Table 4.
Linear regression of generic mental healtha explained by ethnicity, with work stress as potential mediator
| Works stress Models: | 1: Ethnicity + age and gender | 2: Ethnicity + work stress + age and gender | Attenuation %b | Mediation testc | 95 % CI (bootstrap) | ||
|---|---|---|---|---|---|---|---|
| b | 95 % CI | b | 95 % CI | ||||
| Dutch (reference) | - | - | - | - | - | - | - |
| African Surinamese | −0.40 | (−1.02 to 0.23) | −1.15 | (−1.75 to −0.55) | 188 % | 0.81 | (0.58 to 1.05) |
| South-Asian Surinamese | −2.33 | (−2.98 to −1.66) | −2.47 | (−3.09 to −1.84) | - | 0.12 | (−0.14 to 0.37) |
| Turkish | −4.25 | (−4.93 to −3.58) | −4.37 | (−5.00 to −3.73) | - | 0.14 | (−0.12 to 0.40) |
| Ghanaian | −1.35 | (−2.10 to −0.60) | −2.90 | (−3.61 to −2.18) | 115 % | 1.61 | (1.33 to 1.91) |
| Moroccan | −2.81 | (−3.58 to −2.04) | −3.41 | (−4.14 to −2.67) | 21 % | 0.56 | (0.26 to 0.86) |
aScores in generic mental health range from 11 to 70, higher scores reflect better generic health
b% Change in B calculated as (Bethnicity+workcondition-Bethnicity)/(Bethnicity); only for models where the ethnic group showed a statistically significant higher risk of mental health problems and the mediation test for that ethnic group was statistically significant. Negative signs (−) are used for changes towards non-significance (zero B)
cBold printed figures represent statistically significant mediation for that ethnic group