You are on page 1of 39

See discussions, stats, and author profiles for this publication at: https://www.researchgate.

net/publication/262081299

Domestic Violence against Women: Statistical Analysis of Crimes across India

Article  in  Journal of comparative family studies · January 2011


DOI: 10.2307/41604436

CITATIONS READS
8 5,632

1 author:

Michelle Hackett
Murdoch University
7 PUBLICATIONS   113 CITATIONS   

SEE PROFILE

All content following this page was uploaded by Michelle Hackett on 27 May 2016.

The user has requested enhancement of the downloaded file.


Domestic Violence against Women:
Statistical Analysis of Crimes across India

This is the final draft of the paper published by the Journal of Comparative Family
Studies in 2011:
Hackett, M. (2011) “Domestic violence against women: Statistical analysis of crimes
across India”, Journal of Comparative Family Studies, Vol. 42, Iss. 2, pp. 267-292.

Author:
Michelle T. Hackett
PhD (Politics), MA (International Studies), BSc (Hons)

Acknowledgment:
The author would like to acknowledge the contribution and support of Dr Peter Mayer
from The University of Adelaide.
Domestic Violence against Women:
Statistical Analysis of Crimes across India

Michelle T. Hackett*
INTRODUCTION

“Charanpreet Kaur, 19, had been married less than nine months when her
husband and his family decided it was time for her to go.

According to a police document, her husband and his father trapped her
in the bathroom. While her husband clamped his hand over her mouth,
his father drenched her with kerosene and lit a match, setting his
daughter-in-law on fire. Charanpreet lost consciousness and was taken to
the hospital by her husband and father-in-law, who apparently believed
she was close to death and would not be able to incriminate them, the
young woman's relatives said. But Charanpreet regained consciousness a
few hours later and gave her statement to a magistrate; her in-laws were
arrested the same day. Charanpreet died five days later.” (“India”, 2005)

After generations of silence, domestic violence is slowly becoming a topic of popular


media and academic scrutiny in developing countries. Even in India, where gender
discrimination is still legally condoned1, the press coverage and public outcry against
“dowry death” murders of young wives is rising (Borah, 2008). The more private nature
of less fatal forms of domestic violence, however, means that familial abuse still remains,
more often than not, hidden behind closed doors.

The statistical literature on domestic violence in India to date has focused on the sub-
national scale, such as studies over a few villages in a subdistrict or slum areas in a
metropolitan city. With the exception of Atal and Kosambi (1993), who attempted to use

*
The author would like to acknowledge the contribution and support of Dr Peter Mayer from The
University of Adelaide.
media reports to construct India-wide statistics on domestic violence, no other national
analysis could be found by the author. Indeed, many researchers have noted that domestic
violence has not been systematically studied with both small-scale and large-scale
surveys in India and other developing countries (Jejeebhoy, 1998: 300; Koenig et al.,
2006: 132; Krishnan, 2005: 88; Rao, 1997: 1169; Verma & Collumbien, 2003: 73). In
order to supplement the existing research, then, this paper will attempt a large-scale
statistical analysis of domestic violence against women in India, using nation-wide,
aggregate “Crimes Against Women” data from the National Crime Records Bureau of
India (Government of India, 2000).

As this paper will show, multivariate linear regressions of the crime records yield some
interesting and statistically significant results. By situating these results within an
ecological framework of domestic violence, we can attempt to construct some hypotheses
about domestic violence in India. This paper will propose that, in India, Dowry Death
(wife murder) and Cruelty (wife abuse) crime rates depend on a state’s level of
development. Firstly, we find that the higher the levels of gender equality, urbanisation,
health and education in a state, the lower the rates of Dowry Death crimes (and vice
versa). Secondly, whilst the Cruelty crime regression analyses will also reveal a
connection with development, it appears as a more complex relationship between and
within the Indian states. This paper will use a gendered resource theory to support the
hypothesis: Social development change (e.g. changing gender roles) can exacerbate
domestic violence, as observed in the discrepancies between rural and urban Cruelty
crime rates across India.

In summary, this study concurs with the view that domestic violence, in its various forms,
has a multifaceted relationship with the “stages of development” in developing countries
such as India (Bates et al., 2004: 197; Naved & Persson, 2005: 299). The specific
hypotheses presented here, however, must continue to be treated as provisional until they
are supported by further large-scale studies and the confounding influences of under-
reporting are better understood. Regardless, the strong statistical results in this study will
hopefully lead to further analysis at local and national levels, and present an informative
line of enquiry in the campaign against domestic violence.

The paper will begin with a brief overview of the prevailing approaches in domestic
violence theory, and relate these to the specific environment and existing research in
India. The next section will then present a statistical analysis of the Dowry Death and
Cruelty crime data, including the methodology and results. Following this is a discussion
of these results, with detailed scrutiny of the reservations due to the data source (i.e.
police records). And finally, the paper will conclude with an exploration of the possible
consequences for women and development in India more generally.

DOMESTIC VIOLENCE PARADIGMS

Current theories and debates in the domestic violence discourse

A “theoretical” division in the literature

In the 1980s, Gelles (1985: 359) identified eight domestic violence theoretical streams.
Throughout the last few decades, however, the literature has increasingly become divided
into two distinct opposing perspectives (Archer, 2006: 133): those who support a
“feminist” approach, and those who support a “family violence” approach. The central
debate between these two camps is over whether research should focus on the underlying
patriarchal causes of violence against women (Dobash et al., 1992; Mann, 2008), or on
the structural inequalities and individual dynamics of violence perpetrated by both
partners in the home (Dutton & Nicholls, 2005; Straus, 1991). Romans et al. (2007:
1496) have suggested that this theoretically based division is exacerbated by narrowly
enacted research agendas, on both sides; specifically, “different research methodologies
capture different types of interpersonal violence preferentially [italics added]”. The two
paradigms and their advocates, therefore, appear to conflict on both theoretical and
methodological grounds.
In opposition to this trend, several authors such as Anderson (1997) and Heise (1998)
have attempted to bridge the gap by integrating the determinants of abuse. Anderson
(1997: 668) has suggested the addition of gendered components to family violence
theories, such as resource theory. Heise (1998: 263) has proposed the adoption of an
ecological framework: “An ecological approach to abuse conceptualizes violence as a
multifaceted phenomenon grounded in an interplay among personal, situational, and
sociocultural factors”. This approach acknowledges the influence of larger societal
factors such as patriarchal norms, through to family and individual level factors such as
class and alcoholism. Thus an ecological model enables us to assimilate the simultaneous
effects of wider-society gender norms (a feminist approach), and individually-felt
structural inequalities (a family violence approach). The approach of both authors
acknowledges that “gender interacts with structures of race, marital status, and
socioeconomic status to influence power within relationships and propensities for
domestic violence” (Anderson, 1997: 667).

Domestic violence in India

Domestic violence in developing countries has not been studied as extensively as it has in
western nations (Archer, 2006: 133; Verma & Collumbien, 2003: 61). Correspondingly,
the divide in the literature between the feminist and family violence approaches appears
to be less distinct. In India, the historical, underlying patriarchal system dominates the
domestic violence discourse and appears to be a fundamental premise for the majority of
researchers, implying a feminist bias. For example, whilst it is sometimes acknowledged
that both men and women help propagate India’s patriarchal norms (e.g. the role of the
mother-in-law in domestic violence (Fernandez, 1997: 433)), all studies found by the
author (and presented here) focus solely on female victims. In contrast to this seemingly
feminist trend, the dominant methodologies applied in these Indian studies focus on
household-level structural inequalities (such as caste, education and income), which is
more indicative of a family violence approach. Whilst some studies have attempted to
include gendered components such as female employment, their restriction to a local
scale have prevented the testing of wider gender norms (as noted in Rao, 1997: 1171).
Therefore, it appears that a comprehensive understanding of the current domestic
violence literature for India requires openness to both feminist and family violence
perspectives and, as concluded below, is most enlightening within an ecological
framework.

A brief background on India

There is growing awareness that the influence of culture on domestic violence has not
been adequately acknowledged in western countries (Shirwadkar, 2004: 861). In India,
there exist a multitude of cultural histories influencing the norms and practices relating to
domestic violence, most of which are currently being challenged by various levels of
modernisation. Contrary to popular media depiction, India is not one homogenous
society. Across the sub-continent there exist a variety of cultural norms, affecting caste
discrimination, gender roles and attitudes towards domestic violence. The over-arching
regional pattern is of more patriarchal and “traditional” states in the north, such as Uttar
Pradesh and Punjab, and relatively more egalitarian and educated states in the south, such
as Kerala and Tamil Nadu (see Mayer, 2006). Correspondingly, within each state, there is
an increasing disparity between the experiences of those who live in rural areas and those
who live in the major cities, as industrialisation and modernisation change the face of
India.

In India recently, industrialisation and economic development have caused many changes
in the (gendered) landscape. For example, industrialisation has led to many women
beginning employment outside the family home for the first time, especially in the cities.
This new role has the direct effect of increasing female education, training, access to
money and movement into the public sphere. “Paid employment enabled women to meet
other people from the wider society and enabled them to participate in a larger social life”
(Dutta, 2000: 223). However, the traditional role of the female as the primary care giver
and house-worker is still expected to be fulfilled, and changes to the patriarchal family
power structure are often strongly contested (Dutta, 2000). Changes have occurred at
different rates across India, and as expected, gender role changes have been assimilated
with varying degrees of resistance. For instance, in the rural areas of the least developed
states, such as Uttar Pradesh, women’s roles have changed little and patriarchal practices
are basically unchallenged (Dreze & Gazdar, 1996: 108); whereas in major cities of more
developed states, such as Kochi in Kerala, women have more readily been assimilated
into public non-traditional roles (Jeffrey, 1987).

Industrialisation and rural-to-urban migration have also affected households in terms of


increased economic and psychological pressures. Migration has led to slums and
overcrowding in the cities and industrial competition has led to loss of livelihood for
many villagers (e.g. for “potters” in rural Karnataka (Rao, 1997: 1171)). The entrenched
caste system in India also plays a role in how these changes affect individuals. Whilst the
lowest castes do have employment quotas within the public service, the lowest
“Scheduled Castes” are generally under greater economic pressure and social
ostracization than other sectors of society, in both the cities and the villages (Heller,
2000). Hence, there are a variety of pressures on Indian households, varying from state to
state, villages to cities, and family to family.

Studies of domestic violence in India – a feminist approach

Whilst the degree of gender inequality and types of socioeconomic pressures do vary
across India, patriarchal norms are historically embedded in the broader Indian culture
(Dreze & Sen, 2002). Corresponding to this is the wide-spread acceptance of some level
of domestic violence against women. In one Karnataka village, for example, 88% of the
women surveyed would “accept it quietly” if beaten by their husbands (Rao, 1997: 1172).
More generally, domestic violence against women is socially tolerated in India provided
that the cause is considered “legitimate”: “Dispute over dowries, a wife’s sexual
infidelity, her neglect of household duties, and her disobedience of her husband’s dictates
are all considered legitimate cause for wife-beating” (Rao, 1997: 1174). Domestic
violence involving sexual abuse, economic deprivation, verbal abuse, incarceration and
psychological abuse, moreover, are rarely publicly acknowledged (“Know Your Rights”,
2003). The consequence of this is acute under-reporting of domestic violence to police,
especially in regions where women’s rights are neglected and their social networks are
limited. According to Bush (1992: 598), police in India are generally unsupportive of
domestic violence reporting due to their traditional belief in the sanctity of the family
unit. Many rural women, furthermore, have little access to police and no knowledge of
the laws set to protect them (Dutta, 2000: 227). Though severe beatings are more likely to
be publicly and legally condemned, the vast majority of domestic violence is either
ignored, deemed a husband’s right, or accepted as an inevitable part of married life.

As previously mentioned, most of the quantitative studies on domestic violence in India


to date have tended to focus on structural inequalities, rather than gender norms.
However, a few have attempted to incorporate gendered analysis of factors such as
female employment and dowry. The feminist approach to domestic violence in India sees
a connection between gender-discriminating laws (e.g. inheritance laws), dowry and
domestic violence; all are manifestations of the gender inequality in India, perpetuated by
the traditional patriarchal social system. Beating or harassing a young wife in order to
extract more dowry payment from her family has been recognized as a significant issue in
India (Rao, 1993). Dowry involves the bride’s family giving gifts and money to the
groom’s family as a part of the marriage arrangement2. A study conducted by Rao (1997)
in rural Karnataka, found the most robust (significant) determinant of domestic violence
to be an “inadequate” dowry (independent of wealth, education etc.). Verma and
Collumbien (2003) found a similar result in the slums of Mumbai. The reward of further
dowry from the wife’s family by beating her may be considered an economic motivator
for domestic violence. However, it also exemplifies the wife’s lack of power (Rao, 1997:
1178), the patriarchal belief in a woman’s lesser worth and a husband’s right to abuse his
wife for material gain.

We can develop a similar hypothesis from analysis of other “assets”, such as land
ownership and education. Most of India has discriminatory inheritance laws (Kishwar,
1998: 269). One exception to this rule is in the Indian state of Kerala. Kerala, unlike the
majority of Indian states, has a matrilineal history, such that property or land can be
passed down the female-line (Panda & Agarwal, 2005: 825). Panda and Agarwal (2005),
in a study in Kerala, found that a woman’s ownership of property or land was a
significant deterrent to wife-beating (independent of wealth, education etc.). In terms of
less tangible assets, Verma and Collumbien (2003) found that in Mumbai slums women
with some education were less likely to be beaten than those with no education. In Uttar
Pradesh, Koeing et al. (2006) similarly found that women with secondary education (but
not primary) experienced less domestic violence.

One feminist interpretation of these results is that women with more assets (property,
dowry, education) have more value and power within their family, which leads to less
domestic violence. However, in apparent contrast to this, a study by Krishnan (2005) in
rural Karnataka found that women who “controlled their own income” were the ones who
experienced significantly more violence than those who had no income or handed their
income over to their spouse. This implies that we need to consider a more complex
conception of how female empowerment influences domestic violence, possibly by
drawing from other theoretical frameworks.

Studies of domestic violence in India – a family violence approach

The family violence approach to domestic violence encompasses a wide spectrum of


theories (Gelles, 1985), including the “social-structure model” which is particularly
applicable to the Indian context. This social-structure family violence interpretation of
domestic violence, as defined by Gelles (1985: 361), firstly states that stress is unevenly
distributed in societies, specifically as a result of unequal employment opportunities and
income, exposure to illness and other indicators of disadvantage. When violence is the
“accepted response or adaptation to stress” by the individual or community involved,
then this “socially structured stress” can lead to varying degrees of domestic violence
(Gelles, 1985: 361). Thus a social-structure model in India may lead us to expect greater
levels of domestic violence among the lowest castes and tribes, those in rural poverty,
urban slums and those who are socially outcast.
In support of this theory, Koeing et al. (2006) found that economic pressure, lack of
assets, low education, and family violence history were all determinants of domestic
violence in Uttar Pradesh. The study also found that community norms that tolerated
violence against women were strongly correlated with higher rates of abuse (Koeing et
al., 2006: 135). Similarly, Martin et al.’s (1999) study in Uttar Pradesh found that men
reported stress-related factors, such as household poverty and a husband’s low education,
as the most significant precipitators of violence. Direct evidence of caste influence in the
occurrence of domestic violence was found in a study by Visaria (1999) in rural Gujarat.
This study indicates that economic stress precipitates marital violence, as experienced
more by lower caste women. A study over three Indian states by Duvury et al. (2002,
cited in Krishnan, 2005: 89), however, found that while physical violence decreased with
household economic improvement, incidents of sexual violence increased, a dimension
which most other studies neglected.

Thus, again, we have a variety of evidence from different regions in India. Furthermore, a
review of the literature from both family violence and feminist perspectives appears to
give some contradictory determinants of domestic violence in India. Thus, a broader
analysis is needed in order to situate and explain the variation in results. One way to
grapple with the medley of Indian studies and results is to view them within an ecological
framework.

An ecological model of domestic violence in India

Within an ecological model, the various results and interpretations of domestic violence
do not necessarily conflict. Instead, this model highlights the significance of the context
of the studies. That is, the reason that many of the studies above conflict may be due to
the effect of larger-scale society factors which are not taken into account on such small-
scale studies. Naved and Persson (2005: 289-291) have adapted the ecological model for
their study of domestic violence in Bangladesh, and it is this framework which we will
use for analysis of domestic violence in India (see figure 1). Between different
households we expect to see more immediate structural and individual dynamics
affecting domestic violence (i.e. the inner circles of figure 1), as represented in the
studies above. But between different Indian states we should expect to see the gender
norms and patriarchal histories influencing levels of domestic violence (i.e. the outer
circles of figure 1), which could explain why the studies above conflict. Thus, if we
attempt to take these larger-scale society factors into account, we can attempt to reconcile
the Indian literature’s study results. One theory with which to now test this is the
“gendered resource theory”.

{Figure 1 about here}

A “gendered resource theory” of domestic violence across India

Resource theorists argue that changes to the power dynamics between partners can
influence the level of domestic violence (Goode, 1971). Specifically, physical abuse may
be used as a tool to compensate for lack or lessening of power in the relationship. For
instance, if an Indian wife were to simultaneously begin paid-work (i.e. gain an income
resource) and do less household chores (i.e. challenge the family power norm), this could
be perceived as a threat to the husband’s power in the marriage and lead to physical
violence. A gendered resource theory, such as that developed by Anderson (1997),
situated within an ecological framework of domestic violence, would seek to discover
how shifting gender norms affect the spousal “resources” differently across Indian.

Krishnan’s (2005) research in rural Karnataka found that women who controlled their
own income were the ones who experienced significantly more violence. In contrast,
Verma and Collumbien (2003) found in the slums of Mumbai that although female
employment was positively related to domestic violence, it was not statistically
significant. Furthermore, in rural Kerala, Panda and Agarwal (2005) found that wife-
beating was more prevalent in households where the wife earned more than the husband,
though wives with more land inheritance were less likely to be beaten (independent of
caste etc.). A gendered resource theory can help clarify these apparently contradictory
results.
A gendered resource theory would support the conclusion that “the vast majority of these
reasons [for husbands beating their wives] involve discrepancies between role
expectations and actual behaviour” (Verma & Collumbien, 2003: 61). The contradictions
in the studies above can be explained by looking at how different regions in India have
different gender role expectations. In the more developed state of Kerala, a woman’s
control of wealth (i.e. property) may not be threatening to gender norms, but earning a
higher wage than her husband could possibly be considered that way. In rural Karnataka,
alternatively, women have not traditionally controlled their income or been responsible
for household expenses (Krishnan, 2005: 97), and thus a change to this may be
threatening to the husband’s perceived power in the relationship. In the city of Mumbai,
however, a woman’s employment is relatively more acceptable, and thus is less
influential in domestic violence precipitation. Thus, a gendered resource theory can be
used to explore some of the varying results across India, by analysing differences in
expected behaviours in the husband-wife dyad, in the local community and in the wider
region.

Addressing a gap in the literature

More broadly, use of an ecological model may help resolve some of the discrepancies
between the results highlighted with the feminist and family violence approaches above.
The differences between domestic violence influences in Karnataka, Mumbai and Kerala
are more understandable in the context of “larger society” factors such as gender norms
of the states, and their effect on the “immediate social context” such as female wealth,
power and dowry practices (see figure 1). The results of Martin et al. (1999) and Visaria
(1999) regarding the primary influence of structural inequalities on domestic violence,
such as economic stress and caste, can also be facilitated in the ecological framework.
Economic stress and caste inequalities play a role in the “immediate familial context”,
influencing the “individuals involved” and their propensity for violence in the home (see
figure 1). The variety of results in the literature, therefore, can serve to enrich rather than
confound the Indian domestic violence model.
Whilst some authors have attempted to incorporate community-level norms into their
surveys (Koeing et al., 2006; Verma & Collumbien, 2003), thus far most researchers of
domestic violence in India have had a household-level focus. “A final limitation [of
previous studies] has been an almost exclusive focus on the roles of individual-level
determinants, with much less attention given to the roles of broader community and
contextual factors in precipitating or protecting against violence” (Koenig et al., 2006:
132). Therefore, in order to build a comprehensive ecological model of domestic violence
in India, analysis and collection of large-scale data is needed.

In order to help fill this void, this study will explore the state-wide indicators and
precipitators of domestic violence against women in India. It will use data from the
National Crime Records Bureau of India, and other available social, demographic and
economic variables, to statistically analyse the diversity in domestic violence across the
states of India. The theories and framework analysed above will hopefully then help to
explain the Indian results.

STATISTICAL ANALYSIS OF CRIMES AGAINST WOMEN

Methodology

Data

The primary data utilised for analyses in this paper were sourced from the National Crime
Records Bureau of India, under the section “Crimes Against Women” (Government of
India, 2000). The two crimes against women that relate directly to domestic violence in
India are “Dowry Deaths” and “Cruelty by Husband and Relatives”3. Public recognition
of dowry-related murders in the 1970s led the government of India to create a separate
crime for suspicious deaths of young wives, called Dowry Deaths (Bush, 1992). If a
woman dies within seven years of her marriage, the onus is on the husband to prove that
she did not die at his (or his family’s) hands. Curiously, the cause of this murder, legally,
need not necessarily be related to dowry. There are no known statistical studies on the
motivations for Dowry Death crimes, and thus this study will treat Dowry Death as a fatal
form of domestic violence (similarly to Rao 1997: 1169). According to Manushi (“Know
Your Rights”, 2003) the forms of Cruelty recognized by the Indian courts include
“persistent denial of food; insisting on perverse sexual conduct; physical violence;
taunting, demoralizing and putting down the woman with the intention of causing mental
torture; confining the woman at home and not allowing her normal social intercourse;
threatening divorce unless dowry is given”, among others.

By definition of these crimes, the scope of the statistical analysis in this paper is
restricted. For domestic murder, the data available in the Dowry Death category restricts
our subject to (young) married women, within their first seven years of marriage. The
crime does not specify the reason for the murder and does not include extramarital
murder cases. For domestic abuse, however, the data available in the Cruelty category is
somewhat less restrictive, as it potentially includes all married women. Since non-
married cohabitation is not culturally acceptable in India, we can assume that this covers
the vast majority of women in partnership with men, though does exclude various groups
and circumstances (such as same-sex and extramarital relationships). The Cruelty crime
itself does not differentiate between different forms of abuse: It includes physical
violence, sexual violence, neglect, and economic violence (as defined in Fawole, 2008:
168). Therefore, we can only test a restricted subset of wife-murders and wife-abuse in
India. I will refer to the two crimes as “domestic violence”, with the understanding that
they do not encompass the breadth of the subject and, unfortunately, exclude
consideration of male victims.

The crime statistics are provided as aggregate data sets of rates (per 100,000 people) of
Dowry Death and Cruelty crimes for each state, Union Territory and major city in India
(see table 1 for a data sample). Whilst this means that we are not able to statistically test
household-level determinants of domestic violence, the data may clarify which states
have higher or lower levels of abuse (Cruelty) and murder (Dowry Death) of wives.
Armed with this large-scale data, we can ask a number of pertinent questions: Why do
rates of domestic violence vary from one state to another; is there a fundamental
underlying current that drives this; are there patterns that can be observed and explained?

{Table 1 about here}

Factor analysis

To compare the crime rates across India, we need data to help us distinguish between the
Indian states. This data should convey information about both the patriarchal norms (e.g.
female literacy and female infanticide) and the structural inequalities (e.g. wealth and
caste). For this purpose, a selection of state socioeconomic indicators were obtained,
including education, wealth, urbanity, industrialization, health, male preference and
fertility rates4.

In an ideal scenario, we would regress all of these state indicators against the crimes in
order to determine which have most influence. However, the more variables we use, the
more degrees of freedom this introduces, and the analysis becomes less robust. One
solution to this problem is to employ “factor analysis” to group the socioeconomic
indicators into a few, uncorrelated “factors”. The specific state indicators used to create
these factors were those that gave the best correlations with the crime rates, interpretable
“factor patterns” and optimal regression results5. The socioeconomic indicators used for
the factor analysis in the Dowry Death regression are listed in table 26.

When factor analysis was performed, the resulting factor pattern (i.e. table 2) emerged
with three distinct factors (see table 2 column headings). Each factor represents a
different combination of state socioeconomic indicators, with the weighting indicated by
the “weighting values”. The closer the weighting is to ±1, the more the corresponding
state socioeconomic indicator dominates the factor (as indicated in bold in table 2). The
first factor in table 2 appears to be dominantly influenced by a state’s female literacy rate,
average female age at marriage, fertility rate, infant mortality rate and murder rate. Thus
this factor will be henceforth called the ‘human development’ factor. The second factor in
table 2 is influenced by son preference, child gender ratio and average household size7.
Thus the second factor will be called the ‘gender-equality development’ factor. The third
factor in table 2 is dominated by the percentage of a state’s population with electricity,
urban-living and full-time female workers. Thus this factor will be called the ‘urban
development’ factor.

{Table 2 about here (includes note8)}

Multivariate linear regression

Multivariate linear regressions of the crimes data were performed using Dowry Death,
and then Cruelty crime rates, as the dependent (y) variables, and the development factors
listed above as the independent (x) variables. The form of the multivariate equation used
was

Equation1

where “Crime Y” is a vector of the states’ Dowry Death or Cruelty rates; “N” is the
number of independent variables (in this case N= 3); “b(i)” are the unknown regression
coefficient values to be solved for; “b(0)” is the y-intercept; and “Var X(i)” are the
independent variable vectors, that is, the ‘human development’, ‘gender development’
and ‘urban development’ factors. The purpose of the regressions is to uncover the values
of b(i), and test whether these values are statistically significant. The values of the
regression coefficients b(i=1,2,3) will indicate the type and strength of relationships
between the crimes and the development factors, giving us insight into the domestic
violence crime patterns across India.

Results
Dowry Death multivariate regression

Equation 2
Dowry Death= -0.273 Human Dev + -0.175 Gender Dev + -0.144 Urban Dev + 0.664
(where Dev represents ‘development factor’)

Initial analysis of the regression equation (equation 2) indicates that the Dowry Death
crime rate has an inverse relationship with ‘human development’, ‘gender development’
and ‘urban development’, since all three coefficients are negative. The ‘human’ and
‘gender development’ coefficients are significant to 99% confidence level and the ‘urban
development’ coefficient is significant to the 90% confidence level (two-tailed, see table
3). The regression equation itself is found to be significant also, within a 99% confidence
level in the Fischer’s F test (see table 3). The amount of variance in the data which is
explained by the equation (i.e. the adjusted R2 in table 3) was found to be very high at
75%. This implies that three quarters of the variance in the data (that is, the influences on
the crime rate) are described in the regression equation.

{Table 3 about here (including note 9)}

In summary, the multivariate linear regression for Dowry Death crimes appears to be
statistically robust, and initial observations imply that Indian states with higher ‘human’,
‘gender-equality’ and ‘urban development’ have lower rates of Dowry Death crimes (and
vice versa).

Cruelty multivariate regression

Equation 3
Cruelty rate = 0.460 Human Dev + -0.448 Gender Dev + 0.425 Urban Dev + 5.112

Unlike the multivariate regression analysis of the Dowry Death data, the Cruelty crime
regression results are not robust or significant (see table 4). ‘Human’, ‘gender’ and ‘urban
development’ have not formed any meaningful relationship with Cruelty crimes, as the
coefficients are statistically insignificant. The explained variance (R2 in table 4) is low at
9%, suggesting that important independent variables were not included in the regression.
The regression equation itself, moreover, is not statistically significant according to the
Fischer’s F test (in table 4), which leads us to question the validity of the data.
Confounding factors, such as under-reporting, may be distorting the analysis.

{Table 4 about here}

Cruelty Ratio multivariate regression

To explore this phenomenon further, additional data regarding the Cruelty crime rates for
the major cities in India were also considered for analysis (see table 1)10. Unfortunately,
the socioeconomic indicators (such as literacy) for each major Indian city were not all
available. However, a promising result came from analysing a ratio of city and rural
Cruelty rates, henceforth called the “Cruelty Ratio” crime rate. The Cruelty Ratio
equation given below uses “city” to refer to the major cities only (> 1 million people),
and uses “rural” to refer to the rest of the state population, that is, the villages, towns and
minor cities.

Equation 4

By this definition, if a state has a large Cruelty Ratio (> 1) then it must have higher
Cruelty rates in its major cities than in its rural areas, and conversely, a state with a small
Cruelty Ratio (< 1) must have lower Cruelty rates in its major cities than in its rural
regions. Thus, instead of comparing the total Cruelty crime rates between different Indian
states, we can now compare the Cruelty crime ratio between the urban and rural areas of
each state. And hopefully, by using a relative ratio, and not the absolute number of
crimes, the confounding factors will reduce to give us a statistically robust regression
result. Following, then, is the multivariate linear regression of the three development
factors against the Cruelty Ratio crime rates11.

Equation 5
Cruelty Ratio= -0.836 Human Dev + -0.640 Gender Dev + -0.321 Urban Dev + 1.734

Initial analysis of the results indicates that the Cruelty Ratio does give statistically
significant results (see table 5). The Cruelty Ratio crime rate is inversely related to all
three development factors, with ‘human’ and ‘gender development’ variables being
significant to 99% and 95% confidence levels respectively (two-tailed). The regression
equation itself is found to be significant also, within a 95% confidence level in the
Fischer’s F test (see table 5). The amount of variance in the data explained by the
equation (i.e. adjusted R2 in table 5) was high at 66%. This implies that approximately
two thirds of the variances in the data (that is, the influences on the Cruelty Ratio) are
described in the regression equation, which indicates a note-worthy finding.

{Table 5 about here}

It appears that whilst the regression of total Cruelty crime rates did not lend itself to
explanation, the Cruelty Ratio crime rates do give a regression equation significant
enough to interpret. Initial observations of the inverse regression relationships indicate
that in states where ‘human’ and ‘gender development’ are higher, the Cruelty Ratio is
lower (and vice versa). This implies that in states where social development is relatively
advanced (e.g. higher education, health and gender equality), Cruelty crime rates are
lower in the cities than in the villages. Correspondingly, in states where social
development is less advanced, Cruelty crime rates are higher in the cities than in the
villages.

DISCUSSION OF RESULTS
The confounding issue of police data records

The National Crime Records Bureau data of ‘Crimes Against Women’ is not a
sociological or anthropological survey of peoples throughout India, but rather a count of
the number of crimes that have been reported, and then recorded by the Indian police
(Government of India, 2000: 197). Whilst the statistical analyses above do appear to give
robust statistical results for the crime data, this does not necessarily mean that we can
imply information about the real occurrences of domestic violence across India. For this
we must enquire as to what the connection between recorded crime rate and actual
occurrence rate might be: Specifically, what affect does under-reporting and biased
recording have on the domestic violence crime data?

The first regression result (equation 2) indicates that in Indian states with low ‘gender-
equality development’, the Dowry Death crime rates are comparatively higher. But it is in
these states (with higher discrimination against women) where we also expect there to be
many more cases of wife-murder which are not recorded. Thus, if we were able to count
all of the recorded and non-recorded cases of wife-murder, the ‘gender-equality
development’ factor may become even more prominent in the Dowry Death regression.
We can propose that the Dowry Death crime results reveal a pattern which would be
enhanced, rather than contradicted, by including the unrecorded cases.

A salient distinction between Dowry Death and other gender-specific crimes is that
recording of a Dowry Death crime does not depend on the wife or her family reporting
the incident. If a woman dies within seven years of marriage, the onus is on the husband
to prove she did not die at his (or his family’s) hands. Furthermore, due to the high public
profile and condemnation of Dowry Deaths and the circumstantial structure of the
crime12, the ability and will to legally record wife-murders is arguably more favourable.
The data for Cruelty crimes, alternatively, cannot be expected to truly signify the
prevalence of domestic violence across India. The general public tolerance of wife-abuse
in India implies that crimes of “Cruelty by husband and relatives” will be extremely
underreported. Thus, it is not surprising that the regression of Cruelty crimes across India,
in equation 3, did not yield a statistically significant result.

The Cruelty Ratio, alternatively, does not require the magnitude of wife-abuse to be
accurately represented, but rather focuses on the comparative degree of domestic violence
between major urban and rural areas within a state (see equation 4). Thus the Cruelty
Ratio could reflect the true balance between rural and city-based occurrences of domestic
violence across India, even if we cannot gain a count of the true number of cases.
However, it has been documented that women in villages are less aware of their legal
rights in reporting domestic violence and have less access to police and legal centres than
women in cities (Atal & Kosambi, 1993). Furthermore, since rural communities are
generally considered to be more traditional than those in cities (Dreze & Sen, 2002), we
can expect this to influence the relative degree of both domestic violence and the
reporting of it. Therefore, the Cruelty Ratio may be skewed by rural-urban differences in
reporting and recording, which may be more or less pronounced in different Indian states.

Consequently, it is tempting to simply dismiss all findings pertaining to Cruelty crimes.


However, due to the surprisingly statistically strong and robust regression of the Cruelty
Ratio, we have evidence that an observable relationship does exist between urban and
rural Cruelty crime rates across India. Thus, while we cannot yet interpret this
relationship with any certainty, it is still beneficial to attempt a provisional hypothesis
from the Cruelty Ratio regression results. Let us, then, analyse the Cruelty Ratio
regression equation with an understanding that this will necessarily be adjusted when
more studies and information about rural and city under-reporting become known. The
following section will attempt to interpret the Dowry Death and Cruelty Ratio results
drawing from the domestic violence theories and frameworks explored earlier.

Discussion of Dowry Death and Cruelty Ratio crime regression results

The Dowry Death and Cruelty Ratio regressions used aggregate large-scale data (namely,
the entire sum of recorded crimes divided by the number of people in each state).
Therefore we do not know whether the data is reflecting the culmination of individual
and household level determinants or is instead indicating the effect of community and
societal level norms. For example, does the state crime rate reflect the percentage of
households in financial difficulties (who react with domestic violence), or does the state
crime rate indicate the community’s level of tolerance to domestic violence (when faced
with financial difficulties)? Within an ecological framework these are not necessarily
competing interpretations. The development indicators (in the ‘human’, ‘gender’ and
‘urban development’ factors) would influence, and be influenced by, both societal norms
and individual level determinants of domestic violence. Thus it is beneficial to consider
the Dowry Death and Cruelty Ratio results from both feminist and family violence
perspectives, both of which can be situated in the ecological framework.

Interpretations of the Dowry Death results

The Dowry Death regression results reveal a statistically robust inverse relationship
between Dowry Deaths and the ‘human’, ‘gender’ and ‘urban development’ factors. That
is, the less developed Indian states appear to have more Dowry Deaths (per capita) than
the more developed Indian states.

A feminist approach could propose that in states where women are more valued and
empowered, there would be lower rates of wife-murders; conversely, in states with
stronger patriarchal norms, wife-murder may be more common. This approach is helpful
in interpreting the Dowry Deaths regression results. Firstly, and most directly, the Indian
states with lower ‘gender-equality development’ appear to have greater Dowry Death
crime rates in the regression results. Similarly, states with lower female literacy, health
and employment, present in the ‘human’ and ‘urban development’ factors, were found to
have higher levels of Dowry Deaths in the regression. The feminist interpretation of
domestic violence in India, discussed earlier, highlighted a connection between
empowerment and domestic violence. It appears that a similar though more broad
interpretation can be applied to Dowry Deaths.
Alternatively, a family violence approach, such as the social-structure model, could
predict wife-murder to be more common in states where economic stress is highest and
where social norms for murder are strongest. This family violence interpretation could be
used to help explain the Dowry Death regression results, since Dowry Death crimes were
found to be more prevalent in states where the percentage of murder rates are higher and
literacy and electrification are lower (present in the ‘human’ and ‘urban development’
factors). A family violence approach, however, does not seem as capable of explaining
the presence of the ‘gender development’ factor in the regression equation. For this we
need to introduce a “gendered” component; specifically, through adoption of an
ecological model.

An ecological model of domestic violence could incorporate both the feminist and family
violence interpretations above to help explain the Dowry Death patterns. The regression
equation suggests that both female empowerment and economic factors, for example,
may be involved in precipitation of wife-murder. Due to the aggregate nature of the crime
data, however, we cannot yet elucidate in which levels (of the ecological map in figure 1)
the different determinants fit. Furthermore, due to the multiple socioeconomic
components incorporated in each ‘development’ factor, we also cannot yet judge which
determinants are more dominant at each level. Regardless, the statistical significance of
the ‘gender-equality’ development factor in the regression equation implies that a gender
component, such as patriarchal norms, must be included in any ecological map of Dowry
Death in India. It appears that further nation-wide studies are needed in order to more
comprehensively explore the patterns noticed here, and more local and national studies
are needed to study the specific characteristics of Dowry Death domestic violence.

Interpretation of Cruelty results

In the final regression, the ‘human’ and ‘gender development’ factors (i.e. the “social
development” factors) were found to be inversely proportional to the Cruelty Ratio crime
rate. This means that in Indian states with higher social development, there is less
domestic violence recorded in major cities, than in the rural towns and villages; and in
Indian states with lower social development, there is more domestic violence recorded in
the cities, than in the rural regions. For example, Kerala has relatively advanced literacy
rates, lower infant mortality, less dominant patriarchal norms, and it appears to have more
domestic violence recorded in its villages and rural towns, than in Kochi (its major city).
In contrast, Uttar Pradesh is dominated by traditional patriarchal values with low
education and health levels, and it appears to have more domestic violence reported in its
major cities of Varanasi, Kanpur and Lucknow, than in its rural areas. The Cruelty Ratio
regression results show that there does exist a relationship between social development
(i.e. ‘human’ and ‘gender development’) and the Cruelty crime ratio (urban-to-rural)
across the states. But why should a state’s average literacy or infant mortality rates, for
example, affect the domestic violence level in the cities differently to in the villages?

I have argued that domestic violence in India is best analysed through an ecological
model, which acknowledges the effect of both structural and gender inequalities, at both
societal and individual levels. For the study of Cruelty Ratio, the integrated, ecological
approach of the gendered resource theory is most enlightening. Previously, a gendered
resource theory helped explain why the relative levels of female empowerment affect
domestic violence differently in Kerala as compared to Karnataka and Maharashtra
(Mumbai). This demonstrated how challenging of gender roles precipitates domestic
violence differently from state to state. Hopefully gendered resource theory can now also
help to uncover why Cruelty rates differ between urban and rural areas across India.

A gendered resource theory would support a hypothesis that if the income and
employment of a wife increased, and traditional family structure was altered, a husband
may react with physical violence. In rural areas of states which have entrenched traditions
and negligible social development, and where the traditional “family power norms” are
still generally unchallenged, such as in Uttar Pradesh, fewer husbands may feel
threatened by their wives, and domestic violence may be relatively less common. With
migration to the cities, alternatively, we expect to see the breakdown of traditional family
structure and various other changes to lifestyle, including increased employment of
women outside the family home (see Conklin 1988). Thus it is in the cities of states such
as Uttar Pradesh where domestic violence may be more prominent. In this way, the
Cruelty Ratio results may be partly explained by a gendered resource theory of domestic
violence, which predicts that states with low ‘human’ and ‘gender development’, like
Uttar Pradesh, would have relatively lower domestic violence in its villages compared to
its major cities.

Concurrently, the regression results also imply that Cruelty crime rates are relatively
greater in the rural towns and villages than in the major cities for developed states like
Kerala, which have higher literacy, gender equality and life expectancy. It can be argued
that the employment of women for example, has been assimilated more easily and
quickly in cities such as Kochi, since states like Kerala have relatively more advanced
gender norms and participation of women (e.g. higher literacy, more egalitarian
inheritance laws). Conversely, it is in the rural areas of these more socially developed
states where change is occurring and gender roles are being tested and, according to a
gendered resource theory, this is where we would expect to see more domestic violence.
Thus, again, the Cruelty Ratio results may be explained by a gendered resource theory of
domestic violence, which predicts that states with higher ‘human’ and ‘gender
development’, like Kerala, would have relatively lower domestic violence in its cities
compared to its rural areas.

A similar development-dependent trend has been found in the occurrence of suicide in


India, as explored by Mayer et al. (2008). Suicide was found to be highest in the cities of
underdeveloped states, such as Madhya Pradesh and Uttar Pradesh, and in the rural areas
of more developed states, such as Kerala and Maharashtra. This trend in suicide was
attributed to the stress of changes that accompany urbanisation and modernisation. The
trend in domestic violence proposed in this paper, in contrast, appears instead to depend
more on changes to social and family gender norms (‘human’ and ‘gender development’).
Due to the inherently gendered nature of domestic violence in India, this is not
unexpected.
Thus, it appears that a combination of feminist and family violence concepts can be used
to explain the results for recorded wife-abuse crimes in India, in the form of a gendered
resource theory. The ‘human’ and ‘gender development’ factors represent a range of
gender and structural influences, which may combine to influence the Cruelty Ratio
regression results. Attempting to situate these in the broader context of Indian urban and
rural areas, seen across the different states, requires a broader framework, such as the
ecological model. However, whilst the domestic violence theories do give us a
meaningful way to explain these results, a robust hypothesis requires further data and
analysis of the differences between urban and rural areas, in both domestic violence
occurrence and under-reporting.

CONCLUSION

A transition theory of domestic violence in India

In summary, the only robust hypothesis that can be suggested from the results and the
analyses is that state ‘development’ does appear to affect the levels of domestic wife-
abuse and wife-murder which are recorded in India. In terms of wife-murder, we can
propose that higher levels of state development have a beneficial impact, reducing the
number of dowry deaths. For wife-abuse, however, the influence of development may be
dependent on the process of change. That is, challenge to traditional gender roles may
provoke domestic violence in the family, as observed in the cities of less socially
developed states, and in the rural areas of relatively more socially developed states in
India. These hypotheses have been drawn from Indian crime data, existing studies, and an
integrated theoretical framework, the ecological map. Future research in this area will
hopefully help establish whether the patterns distinguished here do indeed relate to a
development transition.

Further research is also needed to test whether wife-murder is adequately described as a


fatal form of domestic violence, or, instead, whether it requires the construction of a
separate field of study. Unfortunately, the regression of city crime rates for Dowry Death
was not statistically possible due to the low crime rate values (e.g. no recorded Dowry
Deaths in Kochi). Thus, without more data, and information concerning under-reporting
of domestic violence, we cannot properly compare the Dowry Deaths and Cruelty rates.
Similarly, this paper identifies a need to compare the differences in occurrence and
reporting of domestic violence between the urban and rural areas of each state.
Ultimately, a nation-wide study of domestic violence and murder, collected at a house-
hold level, would help clarify the propositions raised here and quantify the confounding
recording bias. Due to the extraordinary resources required for this, however, presently it
appears that an ecological map of domestic violence in India must be created by piecing
together the results from multiple studies, conducted at various levels. Hopefully future
studies will continue to fill the gaps highlighted here.

What the present paper may contribute to this area is to further question the assumption
that domestic violence will wane as a society develops. As noted by Bates et al. (2004) in
the case of Bangladesh, the stress of transition may instead, conversely, provoke
domestic violence: “… changes that somewhat empower women may lead to violence in
the near term. Such changes may become protective only after a critical threshold of
empowerment has been reached and gender roles have shifted substantially” (Bates et al.,
2004: 197). Such scenarios support the suggestion that domestic violence be
independently targeted for elimination, rather than expect or wait for development to
sufficiently advance. Indeed, addressing violence in the home is a step towards greater
development on both societal and family levels, to stop the cycle of violence and
suppression, and to ensure the gendered element of development is fully addressed.
1
See Katakam (2003) for an example of legal gender discrimination in the Indian aviation industry.
2
For more information on dowry and changing dowry practices in India see Bhat and Halli (1999), Dalmia
and Lawrence (2005), and Rao (1993).
3
See Atal and Kosambi (1993) for a detailed definition and appraisal of these crimes.
4
The socioeconomic data were found for as close a date as possible to the year 2000. The two sources used
were the Indian Census data for 2001 (Government of India, 2001) and the statistics cited in Dreze and Sen
(2002) for various years. Data on the average dowry values for each state were not available for the
analysis.
5
Socioeconomic state indicators were available for only the major Indian states, fourteen in total. These
were Andhra Pradesh, Bihar, Gujarat, Haryana, Karnataka, Kerala, Madhya Pradesh, Maharashtra, Orissa,
Punjab, Rajasthan, Tamil Nadu, Uttar Pradesh and West Bengal. This restriction to the major states is not
entirely undesirable in terms of comparable data samples. Some of the Union Territories in particular are
sufficiently small and demographically distinct that they would invariably skew the data distribution. The
negative consequence, however, is that with such a small number of samples, robust results are difficult to
obtain.
6
Following is a brief description of the more ambiguous labels in the table, and a list of indicators which
were trialled but excluded. ‘Child sex ratio’: Number of females aged 0-6yrs over number of males aged 0-
6yrs. ‘Son preference’: Death rate (per 1000) of female children (0-4 years old) over death rate of male
children (0-4 years). ‘% Females visit relatives’: Percentage of ever-married women aged 15-49 years who
do not need permission to visit relatives. ‘Female employment %’: Number of females who are full-time
workers over total number of females. Other state statistics were trialled but were not included in the factor
analysis as they did not yield the most robust regression results. These were ‘% Grade 8, female (male)’,
‘Life expectancy, female/male’, ‘Literacy rate, male’, ‘Maternal mortality rate’, ‘% Muslim in pop.’, ‘%
Newspaper subscription, rural/urban’, ‘% Scheduled Caste/Tribe in pop.’ and ‘state alcohol expenditure’.
Unfortunately, dowry related data were not available for analysis.
7
Whilst the dominant variables for the ‘gender-equality development’ factor appear to be son preference
and child sex ratio, indicating a gender bias, there is also a common inclusion of “wealth” variables in this
factor (i.e. ‘Rural poverty’ and ‘Electricity’ statistics). Further analysis (not shown here) revealed that this
is largely explained by the dominant influence of Punjab on the small sample set. The Indian state of
Punjab is known for its concurrently high gender discrimination and relatively affluent agricultural sector
(Dreze & Sen, 2002).
8
Factor analysis was performed using XLSTAT TM. Principle Components factor analysis was the method
chosen as it offers the assurance of uncorrelated factors, without the assumption of a large sample size.
Varimax rotation was applied in order to better identify the factors. The socioeconomic data were used in z-
score form to enable easier comparison of the factor weightings.
9
Prob >|t| is the Student’s t-test probability that the regression coefficient is from a zero mean distribution.
That is, the lower the probability value, the more likely that the development factor does have a relationship
with the crime rate.
10
The National Crime Records Bureau data on the crime rates are available for the each Indian state and for
the major Indian cities with more than one million people. Only 12 out of the original 14 states have a city
of over one million people, therefore the sample size has become smaller, and a statistically significant
regression will be harder to achieve.
11
Since the Cruelty Ratio data was only available for 12 (instead of 14) of the Indian states, the factor
pattern needed to be reconstructed with two less socioeconomic indicators. The Cruelty Ratio factor pattern
which gave optimal regression results was similar to that chosen for the Cruelty factor pattern, with the
inclusion of ‘Literacy rate, male’, and the exclusion of ‘Female age at marriage’, ‘Household size’ and
‘Infant mortality’. It is available from the author. The resultant factors were still recognisable as the
‘human’, ‘gender’ and ‘urban development’ factors , with ‘Literacy rate, male’ being situated in the ‘human
development’ factor.
12
The Supreme Court of India has recently upheld the position of Dowry Death as a crime that only
requires circumstantial evidence, assuring that the “burden of proof” falls on the defendant/s (“Court”,
2006).
REFERENCES

Anderson, K. L. (1997). Gender, Status, and Domestic Violence: An Integration of


Feminist and Family Violence Approaches. Journal of Marriage and the Family, 59,
655-669.

Archer, J. (2006). Cross-Cultural Differences in Physical Aggression Between Partners:


A Social-Role Analysis. Personality and Social Psychology Review, 10(2), 133-153.

Atal, Y., & Kosambi, M. (Eds.). (1993). Violence against Women: Reports from India
and the Republic of Korea. Bangkok: UNESCO Principal Regional Office for Asia and
the Pacific.

Bates, L. M., Schuler, S. R., Islam, F., & Islam, M. K. (2004). Socioeconomic Factors
and Processes Associated with Domestic Violence in Rural Bangladesh. International
Family Planning Perspectives, 30(4), 190-199.

Bhat, P. N. M., & Halli, S. S. (1999). Demography of Brideprice and Dowry: Causes and
Consequences of the Indian Marriage Squeeze. Population Studies, 53(2), 129-148.

Borah, P. (2008). Examining Media Content: A Case Study of Newspaper Coverage of


Dowry in India, 1996-2006. Asian Journal of Communication, 18(4), 379-395.

Bush, D. M. (1992). Women’s Movements and State Policy Reform Aimed at Domestic
Violence against Women: A Comparison of the Consequences of Movement
Mobilization in the U.S. and India. Gender and Society, 6(4), 587-608.

Conklin, G. H. (1988). Influence of Economic Development on Patterns of Conjugal


Power and Extended Family Residence in India. Journal of Comparative Family Studies,
19(2), 187-205.

Court: Circumstantial Evidence Enough to Prove Dowry Murder. (2006, 25 October


2006). The Hindu. Retrieved 25 October 2006 from
www.thehindu.com/2006/10/25/stories/2006102500561500.htm.

Dalmia, S., & Lawrence, P. G. (2005). The Institution of Dowry in India: Why it
Continues to Prevail. Journal of Developing Areas, 38(2), 71-93.

Dobash, R. P., Dobash, R. E., Wilston, M., & Daly, M. (1992). The Myth of Sexual
Symmetry in Marital Violence. Social Problems, 39, 71-91.

Dreze, J., & Gazdar, H. (1996). Uttar Pradesh: The Burden of Inertia. In J. Dreze & A.
Sen (Eds.), Indian Development: Selected Regional Perspectives (pp. 33-129). Delhi:
Oxford University Press.
Dreze, J., & Sen, A. (2002). India: Development and Participation. Delhi: Oxford
University Press.

Dutta, M. (2000). Women’s Employment and its Effects on Bengali Households. Journal
of Comparative Family Studies, 31(2), 217-229.

Dutton, D. G. & Nicholls, T. L. (2005). The Gender Paradigm in Domestic Violence


Research and Theory: Part 1- The Conflict of Theory and Data. Aggression and Violent
Behavior, 10, 680-714.

Fawole, O. I. (2008). Economic Violence to Women and Girls: Is It Receiving the


Necessary Attention? Trauma, Violence, & Abuse, 9(3), 167-177.

Fernandez, M. (1997). Domestic Violence by Extended Family Members in India:


Interplay of Gender and Generation. Journal of Interpersonal Violence, 12(3), 433-456.

Gelles, R. J. (1985). Family Violence. Annual Review of Sociology, 11, 347-367.

Goode, W. (1971). Force and Violence in the Family. Journal of Marriage and the
Family, 33, 624-636.

Government of India (2000). Crime in India 2000. New Delhi: National Crime Records
Bureau.

Government of India (2001). Census Data Online. Retrieved 16 August 2006 from
www.censusindia.net.

Heise, L. (1998). Violence Against Women: An Integrated, Ecological Framework.


Violence Against Women, 4(3), 262-290.

Heller, P. (2000). Degrees of Democracy: Some Comparative Lessons from India. World
Politics, 52, 484-519.

India: Dowry Murders Continue. (2005). Off Our Backs, 35(11), 6.

Jeffrey, R. (1987). Government and Culture: How Women Made Kerala Literate. Pacific
Affairs, 60(3), 447-472.

Jejeebhoy, S. J. (1998). Associations between Wife-Beating and Foetal and Infant Death:
Impressions from a Survey in Rural India. Studies in Family Planning, 29(3), 300-308.

Kataman, A. (2003). Civil Aviation: A Case of Discrimination. Frontline, 20(21).


Retrieved 22 August 2008 from
www.frontlineonnet.com/fl2021/stories/20031024005413000.htm.

Know Your Rights: Laws against Domestic Violence. (2003). Manushi, 137, 24-25.
Koenig, M. A., Stephenson, R., Ahmed, S., Jejeebhoy, S. J., & Campbell, J. (2006).
Individual and Contextual Determinants of Domestic Violence in North India. American
Journal of Public Health, 96(1), 132-138.

Krishnan, S. (2005). Gender, Caste, and Economic Inequalities and Marital Violence in
Rural South India. Health Care for Women International, 26(1), 87-99.

Kishwar, M. (1998). Rethinking Dowry Boycott. In S. Basu (Ed.) (2005). Dowry and
Inheritance (pp. 268-278). New Delhi: Women Unlimited.

Mann, R. M. (2008). Men’s Rights and Feminist Advocacy in Canadian Domestic


Violence Policy Arenas: Contexts, Dynamics, and Outcomes of Antifeminist Backlash.
Feminist Criminology, 3(1), 44-75.

Martin, S. L., Tsul, A. O., Maitra, K., and Marinshaw, R. (1999). Domestic Violence in
Northern India. American Journal of Epidemiology, 150(4), 417-426.

Mayer, P. (2006). The South Asian Arc of Instability: In Search of Explanations. Paper
presented at the 16th Biennial Conference of the Asian Studies Association of Australia,
University of Wollongong, 26-29 June 2006.

Mayer, P., Steen, D., Bradley, C., & Zianian, T. (2008). Urbanisation and Suicide in
India. In P. Mayer (Ed.) Suicide in India, New Delhi: Chronicle Books, in press.

Naved, R. T., & Persson, L. A. (2005). Factors Associated with Spousal Physical
Violence Against Women in Bangladesh. Studies in Family Planning, 36(4), 289-300.

Panda, P., & Agarwal, B. (2005). Marital Violence, Human Development and Women’s
Property Status in India. World Development, 33(5), 823-850.

Rao, V. (1993). Dowry ‘Inflation’ in Rural India: A Statistical Investigation. Population


Studies, 47(2), 283-293.

Rao, V. (1997). Wife-beating in Rural South India: A Qualitative and Econometric


Analysis. Social Science and Medicine, 44(8), 1169-1180.

Romans, S., Forte, T., Cohen, M. M., Du Mont, J., & Hyman, I. (2007). Who Is Most at
Risk for Intimate Partner Violence?: A Canadian Population-Based Study. Journal of
Interpersonal Violence, 22(12), 1495-1514.

Shirwadkar, S. (2004). Canadian Domestic Violence Policy and Indian Immigrant


Women. Violence Against Women, 10(8), 860-879.

Straus, M. A. (1991). New Theory and Old Canards about Family Violence Research.
Social Problems, 38(2), 180-197.
Verma, R. K., & Collumbien, M. (2003). Wife Beating and the Link with Poor Sexual
Health and Risk Behaviour among Men in Urban Slums in India. Journal of Comparative
Family Studies, 34(1), 61-74.

Visaria, L. (1999). Violence against women in India: Evidence from rural Gujarat.
Washington D.C.: International Center for Research on Women.
Figure 1.

Ecological model of influences on domestic violence in India


(based on the domestic violence model presented in Naved and Persson (2005: 290))
Table 1.

National Crime Records Bureau of India “Crimes Against Women” data,


for the year 2000 (Government of India, 2000)
(‘rate’ is the number of recorded incidents per 100,000 people (1 lakh = 100,000))

Dowry Death rate Cruelty rate


Dowry Death rate Cruelty rate
Indian state major cities only major cities only
(incidents/lakh) (incidents/lakh)
(incidents/lakh) (incidents/lakh)
Haryana 1.320 6.453 N.A. N.A.
Uttar Pradesh 1.295 3.510 1.624 10.365
Bihar 1.079 1.388 2.615 5.231
Madhya Pradesh 0.854 3.854 1.049 6.612
Punjab 0.843 3.882 0.694 11.792
Rajasthan 0.796 10.093 1.176 18.643
Orissa 0.768 3.026 N.A. N.A.
Andhra Pradesh 0.583 7.118 0.392 10.594
Karnataka 0.415 3.228 0.963 4.378
Maharashtra 0.406 7.402 0.094 1.870
West Bengal 0.358 5.074 0.115 1.501
Tamil Nadu 0.308 1.352 0.364 2.015
Gujarat 0.192 7.711 0.453 7.554
Kerala 0.077 7.474 0.000 2.514
Table 2.

Factor Pattern
(used in the Dowry Death and Cruelty regressions8)

Factor 1 Factor 2 Factor 3


weightings weightings weightings
State socioeconomic
(‘Human (‘Gender-equality (‘Urban
indicators
development’ development’ development’
factor) factor) factor)
Female Age at Marriage 0.946 -0.035 0.088
Literacy rate, Female 0.928 0.037 0.234
Infant Mortality rate -0.816 -0.026 -0.156
Murder rate -0.826 -0.218 -0.072
Fertility rate -0.814 -0.275 -0.425
Son Preference -0.359 -0.866 0.035
Child Sex Ratio -0.038 0.967 -0.04
Household Size -0.431 -0.618 -0.454
Rural Poverty -0.378 0.527 -0.522
Electricity % 0.394 -0.493 0.731
Female Employment % -0.037 0.224 0.944
Rural % -0.416 0.187 -0.792
Females Visit Relatives 0.522 -0.095 0.521
Table 3.

Regression results for Dowry Death crime rates

Independent Regression Standard Correlation-


9
Variables, Coefficient Error Prob > |t| Var X(i) and
Var X(i) Values, b(i) 5% Dowry Death
Intercept 0.664*** 0.052 1.7E-07 -
‘human development’ -0.273*** 0.054 0.001 -0.692
‘gender development’ -0.175** 0.054 0.009 -0.445
‘urban development’ -0.144* 0.054 0.024 -0.365
2 2
R =0.81, adj. R =0.75, F=14.23***, sig. F=0.00061
***result is significant to p<0.005, ** result is significant to p<0.01, * result is significant to p<0.05
Table 4.

Regression results for Cruelty crime rates

Independent Regression Standard Correlation-


Variables, Coefficient Error Prob > |t| Var X(i) and
Var X(i) Values, b(i) 5% Cruelty
Intercept 5.112*** 0.766 5.5E-04 -
‘human development’ 0.460 0.795 0.575 0.175
‘gender development’ -0.448 0.795 0.588 -0.169
‘urban development’ 0.425 0.795 0.604 0.162
2 2
R =0.09, adj. R =-0.19, F=0.31, sig. F=0.82
***result is significant to p<0.005, ** result is significant to p<0.01, * result is significant to p<0.05
Table 5.

Regression results for Cruelty Ratio crime rates

Independent Regression Standard Correlation-


Variables, Coefficient Error Prob > |t| Var X(i) and
Var X(i) Values, b(i) 5% CrueltyRatio
Intercept 1.734*** 0.213 3.9E-05 -
‘human development’ -0.836** 0.223 0.006 -0.659
‘gender development’ -0.640* 0.223 0.021 -0.505
‘urban development’ -0.321 0.223 0.188 -0.253
2 2
R =0.75, adj. R =0.66, F=8.13**, sig. F=0.008
***result is significant to p<0.005, ** result is significant to p<0.01, * result is significant to p<0.05

View publication stats

You might also like