Table 4. Summary of multiple mediation analysis.
| Mediator | CLTS | Latrine rebuilding | Indirect effect (95% CI) | Odds ratio for specific indirect effects (95% CI) | ||||||
|---|---|---|---|---|---|---|---|---|---|---|
| B | SE | p | B | SE | p | LL | B | UL | OR | |
| Social capital (social cohesion and inclusion) | 0.399 | 0.163 | .015 | 0.248 | 0.119 | .037 | 0.006 | 0.099 | 0.284 | 1.104 |
| Vulnerability (personal general risk for diarrhea) | -0.580 | 0.161 | .000 | -0.562 | 0.125 | .000 | 0.125 | 0.326 | 0.629 | 1.385 |
| Vulnerability (general health of community members) | 0.042 | 0.076 | .582 | 0.550 | 0.260 | .034 | -0.057 | 0.023 | 0.156 | 1.023 |
| Others’ behavior (Estimated number of other latrine owners) | 0.853 | 0.139 | .000 | 0.972 | 0.159 | .000 | 0.475 | 0.829 | 1.294 | 2.291 |
| Confidence in recovery | 0.049 | 0.109 | .653 | 0.526 | 0.191 | .006 | -0.082 | 0.026 | 0.203 | 1.026 |
CLTS Indirectly Influencing Latrine Rebuilding through its Effect on Several Social and Psychosocial Factors. N = 286. B = unstandardized regression coefficients from linear regressions (CLTS) and logistic regression (latrine ownership); SE = standard error; CI = confidence interval for specific indirect effects; LL = lower limit; UL = upper limit; OR = odds ratio for specific indirect effects. CLTS received was coded ‘1’ and CLTS not received was coded ‘0’; latrine rebuilding was coded ‘1’ and no latrine rebuilding was coded ‘0’. Bias-corrected bootstrap confidence intervals for the specific indirect effects has been computed based on 10,000 bootstrap samples (bold: Significant effects).