Analyzing Intimate Partner Violence and PTSD

This assignment focuses on analyzing the connection between Post-Traumatic Stress Disorder (PTSD) and Intimate Partner Violence (IPV). It utilizes a Chi-Square test to examine the statistical significance of this relationship. The assignment delves into sampling methods, specifically highlighting non-probability sampling techniques, and references relevant research studies that employ similar analytical approaches. The content also emphasizes the importance of understanding competing risks in survival analysis, particularly within the context of trauma research.

Chapter: 1 Paper Review
Contents
1.1 Introduction to the article.............................................................................................................1
1.2 Sample used in the study.............................................................................................................1
1.3 Selection of research methodology..............................................................................................2
1.4 Review of result and discussion...................................................................................................4
1.5 Alternative techniques for the analysis.......................................................................................7
1.5.1 Using Linear Discriminant Analysis.......................................................................................7
1.5.2 Using logistic regression.......................................................................................................7
1.5.3 ANOVA and t test.................................................................................................................8
1.1 Introduction to the article
The main of the current research is to analyze the paper by (Yoshihama & Horrock 2003) on the
basis of the statistical methods used in the paper. Also an alternative analysis which can be used
for such analysis will be proposed at the end of the research paper. There can be different
techniques which can be used for the same analysis. The choice of techniques differs on the basis
of collected data, the aims and objectives of the research and also on individual preferences of
techniques. In this research also an alternative techniques for the analysis of data will be
presented(Rajasekar et al. 2013; Cierniak & Reimann 2011). The use of survival analysis for the
research related to trauma is not very common. Most of the previous researchers have used eithet
the chi square test or the Cox regression where the covariates is not time variant.

This research paper has been divided into different sections. In the first section the review about
the data collection and the sample included in the study has been analyzed; followed by the
research methodology used in the research. After the review of the research methodology, in the
third section the results from the selected articlecurrent study have been reviewed. On the basis
of the research methodology and the results from the data analysis, alternative techniques for the
data analysis have been proposed in the fourth section. Major limitations of the selected paper
have also been discussed before proposing the alternative techniques.
1.2 Sample used in the study
As per the selected article, in the selected paper primary data was collected among the Japanese
women who were either migrated from Japan to United States or Japanese women born in United
States. In other words the target population of the study was the Japanese women living in
United Kingdom. It has been mentioned in the given article that the final sample size used in the
study was only 211. Even though the survey was sent to 407 respondents, only 56 % of them
responded. This shows that the response rate was above 50 %. However the response rate could
have increased if the respondents were given some rewards for providing information (Daniel
2011; Battaligia 2011). In this case the respondents were given $20 each for filling the
respondent’s sheet.
Similarly aAnother reason for the low response rate is may be because the Japanese women do
not want to reveal their partner violence by their intimate partner. Authors of the study has
maintained the privacy of the respondents in the research , however some female do not want to
reveal, that is why the some of them do not respond. According to the article to select the
women for the sample introductory letter was sent to their houses. Also the screening telephone
call was also conducted to identify the eligibility of the women. For the sample selection the
random sampling was used. In case of random sampling each element in the population has same
probability of being selected in the sample. On the other hand in case of non-random sampling
the probability of being selected in the sample differs for each element (Teddlie & Yu 2007).

In terms of the age of the respondents, only the Japanese women between the age group 18 and
49 years were taken into consideration. Since most of the partner violence is faced by the
females in this age group, this group has been selected. Even the appropriate age group was
selected for the study; sample was restricted to specific ethnic group. In this case only the
Japanese women were included in the sample set. If sample from other ethnic group was also
selected for the study it would have been easier to generalize the results. So the research can be
improved in terms of the sample selection and also the sample size can be increased.
1.3 Selection of research methodology
In this section the selection of the research methodology by the authors in the selected research
paper has been discussed.
Use of survival analysis
As discussed in the previous section the paper was aimed to examine the relationship between
posttraumatic stress disorder (PTSD) and the intimate partner violence. Authors tried to study the
time taken for the PTSD (an event), so survival analysis was one of the most appropriate
techniques. In case of the survival analysis correctly identifying the target population and
collecting unbiased data from that population is the most important challenge. In this case
authors were able to identify and collect data from appropriate population which makes this
research paper more important in the existing literature (Indrayan & Bansal n.d.; SABR et al.
2013).
There are mainly three possibilities related to data which can be used in the survival analysis.
The first observation group is the one where the event occurred in the given time period or the
examination time period. For the current research paper it is in the age group of 18-49 where
there PTSD. The second possibility is related to the respondents being dropped from the study in
the given time period or in certain age group. Finally the third possibility is where the selected
event do not occur. This is also known as the censored observation. Censoring of the observation
occurs when the researcher have only some information about the selected event’s time, however
the exact time of the event is unknown. There may be different reasons behind censoring such as

selected person’s withdrawal from the event, selected subject not experiencing the event
etc(SABR et al. 2013; Monem A Mohammed 2014).
In case of survival analysis majorly three functions can be presented:
a) Probability of the selected subject’s survival after the given time t
b) Hazard function h(t), ( it is the rate at which the event occurs, given the condition that the
there are no previous events
c) Cumulative Incidence function (CIF) – In survival analysis it will be 1 minus survival
function
Use of SAS for survival analysis
For the selected research paper authors have used SAS to analyze the data using the survival
analysis technique. Cox proportional hazard models were used by the authors to analyze the data.
Since the Cox regression was used in the research, certain assumptions need to be fulfilled. The
Cox regression assumes that the independent variables are in the non-parametric form. However
it allows the independent function to take any form as long as it is related to survival
function(Lanfranchi et al. 2010; George et al. 2014).
Apart from Cox regression there are some other methods which are used for survival analysis.
One of the methods is the Kaplan –Meirer survival method which is mostly used in case of time
to event analysis. This method is also non-parametric and this method is mostly used in
describing the survivorship of the selected population.
On the other and the Cox regression survival method is considered to be a parametric model.
This method is mostly used in those cases where the research is aimed to find the relationship
between different covariates. The outcome of the survival analysis is the time to some selected
event. Using the survival analysis one can estimate the distribution of the outcome variable and
examines the effect of the independent variable on this distribution.
There are two different ways to visualize the change of an event. The first one is the CIF and the
value of CIF at certain age t indicates the probability that the selected event occurs before time
period t or at t. Another similar method is to use the survivor function. In the selected paper

authors have used the CIF instead of survivor because the study was focused on studying the risk
of developing the PTSD and not on remaining free of PTSD (Gillam et al. 2010).
1.4 Review of result and discussion
Results from the selected paper shows that among the 211 females included in the sample only
115 have said that they have experienced the violence from their intimate partners. Similarly 30
of them have suffered from PTSD in past. The descriptive analysis of the demographic profile of
the respondents included in the study showed that the average age of the women was 37.2 with
standard deviation of 10.2. The high value of standard deviation shows the high variation in the
data set. If the variable is normally distributed then 68% observations lies within the range of
mean ± SD. Similarly results also show that during the time of interview around 57 % of the
women were married whereas 10% of the respondents were either divorced or separated. Out of
211 respondents in the study 27% women were not born in United State. In terms of employment
around 78 % of the respondents were employed and more than 50 % hold college degree. The
average annual income of 36 % of the respondents’ was more than $60000. Furthermore annual
income of around 45 % of the respondents was $15000 or less.
The result from the cox regression and CIF plot are shown below.
Figure 1 Results from the Cos regression

As shown in the results from the cox regression the intimate partner violence is related with
increase in the probability of developing PTSD when other variables included in the study are
controlled. Other variables in the study includes sexual victimization, immigration status and
childhood abuse.
Results from the CIF plots show the cumulative probability of suffering the intimate partner
violence and PSTD. The probability has been shown by age of the respondents in the study. As
the results show with increase in age the probability of PSTD also increases. The sudden jump in
the cumulative probability after the certain age group can be easily seen from the graph. This is
the reason why the age group between 18 and 49 was selected for the current study. However
the probability for the maximum age group (49) also increases rapidly. Even though the increase
was significant no justification has been provided in the research paper to explain such hazard
function. If hazard function was explained with justification, it would have been easier for the
readers to validate the assumptions for survival analysis and cox regression (Lanfranchi et al.
2010; Indrayan & Bansal n.d.). Similarly another problem in the result is that most of the
participants did not experience the selected event in the given age limit. So the appropriate
selection of the age group for the study is contracting the findings from the research. There was
lack of the thorough study while selecting the sample and it also shows lack of background
research on Japanese women. It would have been better if the intimate relationship of the
Japanese women was studied separately before staring the current study.

Figure 2 CIF plot from the study
Also the 28 % of the total population was censored which indicates that the analysis was not
unbiased as the skewness in the data increases after censoring (Donoghoe & Gebski 2017).
Similarly, one of the assumptions in the current research paper was that the PTSD was caused
only due to the intimate partner violence. However there can be other reason for the PTSD
among the samples included in the study. So, there is scope to improve the assumptions in the
research paper and other factor causing the PTSD can be explored. Apart from that discovery of
PTSD among the respondents can be different. In some cases the PTSD can be discovered in
their later age, which was not taken into consideration in the current research paper. Results from
the cox regression shows that the regression used for the time varying covariates was statistically
significant. On the basis of these results it can be argued that the cox regression method is
dynamic and shows both the changing rate of PTSD over the period of time for respondents and
also the number of individuals with changing risk.

Even though the paper shows significant results for the cox regression and time varying
covariate, there were some limitations associated with the paper. The first one is that the reason
behind the cause of PTSD among the respondents. Paper assumes that it only due to intimate
partner violence whereas there may be other reason associated with PTSD. Another limitation is
related to the sample selection in the study. Only the Japanese women were included in the
study. Including sample from other groups can improve the results and also the sample size will
increase. Similarly, the research is based only on the quantitative analysis and no qualitative
analysis is conducted. Qualitative analysis can provide more details analysis about the
relationship between PTSD and the intimate partner violence.
1.5 Alternative techniques for the analysis
Some of the alternative techniques which can be used for the research paper are as follows:
1.5.1 Using Linear Discriminant Analysis
Linear Discriminant Analysis (LDA) is used to find the linear combination of variables in the
model that separates two or more than two classes of events. It is considered as the generalized
model of Fisher’s linear discriminant method. It is widely used in the studies related to
biomedical where the patients are categorized into different section depending on the severity of
the diseases. In this case also the LDA can be used if the PTSD is divided into different
categories such as sexual, physical and verbal trauma. One the category is defined then different
factors can be identified for each trauma group.
1.5.2 Using logistic regression
Logistic regression is one of the most used statistical methods where the dependent variable is
binary. In other words the dependent variable can take only two values 0 and 1. In this case the
number of independent variables can be one or more than one, which will determine the
outcome. The dependent variable is coded as 1 if case of true or success, whereas the outcome as
False or failure are coded as 0. Logistic regression is aimed to identify best model which
establish the significant relationship between the outcome variable (dependent variable) and the
explanatory or the independent variables (Dutta 2012; McCarty & Hastak 2007). However for
the current research the use of logistic regression is not appropriate. This is because the selected

event is happening over the period of time. Since the logistic regression do not take into
consideration the time factor it is not advisable to use.
To perform the logistic regression some assumptions must be fulfilled.
a) The conditional distribution of y/x should follow the Bernoulli distribution. If y/x follows
the normal distribution then it is not appropriate to use the logistic regression
b) The predicted values lies between 0 and 1, this is because the output from the logistic
regression is in the form of probability
1.5.3 ANOVA and t test
Another alternative method is the use of ANOVA and t test. The analysis of ANOVA and t test
is similar. If there are only two variables then t test can be used. Furthermore if the number of
variables is more than two then one should apply ANOVA test. Both ANOVA and t test is used
to measure population’s mean and show whether there is statistically significant difference
between them or not. T test is used t statistics and the degrees of freedom while determining the
probability of two different mean from population (Kuada 2012). There are some important
assumptions which should be fulfilled before conducting the t test or ANOVA test:
a) Variables for the test should follow the normal distribution. If the variables are not
normally distributed then the Wilcoxon rank test is used instead of t test.
b) ps2 follows a χ2 distribution with p degrees of freedom under the null hypothesis,
where p is a positive constant
c) Z and s are independent.
One of the alternative for the selected research paper can be t test. One can perform t test and
examine whether there is statistically significant difference in the occurrence of PTSD with
intimate violence of partner. However the data included for the study is censored so the t test
cannot be applied. The censoring data do not follow the normal distribution, which is one of the
most important assumptions of t test.

1.5.4 Chi square test
Chi square test is used to determine whether any significant relationship exists between
categorical variables. In other words chi square test is to test whether statistically significant
difference exists between observed frequencies and the expected frequency of the categorical
variables. There are certain assumptions which need to be fulfilled for chi square test:
a) Variables used in the chi square test should be categorical. The variable can be either
nominal or ordinal. In case of nominal variable the order of the category does not matter,
for example gender, month of the year etc. However in case of ordinal variable specific
order has to be followed, for example income, age etc.
b) There should be at least two category in the variables included for the chi square test
Many previous researchers have used the chi square test in the traumatic research. In the selected
article also chi square test can be used to examine the relationship between PTSD and the
intimate partner violence.
References
Battaligia, M., 2011. Non Probability Sampling: Enclycopedia of Survey Method, New Delhi:
SAGE Publication.
Cierniak, G. & Reimann, P., 2011. Specification of Research Strategy and Methodology,
Daniel, J., 2011. Sampling Essentials: Practical Guidelines for Making Sampling Choices,
SAGE.
Donoghoe, M.W. & Gebski, V., 2017. The importance of censoring in competing risks analysis
of the subdistribution hazard. BMC Medical Research Methodolog, 17, pp.17–52.
Dutta, A., 2012. Prediction of stock performance in the Indian stock market using logistic
regression. International Journal of ….
George, B., Seals, S. & Aban, I., 2014. Survival analysis and regression models. NCBI, 21(4),
pp.686–694.
Gillam, M.H. et al., 2010. Competing risks survival analysis applied to data from the Australian
Orthopaedic Association National Joint Replacement Registry. NCBI, 81(5), pp.548–555.

Indrayan, A. & Bansal, A., The Methods of Survival Analysis for Clinicians, New Delhi.
Kuada, J., 2012. Research Methodology: A Project Guide for University Students,
Samfundslitteratur.
Lanfranchi, L.M.M.M., Viola, G.R. & Nascimento, L.F.C., 2010. The use of Cox regression to
estimate the risk factors of neonatal death in a private NICU, Taubate.
McCarty, J.A. & Hastak, M., 2007. . Segmentation approaches in data-mining: A comparison of
RFM, CHAID, and logistic regression. Journal of business research, 60(6), pp.656–662.
Monem A Mohammed, 2014. Survival Analysis By Using Cox Regression Model with
Application. International journal of scientific & technology, 3(11).
Rajasekar, S., Philominathan, P. & Chinnathambi, V., 2013. Research methodology, Tamilnadu
India.
SABR, B. et al., 2013. Intimate Partner Violence, Depression, PTSD and Use of Mental Health
Resources among Ethnically Diverse Black Women. NCBI, 52(4).
Teddlie, C. & Yu, F., 2007. Mixed Methods Sampling: A Typology With Examples. Journal of
Mixed Methods Research, 1(1), pp.77–100. Available at:
http://mmr.sagepub.com/cgi/doi/10.1177/2345678906292430 [Accessed July 9, 2014].
Yoshihama, M. & Horrock, J., 2003. The Relationship Between Intimate Partner Violence and
PTSD: An Application of Cox Regression With Time-Varying Covariates. Journal of
Traumatic Stress,, 16(4), pp.371–380.