Date of Award


Publication Type

Doctoral Thesis

Degree Name



Sociology, Anthropology, and Criminology

First Advisor

Maticka-Tyndale, E.


Latent class analysis, Social-ecological model, Violence against women




Approximately one-third of women across the globe have been raped and/or assaulted by an intimate partner in their lifetimes, with these rates of intimate partner violence (IPV) exceeding 50% for women in some areas of sub-Saharan Africa. Despite these prevalence rates, however, there is little empirical work that explores potential differences in experiences of intimate partner violence (IPV). Research tends to oversimplify IPV, viewing it as a unitary homogenous experience, rather than a multifaceted experience, limiting our understanding of the different types or variations of IPV, and the factors related to them. This research begins to address this gap in knowledge using Demographic and Health Survey data collected from women in three countries in sub-Saharan Africa (Kenya, Tanzania, and Uganda), a world region with one of the highest lifetime prevalence rates of IPV. Three exploratory analyses were conducted to model diversity in IPV experiences and to examine the potential covariates for different types of IPV. Descriptive analyses and latent class analyses (LCA) were used to assess the nature of IPV experiences, including whether women experience distinct types of IPV, and what these types look like. In the final analysis, an LCA model was combined with an ecological framework, by adding multi-level covariates to the model to identify the factors in women’s social-ecological environment that were differentially related to the distinct types of IPV. Findings from these analyses support the study of IPV experiences collectively among women in Kenya, Tanzania, and Uganda. For these women, IPV is not a homogenous experience. Results from the latent class analysis demonstrate that there are different types of IPV experienced among women in eastern sub-Saharan Africa. LCA supported the selection of a five-class model of IPV. The five classes include a No Violence class, and four additional IPV classes. The No Violence class (Class 1) included women who had a low likelihood of experiencing violence (< 1%). The remaining four classes included a class of women experiencing Predominantly Control (Class 2), two classes of physical IPV, distinguished by the severity of abuse (e.g., Less Severe Physical IPV (Class 3), and Severe Physical IPV (Class 4)), and a class of women experiencing Sexual IPV (Class 5). So, while class 1 represents experiences of women not experiencing violence, the remaining four classes describe different types of IPV. Class 2 and Class 5 both characterize types of abuse that occur in the absence of or very low likelihood of physical violence, while Class 3 and Class 4 describe abusive experiences that include elements of emotional abuse and physical abuse. Noteworthy here is the pervasiveness of control across all classes, even in the No Violence class, the finding that Sexual IPV is distinct from physical IPV (as most research tends to combine physical and sexual IPV), and the two distinct forms of physical IPV that suggest that not all physical forms of IPV are the same. These different classes of IPV were influenced differentially by the multi-level (micro-, meso-, exo-, macro-) factors across the social-ecological context, illuminating the need to include not only individual-level covariates but also relationship-level and community-level covariates when assessing factors related to IPV experiences. Implications for future research, including the need for research that seeks to understand in-depth what these types of IPV mean to women, and whether these types are evident in other locations, or if they are regionally specific are discussed.