Patterns of victim-survivor utilisation of domestic violence support services

主要的研究问题是：什么是家庭暴力的妇女使用家庭暴力支持服务的决定因素？这包括几个分问题：（a）遭受家庭暴力的妇女目前如何利用支助服务？B）获得家庭暴力服务的妇女的利用模式是什么？（c）家庭暴力的轨迹与家庭暴力支助服务的利用之间有什么关系？（d）提供家庭暴力服务对降低家庭暴力发生率有何贡献？（e）不同的服务使用模式对理解性别不平等与暴力之间的联系以及对社会学学科有何影响？拟议的项目采用定量方法分析合作伙伴提供的两个大型数据集。它将应用贝叶斯建模，事件发生时间（TTE）模型和机器学习模型（Mohri，Rostamizadeh和Talwalkar，2012; Russel和Norvig，2013）。虽然主要是定量的，该项目将通过一个系统的文献综述，在其中收集的数据，和理论的变化的thecatories的建设的理解，系统的相关文献综述学生将产生一个文献的系统性审查的帮助，受害者幸存者的服务使用，和家庭暴力支持服务的影响，在社会学和相关学科。系统审查将基于数据库搜索、灰色文献以及通过妇女援助组织成员组织、对妇女和女童的暴力行为研究网络和欧洲性别与暴力网络传播的证据呼吁。这将为后续阶段的研究提供信息，包括识别数据集中要探索的变量。通过系统综述，学生将提供对服务有效性措施的反思性分析。学生将确定服务提供的有效性的不同措施，作为减少家庭暴力的指标。这将涉及对数据集的限制和推论的批判性反思。这一阶段的方法将产生一个理论的变化，以确定影响，结果和干预措施，以改善对受害者幸存者的支持。这将与性别不平等和暴力的相互联系的社会学理解。纵向数据的定量分析学生将重新组织的原始数据，以确定关系表，识别缺失和冗余数据以及数据集之间的重叠点。数据将被提取，清理和组装成一个工作数据集。数据将按照GDPR安全保存。这将涉及家庭暴力支持服务，其他机构和受害者幸存者的数据相结合。数据中的缺失模式将通过链式方程和大量重复（mi<30）的多重插补来解决，以减少偏见（Goldstein，Harron，& Wade，2012），当有必要创建一个完整的数据集时。敏感性分析将用于比较完整病例和完全插补数据。

The primary research question is:What are the determinants of domestic violence support service use for women experiencing domesticviolence?This includes several sub-questions:a) How do women experiencing domestic violence currently utilise support services?b) What are the patterns of utilisation amongst women who access domestic violence service provision?c) What is the relationship between trajectories of domestic violence and the utilisation of domesticviolence support services?d) What is the contribution of domestic violence service provision to reductions in the rates of domesticviolence?e) What are the implications of the different patterns of service use for the understanding of theconnection between gendered inequalities and violence, and for the discipline of sociology?The proposed project employs a quantitative methodology to analyse two large datasets provided by thecollaborative partner. It will apply Bayesian modelling, time-to-event (TTE) models and machine learningmodels (Mohri, Rostamizadeh, & Talwalkar, 2012; Russel & Norvig, 2013). Although primarily quantitative,the project will be informed by a systematic literature review, understanding of the construction of thecategories in which the data is collected, and the theory of change.Systematic review of relevant literatureThe student will produce a systematic review of literature on help-seeking, service use by victimsurvivors,and the impact of domestic violence support services, in sociology and related disciplines. Thesystematic review will be based on database searches, grey literature, and a call for evidence circulatedvia Women's Aid's member organisations, the VAWG Research Network and the European Network onGender and Violence. This will inform subsequent phases of the research including the identification ofvariables to explore in the datasets.Identify a theory of changeInformed by the systematic review, the student will provide a reflexive analysis of measures ofeffectiveness for services. The student will identify different measures of effectiveness of serviceprovision as indicators for reductions in domestic violence. This will involve critical reflection upon thelimitations and inferences made about the dataset. This phase of the methodology will produce a theoryof change to identify impacts, outcomes and interventions necessary for improving support for victimsurvivors.This will engage with sociological understandings of the interconnections of genderedinequalities and violence.Quantitative analysis of longitudinal dataThe student will re-organise the raw data from On Track in order to define the relational tables, identifymissing and redundant data and points of overlap between the datasets. Data will be extracted, cleanedand assembled into a working dataset. The data will be held securely in compliance with GDPR. This willinvolve combining data on domestic violence support services, other agencies and victim-survivors.Patterns of missingness in the data will be addressed by multiple imputation by chained equations and alarge number of repetitions (mi<30) in order to reduce bias (Goldstein, Harron, & Wade, 2012), whennecessary to create a full dataset. Sensitivity analysis will be used comparing complete cases and fullyimputed data.