Transcript The impact of intimate partner violence against women in Peru

Measuring the impact of intimate
partner violence against women on
victim’s and children’s well-being:
An application of Matching
Decomposition Techniques
Andrew Morrison
Maria Beatriz Orlando
Georgina Pizzolitto
September, 2008
Outline: IPV Impacts-MDT in Peru






MDT : non-parametric method –
victims and comparison group nonvictims
Prior research on impacts of IPV
Advantages of MDT to gauge impacts
DHS data for Peru: prevalence and
characteristics of victims/non-victims
MDT description and results
Conclusions
Prior Research Definitions of IPV
Types of violence



Psychological
Physical
Sexual
Timing of violence


Current
Lifetime
Prior Research Methods
Methods





Comparison of
means
Simple correlations
Bivariate and
multivariate logit/
probit regresions
Propensity score
matching
Focus Groups and
In-Depth interviews
Data



Victimization
Surveys
DHS
WHO surveys
Prior research on the Impact of IPV
Women’s control of reproduction and
unintended births
 Use of Contraception (by type of
violence)
 Unintended births
 STIs including HIV/AIDS
 Evidence for Africa (Kishor and Johnson,
2004) –but direction of causality
unknown


Women’s mental and physical health
Problems walking, problems carrying
out daily activities, pain, memory
problems, dizziness, vaginal discharge,
emotional distress
Visit a doctor, be hospitalized, or
undergo surgery (Nicaragua). Effects
are country specific
Infant health -Kishor and Johnson
(2004)
 Lower use of antenatal health care
services
 Increased the probability of a non-live
birth (miscarriage, abortion or
stillbirth)
 May produce increases in infant
mortality rates

Advantages of using MDT:
attribution, modeling, precision




Comparing women who suffer IPV with a
control group of women who do not suffer
IPV—but who are nearly identical over a
range of measurable characteristics
MDT does not require assumptions about
functional form required by multinomial logit
MDT does not assumes a causal relation
between IPV and the outcome variables
More precise measurement of the explained
and unexplained components of differences
Data




DHS Peru (2000)- violence modulenationally representative.
All women between 15 and 49 years old
who are present in the household
Focused on violence by intimate partners
and relatives (no questions about sexual
violence)-physical violence.
The survey did not ask about the timing
of the episodes -lifetime violence
Prevalence of domestic violence in Peru (2000)
Women aged 15-49 currently married or living with a partner
Intimate partner violence
Prevalence
(%)
Ever experienced physical violence by partner
39.80
By age groups (years)
15-19
20-24
25-29
30-34
35-39
40-44
45-49
28.36
32.52
39.49
41.79
42.24
41.92
43.65
Decreases with
age
By educational level
No education
Primary school
High school
Tertiary, College or more
42.04
42.80
41.21
28.93
Decreases with
education
Frequency of Husband getting drunk
Never
Sometimes
Frequently
24.59
41.08
76.84
Punished or hurt by father as a child
67.72
Increases with
alcohol
consumption
Source: Own estimations based on DHS, Peru 2000
Descriptive Statistics
Women victims and non-victims of physical violence
Difference
Age
*
Education
*
Punished as *
Child
Partner
Employed
Husband
*
Drunk
Source: Own estimations based on DHS, Peru 2000
*** Significant at 1%, ** significant at 5%, * significant at 10%
Descriptive Statistics – Outcome Variables
Women
Outcome Variables
Women's Health
Weight *Height (centimeters* kilograms)
Amenia (severity degree)
Number of Children
Number of children ever born
Last child wanted (%)
Terminated Pregnancies (%)
STD (%)
Delivey Complications (%)
Women's use of health facilities
Visit health facility (%)
Antenatal care (%)
Births assisted by health Care Professional (%)
Unmet family planning needs (%)
Contraceptive use (%)
Victims of
Pyisical Violence
Non victims of
Physical Violence
Difference
(Victims-Non Victims
12328
29.88
3.27
3.70
1.99
26.57
21.54
42.75
12274
32.50
2.74
3.02
1.79
16.99
20.39
32.86
54.64
-2.62
0.53***
0.68***
0.19***
9.57***
1.14*
9.89***
47.01
97.80
52.20
13.16
90.21
48.53
97.45
54.20
15.32
86.00
-1.52*
0.34
-2.00*
-2.15***
4.20***
70.35
64.05
6.29**
Source: Own
estimations based on DHS, Peru 2000
Women's
Employment
*** Significant at 1%, ** significant at 5%, * significant at 10%
Employed (%)
Descriptive Statistics – Outcome Variables
Women
Outcome Variables
Children's health
Diarrhea (%)
Anemia (%)
Height*age
Weight* height
Inmunization (%)
Under 5 year mortality (per 1000 bitrths)
Children's educational achievement
Educational Gap (%)
Schooll attendance (%)
Victims of
Pyisical Violence
Non victims of
Difference
Physical Violence (Victims-Non Victims)
20.17
75.06
2009
6023.57
40.96
0.66
13.69
73.50
2384
6216.14
33.73
0.70
6.48***
1.56
-375.0***
-192.5**
7.22**
-0.04
54.89
88.23
61.32
85.46
-6.43**
2.76***
50.13
37.26
12.86***
Mother's using violence to discipline Child
Use violence to discipline child (%)
Source: Own estimations based on DHS, Peru 2000
Matching Decomposition Technique
(MDT) Nopo 2004
Using MDT women who experienced
violence are matched to those who did not
on the basis of their observable
characteristics. The resulting matched
females have exactly the same observable
characteristics
Matching Decomposition Technique
(MDT)




Step 1: Select one victim from the sample
(without replacement)
Step 2: Select all the non-victims that have
the same characteristics x as the victim
Step 3: With all selected in Step 2,
construct a synthetic individual whose
characteristics are equal to the average of
all of them and “match” her to the original
victim.
Step 4: Put the observations of both
individuals (the synthetic non-victim and the
The result is the generation of a partition of the dataset.
Matched victims and non-victims have the same
empirical probability distributions for characteristics x.
Unmatched non-victims
(∆NV)
Unmatched victims
(∆V)
Matched Victims and Non-victims (∆X ,
∆0)
Variables included in control groups
used in the matching decomposition
Variable
1
Age
x
Number of Children
x
Years of education (women)
x
Was hurt by father or punished as childx
Spousal Age Difference
Spousal Education Difference
estimations based on DHS, Peru 2000
HusbandSource:
getOwnDrunk
Income level
Control
2
3
x
x
x
x
x
x
x
x
x
x
4
x
x
x
x
x
x
x
x
Results from MDT- Victims
Outcome
Difference
(Victims-Non
Victims)
Delta
Decomposition
(X)
Control 4
Women's Health
Weight *Height (centimeters* kilograms)
Amenia (severity degree)
Number of Children
Terminated Pregnancies
Last child wanted
(index: 1=wanted - 3=did not want more
children)
STD (%)
Delivey Complications (%)
Women's use of health facilities
Visit health facility (%)
Antenatal care (%)
Births assisted by health Care
Unmet family planning needs (%)
Contraceptive use (%)
Women's employment
Employed and earnning cash
(probability)
54.64
-2.62
0.53***
9.57***
-0.002
0.034
0.047**
0.080**
0.19***
1.14*
9.89***
0.006*
-0.083*
0.020*
-1.52*
0.34
-2.00*
-2.15***
4.20***
-0.030**
-0.010
-0.004
0.024
-0.007*
6.29**
0.002
Results from MDT- Children
Outcome
Difference
(Victims-Non
Victims)
Delta Decomposition
(Delta X)
Control 4
hildren's health
Diarrhea (%)
Anemia (%)
Height*age (centimeters* age in months)
Weight* height (centimeters* kilograms)
Inmunization (%)
Under 5 year mortality (per 1000 births)
hildren's educational achievement
Educational Gap
Schooll attendance (%)
6.48***
1.56
-375.0***
-192.5**
7.22**
-0.04
0.067*
0.155
0.016*
0.015*
-0.029
0.012
-6.43**
2.76***
-0.002*
0.015*
12.86***
0.100**
other's using violence to discipline Child
Use violence to discipline child (%)
Conclusions

In general, results are not robust to the use
of different methods

The MD technique is our preferred
methodology.
–
–

allows separating the impact of observable and
unobservable factors
takes into account that women who do and do
not suffer violence and female no violence have
characteristics that are distributed differently in
their common support (Delta X).
Naive comparisons shouldn’t be used to
formulate policy
Based on the MD technique, IPV has:



A strong negative impact on victim’s
reproductive health
Negative impact on visits to health facilities
and use of contraceptives
Negative impact on children’s health with the
exception of immunization

Children of women who are victims are more
likely to be in school

Strong evidence of intergenerational
transmission of violence