Analyze GSS Data on Crime
Step 1: Research Scenarios
Research suggests that the experience of victimization may be linked to poor psychological well-being. Using any form of victimization captured in the 2009 GSS on victimization (e.g., personal victimization, household victimization, intimate partner violence/abuse, and cyber-bullying), compare those who reported victimization to those who did not on any psychological outcome (e.g., general mental well-being, life satisfaction, stress, substance use).
To choose independent variables, search under the headings: “Most Serious Crime”, "Emotional/Financial Abuse by Ex-spouse/partner (EFX)", "Physical/Sexual Violence by Ex-spouse/partner (PSX)", and “Cyber Bullying of Respondent (CBR)” To choose dependent variables, search under the heading: “Self-Rated Health (SRH)”, "Drinking of Respondent (DRR)", and "Drug Use of Respondent (DUR)"
Once you have created your proposal, received feedback, and refined the direction of your research project, the next step is to begin preparing and analyzing your data. This is the work that occurs ‘behind the scenes; it is an iterative process of preparing variables, analyzing the data, and refining the analytical approach until a final product is reached. This is the component of quantitative research we are not subjected to, as we often only read the published article which lacks any indication of the hard work that occurs prior to developing the final product. Your main objective for this assignment is to draft a semi-formal report explaining, in detail, the step-by-step process that was taken to conduct statistical analyses aimed to address the particular research scenario you have chosen (based on Assignment One) and your proposed research question.
Intimate Partner and Spousal Violence Effect on Victim’s Mental Health and stress
Intimate partner and spousal violence (IPV) is a global problem affecting more than 2 million women and 800,000 men worldwide1. (Garcia-Moreno, Heise, Jassen, Ellsberg, & Watts, 2005). It impacts significantly the victims’ ability to have a happy and productive life. According to a 2002 report by the World Health Organization, IPV has huge physical, emotional, and economic costs and may lead to death in some cases. The World Health Organisation in its 2010 report defines IPV as “behavior within an intimate relationship that causes physical, sexual or psychological harm, including acts of physical aggression, sexual coercion, and psychological abuse and controlling behaviors” [1, page 11]. There is numerous mental health consequence of IPV which includes post-traumatic stress disorder (PTSD), depression, low self-esteem, anxiety, and substance abuse (Karakurt, Smith, and Whiting (2015)). There may also be mental health consequences for children who witness such abuse (Golding 1999) but will not be explored in this study. Based on the stated consequences of IPV on health-related outcomes, this study seeks to determine the effect of IPV on two health-related outcomes: Mental health and stress. Specifically, we will determine if mental health and stress differ based on whether one has suffered/is suffering from domestic violence.
Even though there is a growing body of literature have studied the relationship between IPV on health-related outcomes (Golding (1999), Garcia-Moreno et. al 2005, Karakurt et. al 2015, Dillon 2013), most of the articles focused only on women because the prevalence rate is higher because violent behavior is more directed at women than men (Karakurt et. al 2015). Even though we will not explore the gender differences in IPV, we are going to use samples of both men and women for this study.
The significance of this problem is that knowing that people who suffered from domestic violence face worse mental health and increasing stress will point out the menace of domestic violence. Therefore, all hands will be on deck to curb the menace of domestic violence so that all associated costs and death will reduce.
Studies have examined various aspects of mental health implications of Intimate partner and spousal violence (IPV). Karakurt et. al (2015) identified comparisons and contrasts in the mental health needs of women residing in violence shelters. Using hierarchical clustering on data from 35 women, they differentiated the women into 3 clusters which are ready to change, focus on negative symptoms, and focus on feelings of guilt and self-blame. Moreover, stress and physical abuse were the commonly reported problem for the ready-to-change group. The focus on negative symptoms report problems relating to depression, sadness, unhappiness, fears, and stress. The third group reported sleep problems, nervousness, temper, guilt, and shyness as their own problem. Dillon (2013) undertakes a review of 75 papers published from 2006 to 2012. They found that health issues that were associated with domestic violence include depression, PTSD, sleep disorders, self-harm anxiety. Cooker (2002) studied both physical and mental health implications of intimate partner violence among men and women using data for women and men aged 18 to 65. They found that IPV was associated with an increased risk of poor health depressive symptoms, substance use, chronic mental illness, and injury. Garcia-Moreno et al. (2008) studied the effect of domestic violence on women’s health. The data was collected through interviews in ten countries. Logistic regression analysis was carried out on data from 24,097 women to determine the association between health outcomes and IPV. They found out that those women who had life experiences of physical or sexual violence, or both, by a partner were more likely to report poor or very poor health. All the evidence pointed to the fact that intimate partner violence is associated with poor health conditions. However, all cited papers focused on women, while Cooker who studied men and women studied them separately. It will be interesting to determine whether the relationship will be maintained or change when data for both men and women are aggregated. Secondly, the methods employed in those works do not compare the health outcomes of those who had to experience domestic violence and those that have not which I believe is more straightforward and intuitive. The logistic regression methods of Ellsberg (2008) compare groups but the odds ratios are not so intuitive. Therefore, this study compared the mental health and stress of respondents who had experienced domestic violence and those that have not. Specifically, the study answered the following research questions:
- Is there a significant difference in the mental health of those who had suffered lifetime emotional abuse and those that have not?
- Is there a significant difference in the stress of those who had suffered lifetime emotional abuse and those that have not?
- Is there a significant difference in the mental health of those who had suffered physical abuse in the last five years compared to those that have not?
- Is there a significant difference in the stress of those who had suffered physical abuse in the last five years compared to those that have not?
The participants for the research will be respondents from the GSS 2009 survey, with which there are 19,422 respondents. The survey included all persons 15 years of age and older in Canada, excluding residents of Yukon, Northwest Territories, Nunavut, and full-time residents in institutions. The sampling technique is probability sampling where respondents are contacted and interviewed through randomly sampled landlines. The strengths of using this survey is that it is representative of those 15+ in the Canadian provinces who have landline in their homes. The limitations of using this survey are that it does not account for Yukon, Northwest Territories, Nunavut. Moreover, since it involves random digit dialing of landlines, those with only a cell phone or those without a house are excluded.
The variables I intend to use are two independent variables and 2 dependent variables. The independent variables are “Emotional abuse (lifetime) by ex-spouse/ex-partner and “Respondent has received physical abuse (during the past 5 years) by ex-spouse/ex-partner”. These will be coded as EPQ1 and DRPQ1 respectively. They are categorical variables, so respondents will fit only into one category of response; Yes, no, don’t know, not asked, not stated. However, for this analysis, only those who answered yes or no will be considered. There are two dependent variables that will be used: “In general, would you say your mental health is” and “thinking of the amount of stress in your life, would you say that most days are”. These variables are ordinal variables with 5 categories for mental health – excellent, very good, good, fair, poor – and 5 categories for stress also: not at all stressful, not very stressful, a bit stressful, quite a bit stressful, and extremely stressful.
The dependent variables are ordinal variables that do not have the central tendency and have no normal distribution. The values of ordinal have no midpoint but are evenly distributed which defeats the assumption required for an Independent t-test which would have been appropriate for a continuous, normally distributed dependent variable. Consequently, we will use the Mann-Whitney U test which is the non-parametric equivalent to the independent t-test. The Mann-Whitney U test is appropriate for the ordinal dependent variable, categorical independent variable with two groups, and independence of observations which these data meets.
For the data analysis, mental health is coded such that excellent has the lowest value while poor have the highest value. Similar to stress, not at all stressful has the lowest value while extremely stressful has the highest value. I will recode both dependent variables such that for mental health, excellent has the highest value while poor has the lowest value while for stress, not at all stressful has the highest value while extremely stressful have the lowest value. Missing values will be identified by frequency table and excluded list-wise. List-wise deletion means if one of the variables has a missing value on observation, the whole data for such respondents will be removed from the analysis.
Some challenges that I may face during my research project may include how to account for response data that was not either “Yes” or “No” because I am using a two-group method which is what is appropriate given my research question. There is no other option than to remove them from my analysis. Unfortunately, only 14.8% (2,877) of the respondents answered “Yes or No” on the independent variables, and 85% were not asked. This means the highest number of observations that can be used for this analysis is 2,887 which is only 14.8% of the dataset but it is large enough for the analysis we are interested in.