Much of the scholarship examining culture in psychology has compared people of the Asian diaspora with people of the Euro west (Vignoles et al. 2016) in attitudes and behavior related to group orientation. When Latin American groups are included, they have generally been deemed high in group orientation (Hofstede 1983; Triandis 1983; Triandis et al. 1986; Triandis et al. 1988; Delgado-Gaitán 1994; and others), although recent work by Vignoles, et al., (2016) found that Latin Americans in their sample, while complexly situated on a 7-factor model of self-construal, as a group, were above average in independence. That study included participants from 5 nations but did not include Costa Rica. Relatedly, in the well-known meta-analysis by Oyserman et al. (2002), Costa Rica and Venezuela were excluded from the Latin American cultures categorized as high in group orientation. No serious theoretical or practical explanation has been offered on the question of why or how either of these cultures would come to diverge so significantly from the rest of the region.

In fact, despite a significant popular narrative framing Costa Rica as one of the “Whiter” (eg: more European) nations in Latin America, the sociological evidence suggests early and regular mixing of racial/ethnic populations there (Perez, 1997). The ongoing presence of indigenous and African descended peoples, combined with the colonial rulers’ attitudes concerning mestizaje (Spanish for the mixing of races), has resulted in a predominantly mestizo populace,Footnote 1 though these influences are largely excluded from national narratives (Fernández Guardia 2005; Lobo and Meléndez 1997). Beyond shaping phenotypes, the history of mestizaje can be expected to have influenced Costa Rican cultural values, likely in the direction of elevated group orientation. Especially given that the primary colonial inputs came from Spain, a nation also of significant African influences (Toasije 2009) and one of the few in Europe typically categorized as high in group orientation (Oyserman et al. 2002).

A closer look at the empirical record amplifies these questions. For example, the single study on which Oyserman et al., (2002) categorized Costa Ricans as low in group orientation was concerned with differences in emotional expression between US and Costa Rican college students (Stephan et al. 1996). Costa Ricans scored lower on the subscale of interdependence that assessed maintaining self-other bonds, whose items the authors describe as emphasizing reciprocity, but not different on the other dimension of interdependence assessed (concern with other’s evaluations). Means were not reported. Concerning independence, Costa Ricans reportedly scored lower in self-knowledge and there was no difference on a measure of self-other differentiation. The authors reported that US participants were more comfortable expressing both independent and, counter to their predictions, interdependent emotions than were Costa Ricans. These results seem more mixed than indicating low group orientation. These findings are also difficult to interpret because there is fair consensus in the literature that individualism and collectivism are best examined as orthogonal, multidimensional constructs that can coexist simultaneously in individuals (Vignoles et al. 2016). The choice of US samples (psychology students at New Mexico State University) also creates ambiguity, since there are a range of subpopulations in the US based on ethnicity and region, some of which are noted for their collective orientations, particularly in the heavily ChicanX southwest (Schwartz et al. 2010). Finally, despite meeting the quality threshold established for inclusion in the meta-analysis, reliability estimates for the interdependence sub-scales on which the difference was reported were marginal at .65 (Oyserman et al. 2002).

Another cross-cultural study conducted with Costa Ricans looked at parenting behavior (Keller et al. 2005). No self-report measures of group orientation were employed but the authors predicted that middle-class Costa Rican parents would score between German middle-class parents and Cameroonian Nso farmers on measures of interdependent parenting practices. In their results, the only significant difference between Cameroonians and Costa Ricans was on the dimension of body contact, with Cameroonian parents showing a greater frequency. Costa Ricans showed more body contact, less object stimulation and more face-to-face interactions than did German parents. The authors interpreted these observations as suggesting a trend toward independence among Costa Ricans, however, given the nature of the evidence, the question certainly remains open.

The history and demographic profile of Costa Rica argues against the likelihood that it would stand out significantly from the interdependence that characterizes most of Latin America. Given the scarcity of empirical work examining this question there, and the fact that existing research has not employed reliable and standardized measures, we believe that further exploration of group orientation in Costa Rica is warranted.

The present study sought to address these shortcomings in the previous research using both hypothesis testing and exploratory approaches. To determine whether employing better established measures would clarify the importance of group orientation in Costa Rican culture, we employed two well-established explicit attitudes measures of group orientation. These were the Scales of Interdependent and Independent Self-Construal (Singelis 1994) and the Communalism Scale (Boykin et al. 1997). Derived from the east-west dichotomy common in much of the literature, interdependent self-construal emphasizes the interconnectedness and group deference often found in eastern cultures, while independent self-construal emphasizes the separateness and uniqueness of the individual often associated with the Euro-West. The accompanying scale has been employed and validated in a variety of European, American, Asian (Markus and Kitayama 1991; Singelis 1994) and less frequently in Latin American and other populations (Oishi 2000; Cross et al. 2011). Scale items refer to respect for authority and deference to group goals. Given the African population inputs in Costa Rica, we also employed the Communalism Scale, a measure of group orientation that is emic to peoples of the African Diaspora. Related to collectivism, communalism is defined as a person’s sensitive appreciation for the interconnectedness among people, social bonds, and the importance of sharing (Boykin et al. 1997). Those high in communalism are said to see themselves as inextricably linked with their social milieu. Scale items include references to the social and interactional dimensions of group orientation, which by reputation may align well with South American forms. The Communalism Scale has shown psychometric stability in a variety of sample groups of the African diaspora including African Americans (Wallace and Constantine 2005), West-Indians and South Africans (Hurley and Hurley 2011), but also in White, White-Latinx and other non-Black populations (Schwartz et al. 2010).

As exploratory assessments, and to evaluate the validity and predictive utility of the two explicit attitudes measures, we also included the Learning Context Scenarios (Sankofa et al. 2005). In cross group research, measures of explicit attitudes suffer the disadvantage of relying on respondents’ subjective theories about the ‘typical people’ who populate the implied comparison group under the implied ‘normal circumstances’ (Schwarz and Oyserman 2011). The likelihood that participants from different groups will answer with reference to systematically different ‘typical people’ and ‘normal circumstances’ is a serious threat to the validity of subsequent comparisons. By contrast, scenario measures simply ask respondents to offer their own assessments of the specific behaviors of a target, exhibited in a specific circumstance. Scenario measurement is also better aligned with the idea of implicit psychological and behavioral tendencies as defined for cultural task analyses (Kitayama and Imada 2010). A cultural tasks analysis contends that individuals learn to embody the priorities of their culture by engaging in cultural tasks that are made available in the cultural context, and that these behaviors become habitual (implicit) over time. Consequently, the behaviors may or may not align well with an individual’s more explicit or declarative beliefs about the self as interdependent or independent (Kitayama et al. 2009).

The Learning Context Scenarios (LCS) present brief descriptions of hypothetical peers whose learning behavior is marked by one of four cultural themes. Respondents are asked to indicate how similar they feel to each peer. Previous research employing the LCS has found a strong preference for communal over peers described as competitive or individualistic among African American students (Marryshow et al. 2005). Boykin et al. (2005) found that the LCS differentiated among Black and White students in the US, with the former reporting significantly more positive attitudes toward the communal peer. White students in the study preferred the individualistic and competitive peers significantly more than did their African descended counterparts.

We examined scores on these measure in a sample of Costa Rican college students and compared them with those of African American and European American college students, two populations whose group orientation status is well established on at least one of the measures employed. For example, European Americans are a well-established and empirically documented standard bearer for cultural individualism (Oyserman et al. 2002). African Americans are the population in which the communalism scale was originally designed and normed (Boykin et al. 1997). Scores on the scale are known to be predictive of various self-reported (Hurley and Hurley 2011; Schwartz et al. 2010) and actual (Hurley et al. 2018) outcomes in African American and other samples.

Method

SampleFootnote 2

Costa Rican participants were 69 female and male students at a large urban public university. Among them, 63 self-identified primarily as Latinx, Hispanic, Spanish or Latin American, 4 identified as Caucasian, White or European American, and 1 each identified as African American, Black or Afro-Caribbean, and Asian, Asian-American. The 6 participants whose primary identification was not Latino/a, Hispanic, Spanish or Latin American did report being of Latin, Hispanic or Spanish heritage when subsequently asked to describe their ethnicity in their own words and so were included in our analyses. Their mean age was 21.96 years (SD = 4.137) with a range of 18 to 39.

The American sample was composed of 63 university students from a large historically black university (HBCU) in the southeastern and 40 European American students from a large public university in the northeastern United States. Of the HBCU students, 59 self-identified as Black, African American, Afro-Caribbean, Black African, or other in the general category, 1 as East Asian, Asian American or Asian-Caribbean, 2 as Latino(a), Hispanic, Spanish, Latin American or of Spanish origin and 1 as South Asian or South Asian American. The 2 participants who identified as Latino(a), Hispanic, Spanish, Latin American or of Spanish origin did report being of Afro-Latino or Hispanic and black heritage when allowed to describe their ethnicity in their own words and were kept for analysis. However, the two participants who identified as South Asian or East Asian identified only with those categories and were excluded from further analyses because they did not meet our criteria of identifying with African heritage. This left a total of 61 participants in our African American sample group. All 40 European American participants described themselves as Caucasian, White or European American in the forced choice format and when allowed to use their own words. The mean age for the US sample was 19.41 years (SD = 1.76) with a range of 18 to 28. Gender was about evenly distributed in all three sample groups.Footnote 3

Instruments

All measures were administered in Spanish to the Costa Rican sample and in English to the U.S. samples. English-Spanish bilinguals translated all measures and scales from English to Spanish. Two translators were used, each of which translated a measure, and then forwarded it to the other for back-translation to English for comparison. As a final step, we employed a professional translator who reviewed and made minor revisions to the Spanish versions.

Scales of Interdependent and Independent Self-Construal

The 24 item self-construal (SC) scales were also included to assess group orientation as interdependent and independent construal of the self (Singelis et al., 1994). Participants responded on a six-point Likert-type scale with options from 1 “completely false” to 6 “completely true”. The scale yields two scores, obtained by calculating the mean of 12 interdependent and 12 independent items.

Communalism Scale

Group orientation was also measured using the 31-item Communalism Scale introduced earlier (Boykin et al. 1997). Participants responded on a six-point Likert-type scale with options from 1 “completely false” to 6 “completely true”. The scale yields a single score, obtained by calculating a mean after omitting filler items and reverse coding as appropriate. The resulting scores range from one to six.

Learning Context Scenarios

Also introduced earlier, the four subscales of the LCS were included (Sankofa et al. 2005). Based on the attitudes and behaviors described in the ~ 90-word scenarios, respondents are asked to indicate how similar they feel to peers described as exhibiting communal, individual, competitive and high verve learning preferences and behaviors.

In the communal scenario, the peer is described as preferring to share ideas and materials with other students. This hypothetical student feels “it is a good idea for students to help each other learn.” The student in the individualistic scenario is depicted as preferring learning and instruction that provides opportunities to achieve via individual accomplishments and for example, as believing that they “perform better on school tasks when working independently”. The student in the competitive scenario is described as preferring to compete with others. This student is characterized as “not liking to get the second highest test score”. Although not a focal construct in this study we did include the high-verve scenario that typically accompanies the others. Verve describes the level, intensity and variability of immediate and environmental sensate stimulation that a person prefers and in which they are optimally able to function (Sankofa et al. 2005). The high-verve scenario describes a student who prefers that the teacher use many different ways to teach. He/she also “enjoys working on several different subjects within a class period” and would “not be bothered if music were playing in the background while she/he was working”. In African American populations, responses to the verve scenario have been positively associated with behavioral variety and also positively correlated with communal orientation (Marryshow et al. 2005). Participants respond to the LCS on a 6-point Likert-type scale with options from 1 “strongly disagree” to 6 “strongly agree”.

Procedures

All participants were surveyed through their universities, either in person on campus or via a link to a web survey. The students gave informed consent and were given course credit or a small cash payment for their participation. When the survey was complete, researchers collected the surveys and then debriefed and thanked the students for their participation.

Results

Scale Reliabilities

Internal reliability for all of the measures are displayed in Table 1. For our Costa Rican participants reliability estimates were α = 0.67 and 0.63 for the interdependence and independence subscales of SC respectively. These are somewhat lower than has been reported in other work (α = 0.73–0.74 & 0.69–.70, respectively; Singelis et.al., 1994), and less than ideal for such an established measure. Estimates were better for the Americans in our sample.

Table 1 Internal consistency estimates, means and standard deviations for Self-Construal, Communalism and LCS
Full size table

The Communalism Scale, and the LCS yielded internal reliability estimates of α = 0.78, and 0.86, respectively for our Costa Rican, participants. Previous work has reported reliability estimates between α = .84–.87 for the communalism scale (Boykin et al. 2005) and α = .71–.90 for the LCS scenarios (Marryshow et al. 2005).

Mean Scores

Mean scores for each scale were compared across the three ethnic groups using univariate ANOVAs with ethnicity as the between-subjects factor. Contrasts among the LCS scenarios were analyzed using repeated measures ANOVA and LSD statistics were computed to compare individual pairings. Table 1 also displays the means and standard deviations for each of the measures included.

Attitude Measures

There were no significant differences among Costa Rican, African American and European American participants on the interdependence subscale, F (2, 165) = .403, p = .669, but Costa Rican participants scored significantly higher in independent self-construal than did our African American or European American participants,Footnote 4F (2, 165) = 10.34, p < .001, η2 = .11; p = .001, d = .76, p < .001, d = .70. The mean communalism score for the Costa Rican sample was significantly lower than that for either US group, F (2, 165) = 8.891, p < .001; η2 = .10; p = .002, d = .55, p < .001, d = .34, respectively.

Scenario Measures

All three groups reported feeling highly similar to the communal peer, but pairwise comparison indicated that European Americans felt significantly less similar to the communal peer, F (2, 165) = 2.71, p = .070; η2 = .03; p = .040, d = .41; p = .050, d = .40, than did Costa Ricans and African Americans, who were not significantly different in rating that peer (p = .980). In addition, repeated measures analyses indicated an interaction between cultural theme and ethnicity on participants’ ratings for the communal and individual peers, F (2, 165) = 3.27, p = .040. The difference was significant in favor of the communal peer for Costa Rican (M = .824, SE = .238, p = .001, η2 = .14) and African American (M = .393, SE = .262, p = .050, η2 = .07) participants, but reversed for European Americans (M = −.175, SE = .311, p = .620). Figure 1 depicts the pattern of ratings for similarity to the four hypothetical peers.

Fig. 1
figure 1

Similarity Ratings for the LCS Scenario peers by ethnicity

Full size image

There was a main effect of cultural theme but no main or interaction effects of ethnicity, indicating that participants in all three groups rated themselves significantly less similar to the competitive peer than to the communal, F (1, 165) = 22.46, p < .001, η2 = .25, or individualistic peers, F (1, 165) = 35.287, p < .001, η2 = .18. The majority of Costa Ricans and European Americans rejected the competitive peer while African Americans rated the competitive peer just above that threshold and reporting feeling significantly more similar to that peer than did Costa Ricans, F (2, 165) = 7.21, p = .001, η2 = .08; p < .001, d = .66.

Concerning the high verve peer, African Americans and Costa Ricans both rated the high-verve peer well above the threshold for acceptance. European Americans rated themselves significantly less similar to the verve peer than did African American participants, F (2, 165) = 2.09, p = .012, η2 = .03; M = −.64, p = .043, d = .35, with a slight majority of them actually rejecting the high verve peer (rating that peer below the midpoint of the scale).

Correlations

Table 2 displays the correlations among measures. Communalism scores were positively correlated with interdependent SC scores for all three groups at, r = .48 p < .001; r = .35 p = .003; r = .57 p < .001, for our Costa Rican, African American and European American participants respectively. Communalism (r = .243; p < .001), but not interdependent SC (r = .060; p = .240), was correlated with similarity ratings for the communal peer in the whole sample. Neither was significantly correlated with ratings for the communal peer in the Costa Rican sample alone (r = .17 p = .080; r = .07 p = .276, respectively), despite conceptual relationships. Independent SC was, perhaps counterintuitively, positively correlated with ratings for the communal peer in the whole sample (r = .14, p = .050), likely due to their high correlation (r = .62, p < .001) among European Americans. Overall the convergence between the attitude measures and the scenarios was poor, especially for Costa Rican respondents.

Table 2 Correlations by sample group
Full size table

Discussion

Both subscales of the self-construal scale yielded unfortunately low internal reliability estimates among our Costa Rican participants. This is surprising for such a well-travelled measure, though other work has also reported inconsistencies (e.g. Christopher et al. 2012). Bearing this in mind, the fact that our three sample groups achieved similar mean scores on the interdependence subscale combined with the fact that those scores were similar to means reported for the (deemed highly interdependent) Asian participants in work by the scales authors (Singelis 1994Footnote 5) does not jibe well with conclusion that Costa Ricans are low in group orientation. On the other hand, our Costa Rican participants did achieve significantly higher mean scores on the independence subscale than did both groups of Americans, and their mean scores were also high compared to means reported for the (deemed highly independent) European descended participants in work by the scales authors (Singelis 1994). Given the consensus in the literature that people can be high in both group and individual orientations, that observation could be taken as affirming the assertion of elevated individualism in Costa Rica. As in earlier work with Costa Rican samples however, the marginal psychometric performance of both subscales of the self-construal complicates the question.

Reliability estimates for the Communalism Scale were strong in this study. For Costa Ricans they were similar to those for American participants in this, and in other studies (eg: Boykin et al. 1997; Schwartz et al. 2010). Recall that scores for Costa Ricans on the Communalism Scale were lower than those for the American participants in our sample, and lower than the established norms for (deemed highly communal) African Americans (Boykin et al. 1997). Taken together the pattern of scores does not support our contention that the status of group orientation in Costa Rican culture could be settled by using explicit attitude measures that are well established and psychometrically vetted, or emic to the African Diaspora. Instead they may be taken as contributing additional ambivalent support for the inference that Costa Ricans may be comparatively low in group orientation and high in individualism.

On the other hand, the findings from our scenario measures directly conflict with that conclusion. Reliability estimates for the LCS were good in our Costa Rican sample and also similar to those reported in several previous studies (Boykin et al. 2005; Marryshow et al. 2005; Sankofa et al. 2005). Given that the measure is comparatively less well known, this is an encouraging indication that the values we obtained among Costa Ricans are interpretable.

Some telling group differences emerged. In ratings of similarity with the communal peer, the typical pattern of separation between African Americans and European Americans reported in other work did manifest. Costa Ricans too gave high ratings for communalism and low ratings for individualism when these were represented in the specific behaviors of a peer. Further, Costa Ricans also scored with African Americans in feeling more similar to the communal than to the individual peer, a trend that was reversed for European Americans. On their own these observations support our theory-based prediction that Costa Ricans would be relatively high in group orientation and low in individualism. But, what are we to make of the difference between scenario ratings and the attitude measures? Perhaps it signals a larger issue concerning role of context and anchoring in the use of self-report measures for cross group comparisons.

The scales of Self-Construal and the Communalism scale, like other explicit attitudes measures, ask about attitudes and beliefs, largely in the abstract, employing phrases like “it is important to me to…”. Given that the things named will have some importance to most people, respondents are forced to make a judgment based on subjective theories (Schwarz and Oyserman 2011) eg: self-generated reference standards. Such standards are bound to be grounded in the mores of the society or culture in which respondents have been socialized. That is, the phrase “compared to other people I know” is implied. In cross group measurement, then, the fact that we “know” different people becomes important.

For example, in a society that prioritizes group orientation, college students may tend to be, or be perceived to be, comparatively independent/individualistic,Footnote 6 and as such may receive regular messages (accusations) to that effect, for violations of community standards and despite that by the anchoring standards of another society, their attitudes and behavior may be quite interdependently/collectively oriented.Footnote 7 We can expect such persons to rate themselves lower on most explicit attitude measures of group orientation than they would if they evaluated their same attitudes and beliefs with reference to a more individualism centered standard.

An advantage of our scenarios then, is in having had respondents rate themselves in relation to a pre-anchored standard - the specific relevant behaviors of a peer in a specific context. Respondents only needed to decide whether they would likely feel and do the things described, in the specific situation described. To the extent that they would, they can rate themselves as similar to the peer. If we believe that the behaviors described are communal (or competitive, etc.) then we must accept that differences across our sample groups represent meaningful differences in the orientations measured.Footnote 8 The observed correlation between Communalism scores and ratings for the communal peer for our American but not our Costa Rican participants is consistent with this idea.

Our scenario measurement is also better aligned with the idea that implicit behaviors become habitual and so less pliable than peoples’ perceptions of their own attitudes and intentions relative to others (Kitayama et al. 2009). Scenarios also likely take advantage of the habitual nature of such behaviors in so much as our participants, as current students, would have ready access to concrete instances of their own related behavior as a reference.

Conclusions, Limitations and Future Directions

We began by observing that characterizations of Costa Rican culture as comparatively low in group orientation are under justified in the empirical record. The nature of our questions led us to pursue them using multiple measurement approaches. We therefore tested the hypothesis that employing better established measures might clarify (with evidence supporting or undermining) this view. We also took the more exploratory step of including an explicit attitude measure emic to and well explored among cultures of the African Diaspora, in case Costa Rican forms might be better captured on a measure emphasizing the social and interactive elements of group orientation. Including the scenario measures allowed us to explore the influence of anchoring vs. subjective theories on participants’ patterns of response. These scenarios also allowed us to take a structure-oriented look at the relationships among participants’ group orientation and other dimension of culture (Matsumoto and Juang 2016).

All in all, we are inclined to trust more the observations of our scenario measures which performed better psychometrically, are clearly anchored, and are less subject to participants’ subjective theories. Those measures suggest high levels of group orientation and a rejection of individualism among Costa Ricans, at least in academic settings. Other activity settings, such as work and family should be examined in future studies.

Within that, we also found that the two measures created for and normed in African American populations yielded better psychometric performances among Costa Rican participants than did the those created with reference to cultures of the Asian Diaspora. This bears on our suggestion, based on socio-anthropological evidence, that manifestations of group orientation in Costa Rica might be more similar to those of the African than the Asian Diaspora. Relatedly, Costa Ricans were not different from African Americans in similarity ratings for the verve peer, a co-occurrence that also lends mild support for locating Costa Rican culture in proximity with cultures of the African Diaspora. It is also noteworthy considering that it may be the first documentation of a high-verve orientation in any Latin American cultural group.

The more exploratory elements of this study allowed for useful breadth but limit our ability to unpackage these observations. They raise a number of notions and questions that should be pursued in further empirical research. We look forward to additional work pursuing these and related issues further in Costa Rica and among other Latin American cultural groups. Another limitation of this study is in our choice to employ college samples. Our point about anchoring via subjective theories, while a general concern with cross cultural use of attitude measures, seems likely to have been intensified for our Costa Rican participants, and perhaps in the other groups as well. Additional work employing community samples could help to clarify this issue.

This exploratory study pushes back against the conclusion that Costa Ricans are highly individualistic and low in group orientation. It also raises larger questions about the structuring of self-report measures of group orientation and about the measurement of culture in general. We hope this work encourages and contributes momentum to the empirical study of culture in Costa Rica and in all of Latin America.