Skip to main content
. Author manuscript; available in PMC: 2019 Jan 1.
Published in final edited form as: Soc Sci Med. 2017 Dec 6;197:127–135. doi: 10.1016/j.socscimed.2017.12.003

Table 3.

Fixed-effects model estimates and 95% confidence intervals for associations between changes in CES-D-10 scores and changes in personalized and generalized trust

Wave-adjusted Model 2 Model 3 Model 3+
Personalized trust (Trust of neighbors)
Individual level
Very likely 0.03 (−0.45, 0.51) 0.03 (−0.45, 0.52) −0.02 (−0.47, 0.44) 1.13** (0.37, 1.90)
Somewhat or not likely Ref
District level
% Very likely −0.01 (−0.05, 0.02) −0.00 (−0.04, 0.03)
Cross-level interaction
% Very likely * Very likely −0.06** (−0.10, −0.02)
% Very likely* Somewhat or not likely Ref
Generalized trust (Trust of strangers)
Individual level
Very likely 0 99*** (0.49, 1.48) 0 97*** (0.47, 1.47) 1.07*** (0.62, 1.53) 0.89** (0.27, 1.50)
Somewhat or not likely Ref
District level
% Very likely −0.00 (−0.04, 0.03) −0.01 (−0.04, 0.03)
Cross-level interaction
% Very likely* Very likely 0.01 (−0.02, 0.05)
% Very likely* Somewhat or not likely Ref

Notes: Models use individual fixed effects and adjust for clustering by district. Wave-adjusted model controls for wave. Model 2 adds employment status, log household income, household size, receipt of government grants, and marital status. Model 3 adds district variables percent very likely to trust, mean age, percent female, percent rural, percent African, percent unemployed, percent not economically active, percent with no formal education, percent with completed further education, percent with tertiary education, percent of households receiving government grants, and Gini coefficient to Model 2. Model 3+ adds an interaction between district percent very likely to trust and individual-level trust to Model 3. Bold indicates significance at the p<0.05 level.

*

indicates p<0.05.

**

indicates p<0.01.

***

indicates p<0.001.