All study procedures were approved by the University of Michigan human subjects research committee in accordance with the Helsinki Declaration of 1975, as revised in 2000. Informed consent was obtained from all participants included in the study.
The MENA-IM was adapted from prior ethnic identity measures developed for African Americans [14, 28, 29] and Latinos [18, 19] (see Table 1). The key domains of EI we aimed to assess were (1) connection to prior country/region of origin (sample item: I keep up with political activities in the Middle East and North Africa), (2) desire to preserve and transmit family culture and traditions (sample item: It is important for Arab American people to educate their children about Arab/Arab American art, history, music, and literature), (3) centrality of Arab American and Chaldean identity (sample item: Both in my public and private thoughts, being Arab American is an important part of who I am), (4) respect for other cultures or multiculturalism (sample item: I care deeply about the needs of other groups such as Native Americans, etc.), and (5) Arab American media use (sample item: When I watch television, I usually watch Arabic television shows, such as ART and MBC). Modifications from the African American and Latino versions were required to make the measure applicable for Arab Americans in general, and in some instances to tailor survey items for Michigan. For example, media consumption included popular international Arabic television networks such as ART and MBC but also local Michigan radio stations (e.g., CINA) and regional newspapers (e.g., Arab American News). All items were answered along a four-point continuum ranging from strongly disagree to strongly agree. Higher scores indicate stronger ethnic identity.
The measure was translated into Arabic using an iterative process, as per recommended methods . We began by having a bilingual Arabic language expert translate the English version into Modern Standard Arabic. Translations were reviewed by bilingual professionals from our community partners. After each review, appropriate modifications were made.
After the professional reviews, we conducted cognitive pretesting of the MENA-IM and other new measures among five primarily Arabic-speaking individuals: two men and three women, ages 45–60. The interviews were conducted in both Arabic and English and were facilitated by study team members, one of whom was fluent in both languages.
Participants of the cognitive interviews suggested several modifications to the ethnic identity items. Specifically, they felt that several items could better distinguish between internal manifestations of identity (thoughts and feelings) and external behavioral items (outward expressions). To address this recommendation, we added the phrase “in my private thoughts” to relevant items. They also recommended adding “North Africa” or “North African” in addition to Arab American and Middle Eastern.
After the cognitive interviewing, the full electronic version was tested among three participants (2 in English and 1 in Arabic) and the paper version was tested among two participants (1 English, 1 Arabic). Further minor revisions were made based on the second round of pretesting feedback.
To explore concurrent validity, divergent validity, and predictive validity of the MENA-IM scale, we compared scores with other identity-related variables as well as health behaviors and health status.
Racial mistrust was measured using the mean of two items adapted from a prior measure . The two items, each answered along a 1 (strongly disagree) to 4 (strongly agree) continuum were (1) When I think about culture/race relations in America, I get upset and (2) The United States government is trying to make things better for Arab Americans. The later item was reverse-coded prior to computing the scale score so that higher values for the two-item mean indicated greater mistrust.
Centrality of MENA identity was assessed with a single item, adapted from our prior work . The item queried, “how important is being Middle Eastern/North African to your overall identity?” Responses ranged from with zero (not at all important) to 10 being (very important). This item is intended to serve as a brief measure of global ethnic identity and was assumed to be positively correlated with MENA-IM scores.
Mental health symptoms were assessed with the four-item PHQ-4 , which asked, “how often have you been bothered by any of the following problems?” (1) Little interest or pleasure in doing things; (2) feeling down, depressed, or hopeless; (3) feeling nervous, anxious, or on edge; and (4) not being able to stop or control worrying, all of which were answered with a 1–4 scale, with 4 being “not at all” and 1 being “nearly every day.” Higher scores are indicative of better mental health status. Alpha for the four items in our sample was 0.91.
Religiosity was measured with a three-item scale adapted from Krause ; 1) God/Allah put me in this life for a purpose, 2) God/Allah has a specific plan for my life, and 3) God/Allah has a reason for everything that happens to me. Responses were 1 “strongly disagree” through 4 “strongly agree”. Alpha for the three items in our sample was 0.96.
Health behaviors assessed included current cigarette use, defined as consuming at least 100 cigarettes lifetime and currently smoking on at least some days in the past month ; past month alcohol use defined as consuming at least one drink of any alcoholic beverage at least once in the past 30 days ; past month marijuana, defined as any use in the past 30 days; past year Hookah use, defined as use on greater than 2 occasions in the past year. [8, 34] Lifetime pain medication use was queried with an item from the 2017 Youth Risk Behavior Survey ; “During your life, how many times have you taken prescription pain medicine without a doctor’s prescription or differently than how a doctor told you to use it?” (count drugs such as codeine, Vicodin, OxyContin, hydrocodone, and Percocet). Use was considered more than 2 times in one’s lifetime.
Self-reported medical history was assessed by asking if the respondent had ever been diagnosed with (1) cancer, (2) diabetes or high blood sugar, (3) high blood pressure or hypertension, or (4) heart condition such as heart attack, angina, or congestive heart failure. Each variable was answered NO (0) or YES (1).
Demographic variables assessed included age (collapsed into 4 groups: 18–35, 30–45, 45–65, and > 65), household income (collapsed into 4 groups: under $10,000; $10,000 to $49,999; $50,000 to $99,999; and > $100,000), education (collapsed into 4 groups: high school or less, some college, college graduate, graduate school or higher), religion (Muslim or Christian), native country (US-born and not US-born), Arabic spoken at home (yes or no), and Arab or Chaldean identity.
Surveys were distributed at 12 settings, across three Michigan counties, that included two supermarkets frequented by the MENA community, one health clinic serving as a predominantly MENA population, one health clinic serving a predominantly Chaldean population, a state university with a high number of Arab American students, four mosques with a high proportion of Yemeni and Lebanese worshippers, two Chaldean churches, and a recreation center frequented by Lebanese youth.
Participants were given the option of completing surveys using pen and paper or online forms (tablet provided), with or without assistance, in English or Arabic. For those opting to complete surveys at home, we provided a self-addressed stamped envelope or a web address to complete the online version. Both paper and electronic surveys required active consent and testament that the respondent was over 18 and self-identified as Arab or Chaldean. Data collectors, many of which were fluent in both English and Arabic, were trained in interviewing by study staff. Participants received a $25.00 gift card after completing their survey.
Scale Construction and Psychometric Properties
The 20 MENA-IM items were examined initially through exploratory factor analysis using principle components extraction, with Varimax rotation in SPSS 25 . One of our goals was to create a scale that could be used for both Arab and Chaldean populations, i.e., we wanted to create a factor structure that was invariant between Arab and Chaldean respondents. We therefore identified an initial three-factor solution that appeared conceptually and statistically to fit for both the Arab and Chaldean respondents. This solution utilized 11 of the original 20 items. This three-factor solution was tested for invariance between Arab and Chaldean respondents through confirmatory factor analysis (CFA; conducted in Mplus 7).
For the CFA, items were all treated as ordinal, and measurement invariance was compared between a configural invariance model (Chi-square = 277.18, df = 82, RMSEA = .09, TLI = .98, CFI = .98) and a model which simultaneously constrained factor loadings, item thresholds, and scaling parameters (Chi-square = 218.41, df = 120, RMSEA = .08, TLI = .98, CFI = .98). While the constrained model fit was somewhat weaker than the configural invariance model (Chi-square = 82.47, df = 38), the factor structure parameters were essential similar for both groups. Given the empirical fit data, combined with the high interpretability of the three-factor solution, we proceeded with the common three factors for both the Arab and Chaldean respondents. Additional results of the exploratory and confirmatory factor analyses are available from the first author upon request.
The three factors were named MENA CULTURAL AFFILIATION (5 items), MENA MEDIA USE (3 items), and MULTICULTURAL AFFILIATION (3 items). Individual items and their factor loadings can be found in Table 1. Correlations between the three subscales ranged from .13 (between Factors 2 and 3), .36 (between Factors 1 and 2) to .58 (between Factors 1 and 3).
In addition to presenting results for the common three-factor solution, we also present results for the full 20-item scale, as some practitioners and researchers may not be working with both Arab and Chaldean populations in the same study and because the broader 20-item measure may be preferable under some situations. Scale scores for the three common factors and the overall 20-item scale measure were computed by taking the mean of items in that scale.
We hypothesized that MENA-IM scores would be positively associated with the EI centrality item as well as the religiosity and mental health scales. Because we conceptualize ethnic identity as only partially overlapping with these constructs, we hypothesized that the associations, while significant and positive, would be of modest magnitude, thereby indicating both convergent and discriminant validity. We did not have a priori directional or magnitude hypotheses regarding the relationship between MENA-IM with cultural mistrust. We also hypothesized that higher MENA-IM scores would be associated with lower rates of substance and self-reported disease.
We first present correlations between continuous variables (e.g., other identity measures and mental health symptoms) with MENA-IM scores (Table 3) and then multiple regression results adjusting for income, education, age, gender, and place born (Table 4). For categorical independent variables, we present mean MENA-IM scores by sociodemographic variables, without covariate adjustment (Table 5). For health behaviors and medical history, which were all dichotomous variables, we report multivariate regression, using MENA-IM scores as the primary dependent variable and substance use and health history as the independent variables, adjusting for income, education, age, gender, religiosity, and place born (Table 6).