Development of a Bi-Dimensional Simpatía Scale for Use with Mexican American, Puerto Rican, and Cuban American Adults

Abstract

Background:

Through its influence on social interactions, simpatía may have a wide-ranging influence on Latinx health. Simpatía—which does not have a direct English translation—refers to being perceived as likeable, pleasant, and easy-going. Research to investigate the influence simpatía on Latinx health is limited; however, it is due to a lack of options for measuring simpatía among diverse Latinx populations.

Objectives:

The goal of this research was to develop a bilingual, survey-based simpatía scale for use among ethnically diverse Latinx adults in health-related settings.

Methods:

Data were obtained through a telephone survey data of 1,296 Mexican-American, Puerto-Rican, and Cuban-American adults living in the U.S. Interviews were conducted in English and Spanish. Exploratory factor analysis, item response theory analysis, confirmatory factor analysis, and computation of estimates of internal consistency reliability were conducted to inform the development of the final simpatía scale.

Results:

Results indicate that the final, 9-item, simpatía scale has high internal consistency (α = 0.83) and measurement invariance among Mexican-American, Puerto-Rican, and Cuban-American adults. Two dimensions were identified, as indicated by a perceptions subscale and a behavior subscale. Cuban Americans were found to have the highest simpatía scores, followed by Puerto Ricans and Mexican Americans.

Discussion:

Culture is often identified as a powerful potential influence on health-related behaviors, but measures are often not available to assess specific cultural traits. By developing a new tool for measuring simpatía, this research advances opportunities for understanding and promoting Latinx health.

When providing care to U.S. Latinxs, health professionals are often encouraged to attend to simpatía as a core Latinx cultural value (Flores, 2000; Juckett, 2013). Simpatía does not have an English counterpart. However, according to Triandis, Marin, Lisansky, and Betancourt (1984), it can be approximately translated as “a permanent personal quality where an individual is perceived as likeable, attractive, fun to be with, and easy-going … and seems to strive for harmony in interpersonal relations” (p. 1363). A person who is simpatico/a expresses emotional intelligence, is courteous to others, and avoids making other people feel embarrassed, uncomfortable, or disrespected (Triandis et al., 1984). Simpatía has been associated with Latinx culture. However, as with other cultural constructs, Latinx individuals may vary in their endorsement of this specific cultural value. Simpatía is an individual-level personal characteristic, but it also informs how individuals interact with other people at the interpersonal and group levels. From this perspective, simpatía is an individual-level cultural trait that guides both one’s own behaviors and the expectations that one may have for others’ behaviors during interpersonal interactions.

Simpatía may have a wide-ranging influence on Latinx health. Culture may be defined as a shared set of normative behaviors, values, beliefs, and assumptions that guide individuals’ interpretations of social situations and ensuing behaviors (Triandis, 1995). As a cultural script (Triandis et al., 1984), simpatía may inform how Latinxs interpret and behave in social situations, thereby influencing the extent to which they engage in specific health-related behaviors. Nevertheless, empirical research to explore the influence of simpatia on Latinx health is currently constrained by limitations in the tools available for measuring this potentially important cultural construct. The goal of the current research was to develop a psychometrically sound, bilingual, survey-based measure of simpatía that could be easily administered to diverse Latinx populations in health settings.

Although quantitative investigations of simpatía are limited, at least four studies suggest that simpatía influences health-related attitudes, behaviors, and outcomes among Latinx populations. Endorsement of simpatía has been associated with higher substance abuse treatment satisfaction among Latinx and non-Latinx patients undergoing methadone treatment (Griffith, Joe, Chatham, & Simpson, 1998); reduced likelihood that ethnically diverse sixth-graders had smoked a cigarette (Johnson et al., 2006); increased sexual abstinence among Latinx boys and girls (Ma et al., 2014); and increased age at first intercourse for Latina girls (Ma et al., 2014). Simpatía has also been associated with a 36% reduction in the odds of having hypertension among Latino men; although no association was found among Latina women—potentially due to gendered differences in social behavior (Merz et al., 2016). Collectively, the results of these studies suggest that simpatía may influence health behaviors that include a social interaction and/or self-presentation component.

Researchers have observed additional ways in which simpatía may influence Latinx health. For example, Flores (2000) asserted that Latinx patients who endorse simpatía may prefer to obtain care from health professionals with a positive demeanor and would be less likely to share health information, follow treatment recommendations, or return for medical visits when health professionals are not perceived as simpático/a. In a qualitative study, Jones (2015) found that hospitalized Mexican-American patients were more likely to trust their nurses when the nurses exhibited behaviors consistent with simpatía, such as smiling, being friendly, and engaging in pleasant conversation. In contrast, some patients avoided and were reluctant to request or receive assistance from nurses whom they perceived to be rude or uncaring (Jones, 2015). Simpatía may also motivate Latinx patients to seek care from health professionals whom they already know, even if such professionals are less competent than those outside their social network (Triandis et al., 1984). In nonclinical settings, simpatía may discourage Latinx patients from sharing sensitive health information with family members, such as sexual orientation or HIV status, to avoid conflict or causing discomfort (Mason, Marks, Simoni, Ruiz, & Richardson, 1995). Simpatía may also play a role in health research by invoking higher study consent and no-show rates, as refusals may be considered more disrespectful than failing to participate in future research activities (Marín & Marín, 1991).

Although simpatía may have a broad influence on Latinx health, health-related research is constrained by the lack of a psychometrically sound, survey-based measure of simpatía that can be quickly and easily administered to general adult Latinx populations in a variety of health-related settings. To date, researchers have used different methods for assessing simpatía, making it difficult to make comparisons across studies. At least two groups of researchers have focused on the social interaction components of simpatía, which are at the core of this cultural trait. In the first of these studies, Triandis and colleagues (1984) found that Latinx U.S. Navy recruits were more likely to report reactions to social situations that were congruent with simpatía than non-Latinxs. More recently, Ramírez-Esparza, Gosling, and Pennebaker (2008) observed that bilingual adults were more likely to exhibit nonverbal expressions of simpatía when speaking Spanish than when speaking English. Both studies used methods that creatively assessed social interactions and may be appropriate for in-depth examinations of simpatía. However, these approaches are impractical for use in the contexts in which many health-related studies conducted, wherein a short, survey-based measure of simpatía would likely be most useful for assessing simpatía as one of a number of variables of interest.

Of the three survey-based measures of simpatía, the most widely used measure is a 17-item scale developed by Griffith and colleagues (1998) for use with Latinx patients in substance abuse counseling. Griffith et al. (1998) observed high internal consistency reliability (α = .80). Though, as 10 items queried attitudes about counselors, only about half of the participants were Latinx, and few participants spoke Spanish; this scale was not designed to measure simpatía as a generalized personal trait with Spanish-speaking populations or in noncounseling settings. An adapted version of the Griffith scale was used in the national Study of Latinos (SOL), for which Merz and colleagues (2016) found measurement invariance across interviews conducted in English and Spanish and good internal consistency reliability among a large sample of Latinxs (α = .75). However, it is unclear how the Griffith scale was adapted, and the SOL study items have not been published, which makes it impossible for researchers to make their own determination of face validity. Johnson et al. (2006) also developed a scale to evaluate simpatía. This scale was developed for use with adolescents, measures a combined construct of simpatía and saving face, and consists of only four items with limited content validity, which yielded low internal consistency reliability (α = .62). Thus, this scale may not be appropriate for measuring simpatía as an individual-level trait among a general population of Latinx adults.

Existing simpatía scales have several additional limitations. First, pretesting and thorough psychometric analyses have either not been conducted or reported, making it difficult for other researchers to make informed decisions when selecting a simpatía measure. Second, with the exception of the SOL study (Merz et al., 2016), previous studies have focused on unique populations for which participants were recruited from a single Latinx subgroup; yet, Latinx ethnic heritage was not described and/or psychometric data were not reported separately by Latinx subgroup. As a consequence, it is difficult to predict whether existing measures will have adequate validity and reliability when administered to other Latinx populations. Lastly, there are few existing measures of simpatía, and these measures focus on different aspects of this complex, multidimensional construct. Researchers need more options to choose from when selecting a measure of simpatía to study the aspects that they believe to be most relevant to their populations and outcomes of interest. For these reasons, the goal of this study was to develop a brief, psychometrically sound measure of simpatía that could be easily administered through surveys of ethnically diverse, general adult Latinx populations in health-related settings. The Latinx population is the largest and second-fastest growing racial/ethnic minority group in the U.S. (Flores, 2017). This research was conducted using data from a bilingual telephone survey of Mexican-American, Puerto-Rican, and Cuban-American adults who represent three of the largest Latinx subgroups and constitute approximately 77% of the U.S. Latinx population (Flores, 2017).

Methods

Participants

Data were collected between April and November 2016 through a telephone survey of 1,296 Mexican-American, Cuban-American, and Puerto-Rican respondents as part of a study of cultural influences on survey responding. Participants were recruited by randomly selecting numbers from a list purchased from a commercial sampling vendor of landline and cellular telephone numbers associated with addresses in the U.S. or Puerto Rico for individuals with 12 years of education or less and a household income of $25,000 or less to improve the efficiency of locating eligible participants (Pasek, Jang, Cobb, Dennis, & Disogra, 2014; Valliant, Hubbard, Lee, & Chang, 2014; West, Wagner, Hubbard, & Gu, 2015). Participants’ education and income levels were also directly assessed through the telephone survey, and only these latter data were used in data analyses. Eligible participants were between the ages of 18 and 90, spoke English or Spanish, and self-identified with one of the three targeted Latinx subgroups. For the purposes of the parent study, participants were also stratified (50%/50%) by their tendency to acquiesce using a 20-item screener (Lee, Davis, Resnicow, & Johnson, 2017). The response rate was 21.3% (American Association for Public Opinion Research [AAPOR], 2017; Response Rate 3).

Data Collection

Informed consent was administered over the telephone immediately after eligibility screening. Participants who agreed to participate subsequently completed an approximately 35-minute, computer-assisted, telephone interview in English or Spanish, during which the interview questions were orally administered without visual aids. All interviews were conducted by a team of professional, bilingual interviewers employed by a national telephone survey company and located at call centers in the Los Angeles area. In addition to their previous interviewer training and experience, the interviewers participated in a one-day training with members of the study team to prepare them to administer the survey. Interview language was determined during eligibility screening using responses to three language use items adapted from the National Latino and Asian American Study (Alegria, M., Center for Multicultural Mental Health Research at Cambridge Health Alliance, & Takeuchi, 2002) and confirming whether or not participants were comfortable continuing in the indicated language. When preexisting translations were unavailable, study materials were developed in English, translated into Spanish by a professional translation company specializing in Spanish dialects, and reviewed and adjusted by bilingual members of the study team. Alterations were concurrently made to the English text to enhance consistency in meaning across the two languages. The research team was ethnically diverse and included several members who were bilingual, Latinx, and/or had extensive experience working with diverse Latinx populations in the U.S. and abroad. A $20 gift card was mailed to participants who completed the survey. This study was approved and monitored by a university-affiliated institutional review board.

Measures

Simpatía.

The simpatía items were pretested using cognitive interviewing, which is a method for identifying potential problems with respondents’ processing of survey items (Beatty & Willis, 2007). Two pretests were conducted in Chicago, during which an approximately 90-minute, audio-recorded, cognitive interview was administered in English or Spanish using a semi-structured interview guide. Consistent with standard cognitive interviewing practice (Beatty & Willis, 2007), probes were used to explore participants’ comprehension, response formation, and response processes. Prescripted probes allowed for direct probing about aspects of questions that were anticipated to be potential sources of error (Willis, 2005), while unscripted probes were primarily used to improve content validity by delving deeper into ways in which the items accurately or inaccurately captured how simpatía influenced participants’ thoughts and behaviors.

Pretest 1.

Ten items were pretested as part of a pilot study examining cultural influences on diabetes management among Mexican Americans. A convenience sample of 33 Mexican-American adults with type 2 diabetes was recruited from a university-affiliated medical practice, a community-based clinic, and a diabetes intervention trial. One item was adapted from a measure of independent versus interdependent self-imaging (“Even when I strongly disagree with someone, I avoid an argument”; Singelis, 1994). Based on a review of the simpatía literature, nine items were created by the study team, who possessed expertise in survey methodology, public health, and healthcare, as well as experience working with Latinx populations. The results indicated that the items were generally well understood, though several issues arose. The most pervasive issue was that respondents held in mind different referent groups (e.g., friends, family, coworkers, strangers) both across respondents and within respondents across questions. Transcripts of the interview recordings indicated that a few respondents had difficulty mapping their answers to the response scale, as they made comments indicating that their responses should be selected from one end of the response scale but chose answers from the opposite end of the scale. The interviews also elicited comprehension problems with specific items. Of the 10 pretested items, the data indicated that six items needed revisions and four items were unrepairable and needed to be dropped. For all 10 items, respondents’ comments provided valuable information about how simpatía was enacted during social interactions, which was used to create new items to enhance content validity.

Pretest 2.

Data were collected with a convenience sample of 61 Mexican-American, Puerto-Rican, Cuban-American, and non-Latinx White adults. The 20 items pretested included revised items from Pretest 1 and additional items created by the study team to improve content validity. To remedy issues identified in Pretest 1, the response scale was changed, and the items were divided into two sets. The introductory text for each set directed participants to specifically think about strangers (“think about times when you are talking in public with people who you don’t know very well, such as people who work at the post office, the electric company, the library, the grocery store, or the bank”) or friends and family (“think primarily about friends and family that you know well”). Periodic reminders were also added to encourage participants to focus their thinking on the intended referent group. However, the data indicated that respondents often thought about people who were not in the referent group, despite these instructions and reminders. In addition, some Spanish-speaking participants had difficulty mapping their answers onto the proffered response options (“Not important, not very important, somewhat important, very important”). In particular, participants seemed to struggle when distinguishing between “not very important” and “somewhat important,” which suggested that these labels were more distinct in English than in Spanish. To address this problem, the response options were changed to “not important, a little important, important, and extremely important” during fielding, which appeared to reduce response mapping difficulties in subsequent interviews. Although, some Spanish-speaking respondents said, “very important” when choosing “extremely important,” implying that the “very” wording was more familiar. Of the 20 items pretested, seven items needed minor wording changes, five items were recommended for deletion, and eight items had no major problems and were recommended for retention without revision.

Telephone survey items.

The telephone survey included 16 simpatía items. Fifteen items were revised from Pretest 2, and one item was newly created by the study team. To address problems identified in Pretest 2, two global changes were made. Language was added to each item stem to remind participants to think about the intended referent groups, as well as preserving the introductory language and referent group reminders used in Pretest 2. In addition, the response scale labels were revised to better match the question content. As research indicates that response labels that are tailored to the construct being assessed may enhance validity (Saris, Revilla, Krosnick, & Schaeffer, 2010), the items were divided into two sets to pair them with more appropriate response labels: (a) “Not important, a little important, important, and very important”; and (b) “Not likely, a little likely, likely, and very likely”).

Other Measures

Other variables assessed included age, gender, education, income, Latinx ethnicity, nativity, and years lived in the U.S. acculturation was measured using a 27-item version of the Acculturation Rating Scale for Mexican Americans II (ARSMA-II; Cuellar, Arnold, & Maldonado, 1995), which, based on the study team’s expertise in survey methodology, was slightly adapted by adding one item to balance the content of the two subscales, revising the wording for several items, and changing the response scale labels. These changes were made to improve question comprehension, better match the response scale labels to the question content, and facilitate telephone administration. The ARSMA-II contained two subscales:

1. a Latinx orientation subscale (LOS) exploring use of Spanish and identification and interaction with the people and culture from each participant’s Latinx subgroup (15 items; α = .86); and

2. a non-Latinx White, or Anglo, orientation subscale (AOS) querying respondents’ use of English and identification and interaction with non-Latinx White people and culture (12 items; α = .89).

Data Analysis

SAS® was used to compute means, standard deviations, and alpha statistics for the full sample and relevant subgroups. Several R packages were used: “mirt” (Chalmers, 2012) for item response theory (IRT) analysis to examine the discrimination of simpatía items; “stats” for exploratory factor analysis (EFA) to inform the allocation of items with reasonable discrimination to similar factors; and “lavaan” (Rosseel et al., 2018) for confirmatory factor analysis (CFA) to test the models informed by EFA. Additionally, multigroup CFA (MGCFA) was conducted to ascertain measurement invariance across selected subgroups, such as Latino-specific ethnic subgroups (Mexican Americans, Puerto Ricans, vs. Cuban Americans), gender (male vs. female) and interview language (English vs. Spanish). In MCGFA, full invariance was assessed from configural, metric (weak), scalar (strong), and strict invariance models.

Results

Characteristics of the sample are shown in Table 1. Most study participants were female (76.5%), Spanish speaking (84.2%), and born outside the mainland U.S. (83.5%). Participants had a mean age of 59.5 years, affiliated more with Latinx than Anglo culture, and were lower income (58.0% had a household income of less than $20,000). Most participants (62.1%) had a high-school-level education or less.

Table 1.

Telephone survey participant characteristics (n = 1,296)

Age in years M (SD)	59.5 (18.1)
Gender n = female (%)	990 (76.5)
Ethnicity n (%)
Mexican American	446 (34.4)
Cuban American	424 (32.7)
Puerto Rican	426 (32.9)
Born in the mainland U.S. n (%)	212 (16.5)
Not born in the mainland U.S. M years lived in the U.S. (SD)	33.3 (16.9)
Interview language (Spanish) n (%)	1,088 (84.2)
Acculturation M (SD)
Latinx Orientation Subscale Score	3.3 (0.5)
Anglo Orientation Subscale Score	2.2 (0.6)
Education n (%)
12th grade or less, no diploma	474 (37.5)
High school graduate or equivalent	311 (24.6)
Some college or technical/vocational school	236 (18.7)
4-year college degree	151 (12.0)
Graduate degree	91 (7.2)
Annual household income n (%)
Less than $20,000	552 (58.0)
$20,000-$39,999	239 (25.1)
$40,000-$59,999	87 (9.2)
$60,000 or greater	73 (7.7)

Note. M = mean; SD = standard deviation.

Initial estimates of internal consistency reliability indicated that 3 of the 16 simpatía items should be dropped from the scale. These items tapped into expressing negative thoughts and emotions (e.g., “When talking with people you don’t know well, how likely are you to tell them when you disagree with them?”), had low correlations with the remaining items, and did not form a strong subscale. As a result, these items were removed from subsequent analyses.

Results from IRT analysis (data not shown) indicated that when using the 4-point response scale, the second category, P2 (“A little important/likely”), was often indistinguishable from its adjacent category, P3 (“Important/likely”; e.g., Q2, Q4, Q9). In addition, the thresholds between the second category and its adjacent categories were estimated illogically (e.g., Q8 where the threshold between P1 and P2 was higher than that between P2 and P3). Hence, the second and third response categories were collapsed into one, and item discrimination was examined again using three response categories. This recoding improved item discrimination. However, this revised analysis indicated that four items (Q10-Q13) showed less than moderate discrimination (i.e., estimates less than 1; Baker & Kim, 2017). Hence, these four items were removed from further consideration for the simpatía scale.

EFA conducted with the remaining nine items indicated that two factors explained a larger percentage of the total variance than a single factor (44.0% vs. 35.5%) as seen in Table 2. A 3-factor model only explained a slightly larger percentage of the variance (46.8%) and left some factors delineated with too few items. Under the 2-factor model, five items (Q2, Q3, Q4, Q8, Q9) were associated with Factor 1, and four items (Q1, Q5, Q6, Q7) with Factor 2. These item groupings corresponded with a conceptualization of simpatía as having two components: Factor 1 represents the importance of engaging in simpatía-consistent behaviors toward others (i.e., avoiding causing discomfort in others), while Factor 2 represents the importance of being perceived as simpático/a by others (i.e., being perceived as friendly, fun, likeable, and easy to get along with). These two subscales were henceforth labeled as the behaviors subscale (Factor 1) and the perceptions subscale (Factor 2), respectively.

Table 2.

1-factor and 2-factor exploratory factor analysis with no rotation (n = 1,296)

	1-Factor EFA Factor Loadings	2-Factor EFA Factor Loadings
	Factor 1	Factor 1	Factor 2
Q1	0.663	0.274	0.669
Q2	0.552	0.590	0.211
Q3	0.629	0.549	0.343
Q4	0.572	0.606	0.225
Q5	0.544	0.126	0.650
Q6	0.608	0.222	0.642
Q7	0.703	0.353	0.635
Q8	0.463	0.546	0.129
Q9	0.591	0.643	0.224
% Variance	35.5%	22.2%	21.8%
% Cumulative Variance	35.5%	44.0%

When subsequently testing the 2-factor model with CFA, as seen in Table 3, the factors loaded onto the respective items well with reasonably high factor loadings. The behaviors subscale loadings were close to 0.7, while the perceptions subscale had slightly lower loadings ranging from 0.539 to 0.670. The subscales were positively correlated at 0.68. Overall, the data fit this model very well as indicated by various model fit statistics (CFI = .981; TLI = .973; RMSEA = .043 (90% CI [0.033, 0.054]); SRMR = .03). When testing the comparability of this factor structure across the three Latinx subgroups through MGCFA, the analysis indicated strict measurement invariance of the selected 2-factor scale (χ² = 208; df = 124; CFI = .972; RMSEA = .041 (90% CI [0.031, 0.051]; SRMR = .047). This means that the same factor structure held across the three Latinx subgroups where the reasonably invariant factor loadings, intercepts, and error variance apply. The same was true for MGCFA by gender (strict invariance model: χ² = 128; df = 75; CFI = .982; RMSEA = .034 (90% CI [0.024, 0.044]; SRMR = .039) and by interview language (strict invariance model: χ² = 196; df = 75; CFI = .960; RMSEA = .052 (90% CI [0.043, 0.061]; SRMR = .041).

Table 3.

2-factor model confirmatory factor analysis (n = 1,296)

	Factor loading (Standardized)		Error Variance (Standardized)
	Estimate	SE	Estimate	SE
Factor 1 – Behaviors Subscale
Q2	0.631	.022	0.601	.028
Q3	0.663	.021	0.561	.028
Q4	0.636	.022	0.595	.028
Q8	0.539	.025	0.710	.027
Q9	0.670	.021	0.550	.028
Factor 2 – Perceptions Subscale
Q1	0.726	.019	0.474	.027
Q5	0.627	.022	0.607	.027
Q6	0.668	.020	0.553	.027
Q7	0.735	.018	0.460	.027
Subscale 1 ~ Subscale 2	p = .681 (SE =0.025)

Note. SE = Standard error.

The final, 9-item simpatía scale had high internal consistency reliability for the full sample (α = .83), as well as within each of the three Latinx language and ethnic subgroups (α = .81-.84) as seen in Table 4. Alpha coefficients for the subscales were also high and comparable across each ethnic subgroup. Final item wording is presented in Table 5.

Table 4.

Simpatía scale alphas by subgroup (n = 1,296)

		Ethnic Subgroup			Interview Language
	All Respondents	Mexican Americans	Puerto Ricans	Cuban Americans	English	Spanish
Final scale (9 items)	0.83	0.84	0.83	0.82	0.81	0.83
Perceptions Subscale (4 items)	0.79	0.81	0.78	0.77	0.83	0.77
Behaviors Subscale (5 items)	0.77	0.76	0.76	0.77	0.70	0.78
Full scale before IRT analysis (13 items)	0.82	0.83	0.81	0.79	0.81	0.81

Table 5.

Item wording and item-total correlation coefficients for the final, 9-item simpatía scale (n = 1,296)

English Instructions: We’d like to know about your interactions with other people. For these questions, think about times when you are talking in public with people whom you don’t know well, such as people who work at the post office, the electric company, the library, or the grocery store. When interacting with these kinds of people, tell me how important the following things are to you:
Spanish Instructions: Nos gustaría saber sobre sus interacciones con otras personas. Para estas preguntas, piense acerca de las veces en que habla en público con personas que no conoce muy bien, por ejemplo, personas que trabajan en la oficina de correos, la compañía eléctrica, la biblioteca o el supermercado. Dígame qué tan importante son las siguientes cosas cuando interactúa con este tipo de personas:
Each ítem begins with “When talking with people you don’t know well, how important is it to you…” (Cuando habla con personas que no conoce bien, ¿qué tan importante es…)
Item Number		Item-Total Correlation
Q1.	… that they think you are friendly? (… para usted que piensen que usted es amigable?)	.586
Q2.	… to avoid arguments? (… para usted evitar discusiones?)	.511
Q3.	… to say things in a positive way? (… para usted decir las cosas de una manera positiva?)	.579
Q4.	… to avoid saying something that would embarrass them? (… para usted evitar decir algo que les avergonzara?)	.518
Q5.	… for them to think you are fun to be around? (… para usted que piensen que es divertido estar con usted?)	.477
Q6.	… that they like you? (… para usted caerles bien?)	.538
Q7.	… that they think you’re easy to get along with? (… para usted que piensen que es fácil llevarse bien con usted?)	.621
Q8.	… to always treat them with respect? (… para usted tratarlas siempre con respeto?)	.424
Q9.	… to avoid hurting their feelings? (… para usted evitar lastimar sus sentimientos?)	.541
Perceptions Subscale = Q1, Q5, Q6, Q7 Behaviors Subscale = Q2, Q3, Q4, Q8, Q9
Response options: Is this not important, a little important, important, or very important? (¿Esto no es importante, es un poco importante, es importante, o es extremadamente importante?)
Combine the middle two response options for analysis.

Simpatía scores were not significantly correlated with Anglo orientation subscale scores (r = .05, p = .06); yet, simpatía was significantly but weakly correlated with the Latinx orientation subscale (r = .11, p < .0001). Though the differences were small, mean simpatía scores in bivariate analysis were significantly different across the three Latinx subgroups (ρ = .02). Mexican Americans had the lowest simpatía score (M = 2.34, SD = 0.40), followed by Puerto Ricans (M = 2.39, SD = 0.38) and Cuban Americans (M = 2.41, SD = 0.38). However, only Cuban Americans (b = .07, SE = .04; p = .04) had significantly higher simpatía scores than Mexican Americans when sociodemographic variables were added to the model (model p = .0002). In the multivariate model, simpatía was positively associated with participants’ Latinx orientation subscale scores (b = .09, SE = .03; p = .001), inversely associated with female gender (b = −.06, SE = .03; p =.03), and marginally associated with lower income (b = −.02, SE = .01; p = .09). No relationships were found for age or education. Interview language was moderately to highly correlated with nativity and the acculturation subscale scores; thus, these variables were assessed in separate models (data not shown). These analyses indicated that simpatía was not associated with interview language, Anglo orientation, or being born in Mexico versus the U.S. However, being born in Cuba or Puerto Rico was associated with increased simpatía.

Discussion

When striving to provide care to Latinxs, health professionals are frequently encouraged to be attentive to and respectful of Latinx cultural norms (Altarriba & Bauer, 1998; Flores, 2000; The National Alliance for Hispanic Health, 2001; Juckett, 2013; Rothschild, 1998; Warda, 2000). These efforts are not only made to enhance patients’ comfort, but also because cultural values such as simpatía are believed to influence a range of health-related attitudes, behaviors, and beliefs that affect health outcomes. As Jones (2015) observed, simpatía-consistent behaviors serve to build trust between Latinx patients and nurses. Similarly, Sobel and Metzler Sawin (2014) found that Latinx patients felt more connected to nurses who smiled and expressed a positive attitude. Participants in both studies indicated that simpatía-consistent behaviors enhanced trust between Latinx patients and nurses, regardless of nurses’ own ethnicities or ability to speak Spanish (Jones, 2015; Sobel & Metzler Sawin, 2014). Assessments of patients’ value for simpatía may be particularly useful in clinical settings, as such assessments could be used as indicators of potential patient behaviors (e.g., withholding personal health information). These assessments may also be helpful in understanding the types of behaviors patients might expect from healthcare professionals (e.g., friendliness, presenting medical information in a positive manner) that are critical for developing trust and open communication, as well as encouraging better adherence to recommended treatments, regular follow-up care, and positive health outcomes for Latinxs. Simpatía could also be measured among nurses, doctors, community health workers, and other healthcare providers when hiring for positions working with Latinx populations, both as a means of evaluating cultural compatibility and/or providing feedback during cultural competency training. However, attending to simpatía in health research and practice with Latinx populations requires the availability of usable tools for measuring simpatía, which have been previously limited.

This research developed and evaluated a new, bilingual measure of simpatía to address the need for a brief, psychometrically evaluated, survey-based measure of simpatía that can be used with ethnically diverse Latinx populations in a wide variety of clinical and nonclinical settings. The scale appears to have high internal consistency reliability across and within Mexican-American, Puerto-Rican, and Cuban-American subgroups, and findings from confirmatory factor analysis indicate that the scale performs similarly across these three Latinx subgroups. As such, this measure appears to be appropriate for use in surveys with Mexican-American, Puerto-Rican, and Cuban-American respondents.

Consistent with the conceptualization of simpatía as a multidimensional construct, support was found for two dimensions in the final, 9-item simpatía scale. The first dimension is represented by the perceptions subscale, which assesses how important it is to be perceived by others as friendly, fun, likeable, and easy to get along with. These characteristics are congruent with Triandis et al.’s (1984) description of being perceived as simpático/a. In contrast, the behaviors subscale measures the extent to which it is important to engage in simpático behaviors toward others by avoiding causing embarrassment, disrespect, or other forms of discomfort in others. The behaviors subscale is also consistent with respeto, or respect for others, which is believed to be another generalized Latinx cultural value (Flores, 2000; Halgunseth, Ispa, & Rudy, 2006; Marín & Marín, 1991). Though, where respeto encourages individuals to be deferential in social interactions with persons from higher social levels (The National Alliance for Hispanic Health, 2001), the behaviors subscale focuses on avoiding discomfort among all persons, regardless of their social status. Together, the perceptions and behaviors subscales measure the two halves of simpatía: Being perceived as person who is simpatico/a, and behaving in a manner that conveys simpatía toward others (Triandis et al., 1984).

Findings from this study indicate that there is broad endorsement of simpatía across three of the largest U.S. Latinx subgroups. However, this research also provides evidence of cultural heterogeneity among Latinx subgroups, which is frequently noted (e.g., Altarriba & Bauer, 1998) but rarely empirically tested. Small but significant differences were observed across the three subgroups, with Cuban Americans having the highest simpatía scores, followed by Puerto Ricans and Mexican Americans. These findings are consistent with those by Merz and colleagues (2016). Although differences between Puerto Ricans and Mexican Americans disappeared after controlling for additional variables, the observation that Cuban Americans continued to evince higher endorsement of simpatía than Mexican Americans in the multivariate model suggests that there may be additional cultural variability among these and other Latinx ethnic subgroups.

The use of item response theory yielded a methodological finding that may be relevant to other cross-cultural survey research involving Latinxs: The lack of discrimination between the two middle options in the 4-point response scale. In social science research, a larger number of response options is often believed to be desirable to increase variability (Alwin & Krosnick, 1991; Weng, 2004), although contradictory findings (Revilla, Saris, & Krosnick, 2014) and curvilinear relationships (Weijters, Cabooter, & Schillewaert, 2010) have also been reported. If respondents cannot distinguish among response option labels, however, people with similar true scores may report different survey responses. In such situations, observed variability may arise from measurement error. In this research, the lack of distinction between the middle response options was likely due to difficulty distinguishing among the response option labels. As became evident during cognitive interviewing, it is often difficult to identify response option labels that are equivalent in meaning and valence in English and Spanish. Pasick and colleagues (1996) reported similar difficulties in a study with Asian and Latinx survey respondents. More research is needed to identify response options with scalar equivalence across multiple languages.

This simpatía scale was developed with an ethnically and sociodemographically diverse Latinx population, which is a strength of the research. Nevertheless, some characteristics of the study participants should be noted when considering future applications of the scale. For one, this research only involved participants from three Latinx subgroups, and research is needed to test the validity and reliability of the scale among other ethnic subgroups. Though simpatía has been described as general Latinx cultural trait (Marín & Marín, 1991; Triandis et al., 1984), other Latinx subgroups may have different language norms and cultural values that might affect their interpretations of the items and their conceptualizations of simpatía. Second, this scale was exclusively developed with adults and may be inappropriate for use with youth populations for whom different stages of cognitive development may differentially affect respondents’ ability to process survey items (Borgers, de Leeuw, & Hox, 2000). As the majority of the telephone survey participants were female, more oriented toward Latinx culture, Spanish-speaking, lower income, and relatively limited educational attainment, the simpatía scale may also present different psychometric properties when used with populations that include more participants who are male, oriented toward Anglo culture, English-speaking, higher income, or more educated. However, since a substantial proportion of health-related research with Latinxs focuses on underserved populations, who tend to be lower income and less acculturated, this scale should have broad applicability in health-related research.

Conclusion

The health literature is replete with references to culture as a powerful influence on health behaviors and outcomes, particularly in discussions of study findings. Often, though, culture has not been directly assessed or was measured using proxy measures such as language use or acculturation. As noted by the U.S. Office of Behavioral and Social Sciences Research, there is a need for improved measures of cultural traits to enable research to understand the specific ways in which culture influences health. The simpatía scale presented here can be used to better understand how culture influences health-related variables, assess cultural diversity among Latinx populations, and inform how nurses and other health professionals should interact with patients to provide culturally respectful care. Through these means, the simpatía scale developed through this research may serve as an effective tool for preserving and promoting Latinx health.