Abuse and Pregnant Women

Violence against pregnant women is a commonplace phenomenon and this research paper will explain the background of violence against pregnant women. Women undergo different forms of violence for instance, beating, threats, raping and unwilling prostitution. Some years back, it wasn't a big issue as approach towards women was a tad bit different back then. Men were treated as the dominant sex due to their physical strength. So is the case with education and men. It is common for men to get educated while women remain undereducated. With the passage of time, this created a genuine gap for women in the society as they were being undermined. Now violence against women, in general, is red flagged especially in Asia where women are treated very poorly. Now the circumstances for women are improving slowly and gradually in Asia but not completely (Jasinsk, 2004).

Violence against pregnant women isn't just limited to developing countries; it is commonly seen all over the world, even in western countries. UN Secretary General Kofi Annan spoke way back in 1999, that, 'violence against pregnant women is a crime which has no boundaries and spreads over boundaries, geographies, cultures and entails wealth. It is a shameful form of violation of human rights'. Then he went on further to say that, 'it's quite pervasive'. Women are vulnerable to violence and many resort to activities even they didn't deem fit for themselves for instance, prostitution to work for basic human needs such a food, clothes and shelter. Men have sex and pay them for their services as such women have no alternative choice. Women have been subjected to different forms of domestic violence as La Shawn R. Jefferson states that, 'There isn't proper policy regarding violence against women in the developing world which poses a serious threat (Jasinsk, 2004).

In the current scenario, which is presented by UNHCR, the primary principle is not to negate the violence against pregnant women at all. The human rights groups have criticized the governments in the developing world and policy makers for being incompetent and negating an escalating issue. Governments should protect its public against the dangers of sexual violence and physical assault. Some families have a track record of generations and generations of sexual and physical violence. Now both women and children are accustomed to it. Children watching their mothers getting beaten up by their fathers leave an indelible impression in their minds. It's a memory which haunts them day and night all their life. In third world countries, domestic violence is a critical problem kept on the sidelines (Jasinsk, 2004).

Background of the study

Domestic violence commences from family. Father in this case is the instigator of violence as he beats his wife regularly when his anger button is pressed. The man is solely responsible for intimidation, bullying, economic violence and threats. Physical assault is the most commonly occurring form of violence. Many forms of domestic violence exist such as (Jasinsk, 2004):

• Sexual violence

• Physical violence

• Emotional violence

• Rape

• Violence (Jasinsk, 2004).

A very silly concept among the less educated population is that men beat women because they have paid dowry to their wife, who makes a women their rightful property and now she is at the mercy of her husband. Sexual violence and physical violence still falls under domestic violence and it's a criminal act. Domestic violence affects the society on the whole. Not just women are affected but the entire society is. Women are passing through a tough time and they don't like revealing the conditions they are living in due to fear of being exposed. Two thirds of women married in third world countries face this dilemma with Africa and India adding generous numbers (Wuest et al., 2008).

Women, all over the word, are target of countless offences and in order to avoid them they succumb to prostitution. Since poverty is widespread, women work to pay for the house (it's her own choice to engage in sexual affairs, not that the husband is promoting it). Women need food and shelter as well as feeding their young; hence they are out of options. A survey conducted by World Health Organization (WHO), says that women deem it rightful and legal to get beaten by their husbands for instance when a man wants to have sex with his wife and she refuses, he has a right to beat her. Apart from that when she uses foul language and misbehaves, he is right to beat her. According to the survey conducted by World Health Organization (WHO), millions of women are impacted by domestic violence in Asia alone. A survey was done by World Health Organization (WHO) in 2005, which found that 50% of the surveyed women in India were beaten or exposed to various forms of domestic violence, while the percentage from Ethiopia was 71% (Wuest et al., 2008).

Prostitution is common and it's a form of an exit route from domestic violence. The amount of populace with STD has also risen. Now more people have HIV and AIDS. Children aged six are testing positive for AIDS and HIV. Women are more prone to catch the sexually transmitted diseases than men due to increased prostitution. Women are quite backward in that sense. Gender equality denotes monetary and social equality. The husband is the only powerful figure as the children and women take a backseat in the social framework. Hence the violence of women, especially pregnant women are observed (Wuest et al., 2008).

Purpose of study

This research paper will highlight the influence of violence against pregnant women and its resulting consequences. The rate of violence is skyrocketing with no rules and regulations in place to tackle it. The women impacted are seen fighting to live shoulder to shoulder with men. Such cases bring shame and disgrace to the entire country, not to mention leaving an indelible mark on women and girls in the social fabric of the society including children (Jasinsk, 2004). The aim of this study was to explore whether or not there is an association between intimate partner violence (IPV) and induced abortion among women of all age (teenagers to elderly). Demographic factors were taken into account as possible moderators in our study. Four research questions addressed important matters which the meta-analysis tried to answer.

Questions to be answered

The first research question: Is there any history of forced pregnancy or contraceptive sabotage, or coerced decision-making, which can be associated with abortion?

The second research question: Do those who commit violent acts against pregnant women aware of the implications of violence?

The third research question: Are there any demographic or socioeconomic trends, which can serve as a mediating factor?

Forth Research Question: How do women perceive disclosing what they have gone through during pregnancy and about the social therapy/Intervention they are offered?

Assumptions

Commonly the violence against women isn't reported even though it occurs at a high rate and on a daily basis. Women don't report this crime due to sheer shame and fear which would befall her family. However the numbers reported by the hospital, police, survey and research states that these crimes are spread far and wide. This study proposes that the numbers reported in the meta-analysis sample of our study are accurate and precise. The study also presupposes that the methods adopted by these studies are relevant and strong.

Theoretical framework

The first independent variables in this study are forced pregnancy or contraceptive sabotage, or coerced decision-making. The second independent variable is awareness of the abuser about the implications of the abuse. The third is a mediating variable, which are demographic and socioeconomic trends. The forth variable is perception of women about disclosing what they have gone through during pregnancy and about the social therapy/Intervention they are offered.

Scope of study

This research paper will highlight the influence of violence against pregnant women and its resulting consequences. Specifically, this paper will focus on the following four aspects. Firstly, forced pregnancy or contraceptive sabotage, or coerced decision-making. Secondly, awareness of the abuser about the implications of the abuse. Thirdly the demographic and socioeconomic trends serving as a mediating effect. And fourthly, perception of women about disclosing what they have gone through during pregnancy and about the social therapy/Intervention they are offered.

Chapter2

Historical and general background

Meta-analysis is 'the statistical analysis of a large collection of analysis results from individual studies for the purpose of integrating the findings' (Glass,1976).

Pearson (1904) also applied a series of methods for summarizing correlation coefficients, Tippet (1931) and Fisher (1932) presented a series of methods for combining p-values (significance levels), and Yates and Cochran (1938) have considered a combination of various estimates from different experiments. However, the introduction of a name for this collection of techniques appears to have led to an upsurge in development and application. Meta-analyses became popular in the '80s in medical fields. Meta-analyses provided answers concerning the treatment of heart disease and cancer. In 1990s published meta-analyses were ubiquitous. Chalmers and Lau (1993) claimed: 'It is obvious that the new scientific discipline of meta-analysis is here to stay'.

A significant growth in meta-analysis procedures was identified, from 18 in the 1970s to 406 in the 1980s. It is difficult to estimate what those numbers look like in 2014. Although Altman (2000) noted that Medline contained 589 meta-analyses from 1997 only. The rapid increase in meta-analyses being conducted in the past 10 years is due to greater emphasis on evidence-based studies and need for reliable summaries of vast, expanding volume of clinical research. Evidence-based medicine has been defined as 'integrating individual clinical expertise with the best available external clinical evidence from systematic research' (Sackett et al., 1997).

A systematic review of relevant external evidence provides framework for integration of the research while meta-analysis offers a quantitative summary of the results. In many cases a systematic review will include a meta-analysis, although there are some situations when this will be impossible due to lack of data or inadvisable due to unexplained inconsistencies between studies.

Existing Studies

Defining violence against pregnant women

Violence against pregnant women is also termed as:

• Intimate partner violence

• Battering

• Wife beating

It basically points out to all forms of domestic violence committed against women such as:

• Physical violence

• Psychological violence

• Economic violence

• Sexual violence (Asling-Monemi, 2009)

It occurs during a failed marriage. Its gender-based violence and often categorized by long seen patterns of abusive behavior (Asling-Monemi, 2009).

Violence means the usage of verbal violence or physical violence by one person on another person. In this case, it's usually the husband abusing the wife. The factors which instigate the violence include the following (Asling-Monemi, 2009):

• Verbal violence

• Sexual violence

• Physical violence

• Property damage

• Harassment

• Monetary violence

• Threatening consistently

The domestic relationship consists of married together / living together, spousal relationship and intimate personal relationship. It's between two people of similar interests and minds (Yoshihama and Horrocks, 2010).

Family relationship consists of two people connected by blood and families. People who deem themselves as cousins / relatives fall under family relationship. Informal care relationship is when one person assists another in daily activities. The domestic violence against pregnant women is quite prevalent in HIV positive women as compared to HIV negative women because many researches have concluded that women under age 30 years being HIV positive would certainly be violenced by her partner is 95%. These crimes aren't reported as evidence remains paper thin. Domestic violence against pregnant women impacts both the women and children alike. The child suffers from domestic violence as third parties. Domestic violence against pregnant women isn't discriminatory; it's not bound by particular age, salary, caste, creed and race. It kills the coziness of a house and generates hatred in families. The children who observe their fathers abusing their moms grow to despise their dads and end up just like them. Non-one should be violenced both physically and mentally. Apart from that physical violence, domestic violence, sexual violence, social isolation, monetary violence and so many social vices persist. The influence of domestic violence against pregnant women is beyond measure (Yoshihama and Horrocks, 2010).

'The term domestic violence against pregnant women against women has been used to explain many forms of crimes, which includes rape, sexual assault, murdering, physical assault, expressive violence, prostitution genital mutilation, stalking, harassing sexually and pornography (Yoshihama and Horrocks, 2010).

From the above quotation, violation consists of multiple factors which include sexual harassment as the top leading contenders. Women are sexually violenced in homes and in other places. Men are responsible for sexual harassment solely due to many factors. It happens when sometimes, threats are received pertaining complaints and threats of getting fired and suspended from office are the reasons (Yoshihama and Horrocks, 2010).

Reason for violence against pregnant women

Loss of control

The violencer has totally gone overboard when he starts abusing his wife. He should control it as it's his choice. It's always on him not the other way around. Hence, controlling and channeling anger is important here (Audi et al., 2008).

Anger management

This often creates domestic violence against pregnant women because the violencer has lost control of his anger as well as he didn't find a way to control it. For instance, a man can negate his anger while at office, with family and out with colleagues. He should beat his wife and kids just because there's some misunderstanding. Coming home, he decides to beat his wife and channel out his anger through her. That's one form of violence (Audi et al., 2008).

Managing stress

People are usually undergoing stress in their life and it would be wonderful to manage that stress whilst not taking out anger on people close to you. Hence, men should find out ways to vent out their emotional and physical stress (Audi et al., 2008).

The communication barrier

The violencer won't violence his wife as long as he is getting what he wants, but if poor communication exists on both sides, then physical violence is bound to occur. The wife takes the heat in this regard. The violencer needs to improve his communication in this case and restrain him from beating his wife (Audi et al., 2008).

Violence against pregnant women and children

Violence against pregnant women violence is a Pandora box opening as it creates future problems as well problems in the present as well. The active problems materializing are (Audi et al., 2008):

• Death

• Trauma

• Miscarriage

• Suicide

Violence against pregnant women has adverse effect on children because they are the third party victims in this case and they carry this anchor all their lives haunting them. It's a haunting memory and an elephant in the room. Not only children in the developing world are affected by it but also children in the first world countries are impacted by it. They can diagnose mental problems and resorting to the behavior done by their adults. It's a cycle of violence (Audi et al., 2008).

Most people who became violencers are now habitual of taking violence since they were taking it since childhood. Many of them were violenced physically, treated like a lesser being, but when they matured they decided to forgive and forget their past. Most of the child who say their moms getting beaten up grew up to be law abiding citizens. But in other cases, children who were exposed to child violence and physical assault were prone to become violencers themselves. Children, who grow up in such abusive environment, grow up to become abusive too. They think it's necessary to do so. Abusing others is now a necessary part of their life now and punishing their children for wrong doings is their justice to them (Audi et al., 2008).

Emotional and physical violence

Emotional and physical violence is a major component of violence against pregnant women. The physical violence has visible marks such as wounds, damages and injuries. The emotional violence is unseen by felt by the victims alone. For instance, emotional violence comes in the form of foul language, degrading people with depressing remarks and using derogatory remarks. The emotional violence is much worse than physical violence as women commit suicide. Many cases of suicide have been seen. Women noted abusive behaviors which fall under the Conflict Tactics Scale. They also recognized quite many abusive behaviors in comparison to non-violenced women. There were particular health differences in women who filed for emotional violence with no record of physical violence and sexual violence. Many women deem that emotional mistreatment is a case which shouldn't be talked about because the data is shady. Whereas in the case of physical violence, the arms are wounded and there are signs of injury and wounds. It's also a commonly seen form of violence (Roberts et al., 2006).

Women who live in such circumstances live in constant fright and panic of getting violenced. They are stressed by their life and have no contact with people whatsoever. They don't hang out a lot as the fear of getting exposed is revealed. This creates low self-esteem and they finally give up on life and finding a solution to their problem. Many victims aren't the drivers of their own life as they slack on taking steps to prevent such heinous acts happening to them. That's the reason the male partners take advantage of and continue abusing their spouses. Both don't have control on themselves. In such cases, the victim can commit suicide due to depressing state of affairs (Roberts et al., 2006).

Verbal violence

Combined with emotional and physical violence, there are many forms of violence. Sometimes, there are marks of physical marks of violence such as bruises, wounds and marks. Verbal violence is usually is made use for undermining other people and render them in severe complex. The violencer can victimize the person from time to time in a joking manner but the offence is taken nonetheless (Roberts et al., 2006).

Making derogatory remarks against someone can cause unhealthy state of mind. This directly affects the personality of a child growing up, the parents using abusive language will deem them useless/good for nothing/unintelligent and the child will start living the lie. This form of verbal taunts is directed towards children and women too. The violencer should become mature and keep away from such insults and distance her from violence. Verbal violence is very insulting and can destroy people's self-esteem (Roberts et al., 2006).

Hindrances cutting off abusive contacts

There are some particular hindrances occurring while abandoning an abusive relationship. Abandoning an abusive relation is quite troublesome and women can't leave due to many reasons for that matter. This paper will highlight some of the hindrances causing the severed relations to continue. 'First people don't admit that they are in an abusive relationship but after accepting it, it's time to move out. The aim is to move out from an abusive relationship / marriage between a man and woman, safety comes first no matter what' by Christina Gleason. The first hindrance is the financial one. Women need their husband's salary to live and working for a living is against their comfort zone. Apart from that, women don't have an alternative place to go in case she leaves her husband. Then the second case for women is to hold on tight to an abusive relationship due to the fact that they were violenced as children. Since childhood, such women were violenced physically / verbally and they have learned to live with it (Woods, 2005).

Another hindrance is fear. The fear of loneliness is terrible for women / men because the mere thought of being lonely is horrendous. It's better to have a relationship than not have one. It could be hard to get into a new relationship. Apart from that, their husband knowing that they are abandoning him might cause more violence against pregnant women is more troublesome to think off. Yet another hindrance is the nonsporting role of friends and family. They won't be such supportive anymore now that the women are looking to seek refuge in their residences. This creates a strained relationship with friends and families too. The popular belief amid the less educated population is that once a marriage is done, it's a closed deal. Now the husband / wife have to make the marriage work no matter and get by. Women don't opt for divorce for abandoning an abusive relationship. They actually think that one day they will see a better day and it's just a rough patch. Apart from that, mothers don't want their children to be raised up without fathers as the society is very unforgiving. Hence, they prefer having a father and take the brunt anyhow. Violence against pregnant women is a commonplace phenomenon all over the world. Here, only limited barriers are discussed, the topic of violence to women is even more complicated for the underdeveloped society. After going through these barriers, people will finally comprehend how tough it is to abandon an abusive relationship (Woods, 2005).

Methodology

The Cochrane Collaboration, launched in 1993, has been influential in promotion of evidence-based science. Reviews are made available online in the Cochrane Database of Systematic Reviews, part of the Cochrane Library (http://www.update-software.com/cochrane).

In pharma, meta-analysis is used to summarize results of a drug development programe and this is recognized in the International Conference on Harmonization (ICH) E9 guidelines (ICH, 1998). Meta-analysis is viewed as a formal evaluation of quantitative evidence from two or more trials bearing on the same question. Specific guidelines suggest that meta-analysis provides ways of summarizing total efficacy results during a drug application and of analyzing less common outcomes in the total safety evaluation. However, a warning was issued, since confirmation of efficacy in a meta-analysis does not necessarily mean it should be accepted as a substitute for confirmation of efficacy from individual trials. Of course, the magnitude of treatment effect can become an important factor in regulatory decision-making. If by any means the treatment effect is smaller than anticipated, then statistical significance may not be reached in the individual trials. Even if statistical significance is reached in the meta-analysis, the magnitude of the treatment effect may not be clinically significant, and thus be considered insufficient for approval (Whitehead, 2002).

Fisher (1999) took into account the two conditions under which one large trial might substitute for the two controlled trials usually required by the Food and Drug Administration (FDA) in the U.S.A. The first relates to the strength of evidence for demonstrating efficacy. He showed that if the evidence required from the two controlled trials is that they should each be statistically significant at the two-sided 5% significance level, then the same strength of evidence is obtained from one large trial if it is statistically significant at the two-sided 0.125% level. The same type of argument could be applied to combining trials in a meta-analysis (Whitehead, 2002).

Instrumentation

It would be plausible to set a more stringent level of statistical significance corresponding to proof of efficacy in a meta-analysis than in the individual trials. Fisher also treats the evidence of replicability by proposing criteria which need to be met by the one large trial. A metaanalysis will always involve at least two trials, and it will be important to assess the consistency of the results from the individual trials. Extents of inconsistencies amongst trials are influential in the choice of model for the meta-analysis and in the decision whether to present an overall estimate (Whitehead, 2002). Retrospective and prospective meta-analyses aim at:

• Providing a more precise estimate of overall treatment effects.

• Evaluating if overall positive results also appear in pre-defined subgroups.

• Evaluating additional efficacy outcome that might require more power than the individual trial.

• Improving estimation of the response relationship

• Evaluating apparently conflicting study results.

There is much to be gained by undertaking a meta-analysis of relevant studies, as it has a useful role to play in the generation of hypotheses for future studies. The Statistician is equipped with technical knowledge, identifying the trials, defining the eligibility criteria for trials, defining potential sources of heterogeneity and interpreting the results.

Over the course of the last twenty years there has been great development in refining statistical methods used in meta-analyses. Various different approaches give the impression that the methodology is a collection of distinct techniques. It is desirable to use general unified framework, and to place this framework within mainstream statistical methodology.

Statistics

The statistical approach in meta-analysis revolves on the idea of increasing the power of findings and reducing the so called "false positive" ones, where researchers believe the results apply to the general population, although only tested on a limited sample size.

Statistical advantages of MA include the fact that results can be actually generalized to a larger sample or population, researcher can use summary data -- which means no sharing of individual-level data is involved -- and can also add precision and accuracy to the estimates of other studies addressing the same or similar problems.

Among pitfalls, we have the boomerang effect: if the researcher(s) does/do not filter the information and studies accordingly, and uses studies with bad design or biased data, the meta-analysis will still result in bias, false estimates. Of course, sources of bias are not controlled by the method, which allow errors to interfere with the findings. Another thing the researcher should account for is the publication bias: meaning studies that show negative or insignificant results, as expected according to the "positivism" approach in research, are less likely to actually see the light of publication. So there are more positive, significant studies published than studies with little to no significance at all.

Specific statistical steps are always followed in a meta-analysis, involving:

Formulation of the research problem

Study selection: criteria, identification of specific studies on well-specified subjects, new subjects that have not been meta-analyzed or have been roughly meta-analyzed before

Decision to summary measures or dependent variables: for e.g., means, differences, relative risk

The Statistical analysis itself and reporting of findings

There are several methods to decide the outcome of a meta-analysis, including Fisher's method, Z score method, analysis of fixed and random effects and more.

Fisher's method combines the p values from independent tests who bear the same hypothesis (overall). When p values are small, the test statistic will suggest the null hypotheses are false for every test by indicating a large value. When nulls are true (generally, all null hypotheses are true), and all p values are independent, we have a chi-square distribution with 2m df (degrees of freedom). As an extension to dependent tests, we can have a scaled chi-squared distribution random variable, too. If the covariance is known or reported, the Brown's method is used. In cases in which covariance is not reported, therefore unknown, Kost's method is used.

The Z score method works with limitations of combining p values -- combining them may be spurious when the studies meta-analyzed do not show consistent direction of effects. It also does not include weights in a straight forward approach. If the researcher wishes to combine Z score, then the following rules must apply:

wi is the square root of sample size of the ith study.

Zi=?-1(1-pi)

Limitations of this method are directly linked to the fact that it cannot provide an overall estimate of effect size, as well as not being able to address heterogeneity issues between studies.

The Fixed Effect model is the most popular among meta-analyzed statistical approaches, as it allows the weighted average of effect sizes from a series of studies. In this case, Yi is effect size of study i, and it can be either a logarithm of Ors (odd ratios), beta-coefficients from regression models, mean difference or standardized mean difference for a continuous phenotype and more. The model treats the inverse of the studies' variance as a commonly used trait. In terms of study weight, the larger the studies meta-analyzed, the more they tend to contribute to significance of findings -- a reason to why there should be a limit of at least 15 studies for performing a meta-analysis. Formulas or principles followed in this model:

wi=1/Vi, Vi is the variance of study i

Z=M/sqrt (VM) is used to test the null hypothesis

The Random Effect model uses and incorporates the "between-study" variance, by considering the total variance. Here, the DerSimonian-Laird method is commonly used for between-study variance. Formulas and principles to look into include:

Wi

*=1/vi

*, vi

* is the total variance of study i.

Which brings us to the Heterogeneity issue. For Fixed effect model, the researcher assumes all studies meta-analyzed share a common true underlying effect and that observed variance is a reflection of sampling within-study error, which is minimized by assigning weights. As sources of potential heterogeneity in a study, these can vary. For e.g., some phenotypes are difficult to identify and standardize, such as the case of behavioral traits. Effect size could also be affected by sample characteristics, such as meta-analyses revealing higher effect size in the case of older, healthier and more educated individuals.

One of the most challenging type of meta-analyzed studies is genetic studies in which different ethnicity groups and different genotyping platforms are concerned. In these particular cases, the underlying true effect size might be different for different studies.

In order to test for heterogeneity, researcher should use Cochran's Q test. Under null, the Q test distribution is approximately a chi-square with k-1 degrees of freedom (condition). However, the test isn't powerful when meta-analyzed study number is smaller than 15 and studies show a within-study variance of large value. The Q test can also not be used as estimate for magnitude of true variance.

In order to quantify the heterogeneity in a meta-analysis, the researcher must expect values of Q test on the assumption that all studies meta-analyzed share a common effect size with df (degrees of freedom). In this case, the Q-df value is perceived as excess variation, and that part is attributed to differences in true effects, depending from study to study. The value of the test describes the percent of total variation across studies, total variation that is due to heterogeneity rather than chance. This particular test is not affected by N (number of studies) directly and it is more a measure of inconsistency across findings on meta-analyzed studies and less of a measure of real variation across underlying true effect sizes.

Values of 0% indicate no observed heterogeneity, while low is between 0 and 25%, moderate between 25-50%, large between 50% and 75% and large if value exceeds 75%. The debate is whether to use random effect model or fixed effect model.

While random effect model is robust, conservative and the heterogeneity test under powered due to poor between-studies variance precision, the fixed effect model is powerful, but not realistic with false positive outcome increases.

In such cases it is ideal to use the Bayesian approach for estimation of between-studies variance from outside of current set of studies, depending on priors. Ideally, a meta-analysis will start with a fixed effect model and switch to random effect model based on heterogeneity test.

The researcher has to keep in mind that heterogeneity tests suffer from lack of power and therefore, the statistical decision should be made based on the understanding of the researcher: whether or not all studies meta-analyzed share a common effect size, and not on outcome of statistical tests.

Significant studies

Studies have shown that pregnant women are afraid to tell about the abusive relationships to other people, as the spouses will find out sooner or later that word got out. The husband may double his aggressiveness. But this doesn't change the fact that the victim should defend herself (Woods, 2005).

Studies also show that victims should comprehend that getting violenced isn't their fault because mostly victims hold themselves responsible for all the violence. The violence isn't the cause of drugs, alcohol, stress and other factors but rather it's the choice of the violencer. He can stop the physical / verbal violence when possible. The violencer can restrain him from beating his wife. Hence the victims should comprehend that abusive relationship isn't their fault after all. Pregnant women should take some pointers when the violencer gets in his abusive form and avoid the situation at all costs. For instance, when the man is drunk arriving at home late, wife should confront him and keep some distance from him. In such cases, unpleasant conversations take place which instigates necessary evil in men. The victims must have an escape plan in case the violencer gets out of control. There are windows and doors to escape from, shut down the room for a while or make a phone call to someone who can rescue. There are some ways of stalling the violencer. Maybe, the violencer and victim should talk it out and resolve matters like educated citizens. As mentioned earlier, there are diverse causes of not abandoning a severed relation due to well-being of children, some don't abandon because they still love their husband no matter what and if violenced wives are trying to make it work, then their partner should do the same. Lack of communication is also a barrier instigating violence against pregnant women. The violencer due to lack of proper communication, a husband gets angry which commences the beating session. It's better to address the issues first and he will refrain when he gets what he wants. Some people don't learn at all and it's better to leave them alone because violencers can't change their nature. Women should take the first step and abandon the relationship because it's wastage of time. In some cases, the situation gets worse and violencer is about the kill the victim or the victim can commit suicide. Women and men for that matter should take prior notice and avoid such uncomfortable situations before it's too late. People living the society have different responses to violence and sometimes matters become more complicated than they were before. Some people think that it's their personal affair which should be resolved by them. No outsiders are allowed and even family members are stonewalled. This makes the situation harder than before as woman has no one to go to as family remains silent and so does the neighborhood. It's the way the world works. The society should learn that domestic violence isn't a private affair but rather treated as a society issue. It's about time that society doesn't bury its head in the sand and open their eyes to reality. Creating the world a better place is primarily important. Another factor is that people think that the violencer has periods of anger which drives abusive behavior. Being a drug addict, drinker, chain smoker and stressful has nothing to do with violence women endure, especially pregnant women (Woods, 2005).

Furthermore, abusing has nothing to do with either drugs or perhaps stress or even anger. This issue has to be controlled by the violencer himself or by the community, which ought to deal very strictly with such individuals, as well as, punish them for their lack of empathy. What policy should the governments adopt in order to prevent avert and stop violence against pregnant women?

Summary of literature

It is clear from the aforementioned studies that people really don't talk about domestic violence as it's a disturbing issue. The norms won't change with respect to physical and emotional violence if people can help it. The government should indulge into educating and counseling with the community pertaining domestic violence against women. Women will become more aware of their rights and defense tactics. When they come across such situations, they can report it as soon as possible. The society can eliminate domestic violence themselves. People should collaborate together and violencer should be severed punished. People with same vision will rise. In case the violencers don't face severe consequences, then people are encouraged to continue as they please. Society should eliminate the norm that it's a family matter because it's not. This affects all the individuals in the society. If children are raped and women are getting thrashed, it touches people and they should counter it. The violencers must be brought to justice (Woods, 2005). But before that actual situation on the ground must be comprehended with accurate and precise statistics.

Chapter 3

Research methodology

The aim of this study was to explore whether or not there is an association between intimate partner violence (IPV) and induced abortion among women of all age (teenagers to elderly). Demographic factors were taken into account as possible moderators in our study.

Four research questions addressed important matters which the meta-analysis tried to answer.

The first research question tested the value of the prevalence of intimate partner violence (IPV) amongst pregnant women. Is there a prevalence of IPV? And if so, what are the values and indices?

The second research question tested if there is a connection between intimate partner violence and Induced Abortion in the case of women ranging from teen moms to fully adults.

The third research question explored the risk factors for IPV among pregnant women and how they impact their lives. Risk factors such as before-after abortion were also taken into account.

The forth research question tested if there was a moderating effect of demographic variables (female sample characteristics) on the association of IPV and abortion among women of all ages.

Eligible studies for the meta-analysis included sample of women who underwent abortion procedures or were looking into performing such a medical procedure. Studies selected had to at least treat one aspect of IPV -- intimate partner violence -, and to be either a control trial based on randomization, case-control study, cohort study, cross-sectional analysis, experimental study, or secondary study. All data should be of interest and preferably peer reviewed. Studies that focused on secondary topics and did not directly relate to the main topic were removed. Restrictions were not place on the setting, time, or language of the studies. Quantitative data formed majority in the inclusion.

The targeted population for this meta-analysis consisted of women of all age who went through abortion or were considering of performing abortion in different settings. IPV was also taken into account, to see if there is an association between the presence and absence of intimate partner violence and if the men -- partners were knowledgeable about how their attitudes would influence the abortion decision of the woman. Control group consisted of women who did not report IPV and women who attended medical clinics, as comparable individual cases.

Basic search strategy included the search of all possible candidates for the meta-analysis by using open guidelines. Studies that truly met our criteria were located and included even highly irrelevant studies. Then, a master candidate list was compiled to prevent same studies from appearing more than once. This fairly happens when using one or more research libraries. Access was gained directly through the libraries or third-party links which allowed the articles to be fully downloaded or full-text online access. No authors were contacted for detailed statistical findings regarding their papers.

Data processing and analysis

A meta-analysis summary was carried out using the selected studies. More than 40 studies between the 2000s and up until recent dates were selected. 20 of them were used in the final dataset, while around other 20 studies were eliminated due to poor findings, incomplete or insufficient data, qualitative nature of approach and less quantitative findings or simply lack of value displays in the papers done by previous authors. A basic and specific search strategy was followed.

Examination was done on the list to check which studies met the criteria for inclusion and which didn't. The ones who met the criteria were kept, the ones that did not were disposed of. Title, year, abstract and keywords were analyzed before moving to the main contents and analyzing Methods and Results section. Most of the times, abstracts roughly contained enough findings summaries in order to decide if they were appropriate for a meta-analysis or not, hence a thoroughly examination of methods and results was needed. Out of the initial list, around 50% of studies were discarded, which is common in meta-analysis procedures. Each article was copied into the final meta-analytic sample.

An excel spreadsheet was used to assist this task, followed by an SPSS datasheet which allowed the coding of variables. Each study in the master candidate list recorded:

Reference to the study (journal name, volume number and starting page number) when permitted

Journal DOI and other identification indices

Current retrieval status of the study (in this case, all studies were requested through online library access

Whether study was included or excluded in the final meta-analysis model

Criterion/criteria used for exclusion, as mentioned above

If any article were unpublished or of private publishing method and not widely accessible to the public. However, no unpublished or private articles were found to add extra value to the meta-analysis and no bias control could be assessed.

Foreign studies were included in the analysis for a wider significance and generalization of the findings.

Specific search procedures included computerized indices collected through online databases available on the Internet. Specific search queries were used to identify the most appropriate relevant studies. Various online libraries with paid and free access were used to search for articles that contained keywords and synonyms such as "abortion," "violence," "young mothers," "termination of pregnancy," "assisted abortion," "teenage pregnancy" and "intimate partner violence." No NOT terms were used for this search method.

Searches were carried out in Medline (2000-2014), PsycINFO (2000-2013), and Ebsco (2000 until recent dates possible). In addition, a search of articles that contained a statistical or quantitative approach and less of a narrative review approach was carried out. Mean, percentage differences, Standard Deviation and confidence intervals were also taken into account when selecting prospective articles. None of the authors of chosen studies were contacted for further information or more articles related to the subject. Language restrictions were applied: only articles and studies presented in English were assessed. If multiple articles based on the same study were identified, duplication was avoided by only using the data reported for different sub-groups. In the same time, other meta-analyses published until the 2000s were accounted for, as a way to compare findings per question.

Intimate partner violence - IPV was coded as percentages and converted to log odds prior to combining. A fixed effects meta-analysis model was carried out on the extracted data, as well as direct effects of variables. Between group effects were carried out using moderating variables (demographic data). Resulting estimates and confidence intervals were displayed in the findings. Forest plots were also used in the meta-analysis. Comparisons between groups used odds ratios (ORs). Estimated heterogeneity (I2) was displayed for all group and sub-group analyses. We investigated the possibility of a sub-group of recent or high-quality studies with consistent methods and consistent results that could be used to give generally applicable results.

The method used in determining heterogeneity assumes that the differences are in part random and in part explainable. Potential sources of heterogeneity were investigated: date of study (year), study design (cross-sectional, cohort, or case-control), and study size (total number of participants as a continuous variable). Other tests were also performed in order to assess potential publication bias.

Methodological assumptions

The current meta-analysis followed assumptions of methodological diversity (heterogeneity) and statistical heterogeneity.

Methodological diversity (heterogeneity) particularly referred to differences between how the studies were filtered and coded, including variables as group, control, crossover trial randomization by clusters or by individuals, study quality, analysis etc.

Questions related to study selection included:

a) Do any of the above mentioned methodological heterogeneity differences make the results show the opposite effect to the one we want?

b) Do these differences make the outcome particularly significant?

When good evidence to back up these suspicions were found, it was redeemed inappropriate to pool all the studies together.

Another methodological heterogeneity assumption was that not every factor that influences prognostic factors will influence the size of the treatment effect.

The most important decision when performing a systematic review is whether or not to combine studies. Each decision should be made based on individual outcome of every comparison in the review. It is possible to perform a meta-analysis for some comparisons, but not for all, depending on individual studies addressing the comparison. The decision to combine studies in a meta-analysis should be made based on the setting, participants, interventions and outcomes of the included trials being sensible to combine. In this case, statistical heterogeneity helped in easing the decision making process.

Statistical heterogeneity considers the estimates of treatment effect of the individual studies. As meta-analysis estimates the combined effect from a group of similar studies, the researcher needs to check that these effects are found in the individual studies and are compatible and similar, in a way in which the combined estimate is a meaningful description of the set of studies.

Statistical heterogeneity claims that individual estimates of treatment effect will vary by chance, due to randomization technique. This also means some variation should be present. The researcher needs to know whether there is more variation than it is expected by chance alone. When excessive variation occurs, it means the study was statistical heterogeneity, or just heterogeneity issues.

Statistical heterogeneity can be analyzed by a Forest Plot, which shows how well CI (confidence intervals) overlaps. In cases in which two studies do not overlap at all, then there is more variation between them than what we would normally expect by chance. Therefore heterogeneity is suspected. Another way to determine heterogeneity is to perform a statistical Chi-squared test. In chi-squared statistic test the degrees of freedom are the results and p-values obtained are analyzed to determine heterogeneity. A small p-value (less than .010 for a sample n=15) indicates evidence of heterogeneity.

In the event that important heterogeneity is present in the review, there are several options to choose:

Use different statistical model for combining studies (random effects meta-analysis)

Investigate heterogeneity by splitting studies into subgroups and looking at the forest plot

Investigating heterogeneity using meta-regression

Fixed and random effects meta-analyses can also offer results that may be similar or differ. Methods of fixed effect meta-analysis work on the mathematical assumption that a single common (or 'fixed') effect underlies every study in the meta-analysis. For e.g., by performing a meta-analysis of odds ratios, the assumption would be that every study is estimating the same odds ratio. Under this assumption, every study is infinitely large and yields an identical or similar result. This is the same as assuming there is no (statistical) heterogeneity among the studies.

A random effects analysis starts by assuming individual studies as estimating different treatment effect, with a distribution with central value and degrees of variability. With a random effects meta-analysis, information is obtained about this distribution of effects across different studies. By convention (but unfortunately) most interest is focused on the central value, or mean, of the distribution of effects. It is also important to know the variability of effects.

If both fixed effect and random effect meta-analyses give identical results then it is unlikely that there is important statistical heterogeneity, and it doesn't matter which method is assessed.

There is a great deal of debate between statisticians about whether it is better to use a fixed or random effect meta-analysis: a stable robust technique with unlikely underlying assumption (fixed effect) or less stable, unpredictable technique based on more likely assumption (random effects).

Methodological limitations

As expected, performing a meta-analysis brings us to various limitations. There is no perfect statistical procedure in data analysis and meta-analysis is far from providing perfect results that can easily be generalized to the entire population. What it can be said for sure is that meta-analysis provides a series of probabilistic findings that can be applied to the general population in particular cases. However, even for a 95% confidence interval, there is still 5% probability the findings will not represent the general behavior or trait. Here are a series of limitations identified during the procedure:

a) A literature review is usually done in order to generalize the differences in primary research techniques. While it is true that the bigger the sample size (N = minimum 15 studies) the more control the researcher has over biased information, this does not mean bias can be fully controlled or eliminated. Overgeneralization of results to an entire population can occur in meta-analyses too, just as they occur in narrative reviews.

b) The fact that researchers focus solely on quantitative data and ignore the qualitative differences between studies does not do justice in the case of meta-analysis. In truth, while meta-analyzed data does not state qualitative differences as being non-existent, it simply combined and treats them as moderating variables, testing their influence in an empirical manner. This is not necessarily a positive outcome in meta-analysis and therefore can be considered as a limitation of the procedure itself.

c) The selection process itself is not necessarily bias-free. The researcher falls to its own subjectiveness when classifying "the good studies" from "the bad studies." The filtering of studies that best represent the object of the study is somewhat of a non-randomized method, where the researcher clearly selects the most appropriate data for the meta-analysis. However, it is easy to separate good meta-analyses from poor ones, which lifts from the limitations of the procedure itself.

d) The procedure itself ignores study quality, by coding the effect of study quality as moderator factor, which helps the researcher understand the difference between good and poor studies. However, this is solely done for the purpose of identifying low quality studies that may interfere with our data and thus, remove them from the analysis. This does not treat study quality at an individual level, but rather treats it as general.

e) Consistency and validity are great limitations in a meta-analysis. We cannot draw valid conclusions because 99% of studies are published only if the findings are significant. While meta-analysis procedure is not so much affected by this bias, it is still present in the statistical outcome and cannot be reduced, removed or avoided. So consistency and validity conditions are not fully met before performing the analysis itself.

f) A meta-analysis procedure deals only with main effects, while the interaction effects are measured through moderator analyses and moderator variables. Instead of following the classic statistic test approach of within and between variables/groups main effects and interactions, meta-analysis focuses only on revealing findings that concern the overall variable, while ignoring the particularity (groups) factor.

g) The very nature of a meta-analysis procedure which implies objectivity is actually inversed, as its proponents are really subjective, the procedure relying more on shared subjectivity.

However, every analysis of statistical nature requires specific subjective decisions, most of which are related to sampling, be it human participants or studies in the same field. It is important for findings to be stated explicitly in a manner which leaves the door open to criticism.

Ethical considerations

The authenticity or validity of a study is what lies behind its failure or success. Even the smallest level of illegitimacy can create the whole research a drowned effort. For any kind of data ethics lays in the depth, integrity and capacity of the information collected along with the targeted sample and the impartial strategy employed by the researcher. In this study, the researcher utilized the meta-analysis with the same process for all studies in the sample. This ensured that the dissimilarities present in the studies are genuine: hence the methodology will not be disapproved for being unethical, subjective or impartial in any way (Trochim, 2006).

Chapter 4

The aim of this study was to explore whether or not there is an association between intimate partner violence (IPV) and induced abortion among women of all age (teenagers to elderly). Demographic factors were taken into account as possible moderators in our study. Four research questions addressed important matters which the meta-analysis tried to answer.

The first research question tested the value of the prevalence of intimate partner violence (IPV) amongst pregnant women. Is there a prevalence of IPV? And if so, what are the values and indices. Findings show that in most cases prevalence of intimate partner violence was present (M=1.62, S.E.09, SD=.49).

The second research question tested if there is a connection between intimate partner violence and Induced Abortion in the case of women ranging from teen moms to fully adults. IPV, including history of rape, sexual assault, contraceptive sabotage, and coerced decision-making, was associated with Induced Abortion (F (1, 2912)=5.293, Sig.=.032, p < .05), with strong observed power indices (.593).

The third research question explored the risk factors for IPV among pregnant women and how they impact their lives. Risk factors such as before-after abortion were also taken into account. Findings showed that risks before abortion were more present than those after abortion. Before abortion risks included depression, anxiety and feelings of guilt. Risks before abortion reported a medium/moderate power of .356, with R. Squared of .096 and adjusted R. squared of .026. A mean of 31.38, S.E. Of 6.270 and 95% CI [18, 49; 44, 27] was found in terms of abortion rates, which means risk factors present in pregnant women's lives can lead to a rate of 31% in terms of abortion on average, ranging from 18% to 44% of women actually terminating their pregnancies.

The forth research question tested if there was a moderating effect of demographic variables (female sample characteristics) on the association of IPV and abortion among women of all ages. Only Socio-economic status was found to have a significant effect (F (1, 4802)=8.726, Sig.=.008, P < .05), with extremely high values of observed power (.804) and partial eta squared values (.294).

Effect size, power and heterogeneity were also tested for the overall model. Levene's test for Equality of Error Variances was significant (F (1,18)=5.22, Sig = .035), meaning the error variance of the dependent variable was equal across groups. Tests of Between-Subjects effects for the dependent variable showed an Observed Power value of .194 and Partial Eta Squared of .264 (F (1,45.69)=1.33, Sig.= .264), which means a weak effect. Results were computed using alpha = .05, with R. Squared = .069 and Adjusted R. Squared = .017. Parameter Estimates indicated a Partial Eta Squared value of .069 (t (2,67)=1.15, Sig = .264) for a confidence interval of 95% (-2,53, 8.70).

Heterogeneity was tested using Chi-square and Kendall's Tau-b. Chi-square value of 21.84 was reported, with df = 15 and p=.112.Likelihood ratio value was 25.78, df=15 and p=.040. Approx. Kendall's tau-b value was -.961, with approx. Sig.=.336 and Asymptotic standard error of .138. No evidence of heterogeneity across data was found. (Table 1)

Table 1. Descriptives, Heterogeneity and Effect Size Overall Model

N

Mean

Std. Deviation

design

20

1.8500

.98809

sample size

20

22757.33672

no of countries in study

20

7.9000

19.56877

% or abortion

20

36.8000

26.30509

prevalence of intimate partner violence

20

1.4500

.51042

intimate partner violence rate

20

1.4000

.50262

risk factors before abortion

20

.8000

.95145

risk factors after abortion

20

.4500

.51042

perceived support

20

1.7500

.78640

Age

20

17.5000

3.95368

Education

20

1.9500

.68633

SocioEconomicStatus

20

.8000

.61559

Residence

20

2.0500

.39403

PressureTOP

20

1.1000

.30779

StatisticTest

20

1.9500

.75915

Valid N (listwise)

20

Heterogeneity: Tau2 = -.961; Chi2 =21.84, df = 15 (P=.112)

Test for Overall Effect: Partial Eta Squared = .069, (p > .05), 95% CI [-2,53, 8.70]

Table 2: Summary Descriptive Statistics for Each Variable

Min

Mean

Mean

SD

Variance

Statistic

Statistic

Statistic

Std. Error

Statistic

prevalence of IPV

1.00

2.00

1.620

.091

.493

.244

perceived support

1.00

3.00

1.827

.140

.759

.576

Age

10.00

28.00

17.689

.691

3.752

14.079

PressureTOP

1.00

2.00

1.103

.057

.309

.096

StatisticTest

1.00

3.00

1.931

.148

.798

.638

Value

.00

22.40

3.856

1.23

6.621

43.883

% of TOP

4.00

90.00

36.620

5.055

27.20

IPV rate

1.00

2.00

1.275

.084

.454

.207

riskfactors before

.00

2.00

.8621

.177

.953

.909

riskfactors after

.00

1.00

.482

.094

.508

.259

Education

1.00

3.00

1.930

.130

.703

.495

SocioEc Status

.00

2.00

.758

.118

.635

.404

Residence

1.00

3.00

2.069

.084

.457

.209

design

1.00

4.00

1.931

.198

1.066

1.138

sample size

8

102394

26146.69

683649736.56

no of countries

1.00

78.00

6.206

3.045

16.39

year

2008

2014

2011.13

.335

1.80

3.26

Table3. Pearson Correlation & Descriptives Q2

Descriptive Statistics

Mean

Std. Deviation

N

Abortion Rate

-.0068

1.03415

20

IPV rate

-.2470

.90497

20

Correlation Pearson between Abortion and IPV

IVP Rate

Abortion Rate

Pearson Correlation

.165

Sig. (2-tailed)

.393

N

20

IVP Rate

Pearson Correlation

1

Sig. (2-tailed)

N

20

Figure 1. Chart with incidence of reported intimate partner violence among women who seek termination of pregnancy (2008-2014)

Step 1

Article Summary

Out of 40+ retrieved studies, only 20 were included in the meta-analysis with a total of 17 variables and 141,756 participants. Majority of studies meta-analyzed were either cross-sectional (n=9) or case-control (n=7), dating from 2008 to present. Descriptive statistics pointed out an average (mean value) of 2011.13, meaning most studies were from 2010-2012, and a standard error mean S.E. Of .335, with standard deviation of 1.80 and a statistic variance of 3.26. Data from 78 countries was included in the analysis (M=6.20, S.E. = 3.04, SD=16.40, VS = 268.96). All studies addressed abortion in women age 10-54. Cases were weighted by prevalence of IPV (intimate partner violence).