Skip to main content

Table 2 Average treatment effects

From: The effect of a savings intervention on women’s intimate partner violence victimization: heterogeneous findings from a randomized controlled trial in Colombia

Dependent Measure

Average Treatment Effect

 

B

[95% CI

95% CI]

p

N

Joint p

Survey Measures

 Formal banking index

0.42

0.31

0.53

0.000

1510

0.000

 Economic status index

−0.04

−0.15

0.07

0.503

1510

0.418

 Confidence index

0.01

−0.09

0.11

0.842

1510

0.711

 Attitudes of social empowerment index

0.07

−0.04

0.18

0.242

1510

0.031

 Perceived norms of social empowerment index

−0.01

− 0.13

0.10

0.865

1510

0.188

 Relationship status (ended: y/n)

−0.01

−0.05

0.04

0.617

1625

NA

 Independent decision-making index

0.11

0.00

0.21

0.028

1325

0.007

 IPV index (0–11 count)

0.06

−0.09

0.21

0.453

1336

0.190

Checkup Measures

 STI test (y/n)

0.03

0.00

0.06

0.134

1129

NA

 Family planning (y/n)

0.03

−0.02

0.07

0.134

1136

NA

 Stress (0–4)

−0.04

−0.11

0.04

0.317

1170

NA

 Depression (0–4)

−0.10

−0.18

− 0.02

0.012

1168

NA

 Anxiety (0–4)

−0.04

−0.14

0.05

0.424

1167

NA

 Systolic blood pressure

−0.71

−1.87

0.44

0.230

1144

NA

 Diastolic blood pressure

0.45

−0.56

1.46

0.384

1144

NA

 Injury (y/n)

0.01

−0.01

0.03

0.317

1170

NA

 Violence (0–7 count)

−0.02

−0.14

0.10

0.741

1165

NA

 Violence frequency (0–4)

−0.03

−0.27

0.21

0.803

1170

NA

  1. We present the coefficient for the effect of the savings treatment on each dependent measure, the 95% confidence interval for the treatment coefficient based on robust standard errors, the p value for the treatment coefficient, and the number of observations for each dependent measure. For dependent measures that are standardized indices corresponding to a family of related survey measures, we analyze the index components using seemingly unrelated regression (SUR) and conduct a Wald test of the significance of the treatment coefficients within the system of equations; the p value for this joint test of significance is presented in the final column. We control for baseline assessment of dependent measures, socioeconomic status, life stage, and project site. Numeric ranges in parentheses refer to the response scale of dependent measures; y/n indicates a binary yes/no response scale