Abstract
Culture influences responses to psychological measurements in ways unrelated to target constructs, thus biasing test scores and potentially contributing to under- and over-diagnosis of mental health problems in populations for which measures have not yet been normed. We conducted a systematic review of publications addressing response style among Latinx population groups in North and South America. In a final corpus of 24 studies, Latinx/Latin American populations were generally found to exhibit higher levels of extreme response style (n = 17), acquiescent response style (n = 10), and socially desirable responding (n = 5). The few publications (n = 3) that investigated midpoint responding reported no differences. Seven publications (29%) attempted to adjust scores to mitigate response style bias, using both scale design and statistical techniques. Findings suggest that researchers and clinicians should directly assess culturally patterned response style as a construct, rather than inferring style indirectly using other measures. For clinicians, knowledge of response style represents another facet of case conceptualization.
Research, evaluation, and treatment recommendations in psychology rely on accurate assessment. However, culture may influence responses to items on psychological measurement tools in ways that are unrelated to target constructs (Triandis, 1972). Culturally influenced patterns of responding, or response styles, may thus contribute to under- and over-diagnosis of mental health problems in populations outside of the primarily non-Latinx White populations in the United States and Canada upon which most psychological research is based (Hinton & Lewis-Fernandez, 2011; Lee et al., 2011; Watkins & Cheung, 1995). Despite the acknowledging that response style affects assessment, knowledge gaps remain regarding response styles differences between cultural groups. This phenomenon is particularly pronounced in the Americas: most of the hemisphere’s population resides in Latin America, and Latinx populations are growing rapidly in the United States and Canada (Lopez et al., 2022; Statistics Canada, 2017), yet Latinx response styles remain underexamined. Understanding response styles among Latinx and Latin American groups is critical to ensure accurate psychological measurement and diagnosis.
Latin America, named for its predominant Spanish- and Portuguese-speaking cultures, includes nations in North, Central, and South America, as well as islands in the Caribbean Sea. People residing in the United States or Canada with family origins in this region are broadly referred to as “Latinx.” As with many pan-ethnic terms, “Latinx” conflates a wide variety of identities, obscuring intragroup cultural differences. “Latinx” has also been critiqued as being inconsistent with Spanish grammar; however, it is useful in contrast to other pan-ethnic labels (Cardemil et al., 2019). Moreover, the use of broad pan-ethnic terms has been the practice of social sciences, including psychology, to examine similarities and differences between variable populations. Consistent with this practice, this review will use “Latinx” as an umbrella term, identifying more specific identities when reported in findings. Our assumption is that such findings have relevance to the literature with Latinx people. We refer to residents of Latin American countries broadly as “Latin Americans,” while acknowledging that residents may prefer and use other terms.
Culture pervades all aspects of human life, including the experience and expression of psychopathology (Kirmayer & Ryder, 2016), and therefore, responses on psychological measures (Triandis, 1972). Both broad cultural orientations and more specific social norms may affect response style. Of particular interest to scholars has been the continuum between individualistic cultures, which emphasize the uniqueness of individuals, and collectivistic ones, which value harmonious relationships within a group (Triandis, 2001). Self-reliance, for example, is a key value in individualistic cultures and has been associated with a higher prevalence of
In contrast, social norms prioritizing modesty may encourage respondents to downplay their symptoms, consistently endorsing choices at the low end of a clinical scale, or
Of course, in addition to the influences of cultural values and norms, sociopolitical and sociodemographic factors also affect response styles. Access to societal power is often contingent on aspects of identity, including race/ethnicity, gender, age, and class, which shape both respondents’ experiences of the world and their comfort with psychological surveys. For example, acquiescent responding may stem from cultural norms that emphasize group cohesion (Davis et al., 2011), the influence of older age (Davis et al., 2019) or lower socioeconomic status (Ross & Mirowsky, 1984), mistrust of the research process born of systemic oppression, or some combination of all these factors. Characterization of response style along broad cultural lines, often operationalized as nationality or pan-ethnicity, therefore necessarily obscures important axes of variance. However, simplification of human diversity into broader categories is sometimes required to make substantive comparisons, while acknowledging that these categories may elide within-group variance.
Most psychological measures have been normed using non-Latinx White populations in the United States and thus predominantly reflect the particular response styles of the historical mainstream of U.S. society: settler populations of Northern European ancestry. Also referred to as “WEIRD” populations in psychology (Henrich et al., 2010), this identity tends to emphasize emotional reserve and individualism, and thus, the response style is not only reserved (i.e., not extreme) but also assertive (i.e., not acquiescent). Because of this historically dominant identity in the United States, investigating response style in non-White populations is an exercise in cross-cultural comparison to these U.S. norms, norms which may not be relevant to people in other cultural contexts. Acknowledging and investigating differences in response tendency has direct implications within several subfields in psychology. For instance, clinical psychologists often use clients’ scores on psychological assessments to inform diagnoses with the assumption that the scores can be interpreted similarly across cultural groups. However, this might ignore group differences in response patterns, biasing assessments and potentiating misdiagnosis, as has been discussed in the context of post-traumatic stress disorder (Pole et al., 2005). Coming from cultural backgrounds that tend to value emotional expression and acquiescence, Latinx or Latin American clients may be incorrectly over-diagnosed with dysfunction and distress by clinicians trained in mainstream U.S. psychology, who generally are not trained in cultural differences in response styles. Alternatively, if mainstream U.S. clinicians acknowledge cultural differences but consider Latinx respondents to be overly expressive or “dramatic,” they may overcorrect and underdiagnose true psychological problems.
The psychological literature demonstrates long-standing and wide-ranging research interest in response style. However, there are no systematic reviews of the literature regarding response styles in Latinx and Latin American populations. The current review aims to address this gap by (a) describing how Latinx/Latin American response styles might differ from the response style implicit in psychology and (b) examining how such differences have been accounted for in attempts to make meaningful cross-cultural comparisons. The purpose of this investigation is to inform U.S.-based psychologists who are interpreting scores from Latinx or Latin American participants and clients on psychological measures about a potentially significant avenue of variance.
Method
Overview of Design
We used the Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) guidelines (Page et al., 2021) to conduct the search (Supplemental Table 2), and a data-based convergent synthesis, in which qualitative and quantitative evidence are analyzed using the same synthesis method (Hong et al., 2017). Our protocol was preregistered with PROSPERO (registration number CRD42022330120).
Population of Interest
Our population of interest was Latinx, Latino, and Hispanic individuals residing in the Americas (see Supplemental Materials for search terms), that is, Latinx and Latin American populations. The population of interest could be compared to either non-Latinx White North Americans, or to existing published norms, which often reflect the response style of non-Latinx White North Americans.
Outcomes
We extracted evidence of response style differences between broad Latinx and non-Latinx White cultural groups in North, Central, and South America. Cultural groups could be defined by nationality, racial or ethnic identity, language, immigration status, acculturation stage, or cultural orientation. Findings of interest were (a) response style differences between cultural groups, including measurement invariance attributable to cultural response styles; (b) the effect of cultural values on response styles; and (c) methods used to adjust for response style differences.
Response styles included socially desirable responding, acquiescent response style, extreme response style, and midpoint/neutral response style. Socially desirable responding occurred when participants chose response options in line with social norms rather than reflecting their true feelings, behaviors, or experiences. Acquiescent response style was reflected in agreement with scale items regardless of content, potentially leading to the endorsement of conflicting responses. Extreme response style was considered to occur when participants tended to select scale endpoints regardless of item content. Midpoint or neutral responding was the inverse, when participants tended to choose the middle option on a scale regardless of item content.
Search Strategy
Search terms were in English and required that publications include terms reflecting (a) extreme, acquiescent, socially desirable, and neutral response styles; (b) culture and race/pan-ethnicity, as well as the comparison group; and (c) any Latin American nation (see Supplemental Materials for the full list). Searches were conducted using PsycINFO, SCOPUS, PubMed, PTSD publication database, and ProQuest Psychology Database in May 2022. Subject-area experts were asked to contribute and reference lists of included studies were examined to identify additional publications.
Inclusion Criteria
We included all studies reporting data on our target outcomes, regardless of methodology. Studies were included if they (a) examined response style, (b) compared group-level differences between cultural groups, and (c) were published in English, Spanish, or Portuguese. Although we anticipated that cultural groups would often be defined based on nationality, pan-ethnicity, and language, we did not define what factors could be considered cultural a priori to be as inclusive as possible of differences that might inform group-level response styles. Previous reviews of response style differences among target population groups would have been eligible for inclusion if they presented relevant data. Studies that discussed but did not measure response style were excluded.
Study Selection
Screening and study selection was completed using Covidence, a web-based tool that provides a clear audit trail of inclusion and exclusion decisions. Each citation was reviewed by one reviewer (C.Z.) with consultation and spot-verification by another (A.R.).
Risk of Bias Assessment
Given the potential heterogeneity of included publications, the flexible Mixed Methods Appraisal Tool (MMAT) was used to assess risk of bias (Hong et al., 2018). Each included study was assessed using quality criteria specific to study design in the form of yes/no questions. For specific MMAT items, see Hong et al. (2018). We did not assign studies a risk of bias score, but we discuss overall quality in the “Results” section (see Supplemental Table 1).
Data Extraction and Synthesis
Data were extracted in Excel to record study bibliographic details (authors, year of publication, country of publication); study methodology (qualitative or quantitative; study design; recruitment method); study population (age range; ethnic identities; nationalities; grouping variables); the type of response style or measurement invariance under investigation and how it was measured; the method used to account for response style (if this was done); and findings relating to response styles or measurement invariance.
In the interest of efficiency, data for each study was extracted once (by C.Z.), with spot-verification of 11 publications by another team member (A.R.). Questionable citations were resolved through team discussion. All extracted data are included in the Supplemental Materials. Given the broad scope of this review, a mixed-methods synthesis was necessary (Hong et al., 2017). We used a narrative synthesis, organized by type of response style under investigation. We compared studies’ findings by operationalization of “Latinx” and subfield of psychology. While we use “Latinx” and “Latin Americans” as umbrella terms in this article, when reporting the results of a specific study, we use that study’s terminology (i.e., Latino, Hispanic, Latinx, national, or ethnic group).
Results
Overview of Included Studies
Our search identified 228 publications for screening: 226 from the scholarly search engines and two additional studies identified through examining the reference lists of included publications. Two publications in Spanish and one in Portuguese reached the full-text review stage; all others were in English. After the process of review and exclusion, 24 publications were included, comprising 21 peer-reviewed publications, including a systematic review, and three dissertations (Figure 1). Twenty-three studies used quantitative methods, and one used mixed methods (Table 1). Most publications assessed extreme responding (n = 17, 71%), followed by acquiescent responding (n = 10, 42%) (Table 2). Socially desirable responding (n = 5, 21%) and midpoint/neutral responding (n = 3, 13%) were less often examined. Included publications represent a broad spectrum of psychological subfields, using scales that assessed measurement (n = 8), cultural/developmental (n = 7), social/personality (n = 6), clinical/health (n = 5), and industrial/organizational (n = 4) constructs. For the purposes of analysis, cultural/developmental and social/personality measures have been grouped together.
Study Flow Diagram.
Characteristics of Included Publications (n = 24).
Note. Groups are reported using the same language as the original publication, for example, “Latino” or “Hispanic.”
Response Style Assessment and Adjustment in Included Publications.
ERS: Extreme response style; ARS: Acquiescent response style; MRS: Midpoint/neutral response style; N/A: not applicable.
Due to the comparison group requirement, all studies were conducted at least in part in the United States. Mexico was also prominently represented (n = 8, 33%) (Table 1). Most other Latin American nations and Canada were included in between one and three publications; many of these samples being accounted for by two studies: an online survey of bilingual adults (Benavidez, 2018) and a systematic review of life satisfaction surveys (Brulé & Veenhoven, 2017). Sociocultural groups were most often based on pan-ethnicity (n = 16, 67%) or nationality (n = 10, 42%); these two categories were conflated in a few cases (Table 1). A quarter of publications (n = 6) compared participants based on acculturation, or cultural orientation, models, or values. The influence of language was assessed in five publications, comparing English to Spanish speakers (21%).
Risk of Bias Assessment
Twenty-three of the 24 included studies were quantitative non-randomized designs under the MMAT classification system (Supplemental Table 1). Most publications assessed non-representative samples (n = 17, 71%). Other common risks of bias were lack of information regarding complete outcome data (n = 12, 50%) or potential confounds (n = 5, 21%). The single mixed-methods study reported its qualitative findings using topic frequencies rather than a more rigorous form of qualitative analysis (Supplemental Table 1).
Extreme Responding
Assessment Methods
Extreme responding was most often assessed by a simple count, namely the proportion of responses at either end of a scale (Table 2). Of the 17 publications reporting extreme responding, the majority used this arithmetic technique (n = 10, 59%), based on responses to a variety of scales. Two additional publications (Benavidez, 2018; Clarke, 2000) used the Greenleaf Extreme Response Scale, which similarly tabulates endpoint responses but is designed so that items are uncorrelated with each other (Greenleaf, 1992). Five other publications assessed extreme responding using a variety of statistical techniques, including differential item functioning (Kim et al., 2009; MacIntosh & Strickland, 2010), latent class analysis (Liu et al., 2017), logistic regression (Weech-Maldonado et al., 2008), and cluster analysis (Zoller, 2018).
Findings
Studies that assessed response style differences based on nationality and pan-ethnicity reported higher levels of extreme responding among Latinx/Latin American populations, both in the United States and in Latin America (Table 3). One of these studies noted that pan-ethnicity interacted with scale design: Hispanic Navy recruits exhibited more extreme responding than non-Hispanic White recruits on a 5-point Likert-type scale, but not on a 10-point scale (Hui & Triandis, 1989). Defining groups by language and cultural factors presented a less clear picture than using group definitions based on nationality or pan-ethnicity (Table 3). In a cross-sectional survey of Mexican Americans with Type II diabetes, for example, extreme responding was positively associated with endorsing Hispanic cultural values familismo, simpatia, and la mujer abnegada, but negatively associated with endorsing machismo among the men in the sample (Davis et al., 2011).
Findings by Response Style and Grouping Variable.
Note. Several studies reported on more than one response style and use more than one comparison category.
Studies using clinical or health psychology measures, as well as those reporting on industrial/organizational psychology measures, found higher rates of extreme responding among Latinx participants, variously defined (Table 4). Studies assessing social/personality or cultural outcomes, or response style itself, generally reported higher rates of extreme responding among Latinx participants as well. Finally, scale design was also related to responding in ways that may obscure response style. Scales with fewer response options (Brulé & Veenhoven, 2017; Clarke, 2000; Harumi & Church, 2011; Hui & Triandis, 1989) and surveys administered face-to-face as compared to online (Liu et al., 2017) were found to potentiate more extreme responding.
Findings by Response Style and Psychological Subfield.
Note. Several studies reported on multiple response styles using more than one type of measure.
Acquiescent Responding
Assessment Methods
As with extreme responding, arithmetic techniques were often used to assess acquiescent responding. Of 10 studies reporting acquiescent responding, four considered the proportion of participants’ agreement responses to indicate their level of acquiescence (Table 2). One of these also used a scale designed to target acquiescent responding, the True Response Inconsistency Scale (Harumi & Church, 2011). Three other publications attempted to tap into the concept of acquiescence as agreement regardless of item content by counting agreement responses on sets of contrasting or dissonant items (Davis et al., 2011, 2019; Lee et al., 2020). One publication employed latent class analysis to assess acquiescence (Liu et al., 2017).
Findings
Compared with non-Latinx White respondents, acquiescent response style was more prevalent among Latinx people in the United States and among Mexicans rather than Americans (Table 3). This response style partially contributed to lower validity on measures of subjective well-being for older Hispanic adults in the United States compared with their non-Hispanic contemporaries (Lee et al., 2020), for example. These national differences, however, may reflect other sociodemographic factors, including age, gender, socioeconomic status, and education (Ross & Mirowsky, 1984). Acquiescence was also generally associated with Latinx cultural values and Spanish-language use (Table 3). Spanish use interacted with pan-ethnicity among older adults in the United States, with more acquiescent responding among Hispanics interviewed in Spanish (Lee et al., 2020). Regarding broader cultural factors, acquiescent responding was positively associated with cultural collectivism among college students across several countries, including Mexico and the United States (Harumi & Church, 2011). Acquiescence was associated with Latinx identity across all types of psychological measures (Table 4).
Socially Desirable Responding
Assessment Methods
Five publications assessed social desirability bias. Unlike the other forms of response bias, social desirability depends on item content and was assessed using a dedicated scale (Table 2). Two publications used the Marlow-Crowne Social Desirability Scale (Ross & Mirowsky, 1984; Warnecke et al., 1997), which asks participants to report on their performance of uncommon but highly socially desirable (i.e., approved) or common but socially undesirable (i.e., disapproved) behaviors (Hart et al., 2015). Three other publications used the Balanced Inventory of Desirable Responding (Benavidez, 2018; Bernardi, 2006; Hopwood et al., 2009), which uses a similar procedure but assesses both conscious and unconscious social desirability (Paulhus, 1991). One of the three studies employed only the Impression Management subscale, which targets conscious efforts to present oneself in a positive light (Bernardi, 2006).
Findings
Comparisons based on pan-ethnicity, nationality, and cultural factors all reported greater socially desirable responding among Latinx participants compared with their non-Latinx White counterparts (Table 3). Compared with 11 other nations, however, business students from the United States scored especially low on social desirability (Bernardi, 2006), which suggests that using non-Latinx White Americans as a comparison group may disproportionately heighten reported international differences. Only one study examined differences in socially desirable responding based on language use, reporting no differences between surveys completed in English versus Spanish among bilingual adults in the United States (Benavidez, 2018). Socially desirable responding was reported to be elevated among Latinx populations regardless of the psychological assessment measure used (Table 4).
Midpoint/Neutral Responding
Of three publications reporting neutral responding, two tallied participants’ proportion of midpoint responses (Harumi & Church, 2011; Hoffmann et al., 2013), and another used cluster analysis to identify midpoint responders (Zoller, 2018) (Table 2). All publications assessed midpoint responding based on nationality or pan-ethnicity and reported no differences between Mexican and U.S. respondents (Table 3), or any differences by type of psychological measure (Table 4).
Studies Adjusting for Response Style
Seven publications (29%) reported attempts to reduce response style variation between groups (Table 2). Three publications assessed the number of response options available to respondents, generally reporting reduced extreme responding with more response choices (Clarke, 2000; Hoffmann et al., 2013; Hui & Triandis, 1989). Clarke found that the use of 5-point scales reduced extreme responding compared with 3-point scales, though expanding the scale to seven response choices offered no further improvement (Clarke, 2000). Hoffmann et al. (2013) advocated for the use of 7-point scales to reduce both extreme and midpoint responding. Finally, a 10-point scale reduced extreme responding among Hispanic Navy recruits compared with a 5-point scale (Hui & Triandis, 1989). A relatively simple statistical method to correct for extreme response style is truncated or collapsed scoring, in which endpoint choices are collapsed into less-extreme categories (e.g., “strongly agree” and “agree” are considered the same response). Among a nationally representative sample of middle-grade students in the United States, however, truncated scoring demonstrated little differential impact (Bachman et al., 2011).
Finally, two studies entered response-style indices into statistical models in attempts to partial out their effects. Ross and Mirowsky (1984) included response styles in regression models as predictors of distress symptoms, finding that acquiescence did not distort the reporting. Controlling for socially desirable responding altered the effects of socioeconomic status and age on distress, but not those of gender or pan-ethnicity (Ross and Mirowsky, 1984). A dissertation sampling college students in several countries opted for analyses of covariance to adjust for the influence of acquiescence on personality measures, which reduced mean score differences between students in the United States and Mexico (Harumi & Church, 2011).
Discussion
Findings from this review are in line with the literature suggesting more extreme, acquiescent, and socially desirable response styles among Latinx populations than non-Latinx White North American respondents. While these patterns were replicated across several psychological subfields, there remains considerable intragroup heterogeneity. Furthermore, despite a long history of research interest in response styles and the bias they can introduce into psychological assessments, we found relatively few relevant publications, particularly regarding socially desirable and midpoint responding. The three publications that assessed midpoint responding did not report differences between non-Latinx White and Latinx/Latin American groups. Although response style was acknowledged as a particular problem for comparing scores across groups, only seven reviewed studies attempted to adjust for these biases.
Assessing Response Style Among Latinx and Latin American Populations
Of the four response styles examined in this review, extreme responding has received the most research attention among Latinx and Latin American populations. While there is general agreement that Latinx people tend to select endpoint responses more often than comparison groups do, the reasons for this difference remain unclear. A small body of research has suggested that Latinx cultural values of simpatía, familismo, and respeto potentiate both extreme and acquiescent responding as group members strive to demonstrate sincere and strong agreement with one another to promote social cohesion (Calzada et al., 2010; Davis et al., 2011; Hui & Triandis, 1989; Marin et al., 1992), and in general, this review supports these hypotheses. However, the use of non-Latinx White North Americans as a default comparison group presents complications, as researchers have come to recognize the “WEIRD” nature of this population (Henrich et al., 2010). Indeed, at a national level, respondents in the United States displayed particularly low levels of socially desirable (Bernardi, 2006) and acquiescent responding (Hoffmann et al., 2013) compared with people from many other countries. In other words, the growing body of research seems to suggest that the unique culturally patterned responses of non-Latinx White people in the United States and Canada should not be considered a neutral “starting point” for analyses.
While drawing on a smaller number of studies, acquiescent and socially desirable responding were also generally reported to be elevated among Latinx and Latin American populations compared with non-Latinx White North Americans. While aforementioned social norms may have affected these findings, an anonymous reviewer made the important point that studies have often failed to consider the contribution of access to power. Structural oppression is likely to influence participants’ trust in research and clinical processes, perhaps potentiating socially desirable or agreeable responses as a self-protective mechanism. Indeed, a study of adults in Chicago found that response bias was reduced when interviewers shared respondents’ ethnicity (Warnecke et al., 1997), potentially indicating more comfort and less need to “perform” socially. In contrast, a study conducted among Mexican, Cuban, and Puerto Rican Americans reported no influence of respondent–interviewer social distance, social deference, or interviewer characteristics on acquiescence (Davis et al., 2019). Ultimately, in response style as in all psychological research, more attention to power dynamics is essential.
Included publications also noted intragroup variation in response styles, which is unsurprising given the broad pan-ethnic groups compared in this review. Cuban Americans demonstrated greater acquiescence than Mexican and Puerto Rican Americans, for example (Davis et al., 2019). While beyond the scope of our focus on the Americas, non-Latinx White samples from different nations similarly exhibited different levels of extreme responding (Clarke, 2000). This within-group heterogeneity is likely generated by cultural differences between national and ethnic groups, as well as myriad sociodemographic factors, including class, age, and gender. For example, extreme responding was associated with less education as well as Latinx identity (Marin et al., 1992). Working with a sample of immigrant, low-income Hispanic women, D’Alonzo (2011) found that extreme responding was reduced by adapting measures to improve readability, suggesting that in some cases response bias may be due not to culturally inculcated response styles but rather to content comprehension. Similarly, acquiescence was associated with lower socioeconomic status and socially desirable responding with older age (Ross & Mirowsky, 1984; Warnecke et al., 1997). These factors make the lack of attention to potential confounds in several included publications worrisome.
Cultural factors also can interact with demographic characteristics to produce distinct sub-cultural response styles (Norris et al., 2001). For example, gendered behavioral norms in Latinx and Latin American cultures, including machismo, caballerismo, and la mujer abnegada, may potentiate distinct response tendencies between men and women (Davis et al., 2011). Work on response style variation within Latinx and Latin American populations is therefore necessary to disentangle cultural and demographic influences. Further investigation is also necessary to determine the extent to which variation between cultures is generated by response style, as opposed to valid cross-cultural and individual differences in target constructs (Smith, 2011; Smith & Bond, 2019). Qualitative, ethnographic, and mixed-methods studies could be useful to elucidate culturally relevant idioms of distress, while continuing quantitative advances will allow these expressions to be related to more traditional measures of psychological functioning.
Toward Improving Psychological Assessment
The few publications that attempted to account for response style biases provided a variety of strategies, including changes to measures themselves and several post hoc statistical methods. Of course, psychological measures assessing content that is not properly culturally adapted can induce bias beyond that of response style if measures do not adequately capture relevant concepts or dimensions (Kirmayer & Ryder, 2016). Setting aside the cultural relevance of content, simple instrument and environmental adjustments may be useful to address response style. Indeed, publications in the current review generally indicated that extreme responding was less likely on scales with more response choices (Hui & Triandis, 1989). Scales with seven ordered response options may best balance conciseness and efficiency against the risk of extreme responding (Harzing et al., 2009; Hoffmann et al., 2013). In addition, acquiescence and extreme responding were heightened in face-to-face interviews compared with paper- or internet-based self-report surveys (Liu et al., 2017), highlighting the importance of survey settings.
Statistical adjustments require that participants’ response styles are measured and then used to alter their scores in some way. There are a variety of approaches in the literature that may be employed to detect differences in response style. Arithmetic methods, such as counting participants’ endpoint responses, offer the benefit of simplicity and face validity. However, such approaches may confound extreme responding with responses that are genuinely high or low. Scales that attempt to directly query response styles have been developed (Crowne & Marlowe, 1960; Greenleaf, 1992; Stöber, 2001), but these are not without controversy, with some scholars discouraging their use in applied settings (McGrath et al., 2010) and others defending their validity (Rohling et al., 2011). Response style has also been gauged by measuring cultural groups’ different probabilities of endorsing survey items (Benítez et al., 2019; Kim et al., 2009; McDonald et al., 2019). There are a wide variety of strategies for operationalizing response styles in a limited literature, making it difficult to identify an approach that is best supported. However, despite controversy surrounding scales measuring specific response styles, we feel that they represent the most direct approach, and therefore, their use is warranted. Given that arithmetic and statistical adjustments may confound response style and genuine responses, conceptualizing response styles as psychological constructs in and of themselves is the most straightforward option presented in the literature.
After data have been collected, there are several potential techniques to account for the impact of response style on analyses. Publications in this review addressed extreme response style using truncated scoring, in which extreme responses are collapsed into less-extreme categories (Bachman et al., 2011; Weech-Maldonado et al., 2008); however, this method did not always reduce response bias (Bachman et al., 2011). Individual items may be combined into “parcels” for analysis, with the assumption that bias on different items will be balanced across item responses within parcels (He et al., 2017). Such procedures have been used in cross-cultural research regarding social axioms (Leung et al., 2011), but may be less suited for multidimensional constructs such as psychopathologies. Alternatively, in some publications, indices of interest were regressed against response-style measures, with the residuals of these regressions used to estimate style-corrected scales, with some evidence of success (Harumi & Church, 2011). Ipsatization, a score standardization process in which an individual’s or a group’s mean value on a range of items is removed from each raw score, was not used in the included publications, but has been suggested as an adjustment method (Fischer, 2016; Geisinger, 1994; Hays et al., 2000). There has been some cross-cultural evidence demonstrating improved consistency with ipsatization (Church et al., 2008); however, removing mean values may obscure genuine differences between groups (He et al., 2017) and precludes any analyses of the remaining scores that rely on covariance (e.g., factor analysis). Ultimately, no one adjustment procedure seems to best account for cultural response styles; a variety of techniques need to be employed in future cross-cultural research.
Clinical Implications
Most of the publications included in this review appeared in the social/personality and cultural psychology literature, with few from the clinical literature. Oversight in the clinical literature is troubling given the lack of population-specific norms on most psychological assessment tools and the real-world implication of misdiagnosis. Correcting for response style bias is necessary not only at the group level to support the internal validity of research studies but also at the individual level of clinical practice to support valid assessments of clients’ psychological well-being. Clients exhibiting extreme or acquiescent response styles may be over- or misdiagnosed with psychological problems, while socially desirable or midpoint responses may obscure symptoms of distress. In the context of the reviewed literature, Latinx and Latin American clients may be at risk of unwarranted diagnosis based on current assessment tools.
Even as assessment-norming projects continue, clinicians attempting to provide care for an individual client face additional challenges. Every individual is a member of numerous cultural groups, and individuals vary in the extent to which they endorse culturally patterned responses (Matsumoto, 2006; Smith & Bond, 2019). Given this within-culture heterogeneity, clinician should assess the extent to which given clients’ responses reflect their cultural backgrounds, perhaps by attempting to directly evaluate response style (Rohling et al., 2011) or using measures of the client’s values orientations, self-construal, or acculturation. Although such measures are more appropriate for assessing group differences, they may provide some indication of the extent to which an individual is representative of group-level responses styles and thus may help contextualize scores from psychological testing (Jayawickreme et al., 2022). To directly assess response style, it may be useful for psychologists to consider extreme, acquiescent, and midpoint responding as part of a continuum rather than as discrete entities (He et al., 2014).
Ultimately, a better understanding of how individuals report their experiences will inform more effective psychological assessments and improve the targeting of clinical treatment. It is possible, and even likely, that response styles will affect findings from not only surveys and Likert-type scales, but clinical and diagnostic interviews as well. While interviews allow for an essential level of nuance that fixed-choice surveys lack, a publication included in our literature review noted that face-to-face survey administration encouraged extreme and acquiescent response tendencies to a greater extent than did paper- and computer-based assessments (Liu et al., 2017). In any case, an awareness of cultural response styles represents another valid tool in a clinician’s proverbial toolkit.
Limitations
Results are tempered by the small number of included studies and intragroup variability. Although we reviewed publications written in Spanish and Portuguese, search terms were only in English. It is possible that a richer body of literature would have been returned with the inclusion of Spanish and Portuguese search terms. In addition, only medical and psychology databases were searched. Given the relevance of this topic to international corporations, publications in business-oriented journals may have been missed. Only one researcher performed screening and extraction, though results were spot-checked by another researcher.
As discussed, previous research has demonstrated variation in response style on the basis of gender, age, and socioeconomic status. The relatively few included studies, however, precluded the comparison of study findings on the basis of more than one or two factors. In this review, we focused on comparing results by cultural/pan-ethnic group, type of measurement instrument, and type of response style. Future reviews, perhaps supported by a greater body of evidence, may be able to better account for additional demographic comparisons.
Conclusion
Cross-cultural psychological research has demonstrated clear variation in response styles between cultural groups. A small but coherent literature across several psychological subfields has further demonstrated differences between Latinx and Latin American populations and non-Latinx White North American populations. We suggest that response styles be assessed as potentially influential constructs in psychological research and clinical practice and that researchers should be specific about the particular cultural groups under investigation. Steps and suggestions for adapting existing measures to account for response style have included scale expansion and post hoc statistical techniques. Such practices have not yet been widely adopted, however, leaving resfigearchers to re-negotiate measurement issues for each study and clinicians to make their assessments using potentially inaccurate tools. Given the importance of psychological assessment in research and practice, developing clear guidance regarding response styles is both warranted and essential.
Supplemental Material
sj-docx-1-asm-10.1177_10731911231194969 – Supplemental material for A Systematic Review of Response Styles Among Latinx Populations
Supplemental material, sj-docx-1-asm-10.1177_10731911231194969 for A Systematic Review of Response Styles Among Latinx Populations by Camille Zolopa, Michelle Leon and Andrew Rasmussen in Assessment
Footnotes
Declaration of Conflicting Interests
The author(s) declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.
Funding
The author(s) received no financial support for the research, authorship, and/or publication of this article.
Supplemental Material
Supplemental material for this article is available online.