10]. More recently, multi-ethnic and multi-dimensional acculturation measures have been developed such as the Multi-Ethnic Acculturation Scale, which assesses Latinx and American cultural orientation separately with response being agreement with statements reflecting various behavioral and social indicators of acculturation (i.e., ranging from completely disagree [1] to completely agree [5]) [18]. While a wide variety of acculturation measures are available, few assess Latinx individuals’ senses of emotional attachment and belonging with their heritage and/or host cultures.

Psychological acculturation scale

The PAS was developed to provide a complementary measure to behavioral-based scales, which focus on emotional-attachment and belonging-based acculturation [6]. The PAS assesses an individual’s psychological attachment and sense of belonging across a unidimensional continuum of emotional identification between Latinx/Hispanic and Anglo-American cultures and incorporates a bicultural middle group on this continuum influenced by Berry’s Model of Acculturation [6,11]. Although bidimensional measures are considered the standard in the field of psychology, meta-analytic research suggests that subscales comprising these measures have a weak negative correlation suggesting some loss of heritage culture through contact with the host culture, consistent with a unidimensional conceptualization [8]. It is also important to note that bidimensional measures often have a greater measurement burden relative to unidimensional measures that are composed of fewer items [4,10]. As unidimensional measures are associated with acculturation proxies (i.e., cultural preferences, knowledge, ethnic identification, generational status) in a similar manner to bidimensional measures, researchers have recommended their use as an inexpensive proxy acculturation measure and recommend bidimensional measures for deeper theoretical exploration [19]. Thus, the PAS can increase the feasibility of research due to its low measurement burden (i.e., 10 items) and should be viewed as a proxy measure for the acculturation process.

As a result of addressing a crucial aspect of psychological acculturation, the PAS has been used in diverse populations (e.g., immigrants from various countries, older adults, caregivers, Puerto Ricans) and referenced in numerous research articles [5]. Overall, psychological acculturation, as measured by the PAS, has been linked to various health behaviors and outcomes (e.g., criminal victimization, intent to breastfeed, prenatal anxiety, allostatic load, acculturation stress) [5,20–23]. For example, in a predominantly Puerto Rican sample of expectant mothers in the United States (U.S.) ages 16–30, Barcelona de Mendoza and colleagues found that higher levels of acculturation was related to lower odds of intent to combination feed (breastfeed and formula feed) [21]. Leveraging a similar sample, Barcelona de Mendoza and colleagues reported that PAS-derived bicultural acculturation was inversely associated with prenatal anxiety [22]. Further, Gassman-Pines and Skinner showed that higher levels of parental acculturation were inversely related to various parenting behaviors (e.g., neglect-indifference parenting) in a sample of Mexican immigrant families [24]. López-Cepero and colleagues also found that individuals with lower psychological acculturation were more likely to have a higher allostatic load (i.e., physiological adaptive response to cumulative stress) in a sample of adults aged 45–75, which is known to increase risk for physical and mental health problems [23]. Lastly, Capielo and colleagues found that Puerto Rican adults with low psychological acculturation living in Central Florida experienced more acculturation stress relative to those with higher acculturation to U.S. culture [5].

Despite its utility and use in the literature, a more comprehensive psychometric evaluation of the PAS has yet to be conducted beyond the Principal Component Analysis (PCA) and construct validation performed as part of the original measure development study [6]. This represents a significant gap in the psychometric literature that limits the impact of this tool in improving our understanding of acculturation processes. The present study sought to address this gap by confirming the structure, internal consistency, language-based measurement invariance, and construct validity of the PAS to aid mental health research with Latinx immigrant populations.

Information regarding the structure and validity of the PAS is largely based on the original development study conducted by Tropp and colleagues, which consisted of three studies published together [6]. The first study included 36 adults ages 13−58 living in the U.S. from various countries of origin (i.e., U.S., Puerto Rico, Mexico, and other Latin American countries). Respondents completed both the English and Spanish forms of the measure, which demonstrated strong mean correlations with each other (r = .94). Notably, although responses on individual items between the language forms were also strongly correlated (r = .70−.92), two specific items had lower correlations between forms (“In what culture(s) do you feel confident that you know how to act?” [r = .37]; “In what culture(s) do you know what is expected of a person in various situations?” [r = .64]). These lower correlations may signal some variability in factor loadings for items between language forms. This first study also found that both language versions had excellent internal consistency (English Cronbach’s α = .85; Spanish α = .83).

The second study performed a PCA of the measure utilizing a sample of Puerto Rican and U.S.-born adults (ages 12–58; n = 107). Results indicated that the PAS scale had a single factor, which explained 51% of the variance, with no additional identified factors (additional factors resulted in eigenvalues less than one). This study also reaffirmed that both language versions of the PAS had excellent internal consistency (English Cronbach’s α = .83; Spanish α = .90). Further, findings demonstrated that the PAS was related to proxies of acculturation in expected ways (i.e., higher acculturation among those born in Puerto Rico relative to those born on mainland; lower acculturation among those who chose to complete the survey in Spanish).

The third study focused on validating the PAS within a sample of caregivers (ages 27–57; n = 228) and their adolescents (ages 13–14; n = 247). In this study, PCAs conducted on the caregiver and adolescent samples also suggested that the PAS was represented by a single factor that explained a range of 55–56% of the variance. Findings further highlighted expected differences in acculturation levels. For instance, lower acculturation on the PAS was related to being born in Puerto Rico relative to the U.S. mainland and selection or preference for Spanish relative to English in the survey.

The foundational evaluation of the PAS conducted by Tropp and colleagues was followed up by Stevens and colleagues, who validated a Dutch language and culturally adapted version of the measure for Moroccan adolescents (M_age = 14.2; n = 783) and adults (M_age = 40; n = 387) living in the Netherlands [6,25]. Stevens and colleagues adapted the scale by selecting six items that assess emotional attachment and belonging, and duplicated these questions to measure psychological acculturation to Dutch and Moroccan culture separately, which added a bidimensional component to the scale [25]. Using this adapted version of the PAS, a Confirmatory Factor Analysis (CFA) established that a bidimensional model representing the aforementioned cultures fit the data better than a single unidimensional factor. Although this study further examined the structure and external validity of the PAS, research is still needed to determine the internal consistency, construct validity, and language-based measurement invariance of the PAS in larger and more diverse samples within the U.S.

The present study

The present study sought to address the need for a comprehensive psychometric evaluation of the original English and Spanish PAS forms. Our aims were to a) confirm the structure of the PAS, b) evaluate internal consistency, c) explore language-based measurement invariance, and d) establish construct validity of the PAS within a diverse sample of Latinx caregivers. Based on prior research by Tropp and colleagues, we hypothesized that the PAS would be best represented by a single factor and would have good internal consistency [6]. Measures of language and generational status were expected to be related to higher acculturation scores in expected ways (i.e., greater preferences for English, selection of the English Survey, and greater generations in the U.S.) [6]. The unidimensional global factor representing the PAS was not expected to have metric invariance due to previously documented differences in item correlations between responses on the English and Spanish forms [6].

Measures

Analytic plan

Measurement structure and internal consistency

First, a global CFA model was fitted with all scale items loading onto a global factor using responses from both language versions of the PAS. The model fit for the global factor solution was mixed with two metrics suggesting marginally poor fit (i.e., RMSEA, TLI), and two suggest a good fit (i.e., CFI, SRMR). Despite the mixed fit indices, the global factor model had excellent internal consistency (α = .96) and strong item loadings (ranging from .75−.88). See Table 3 for overall sample factor loadings and Fig 1 for the global factor model loadings by language version.

Download:

• PNG
  larger image
• TIFF
  original image

Table 3. Fit statistics for global factor, two-factor, bifactor, and language invariance models (N = 511).

https://doi.org/10.1371/journal.pone.0355228.t003

Download:

• PNG
  larger image
• TIFF
  original image

Fig 1. Global factor model loading by language (N = 511).

English (n = 275), Spanish (n = 236).

https://doi.org/10.1371/journal.pone.0355228.g001

Horn’s parallel analysis with the Maximum Likelihood estimator was then conducted to determine whether additional factors may be present in the scale that may improve model fit. The parallel analysis suggested that in our sample, the PAS may be represented by up to two factors (See S2 Fig.). An EFA was conducted to obtain factor loadings and associated eigenvalues for the PAS. The EFA identified two likely factors in the data representing (a) cultural identity and social belonging (items 1–3; eigenvalue = 3.98) and (b) cultural competence and familiarity (items 4–10; eigenvalue = 3.18). A post-hoc EFA scree test was then conducted, which suggested that a third factor may be warranted based on eigenvalues above one (See S3 Fig.). A three-factor EFA was explored but was rejected as the solution produced two factors represented by two items each. The two scales represented by two items introduce issues related to the reliability and stability of the constructs being measured and challenges related to CFA models that typically require at least three items per factor to provide sufficient degrees of freedom to estimate model parameters and latent variables.

A CFA model with the Maximum Likelihood estimator was then used to assess the two-factor solution identified by the EFA. The two-factor model fit was in the recommended range for all approximate fit indices (i.e., RMSEA, SRMR, CFI, TLI). A chi-squared difference test also suggested that the two-factor model fit the data significantly better than the global factor solution, χ²_difference, (1) = 85.58, p < .001. In addition, the two-factor model had excellent internal consistency for cultural identity and social belonging (α = .94) and cultural competence and familiarity scales (α = .89) and had item loadings ranging from .75−.88. However, the covariance for these factors was high (.923), which suggest that they had high overlap and were potentially redundant. See Table 4 for overall global factor loadings and see S4 Fig for loadings by language forms.

Download:

• PNG
  larger image
• TIFF
  original image

Table 4. Overall sample factor loadings for Psychological Acculturation Scale (N = 511).

https://doi.org/10.1371/journal.pone.0355228.t004

A bifactor model was then used to simultaneously load items onto the global and two-factor solutions resulted in negative variance. Item 3 within the cultural identity and social belonging scale was identified as have a large negative loading and variance. Several techniques were used to address this issue, including setting a new starting point and forcing the model to converge on an estimate above zero for item 3. These efforts were unsuccessful in correcting the negative variance, which suggest that the cultural identity and social belonging scale was unstable after accounting for the global factor. The bifactor model converged with a structure composed of the global scale and cultural competence and familiarity scale with fit indices generally in the recommended range (i.e., RMSEA, SRMR, CFI, TLI). A chi-squared difference test also suggested that the bifactor model fit the data significantly better than the two-factor solution, χ²_difference, (6) = 50.79, p < .001. Factor loadings for the bifactor model ranged from .13 – .87 (See S5 Fig). The bifactor fit index suggest that 89.7% of the common variance between items was explained by the global factor (ECV = .897), 96.0% of the total score reflected common variance instead of measurement error (ω coefficient), and 89.4% of variance in total scores was attributed to the global factor (ωH). These findings suggest that the PAS was best represented by a unidimensional global factor as additional factors only accounted for limited unique variance.

Measurement invariance

Language-based performance for the PAS global factor model was assessed in a stepwise manner to determine whether the English/Spanish forms were measurement invariant (i.e., configural, metric, scalar, strict; Jeong & Lee, 2019). First, a configural model with the global factor structure was fitted in both the English and Spanish forms, which had poor fit across indices relative to the freely estimated unidimensional model (RMSEA, SRMR, CFI, TLI). This finding suggest that the latent structure of the measure was different across language version of the PAS. As configural invariances was not achieved, additional invariance test were not conducted.

Construct validity

Differences in PAS scores were examined by participant language selection, language preference, and generational status in the U.S. using ANOVA. Global scores on the PAS increased (i.e., greater Anglo/American acculturation) in concordance with more generations within the U.S., greater preference towards English, and among those who selected the English survey. See Fig 2 for construct validity measures by the PAS global scale.

Download:

• PNG
  larger image
• TIFF
  original image

Fig 2. Psychological acculturation global scale associations with construct validity indicators.

Analysis of Variance (ANOVA) were used for comparisons. Mean and standard deviation are displayed. p > .05 (no stars), p ≤ .05 (*), p ≤ .01 (**), p ≤ .001 (***), p ≤ .0001 (****). PAS = Psychological Acculturation Scale.

https://doi.org/10.1371/journal.pone.0355228.g002

Implications

The findings presented herein add to the existing body of acculturation-related literature indicating that the PAS is a reliable measure. Our research highlights a need for caution when using the different language forms of this tool with Latinx adults. Moreover, the PAS centers on emotional and belonging-based indicators of acculturation, in contrast to many existing measures that predominantly center on behaviors, representing critical yet often overlooked dimensions of this construct. To this end, the novel focuses on emotions and identity, allowing for a more nuanced and comprehensive understanding of the acculturative experience, capturing the internal psychological processes that shape an individual’s sense of belonging, self-concept, and well-being within and across cultural contexts. As such, this tool could be utilized in conjunction with other acculturation tools that focus on behavioral indicators, thereby providing potentially more comprehensive and nuanced data on the acculturation processes among Latinx individuals, which is known to be heterogeneous and multifaceted. However, while the assessment of psychological acculturation may provide important information on internal cultural orientations that may be more proximal to psychological functioning, research is needed to determine the incremental contributions of using the PAS alongside more distal behavioral indicators in explaining within-group differences in mental health and help-seeking outcomes among Latinxs [5].

While bidimensional measures are considered the standard in the field of psychology, research suggests that acculturation and enculturation measures have a weak negative correlation suggesting some loss of heritage culture through contact with the host culture, consistent with a unidimensional conceptualization used in the development of the PAS [8]. The PAS may also increase the feasibility of acculturative research as it is composed of fewer items than most bidimensional measures [4,10]. Thus, while the unidimensional design of the PAS limits its use for research focusing on in-depth theoretical explorations of acculturative processes, this scale may facilitate research as a brief proxy measure of acculturation that can be used to examine within-group differences in mental health and help-seeking outcomes among Latinxs [19]. Future research should also adapt and evaluate a bidimensional version of the PAS that measures emotional attachment and belonging-based indicators of acculturation and enculturation separately to facilitate deeper theoretical research [25] These measures can help elucidate within-group differences in help-seeking behaviors and perceptual barriers to accessing care for youth among Latinx caregivers [3,14,26,27]. Additionally, future research is needed to further assess the validity of this tool, for example, its concurrent and discriminant validity. Lastly, given distinct acculturation trajectories within the heterogeneous Latinx community, future research could utilize latent profile analyses to identify profiles of psychological acculturation.

Strengths and limitations

There are several important strengths of the study that are worth mentioning. First, the present study extended the scientific literature pertaining to the measurement of the acculturation of Latinx people using the PAS tool through the inclusion of structural analyses, construct validity, and language-based measurement invariance. Additionally, this study engaged a large and diverse sample of caregivers and deployed the survey in English and Spanish, which holds important implications for the generalizability of findings.

This study also has several limitations. First, the relatively small sample size for each language version of the PAS constrained our ability to separately validate factors identified by the EFA and subsequently evaluated these factors through a CFA and a bifactor model. In order to address this limitation, future research should confirm the present findings in a larger sample. Additionally, recruiting participants through QuestionPro required that we only use responses from participants who provided complete responses and high-quality data, which may have potentially introduced bias by excluding those hesitant to report on their child’s mental health and those who have difficulties with literacy [27]. Leveraging other recruitment methods such as in-person recruitment through Latinx serving community organizations would help address this limitation in future studies. The present study did not assess concurrent validity between the PAS and other measures of acculturation or associations with contextual factors that may influence participants’ reporting on the measure. Contemporary multi-ethnic acculturation frameworks describe non-linear and domain-specific acculturation processes, and potential mismatches between behavioral and psychological acculturation domains. For example, different domains of acculturation (e.g., cognitive, behavioral, identity-based) may change at different rates and in different directions over time [18]. As a result, it is recommended that future research should examine associations between the PAS and established measures of acculturation and contextual factors known to impact acculturation processes (e.g., geographic region, political climate) [8]. Also, this sample only focused on Latinx caregivers living in the U.S. who had a child with emotional or behavioral problems, limiting the external validity of the present findings. Future studies should investigate whether these results generalize to Latinx caregivers living outside of the U.S. and caregivers whose children do not have emotional or behavioral difficulties. Lastly, acculturation is a dynamic and context-sensitive process, and as a result responses to emotionally and identity-based items included in the PAS may be influenced by contextual or situational factors (e.g., caregiving stress) that reflect the caregiver's current psychological state. Future research should examine potential construct contamination resulting from contextual or situational factors that may influence reporting on the PAS such as caregiver strain.

Conclusions

In sum, the present study advances the field of acculturation research by offering a psychometric evaluation of the PAS. Most measures of acculturation focus on assessing behavior-based indicators of this process [4,6,10]. Instead, the PAS focuses on capturing emotional and belonging-based indicators that serve as a proxy of the acculturation process and has been utilized in numerous research studies as a result [23,24]. However, despite its focus on distinct aspects of psychological acculturation and incremental value, a lack of psychometric evaluation of the PAS limits the measures’ potential inclusion and effectiveness in research. By psychometrically evaluating a measure assessing these understudied aspects of acculturation, the present study offers not only an important tool that may advance our understanding of the phenomenon of psychological acculturation but may also lead to insights that can help clinicians support Latinx caregivers undergoing the process of immigration and subsequent acculturation in the U.S.