Measuring Intimate Partner Violence: A Multi-Study Investigation of Gender Patterns

Abstract

Introduction: Gender patterns in intimate partner violence (IPV) remain a controversial topic. Most behavioral self-report measures produce gender "parity" in rates of perpetration and victimization of IPV. However, not all survey methods produce gender parity. For example, the National Violence Against Women Survey (NVAWS) and the National Criminal Victimization Survey use self-report methodologies which produce more reports of male than female perpetration. Arrests, reports to law enforcement (regardless of whether they lead to arrest), homicide data, and helpseeking data also produce more reports of male than female perpetration. Reports by witnesses (usually other family members) also produce more reports of male than female perpetration, as seen in the National Survey of Children's Exposure to Violence (Hamby et al., 2011) and the original National Family Violence Survey (Straus, 1992). This methodological inconsistency is still poorly understood. This paper will present the findings from three studies.

Method and Results: In Study 1, 238 undergraduates were randomly assigned either the standard version of the CTS2 physical assault victimization items or a version similar to that used in NVAWS, which changed the partner-specific wording to generic wording ("Someone" instead of "My partner"), with perpetrator information collected instead as a follow-up question. A significant Form X Gender interaction was found, indicating that the non-specific version yielded higher rates of female than male victimization. Study 2 compared standard CTS2 items to two alternative phrasings and again found a Form X Gender interaction in a sample of 288 undergraduates. Study 3 implemented one of these alternative forms in a rural community sample of 1000 adolescents and adults (ages 12 to 45) and found that female victimization reports were significantly higher than male victimization reports, for the full scale and for every item.

Conclusions: These studies, taken together, show that the wording of self-report items has a large impact on observed gender patterns. There should be consistency across methodologies in the measurement of any phenomenon. These studies show that it is possible to develop self-report IPV measures that correspond to other data on intimate partner violence.

Full text

Measuring Intimate
Partner Violence:
A Multi-Study
Investigation of
Gender Patterns
Sherry Hamby
Sewanee, the University of the South
Presented at the Society for Prevention Research Annual Meeting,
Washington, DC, May 2014
sherry.hamby@sewanee.edu

The Way Things Were…
and were and were and were
2
Rape of the Sabine Women, event 750BC
depicted by Nicolas Poussin about 1635
Cain & Abel in
Titian’s rendition
Henry VIII,
problematic partner
Pottery from the Moche culture in Peru, ca 600-1100 AD

A Social Science Revolution
•Important innovations in social science
research, such as the U.S. National
Crime Survey and Kinsey’s surveys on
sexual behavior.
•Research established that people would
disclose
–Violent behavior
–Intimate behavior
•Straus, Gelles, & Steinmetz established
that people would also disclose
violence by intimates to researchers.
•Now know that family violence not a
rare act by deviant few but a major
social problem that touches the lives of
many.
1948: An early scientific bestseller
The 1975 National Family Violence Survey

If You Have A Bunch of Scales & A Lump
of Gold You Want to Weigh…..
10 grams
10 grams
10 grams 10 grams
10 grams 20 grams!?

If You Have A Bunch of Scales & You
Want to Assess IPV…..
Witness
surveys: M>F
Arrests:
M>F
Reports to
Police: M>F
Other survey
formats: M>F
CTS-type
checklists:
F=M!?
IP Homicide:
M>F

If You Have A Bunch of Scales & You
Want to Assess IPV…..
Other assault:
M>F
Delinquency:
M>F
Weapon
carrying: M>F
Robbery:
M>F
Sexual assault:
M>F
Gang, hate,
you name it:
M>F

1996: The Revised Conflict Tactics
Scales (CTS2)

8
Brunnenburg Castle, Dorf Tirol, Italy
2004

First forays, 2005 & 2009
These papers had little impact.
My new hypothesis:
Birds gotta fly, fish gotta swim.
Clearly, a new approach was
needed.

Survey Science:
Experiments in Victimization Assessment
•Started with victimization only. Why?
•Victimization is easier to assess than perpetration.
–Less social desirability (Sugarman & Hotaling)
•Victimization is the focus of most national
surveillance.
–NISVS
–NCVS
–NVAWS
–NatSCEV
•Because of NatSCEV and the JVQ, I have conducted
a number of recent studies on development of
victimization items.

Study 1 (n=238 college students)
# of IPV Modes (IPV poly-victimization), p=.039
1.11
0.41
0.63
0.75
0
0.2
0.4
0.6
0.8
1
1.2
Standard CTS Non-Specific CTS
Males
Females

0.25
0.07
0.06
0.16
0
0.05
0.1
0.15
0.2
0.25
0.3
0.35
0.4
0.45
Standard CTS Non-Specific CTS
Males Females
%
Lifetime Rate, Severe IPV (p=.005)

Study 2 (n =251 college students)
23.5
17.9
20
27.2
11.9
36.4
0
5
10
15
20
25
30
35
40
Standard CTS CTS with "angry" stem CTS with "joking" stem
Males
Females

Study 3 (n=1207 community adults):
The Partner Victimization Scale (Hamby, 2014)
Female %
Male % OR
Not including horseplay or joking around……[at
start of each physical
item]
Partner threatened to hurt
*** 23.5 9.1
3.06
Partner pushed or grabbed
*** 27.8 12.6
2.67
Partner hit
** 21.1 14.2
1.62
Partner beat up
*** 12.4 2.3
5.98
Partner did unwanted sexual acts
***
11.5 1.8
7.12
Any partner victimization
*** 34.1 18.7
2.26
IPV Poly
-victimization (mode, α=.85) ***
0.95 (.05)
0.40 (.07)

Construct Validity:
Correlation with Trauma Symptoms
0.00 0.10 0.20 0.30 0.40 0.50
All
Females
Males
0.22
0.20
0.21
All 3 correlations are
significantly different
from zero, p < .001.
The correlations for
males and females
are statistically
similar in magnitude;
z=0.17, p > .80.

Construct Validity:
Exposure to DV in Childhood
0 0.1 0.2 0.3 0.4 0.5
All
Females
Males
0.31
0.32
0.25
The correlations
for males and
females are both
statistically
different from
zero & also are
statistically
similar in
magnitude;
z=1.12, p > .20.

Where We Can Go
•If you want to use a brief, self-report
measure that shows multi-method
convergence & evidence of reliability
and construct validity, use the PVS or
one of the other available strategies that
do not produce gender symmetry.
•Is this the end? Or course not! This is
science; there is no end, there is always
striving for better. Science is novelty,
not convention. The frontier.
•The good news: There is information to
guide us and advances are readily
attainable in survey science.
•We need to push back on the institutional
forces that inhibit the best scientific
practices.
17
A technician preparing
penicillin in 1943
WE CAN HONOR THE
ACHIEVEMENTS OF THE
PAST WITHOUT GETTING
STUCK IN THE PAST.