The Galilee Study had as its goal to examine separately Muslim and Druze adolescents, which the ISMEHA study carried out in 2004–5  had included as one single entity, in order to assess two questions. One, whether ethnicity/religion had a particular impact on prevalence of mental disorders, over and above the effect of other variables, mainly social disadvantage. The other, whether mental disorders were associated with different risk factors among Muslim and Druze adolescents. The fact that these different ethnic/religious groups reside in relatively segregated communities, have different degrees of neighborhood disadvantage and varying perceptions of suffering from discrimination by the majority population, offers a unique opportunity to examine underlying socio-economic factors that may be related to specific mental disorders in minority populations.
Regarding our first question, our findings seem reflected in Dogra’s et al., , statement that there is “a complex interplay between minority status and social class, with terms such as ethnicity being a proxy for multifaceted sociocultural and economic variables” (p. 265). We did find a higher prevalence of internalizing disorders among Muslim than among Druze adolescents. However, we cannot explain this finding without taking into consideration the socio-economic disadvantage of the Muslim population as compared to the Druze population or, in the words of Dogra et al. , the multifaceted sociocultural and economic variables that characterize the distinct ethnic/religious groups. Therefore, we cannot arrive at unreserved conclusions regarding the association between mental health and religion per se, as religion is not a discrete variable isolated from other socio-cultural and economic factors. In our study population, religion was associated with several measures of disadvantage. Muslim adolescents were in the low and lowest socio-economic levels, as compared to Druze who were in the medium and low levels. Muslim families had higher rates of welfare care than Druze families, had higher rates of adolescents with a learning disability, higher rates of maternal GHQ morbidity and much higher rates regarding feelings of discrimination than Druze adolescents. These factors present a picture of poverty and family stress, as exemplified by the fact that 23.3% of Muslim families were in welfare care, as compared to 15.3% of Druze families and by the fact that in 36.1% of Muslim families the mother had a high GHQ score, as compared to 22.8% among Druze families. Other socio-cultural factors that address the challenges facing adolescents belonging to minority groups in a disadvantaged social position are feelings of discrimination. In the case of Israel, decreed by law as a Jewish and democratic state (Basic Law: Human Dignity and Liberty, 17.3.1992), belonging to a religion/ethnicity other than Jewish brings additional discrimination and disadvantage. We also found in this respect a great disparity between Muslim and Druze adolescents, with 30.7% of Muslim declaring they feel discriminated as compared to 1.7% among Druze. Feeling discriminated against by the mainstream population and the negative impact of cultural or political mistrust are important factors affecting mental health, depression, conduct problems and well-being among minorities [6, 7, 28].
Regarding our second question, whether risk factors for internalizing and externalizing disorders differ between these two minority groups, we found marked variations that reflect the different socio-economic conditions between Muslim and Druze and their different relation vis-à-vis the dominant Jewish population. The risk factors for internalizing disorders among the Muslim adolescents were female gender, a very-low socio-economic index, having few siblings, having a learning disability, a high maternal GHQ score and feeling very discriminated. Regarding externalizing disorders (ADHD and ODD), the risk factors for Muslim adolescents were being male, having a medium socio-economic index, having 3–4 siblings having a learning disability and a mother with a high GHQ score. In contrast, these risk factors were found to be not significantly associated with internalizing disorders among Druze adolescents, over and above the effect of each other, while only gender was found to be a strong risk factor for externalizing disorders, with males being 24.6 times more likely than females to have an externalizing disorder.
Our study confirmed the finding of others [14, 29,30,31], regarding higher prevalence of internalizing disorders among females and higher prevalence of externalizing disorders among males. As well, our findings confirm what others have revealed, that socio-economic factors associated with ethnicity are most likely to explain mental health problems in ethnic minorities [5, 32, 33]. The socio-economic index used in this study , showed that nearly all Arab local authorities (N = 84), were ranked below the Israeli average and allowed to distinguish among different degrees of disadvantage, i.e., among the medium, low and very-low SES groups. Like McLeod , we found more internalizing disorders with decreasing socio-economic level among Muslim adolescents: those in the very- low socio-economic level were 2.4 times more likely to have an internalizing disorder than those in the level immediately above theirs. Langton, Collishaw, Goodman, Pickles & Maughan , claim that the gap between the very-low income group and the rest has widened and that the increase in relative inequality “might lead to a disproportionate increase in emotional problems in low-income groups” (p. 1086). This they attribute to the possibility that poorer families are exposed to more risk factors for emotional problems, such as loss of self-esteem and sense of control, and the possibility that factors associated with low income have become “more powerful risks for emotional difficulties over time” (p. 1086). Ford, Goodman and Meltzer  also found a greater gap between the very low socio-economic levels and those a slightly above them regarding exposure to adverse life events and maternal distress and family dysfunction. It is likely that among the Druze adolescents, who were all either in the medium or low socio-economic levels, the gap was less meaningful and the two groups were more homogenous than the Muslim adolescents who were either in the low or very-low socio-economic levels, where this gap was more significant. This would explain why among the Druze, internalizing and externalizing disorders in the bi-variate analyses showed an association with the youth’s particular characteristics (gender, LD and maternal GHQ score) rather than with measures of socio-economic level and neighborhood disadvantage.
Brody and colleagues  address the important contribution of economic hardship, neighborhood poverty and racial discrimination as risk factors associated with children’s and adolescents’ mental health. They acknowledge, though, that family- centered programs can provide health benefits by improving the resilience of those families living in the increased adversity of poverty and discrimination. Thiede and colleges  show that persistent and increasing racial inequality among Hispanic and Black Americans, compared with the White majority, bring America closer to a majority- minority society. They assess the multiple factors underlying race differences, such as having been born into poor families, being over-represented in high poverty areas and having less contact with social safety nets- all as factors associated with higher risk for mental health problems.
Over 30% of Muslim adolescents felt discriminated against by the majority, as opposed to 1.7% among the Druze. It seems that the participation of the Druze in the Israeli military and security services has given them a stronger sense of being appreciated and of being perceived as contributing to the well-being of the state of Israel, as compared with the Muslim citizens. The claim of Frantz Fanon  that dominant groups tend to implant their hegemony by inculcating an image of inferiority – a depreciating self-image - in the subjugated, is of particular relevance when dealing with the relations between the Jewish majority and the institutionally discriminated Muslim Israeli minority.
Regarding externalizing disorders, however, we found different trends: Muslim adolescents in the low socio-economic level were 5 times more likely to have an externalizing disorder than those in the very-low level. This finding may be explained by the excess of ADHD cases in the externalizing category and the socio-economic differential that exists in ADHD diagnoses in Israel where underdiagnoses in minorities might be very much influenced by lower socio-economic level . The ISMEHA showed that in the total Israeli population, lower SES was less likely to be associated with ADHD . Other studies, however, present different results and claim that ADHD is more prevalent among the more disadvantaged population .
We found, like Parry Langdon , higher rates of internalizing disorders among adolescents with fewer siblings: those with 1–2 siblings or 3–4 siblings were 7.7 and 3.1 times more likely, respectively, than those with 5 or more siblings to have an internalizing disorder. A possible link to our findings is the observation that in single-child families, children often report feelings of loneliness, boredom and inferiority . Our finding, however, seems to be particular to the Israeli traditional minorities and seems not to reflect what occurs in other populations .
Regarding externalizing disorders, we found the opposite trend: adolescents in medium- sized families (3–4 siblings) were 3.8 times more likely than smaller families (1–2 siblings) to have an externalizing disorder. These findings may be explained by the fact that in our study almost 60% of medium-sized families were in the higher socio-economic index, where we find more diagnosed ADHD cases.
Comparing adolescents with LD and those without we found, like Prior et al., , a higher prevalence of internalizing disorders (5.2 times more likely) among adolescents with a LD and a much higher prevalence of externalizing disorders (20.8 times more likely), probably due to the co-occurrence of LD and ADHD.
Adolescents whose mothers had a high-risk GHQ score were 4.3 times more likely to have an externalizing disorder and 2.4 times more likely to have an internalizing disorder than those whose mothers had a low risk score. Gonzales et al., , describe how multiple overlapping contextual influences relevant to low-income status, like economic hardship and neighborhood disadvantage, operate together to shape parenting and ultimately affect adolescent mental health.
One of the limitations of the Galilee Study is that the sample includes only students and therefore missed school dropouts or non-attenders who may be suffering from physical or mental problems. However, we estimate that this group is well represented because school attendance in Israel is mandatory for the age group examined. Dropout rates in Localities 1, 2 and 3 were below 2%, and in 8% in Locality 4. Additionally, we found higher refusal rates among males than among females, and this may have underrepresented the prevalence of externalizing disorders, which are more common among males. Refusal rates were also higher among students that were assessed by teachers as having low achievement and, therefore, our results might be under-representing the most severe cases. An unfortunate limitation, which we could not overcome, was that the Israeli Ministry of Education censored and excluded questions related to conduct disorders from the questionnaire. Therefore, externalizing problems in our study lack the conduct component and might be under-represented. Lastly, it is important to stress that we over-sampled for the Druze adolescents in order to match their numbers to those of the Muslim adolescents and thus obtain enough statistical power to compare two representative groups.
Although our R-squared values are somewhat low, this is usual when attempting to predict human behavior. However, we have statistically significant predictors and therefore we can draw important conclusions about how changes in the predictor values are associated with changes in the response value.