Socio-demographic characteristics of the samples
The sample analysed in this study comprised of 14,565 adults in the manual class and 7,891 in the non-manual class; 11,080 women and 11,376 men; and 11,461 people living in high development regions and 10,995 in low development regions (Table 1).
Out of these, 40% of the respondents in high development regions were non-manual workers while in low development regions 30% were non-manual workers. Gender was homogenously distributed by region, but when it comes to social class men tended to belong to manual class more often that women did (68% v 61%).
Intersectional inequalities in material and psychosocial disadvantages
The distribution of material and psychosocial factors displayed distinctive inequalities between, but also within, the indicators of class, gender and regional development (Table 1; Figure 1).
Firstly, considering the three indicators one by one, the largest inequalities were found for social class for which manual class consistently displayed more material and psychosocial disadvantages as compared to non-manual social class. There were eight times higher frequency of material scarcity and more than double the frequency of unstable employment, poor social support and lack of social participation, but with fairly similar prevalence of insecure residential area. Women reported unstable employment 80% more often than men, with the other indicators displaying smaller inequalities (<20% relative difference). The disadvantages for women were material scarcity and insecure residential area and the corresponding for men were social support and social participation. Low development regions reported disadvantages more often (30-90%) than high development regions, except for insecure residential area which was slightly more common in privileged regions.
Secondly, distribution of material and psychosocial factors across intersectional social positions revealed more complex patterns of inequalities not discernible through single indicator inequalities. For example, although the triply disadvantaged group (women in manual class from low development regions) reported 10 times higher material scarcity than the 8 times difference between manual and non-manual social class (Table 1; Figure 1a); it also reported 12 times higher unstable employment than the triply advantaged group (men in non-manual class from high development regions) compared to the moderate 2-3 times higher within each social position (Table 1; Figure 1b). This illustrates how the magnitude of the intersectional inequalities cannot be monotonously predicted from single inequalities, but depended on life conditions.
The complexity become even more apparent when considering intersectional groups with mixed position of advantage and disadvantage; further illustrating the heterogeneity in life conditions not only between, but also within, the crude categories captured by the single indicators. For example, the intersectional position with the overall lowest material scarcity was women and not men, in non-manual occupations and high development regions. Moreover, whereas material scarcity, as noted above, was clearly patterned by social class, women in manual class from low development regions reported twofold material scarcity as men in manual class from high development regions (Table 1; Figure 1a). Additionally, the small relative advantage of women as a group when it comes to psychosocial resources was restricted only to non-manual class (Table 1; Figure 1d; Figure 1e).
Intersectional inequalities in SRH
Descriptive patterns indicating complex inequalities between intersectional positions were also found when it comes to SRH (Figure 2). For example, whereas all manual workers’ intersectional positions displayed higher frequencies of poor SRH than all non-manual positions, there was considerable heterogeneity especially within manual groups, with prevalence ranging between 30% frequency for men in high development regions to 40% for women in low development regions. The most advantageous position was women in non-manual social class from high development regions, while the most disadvantageous position was women in manual social class from low development regions. Differences between gender and regional development were larger among those in manual classes than non-manual classes.
Role of material and psychosocial factors in gender, social class and regional inequalities
Poisson regression analyses were carried out to estimate social inequalities in SRH by social positions of gender, social class and regional development,
The additive approach revealed that social class inequalities was the most remarkable inequality indicator, amounting to 61% higher prevalence of poor SRH among manual compared to non-manual social class. Minor but significant inequalities were found for gender and development regions. As indicated by the explained fraction (EF), psychosocial and material factors partially, but not completely, explained these inequalities. Psychosocial factors (Model B) explained about a fourth of the large class inequalities (EF=26%) and the smaller regional inequalities (EF=26%) in SRH but not substantially gender inequalities (EF=-5%). Material factors (Model C), had a greater relative importance for gender (EF=36%) than social class (EF=17%) or regional (EF=19%) inequalities. As a result, all factors together (Model D) explained a larger portion of social class (EF=34%) and regional inequalities (EF=36%) but less of gender inequalities (EF=23%). All inequality estimates remained statistically significant (p<0.001) even after full adjustment (see Table 2).
Analysing the inequalities according to a contrasting multiplicative approach the reference category was the best-off intersectional position after adjusting by age (the triply advantaged group men in non-manual social class in high development regions) (see Table 3).
The multiplicative approach showed cumulative effects of disadvantages. Whereas the additive approach estimated a 61% higher prevalence of poor SRH among social class groups (Model A) (see Table 2), intersectional social positions revealed up to 111% higher prevalence of poor SRH for manual social class women from low development regions compared to the reference group. Moreover, there was heterogeneity in prevalence ratios; ranging from 1 (reference) to 1.25 within the intersectional non-manual social class groups, and between 1.61 and 2.11 among intersectional manual social class groups (Model A) (see Table 3).
Discrepancies between the additive and multiplicative approach were also evident when it comes to the effect on the estimates when taking the indicators of social processes into account. Overall, psychosocial factors (Model B) explained inequalities mostly involving those intersectional social positions which had a higher relative frequency in both psychosocial factors (see Table 3); EFWML=24%; EFMML=28%; EFMMH=24%. Specifically the increased gender inequality when taking psychosocial factors into account in the additive approach (EF=-5%) (see Table 2) was only evident for women in non-manual social class from high development regions (EFWNH=-25%) (see Table 3). Moreover the sizeable explanation by psychosocial indicator of social class (EF=26%) and regional (EF=26%) inequalities (see Table 2) were in the multiplicative approach comparable only for the specific intersectional position of men in manual social class from low development regions (EFMML=28%) (see Table 3).
Whereas material factors explained a large portion of the overall gender inequalities (EF=36%) (see Table 2) the multiplicative approach showed their importance differed markedly for intersectional social positions within the same gender; from 12.5% to 21.4% for women and from 1.5% to 16.6% for men. A similar variation in explanatory power reflecting the intersectional inequalities was seen when adjusting for material factors: prevalence ratios for non-manual class groups ranged from 1.03 to 1.22 while prevalence ratios for manual class groups ranged from 1.54 to 1.86 (see Table 3).
The full model (Model D) involved the greatest explanatory fractions for all intersectional positions except for women of non-manual class which were better explained by material factors only (EFWNH= 41%; EFWNL=12%) (Model C) and men from non-manual class and low development regions which were better explained by psychological factors only (EFMNL= 17%) (Model B) (see Table 3). Among all intersectional inequalities, the prevalence ratios of manual class groups decreased the most when adjusting by all factors (Model D). However, the considerable explanatory fractions for social class (EF=34%) and regional (EF=36%) inequalities (see Table 2) were only seen for manual social class from low development regions (EFMML=36%; EFWML=36%), (see Table 3).