Gender Differences in Depression Symptoms: Findings From a Population Survey in Kosovo – A Country in Transition

This paper focuses on gender differences in depression symptoms. It takes into consideration relevant contextual factors of a country in transition. This paper’s analyzed data was extracted from European Social Survey, Sixth Round (ESS-6). ESS uses strict probability samples of the resident national population, aged 15 or older, and living in private households. Females reported a significantly higher mean depression on average (M = 8.14; SD = 3.88) compared to males (M = 7.56; SD = 3.86) at t(1247) = 2.604, p ˂ .009. The average for depressive symptoms found in the Kosovar population was higher than the averages reported in other European countries, but corresponded with those in Eastern European countries.

prevalence of major depression than married individuals (Bulloch, Williams, Lavorato, & Patten, 2009). Marriage has been found to be a protective factor against depression (Bebbington, 1996;Kawachi & Berkman, 2001), but other studies that analyzed links between different variables and depression showed that marriage negatively influenced the mental health of men (Almeida-Filho et al., 2004).

Kosovo Context and Background Studies
In the aftermath of war, Kosovo like many other postwar societies is experiencing many challenges, including economic stagnation, widespread poverty, high unemployment, population movement from rural to urban areas, poor quality of life, and pervasive discrimination (The World Bank Group in Kosovo, 2015). All these challenges together with traumatic experiences of war have a negative impact on the well-being of children and their families, ranging from specific phobias to severe mental health problems including PTSD, anxiety, depression and substance use (Jones & Kafetsios, 2005;Thabet & Vostanis, 2005;Thabet & Vostanis, 2011). Findings from one of the first studies, conducted with a population of 1358 Kosovo Albanians aged 15 years or older selected through random sampling, identified that 17.1% of respondents met the criteria for identification with PTSD, whereas women (19.7%) showed higher prevalence rates of PTSD in comparison to men (12%) (Lopes Cardozo, Vergara, Agani, & Gotway, 2000). Another study conducted by Salama, Spiegel, Van Dyke, Phelps, and Wilkinson (2000) in Kosovo reported a high prevalence of social dysfunction and severe depression, especially among the women.  Council, 2006). Epidemiological study findings related to prevalence rates of mental disorders in some Balkan countries (Bosnia-Herzegovina, Croatia, Kosovo, Macedonia, and Serbia) targeted a sample composed of adults who were personally exposed to armed conflict or who continued to live in such an environment. The study shows interesting prevalence rates. For example, anxiety disorder rates ranged between 15.6%-41.8%; rates of mood disorders ranged between 12.1%-47.6% and rates of substance use were 0.6%-9.0% (Priebe et al., 2010).
Specifically, findings from this study showed that 47.6% of participants experienced mood disorders, whereas 41.8% showed symptoms of anxiety disorders (Priebe et al., 2010).
Besides, Kosovo, like many other societies, has maintained a culture characterized to a great extent by patriarchy, evidenced by traditional norms and strictly defined gender roles. Findings from the latest country gender profile (Färnsveden, Qosaj-Mustafa, Farnsworth, & Nordlund, 2014), which aimed to analyze gender differences at all levels with regard to national framework, human rights, politics, and socioeconomic circumstances, show that the situation is rather challenging even 15 years after the war. Specifically, findings show that: 30% of Kosovo citizens lived in poverty; of these, 38% were women heading households, 10% of them under extreme poverty; only 18% of women in comparison to 55% of men participated in the formal labor market; only 8% of properties were owned by women, and women led fewer than 10% of businesses (Färnsveden, Qosaj-Mustafa, Farnsworth, & Nordlund, decreased since 2009(Färnsveden et al., 2014. Moreover, reports on education from the Kosovo Agency of Statistics show an increase over the years in the percentage of females in upper secondary schools, whereas data on higher education show that half of undergraduate students are women (Kosovo Agency of Statistics, 2014). All these conditions, accompanied by the social activism of many non-governmental organizations in promoting gender equality, contributed to the empowerment of women in Kosovo society, which contributed to changes in gender roles and the breaking of traditional and patriarchal attitudes that governed social life (Krasniqi, 2009).
As noted by Krasniqi (2009): tolerant attitudes toward employment of women, and an increased level of women's education in our society, contributed to a shift from traditional division of labor in the family toward the Western model of functioning, where indoor responsibilities are usually not reserved for men (p. 126). However, research evidence shows that changes related to gender roles might cause a certain degree of distress and crisis for men's masculine identity due to difficulties in upholding masculine gender norms (Gallagher & Parrott, 2011;Moore & Stuart, 2005), consequently having a negative impact in risk factors, stress and mental health (Vandello & Bosson, 2013). Therefore, in the context of rapid socio-cultural changes in a post-war society, it was considered necessary to assess the prevalence of gender differences and interaction of education and other socio-economic variables with depressive symptoms in the Kosovo general population.

Method
The data used for analysis in this paper was taken from the European Social Survey (the sixth round -ESS-6); Kosovo participated for the first time in 2012. The sampling methodology consisted of participants of strict probability sampling with a minimum age of 15 years who were part of private households. The data collection method was F2F interviews. 1295 participants were selected by stratified random sampling, with an age range from 16 to 89 and average age 43.22, of which 47.7% were male and 52.3% were female. Analysis of the data for purposes of this paper was restricted to respondents aged 18 to 75 (n = 1249); the average age of respondents resulted in M = 44.22, SD = 17.07, of which 47.6% (n = 592) were male and 52.4% (n = 653) were female from N = 1245 valid cases.

Measures
The brief version of Epidemiologic Studies -Depression (CES-D) was used to measure depression with eight items measuring depressive symptoms, aiming to identify populations at risk for development of depressive illnesses (Radloff, 1977). The translation and adaption of the European Social Survey questionnaire followed the requirements outlined in the specifications for participating countries, and used TRAPD methodology (Translation, Review, Adjudication, Pretesting and Documentation), aiming to achieve equivalence in regard to its translation (Dorer, 2012). Participants answered questions related to whether they had feelings or behaviors of a specific kind (depressive feelings, poor sleeping, feelings of loneliness and sadness, feelings of happiness and enjoyment of life) within seven days before the survey. Participants had to rate frequency of the symptoms' occurrence on a rating scale of 0 -3, the former meaning no symptoms or almost none for the indicated time, and the latter indicating continuance of symptoms during the whole period. Cronbach Alpha for the eight items measuring depression resulted in .80. The individual scores ranged from 0 to 24; higher scores indicated higher levels of depression.
Cases missing more than five items were deleted from the database, and the other missing values were replaced with series means as conducted in the Van de Velde, Bracke, and Levecque (2010)  Other variables were taken from the same database and treated as independent variables, including: gender, age (continuous variable), employment status (employed, unemployed, retired, sick or disabled, doing housework, and others, including military or other forms of service), income (measured in decile 10 as continuous variable), marital status (never married, married, widowed or divorced), years of education (as continuous variable) and children under 12 years of age in the family.

Data Analysis
Initially we looked at the mean differences between gender and education level, whereas the educational level for mean comparisons was measured in three levels: primary (or lower), secondary, and higher education. In the second section of results, findings from the regression model used linear modeling where all independent variables were entered.

Results
The mean depression for the whole sample was M = 7.87; SD = 3.88. The females on average reported a significantly higher mean depression (M = 8.14; SD = 3.88) compared to males (M = 7.56; SD = 3.86), at t(1247) = 2.604, p < .009, Cohen's d = 0.15, which indicates a small effect size. We also looked at education level of the sample and compared the mean scores for depression, based on participants' education. Education was re-coded into three main categories: no primary school or only primary school completed, high school completed, and university or higher degree obtained. Table 1 presents the mean scores for depression based on education level of the whole sample, and separately for males and females. As can be seen in Table 1, participants with lower levels of education generally reported higher depression symptoms than those with more years of education; this remained consistent for both genders. The one-way ANOVA was used to test depression scores among the three education groups for the whole sample. Depression scores differed significantly among the three education groups; F(2, 1242) = 53.80, p < .001, η 2 = 0.13, indicating small effect size. groups. However, no significant mean differences in depression scores resulted when we looked only for men and women with university education t(112) = -0.166, p = .869, Cohen's d = 0.02, indicating no effect size.
To better understand the impact of education and other socio-economic factors on depression, a General Linear Model (GLM) was employed to learn the effect of age, years of education, income reported in 10 deciles as a continuous variable, gender, and marital status of participants.
In Table 2, results are presented from a general linear model (GLM), covering the general population in Kosovo, with separate analysis results for both genders. Unexpectedly, scores derived from applying regression analysis showed no significant association between gender and depression in relation to the general population. Besides, as expected, an increase in education and income significantly decreased depression scores, whereas increasing age increased depression scores. Doing housework, being retired, being sick, disabled and unemployed significantly increased depression scores for the general population, in comparison to being employed. In terms of marital status, being single, widowed or divorced increased depression scores compared to being married in the general population.