Research Article: Variability and validity of intimate partner violence reporting by couples in Tanzania

Date Published: March 8, 2018

Publisher: Public Library of Science

Author(s): Nafisa Halim, Ester Steven, Naomi Reich, Lilian Badi, Lisa Messersmith, Soraya Seedat.


In recent years, major global institutions have amplified their efforts to address intimate partner violence (IPV) against women—a global health and human rights violation affecting 15–71% of reproductive aged women over their lifetimes. Still, some scholars remain concerned about the validity of instruments used for IPV assessment in population-based studies. In this paper, we conducted two validation analyses using novel data from 450 women-men dyads across nine villages in Northern Tanzania. First, we examined the level of inter-partner agreement in reporting of men’s physical, sexual, emotional and economic IPV against women in the last three and twelve months prior to the survey, ever in the relationship, and during pregnancy. Second, we conducted a convergent validity analysis to compare the relative efficacy of men’s self-reports of perpetration and women’s of victimization as a valid indicator of IPV against Tanzanian women using logistic regression models with village-level clustered errors. We found that, for every violence type across the recall periods of the last three months, the last twelve months and ever in the relationship, at least one in three couples disagreed about IPV occurrences in the relationship. Couples’ agreement about physical, sexual and economic IPV during pregnancy was high with 86–93% of couples reporting concordantly. Also, men’s self-reported perpetration had statistically significant associations with at least as many validated risk factors as had women’s self-reported victimization. This finding suggests that men’s self-reports are at least as valid as women’s as an indicator of IPV against women in Northern Tanzania. We recommend more validation studies are conducted in low-income countries, and that data on relationship factors affecting IPV reports and reporting are made available along with data on IPV occurrences. Keywords: Intimate partner violence; measurement; validity; survey research; Tanzania.

Partial Text

Intimate partner violence (IPV) is a global health and human rights problem affecting 15–71% of reproductive-aged women worldwide over their lifetime[1]. Sub-Saharan (30–66%) has one of the highest lifetime physical or sexual IPV rates globally, along with South Asia (42%) [2]. Defined as the threatened, attempted, and completed physical, sexual or psychological abuse that occurs between intimate partners, IPV is one of the most common forms of violence experienced by women across all societies and social hierarchies [3, 4]. IPV is associated with poor physical, psychological, and reproductive health outcomes, and productivity loss among women and poor health and developmental outcomes among children born to IPV-victimized women [1, 5, 6]. In extreme cases, IPV can result in death: 39% of all female homicides worldwide are the result of IPV [7].

Gender-based stratification is a central feature of society in Tanzania, an East African country with an estimated population of 47.4 million and per capita income of $865 [22]. With a score of 0.547 on the Gender Inequality Index, Tanzania ranks 124 out of the 155 countries scored, indicating that disparities exist between Tanzanian men and women in economic status, empowerment, and reproductive health. Underpinning the gender-based stratification is a patriarchal system including patrilineal inheritance that deprives women of critical rights and privileges [23]. In Tanzania, the total fertility rate is 5.3; the adolescent fertility rate is 128.7 births per 1000 women ages 15–19; the maternal mortality ratio is 578 per 100,000; female secondary school enrollment rate is 24%; and land ownership among women of reproductive age is 5% [22]. Patriarchal traditions and institutions govern marriage, childbearing, age-disparate sexual relations, and sexual practices [24, 25]. For example, social pressures for early childbearing discourages condom use [26]; male dominance of sexual decision-making is expected [27]; and bridewealth, a cash or in-kind payment made by the man’s family to the woman’s, restricts women’s ability to negotiate safe sex or to deny sex [28].

Discordant reporting of IPV by couples occurs when (a) the man reports no violence perpetration, but the woman reports victimization (Table 1: Cell 2); and (b) the man reports violence perpetration, but the woman reports no victimization (Table 1: Cell 3).

For convergent validation analysis, the IPV risk factors considered include men’s inequitable gender attitudes; exposure to childhood trauma; multiple sexual partners; condom non-use; and alcohol or drug use. Support in prior literature informed inclusion of these IPV risk factors.

For this analysis, we used the baseline survey data from 450 couples who participated in a three-arm cluster-randomized controlled trial (RCT) to evaluate the effectiveness of an intervention (entitled Together to End Violence against Women (TEVAW)) in reducing men’s IPV perpetration in Tanzania. We conducted this cluster-RCT in Karatu District, one of seven districts in the Arusha region of Tanzania. Our selection of this region and district was based on the high prevalence (>1 in 3) and acceptance (1 in 3) of IPV in Arusha (NBS & ICF Macro 2016), as well as the implementation of an ongoing savings groups (known as LIMCA) with women implemented by our partner, World Education Inc./Bantwana. These groups aimed to empower women participants through savings and credit activities to increase their economic independence and expand social networks. The groups also aimed to improve women’s knowledge about the physical, mental and emotional harms of IPV on women, men and children.

We estimated the crude percentage agreements and chance-corrected agreements between couples’ reports of physical, sexual, emotional, and economic violence in the last three and twelve months prior to the survey, ever in the relationship, and during pregnancy. We cross-tabulated men’s reports (Yes/No) of physical, sexual, emotional and economic violence perpetration in the last three and twelve months prior to the survey, ever in the relationship, and during pregnancy and women’s reports (Yes/No) of victimization. Further, we estimated the chance-corrected agreements between couples’ reports of each of the 48 violent events considered in the analysis (see Measures).

In recent years, the Government of Tanzania has amplified its effort to redress the consequences faced by the victims of intimate partner violence in the country. These efforts, albeit important, may become those of limited impact if assessment tools are neither specific nor sensitive in correctly classifying women as victims or otherwise. Self-reports by the victim are commonly used for IPV assessment. Yet, self-reports are prone to reporting bias. The validity of self-reports by the victim is contingent on victims recognizing an act as violent as and when it happens, remembering when violence occurs in the relationship, and not concealing violence. Self-reports by the victim and the perpetrator are believed to mitigate reporting biases (e.g., underreporting).




0 0 vote
Article Rating
Notify of
Inline Feedbacks
View all comments