“…Columns 2 and 3 of Table 7.6 conduct a similar analysis using social capital and religious fragmentation, both measured at the TU level, again considering above and below median values for each. Column 2 suggests that CLTS had an impact of 5pp among households living in communities with high levels of social capital, and no impact in areas with low levels of this index, consistent with the findings presented in Cameron, Olivia and Shah (2015). Nonetheless, in our case, the evidence is not extremely robust: the difference between the coefficients for these two groups is small, just 4pp, and not statistically significant.…”