Method

Data Analysis

Data analysis was carried out using the SPSS program version 22 and EQS 6.2 for Windows. First, to determine the appropriateness of the items in each of the instrument's subscales (physical, psychological, and sexual violence as well as controlling behaviors), an inter-judge concordance was conducted. Then, Cronbach's alpha index was applied to verify the internal consistency of the instrument and subscales, adding the Rho (ρ) composite reliability coefficient with the aim of making up of some of the possible limitations of alpha. After that, construct validity was calculated using confirmatory factor analysis (CFA) by Maximum Likelihood method to test the theoretical structure of the instrument (Figure 1). Lastly, to explore the prevalence of IPVAW throughout life and in the last 12 months in our sample, descriptive statistics (mean and standard deviation) were performed.

Note. VAW = Violence Against Women; PhyV = Physical Violence; PsyV = Psychological Violence; SexV = Sexual Violence; CBV = Control Behaviours Violence; Independence Model χ2 = 12260.446 (378 d.F.); χ2 = 2187.235 (p < .01): Bentler-Bonett Normed Fit Index = .82; Bentler-Bonett Non-Normed Fit Index = .83; Comparative Fit Index = .84; Root Mean Square Error of Approximation [90% CI] = .10 [.09, .10]; e = error.

Figure 1. Theoretical Model of WHO VAW Instrument 

Results

Reliability and Validity

Concerning the validity of internal structure of the instrument, a CFA was performed using EQS 6.2 for Windows, finding the fit indices and saturations for all items except 2, 17, and 25, which were deleted, obtaining an adequate instrument' reliability (ɑ = .95 [.94, .96]; ρ = .96; Figure 2). Intercoder reliability to determine final items and the response format of the Spanish version of the WHO VAW instrument were adequate, resulting in a kappa coefficient of .83. The descriptive statistics of the initial items of the instrument can be seen in the Table 4. Similarly, the final instrument, composed of 25 items, revealed high consistencies among subscales, resulting in a Cronbach's alpha of .95. Table 5 shows Cronbach's alpha for the total scale and subscales of the initial version (28 items) and the final version (25 items).

Note. PhyV = Physical Violence; PsyV = Psychological Violence; SexV = Sexual Violence; CBV = Control Behaviours Violence. All coefficients are significant. The fixed parameters were marked with “*”. Independence Model χ2 = 11211.85 (300 d.F.); χ2 = 1238.20 (p < .01): Bentler-Bonett Normed Fit Index = .89; Bentler-Bonett Non-Normed Fit Index = .90; Comparative Fit Index = 0.91; Root Mean Square Error of Approximation [90% CI] = .08 [.07, .08]; e = error.

Figure 2. Final Confirmatory Factor Analysis Model Obtained in Standardized Values 

Table 4. Items Descriptive Statistics 

Items	M	SD	Asimetrya	Curtosisb	DI
Item 1	1.96	1.14	.61	-1.21	.73
Item 2	2.68	1.19	-.33	-1.41	.66
Item 3	1.88	1.12	.72	-1.07	.70
Item 4	1.68	1.05	1.11	-0.33	.72
Item 5	1.65	1.09	1.28	-0.02	.79
Item 6	1.66	1.10	1.26	-0.10	.81
Item 7	1.34	.84	2.26	3.72	.75
Item 8	1.21	.69	3.17	8.86	.64
Item 9	1.13	.48	4.19	18.17	.56
Item 10	1.20	.61	3.05	8.49	.65
Item 11	1.42	.83	1.88	2.39	.72
Item 12	1.08	.41	5.54	31.59	.51
Item 13	1.12	.49	4.17	17.38	.59
Item 14	1.09	.41	5.06	27.01	.51
Item 15	1.11	.45	4.37	19.14	.56
Item 16	1.06	.36	6.62	46.30	.35
Item 17	1.01	.16	15.80	263.65	.22
Item 18	1.05	.30	6.32	45.80	.47
Item 19	1.91	1.24	.79	-1.15	.77
Item 20	1.49	.98	1.68	1.20	.75
Item 21	1.95	1.26	.73	-1.25	.75
Item 22	2.38	1.28	.09	-1.70	.63
Item 23	2.07	1.25	.51	-1.47	.72
Item 24	1.82	1.15	.95	-0.75	.71
Item 25	1.15	.60	3.90	13.94	.55
Item 26	1.71	1.05	1.06	-0.41	.65
Item 27	1.50	.99	1.68	1.21	.68
Item 28	1.43	.88	1.86	2.08	.61

Note. a: Standar Error = .106; b: Standard Error = .211; DI = Discrimination Index.

Table 5. Internal Consistency of the WHO VAW Instrument Subscales 

		WHO VAW instrument
Subscales	Nº items initial	αinitial	Nº items final	αfinal
Complete Scale	28	.95 [.95, .96]	25	.95 [.94, .96]
Psychological Violence	8	.92 [.91, .93]	7	.91 [.91, .92]
Physical Violence	10	.91 [.90, .92]	9	.92 [.90, .93]
Sexual Violence	3	.86 [.83, .88]	3	.86 [.83, .88]
Control Behaviors Violence	7	.91 [.89, .92]	6	.92 [.90, .93]

Note. WHO VAW = World Health Organization violence against women; Nº items initial = 28 items of the initial version of the instrument; Nº items final = 25 items of the final version of the instrument, deleting 2, 17, and 28 items.

Discussion

IPVAW's effects on the victims' physical and mental health underline the importance of having appropriate tools for early detection. The WHO has recommended the WHO VAW instrument (^{WHO, 2005}) to assess physical, psychological, and sexual IPVAW as well as controlling behaviors by a partner or former partner during the past year and throughout life among women worldwide. However, it has not been adapted and validated in Spain. The present study is the first to do it with a sample of Spanish-speaking women, providing effective psychometric properties and allowing its usability in Spain. According to previous validation studies in other countries (^{Nybergh et al., 2013}; ^{Schraiber et al., 2010}), with its high internal consistency, the instrument discriminates between victims and nonvictims of IPVAW as well as among different forms of violence (physical, psychological, or sexual violence and/or controlling behaviors) among Spanish-speaking women in Spain. These findings revealed that the Spanish version of the WHO VAW instrument is an effective tool to evaluate IPVAW in Spain.

Compared to the original instrument (^{WHO, 2005}), the Spanish version of the WHO VAW instrument included some improvements. First, according to ^{Muñiz and Fonseca-Pedrero (2019)} , items should be representative, relevant, clear, and simple and avoid excessively wordy or ambiguous statements. Therefore, research experts in IPVAW discussed aspects that could improve the instrument's usability in Spain; consequently, we deleted double questions and the dichotomous response format (e.g., “Has your partner or former partner put you down or humiliated you in front of other people?”). Instead, the inclusion of frequency response allowed more precise information to be obtained on the incidence of occurrence of the different forms of IPVAW manifestation. Also, results suggest that IPVAW is not a singular behavior; rather, it tends to be repeated over time. Specifically, the majority of women victims of IPVAW reported suffering it more than once. These findings are consistent with the theory of the cycle of violence (^{Walker, 2009}), which indicates that without intervention and treatment, most aggressors will repeat the cycle of violence, which incidents becoming increasingly frequent and severe. Considering it will be critical because IPVAW does not always, or immediately, lead to women's death, however, women's health is progressively declining inadvertently (^{Zara & Gino, 2018}), posing a danger to their lives. Similarly, alarming results were obtained regarding the perpetrators of violence, so approximately half of the included women were involved in a violent relationship by their current partner or were re-victimized with another partner. The Spanish version of the WHO VAW instrument allows us to analyze the above-mentioned aspects, facilitating the collection of relevant data to address IPVAW.

Concerning psychometric proprieties, although an adequate internal consistency was demonstrated for the WHO VAW instrument, the methodological approach conducted in the present study has not been reported. Initially, inter-coder reliability was adequate, obtaining a kappa coefficient of .83. Then, a high Cronbach's alpha coefficient was obtained for the final instrument composed of 25 items, indicating that the Spanish version of the WHO VAW instrument seems an effective tool to assess IPVAW. Similarly, the present instrument differentiated among four subscales to differentiated types of IPVAW (physical, psychological, and sexual violence as well as controlling behaviors), showing high internal consistency. In this regard, differentiating among the various forms of IPVAW is essential because although IPVAW entails any physical, psychological, or sexual violence or controlling behavior abuse from a current or former intimate partner (^{WHO, 2021}), the predominant type of violence manifested (e.g., physical violence) may influence key aspects of victims' decision-making processes such as their perception of severity, assessment of risk, or help-seeking behaviors (^{Cho et al., 2020}; ^{Novo et al., 2016}; ^{Wilson & Smirles, 2022}). Ultimately, CFA yielded fit indices and saturations for the final 25 items included in the Spanish version of WHO VAW, obtaining a high internal consistency (ɑ = .95). It shows an adequate internal structure of the different items in each subscale, concluding the appropriation of this instrument for use in Spain.

Morever, it is noteworthy that compared to previous studies in Spain (^{Ministry of Equality, 2019}), our results displayed higher rates of reported physical, psychological and sexual violence. Likewise, the majority of our sample were women under 30 years of age, and the manifestation of IPVAW through controlling behavior was the most common form of violence experienced by women (68.4%) and repeated over time. However, ^{Sánchez-Hernández et al. (2020)} found that although young women considered it common among young couples, 82.9% had never or hardly ever suffered from these behaviors in their relationships, possibly due to the technological context, which would facilitate the acceptance of controlling behaviors in the relationship as normal. Therefore, more studies are required to identify the different forms of IPVAW manifestations, using a universal detection instrument with adequate psychometric properties that guarantee its applicability and facilitate comparisons. The WHO VAW instrument has been developed by leading professional members of the WHO, a referral organization for addressing health problems such as IPVAW, and its adaptation and validation for use by Spanish-speaking women sample allows for its application in any context of intimate relationships.

This study was subject to limitations. Specifically, although participants were women from the general population, most of them had high levels of education. Future researchers could improve this instrument accounting for age groups (e.g., adolescents), clinical samples (e.g., victims in women's centers), residence (e.g., rural area), and level of education (e.g., university studies), among other factors. Comparative studies could establish specific intervention programs to prevent IPVAW escalating and worsening. Furthermore, although we asked women whether the violence was perpetrated by a former partner or their current partner to assess the risk in their situation, we could not take protective measures due to the surveys' anonymity. In future studies and contexts of application of the Spanish version of the WHO VAW instrument, researchers should offer women the opportunity to provide contact information in case the investigators believe participants may need help. Likewise, it would be convenient to include at the end of the instrument the resources available to women in Spain who suffer IPVAW.

Lastly, some methodological points should be noted. On one hand, although the validation of the instrument conforms to the established minimum standards (^{Kline, 2015}) and largely maintains the original structure of the instrument, future studies should make efforts to test other models which can fit better, such as taking account the interactions among factors. On the other hand, three items were eliminated; items 2 and 17 corresponded to double questions in the original instrument, while item 28 was a complete item for assessing control behaviors. In this regard, the elimination of items 2 and 17 would not seem to affect the original structure of the instrument. However, it would be advisable for future studies to create an alternative item to 28 (“Did you expect me to ask your permission before seeking medical attention?”), taking account the international test commission guidelines for test adaptation (^{Hernández et al., 2020}).

Despite these limitations, this study bridges a gap in the application of research findings on the psychometric properties of the WHO VAW instrument as well as its adaptation for a sample of women in Spain, involving relevant implications. Specifically, the Spanish version of the WHO VAW instrument has demonstrated suitable psychometric properties and can be used in IPVAW research, promoting comparative studies. Also, professional contexts could benefit from the use of this instrument, especially health care centers, where women sometimes go with a covert reason and symptoms that are difficult to diagnose. Therefore, the instrument may be applied among women suspected of experiencing IPVAW, facilitating its detection to help victims.

In sum, an instrument to facilitate IPVAW detection as a whole, taking into account its manifestation through physical, psychological, and sexual violence as well as controlling behaviors by a partner or ex-partner, was adapted and validated in a sample of women in Spain. It was based on the reference tool developed by the WHO, which is widely used around the world. The adequacy of the psychometric properties of the Spanish version of the WHO VAW instrument allows for its use among Spanish-speaking women to identify incidents of IPVAW throughout life and in the last 12 months. Usability improvements included collected information about the frequency of violence (never, once, a few times, or more than once) and perpetrator (past partner, current partner, or both). This instrument will be essential as an early screener of IPVAW in various contexts (e.g., health care centers, studies, or education centers) and to make comparisons.