The objective of this study is to ascertain the attitudes of individuals aged 18 and over towards violence against women and to evaluate these attitudes in terms of various psychosocial factors. The phenomenon of violence against women is multifaceted, with numerous factors contributing to its perpetuation, including but not limited to gender inequality, patriarchal norms, and individual beliefs. Recent international studies highlight the importance of developing culturally adapted measurement tools to assess violent and aggressive behaviors in different societies (Sanjari et al., 2025). A review of the extant literature reveals that there are various measurement tools aimed at measuring attitudes towards violence against women. One such measurement tool is the Attitude Scale Toward Violence, which was developed by Gömbül in 2000 (Gömbül, 2000). The development of this scale was initiated to ascertain the attitudes of health personnel towards violence against women by their spouses within the family. Consequently, its capacity for measurement is constrained to health personnel and is not adequate for general population screening. Another measurement tool related to this topic is the ISKEBE Attitude Scale Toward Violence Against Women, which was developed by Kanbay et al. (2017). This measurement tool is a two-factor, 30-item scale developed for the population aged 18 and over, with an explanatory variance of 47.9%. Despite the scale’s adequate number of items and variability, its comparatively low variability indicates the necessity for a measurement instrument with a reduced number of items and elevated variability for the general population. The objective of this study is to address a significant gap in the existing literature by developing the Sahara Attitude Scale Toward Violence Against Women. The data obtained will be significant in terms of raising social awareness, developing strategies to combat violence, and guiding policymakers.

Since this research aims to develop a measurement tool, it requires the use of analytical techniques and specific methodology. Therefore, the study was designed as a methodological research (DeVellis, 2014).

In this study, the item pool for the scale developed to measure attitudes toward violence against women among individuals aged 18 and older was created based on literature reviews. National and international studies on violence against women and previously developed measurement tools were reviewed. Individual, cultural, and social factors influencing attitudes toward violence against women were identified in the literature. At the end of this phase, an item pool consisting of 30 statements was created.

For content validity, opinions were obtained from 10 experts in the fields of sociology, psychology, measurement, and evaluation. Content validity was achieved using the Lawshe technique (Lawshe, 1975).

The purpose of the pilot study is to test the suitability of the developed scale for the target population. In this process, a sample representing the target population and showing demographic, social, and economic diversity should be selected (Erkuş, 2007; Trochim and Donnelly, 2006). According to Boateng and colleagues (2018), 30–50 participants are sufficient for a pilot application. Therefore, a pilot application was conducted with a sample of 32 participants. During this process, item clarity, relevance, and comprehensibility were evaluated, and minor wording adjustments were made based on participant feedback to ensure the appropriateness of the scale for the general population.

Since this study is a scale development study, factor analysis techniques must be used. In the study, exploratory factor analysis (EFA) was first performed to reveal the structure, and then confirmatory factor analysis (CFA) was performed to validate the structure that emerged. Therefore, two separate samples with similar characteristics were needed in the study. It is stated that a sample size of at least 300 is appropriate to achieve sufficient sample size in scale development studies (Comrey and Lee, 1992). In this study, 696 individuals (360 for EFA and 336 for CFA) were reached as a result of the data collection phase.

51.4% of the EFA sample were male, 29.4% were married, 73.9% were higher education students or graduates, 80.3% had a nuclear family, and 61.9% lived in a city or large city. The mean age of the sample is 28.4 $\pm$ 10.5 (min: 18, max: 72). 53% of the CFA sample are male, 23.2% are married, 67.9% are higher education students or graduates, 75.6% have a nuclear family, and 50.9% live in a city or large city. The average age of the sample is 26 $\pm$ 9.9 (min: 18, max: 72).

During the data collection process, a link created via Google Forms was shared with participants through online platforms and social media accounts. Participants were selected using a purposive sampling method. This method is considered an effective approach in scale development studies to reach the target population and include individuals with diverse socio-demographic characteristics. Online data collection enabled access to a large participant group in a short period of time and allowed us to exceed the minimum sample size recommended for factor analyses.

The inclusion criteria for participation in the study were as follows:

(1) Being 18 years of age or older,

(2) Having the ability to read and understand Turkish,

(3) Voluntarily agreeing to complete the online form.

Participants who met these criteria were included in the study. The data collection process was carried out in two phases. In the first phase, data were collected from 360 participants for EFA. In the second phase, to confirm the obtained structure, data were collected again from 336 participants with similar socio-demographic characteristics for CFA. Missing data (n = 13) were handled using listwise deletion, resulting in complete data sets for the analyses.

Descriptive statistics and the IBM SPSS Statistics for Windows, Version 26.0 (IBM Corp., Armonk, NY, USA) software package for EFA, and the IBM SPSS Amos, Version 23.0 (IBM Corp., Armonk, NY, USA) software package for CFA were used in the data analysis phase. The structure validity was performed using EFA, and the structure obtained was validated using CFA. Item-total correlations were examined prior to factor analysis. Average variance extracted (AVE) and composite reliability (CR) values were calculated for convergent validity and discriminant validity. Cronbach’s $\alpha$ coefficient, split-half reliability, and McDonald’s Omega $\omega$ were calculated for the reliability of the scale.

In order to determine the construct validity of the scale, “principal components analysis (PCA)” was used for EFA, and the “direct oblimin technique” was used as the factor rotation technique based on the assumption that the factors are interrelated. CFA was performed to analyze the fit of the factor structure of the model obtained from EFA, and the fit indices were examined. For the fit indices, 0.05 RMSEA 0.10; 0.90 $\le$ CFI $\le$ 0.95; 0.90 $\le$ GFI $\le$ 0.95; AGFI $>$0.90; 0.90 $\le$ CFI $\le$ 0.95 were considered acceptable limits (Wang and Wang, 2012; Meydan and Şeşen, 2011; Schermelleh-Engel et al., 2003). For convergent and discriminant validity, AVE and CR values were examined. For the convergent validity of the scale to be ensured, AVE values of the items should be 0.50 or above (Bagozzi and Yi, 1988); while the CR value, which is expressed as composite reliability (structural reliability), was determined to be 0.70 and above as acceptable values (Hair et al., 2019). For research scales, it is recommended that the Cronbach $\alpha$ reliability coefficient be 0.70 or higher (DeVellis, 2014). Therefore, if the Cronbach $\alpha$ $>$ 0.70, the scale is considered to have sufficient reliability.

Validity is a concept that expresses the degree to which a measurement tool measures the concept it intends to measure accurately and completely. In this study, the validity of the scale was examined in terms of content validity, construct validity, and internal validity.

Reliability is the degree to which a measurement tool demonstrates consistency and stability. Reliability aims to minimize the margin of error in a measurement tool and ensure that measurements are stable (Crocker and Algina, 2008).

The Cronbach $\alpha$ reliability coefficient for the obtained structure was calculated as 0.94 for the “Justification” factor, 0.92 for the “Control” factor, and 0.94 for the scale as a whole. Gutman values were calculated as 0.94, 0.92, and 0.94, respectively. The Spearman-Brown coefficient calculated for the two-half test reliability of the factors and the scale total ranged from 0.797 to 0.951. These values fall within the limits recommended in the literature (DeVellis, 2014). Based on all these findings, it was concluded that the reliability of the scale is sufficient (Table 5).

3.2.1 Validation of the Structure

The structure of the scale was validated using CFA. CFA is used to test the validity of a previously determined factor structure (Kline, 2016). During CFA, fit indices such as Root Mean Square Error of Approximation (RMSEA); comparative fit indices (CFI); goodness-of-fit index (GFI); adjusted goodness of fit index (AGFI); and normed fit index (NFI) were used as the basis for evaluation.

The two-factor structure of the Sahara Attitude Scale Towards Violence Against Women was tested using first-order multiple factor CFA. When examining the paths related to the structure of the scale, it was determined that both factors contributed significantly to the scale (Fig. 2). In addition, the items included in the factors load significantly on the factor they belong to (p 0.001). When examining the goodness-of-fit indices for the structure, the values obtained were $\chi$²/df: 2.726, RMSEA: 0.072, CFI: 0.97, GFI: 0.93, AGFI: 0.90, and NFI: 0.96. These goodness-of-fit indices are within the levels recommended in the literature (Wang and Wang, 2012; Meydan and Şeşen, 2011; Schermelleh-Engel et al., 2003). The factor loadings ranged from 0.76 to 0.89 for the “Justification” factor and from 0.74 to 0.87 for the “Control” factor. Based on these findings, the two-factor structure of the scale was confirmed.

Fig. 2.

CFA findings as to the factor structure of the scale. CFA, confirmatory factor analysis; RMSEA, Root Mean Square Error of Approximation; CFI, comparative fit indices; GFI, goodness-of-fit index; AGFI, adjusted goodness of fit index; NFI, normed fit index; CMIN, chi-square; DF, degrees of freedom.

In addition to the overall fit indices, Table 6 presents the unstandardized estimates, standard errors, p-values, 95% confidence intervals, and standardized factor loadings for all items. All factor loadings were statistically significant (p 0.001) and ranged from 0.74 to 0.89. The correlation between the two latent factors was also examined, showing a moderate positive association (r = 0.72, 95% CI [0.60–0.84], p 0.001), indicating that while related, the Justification and Control dimensions represent distinct constructs.

Table 6. Standardized and unstandardized factor loadings, standard errors, p-values, and 95% confidence intervals for CFA.

Item	Factor	Unstandardized	SE	p	Standardized	95% CI	95% CI
		Estimate			Loading	Lower	Upper
i16	Justification	1.000	-	***	0.813	-	-
i6	Justification	0.959	0.061	***	0.761	0.839	1.079
i10	Justification	0.842	0.046	***	0.842	0.752	0.932
i7	Justification	0.998	0.051	***	0.885	0.898	1.098
i5	Justification	1.038	0.053	***	0.888	0.934	1.142
i3	Justification	0.898	0.048	***	0.851	0.804	0.992
i11	Control	1.000	-	***	0.804	-	-
i24	Control	1.129	0.069	***	0.797	0.994	1.264
i25	Control	1.157	0.064	***	0.865	1.032	1.282
i27	Control	1.193	0.067	***	0.845	1.062	1.324
i13	Control	1.159	0.078	***	0.740	1.006	1.312
i18	Control	1.201	0.076	***	0.775	1.052	1.350

*** p 0.001.

The test-retest method was used to evaluate the consistency of the scale over time. Test-retest reliability is a technique used to determine whether a measurement tool demonstrates consistency over time. For measurements of medium-term psychological characteristics such as motivation and attitude, a time interval of 2 to 4 weeks is recommended between measurements (Hinkin, 1998). In line with this, the scale was reapplied to a group of 72 participants at two-week intervals within the scope of the study. Test-retest reliability was first examined using Pearson correlation, which yielded a coefficient of r = 0.88 (p 0.001). To provide a more rigorous evaluation of absolute agreement, an intraclass correlation coefficient (ICC) was also calculated using a two-way mixed-effects model with absolute agreement. The ICC was 0.87, 95% CI [0.81–0.91], indicating excellent temporal stability of the Sahara scale over a two-week interval.

This study has several limitations. First, the data were collected through online surveys using purposive sampling, which may limit the generalizability of the findings to the broader population. Second, the sample was restricted to adults aged 18 and above from a single cultural context, and thus the psychometric properties of the Sahara scale should be further examined in larger and more diverse populations. Third, all data were based on self-report, which may be subject to response and social desirability bias. Future studies are recommended to validate the scale in different cultural settings, with more representative sampling methods, and by incorporating multiple sources of information.

“Sahara Attitudes Scale Toward Violence Against Women” was developed to determine the level of attitudes toward violence against women in the population aged 18 and older (Appendix Table 8). As a result of exploratory and confirmatory factor analyses, a two-factor, 12-item structure was obtained. The scale explains 76.3% of the variance, with Cronbach’s $\alpha$ of 0.94, Gutman’s 0.95, and McDonald’s Omega $\omega$ of 0.96.

The factor “Justification” reveals individuals’ tendency to view acts such as violence, harassment, or mistreatment against women as legitimate, understandable, or justified for various reasons. The statements included in this factor reveal the internalization of gender stereotypes, the perception that women’s behavior is the cause of violence, and that the victim is to blame. This attitude is the product of a mindset that explains violence against women not as an individual act but as a result of women’s so-called “wrong” behavior. Such beliefs can lead to the normalization of violence, impunity for perpetrators, and the silencing of victims. Approaches that define violence as “acceptable if minor” also fall within this scope. This sub-dimension represents the belief patterns that should be the main target of social transformation and awareness efforts. This factor consists of items 1–6 and explains 63.2% of the variance, with a Cronbach $\alpha$ of 0.94, Gutman 0.94, and McDonald’s Omega $\omega$ 0.92.

The factor “Control” reveals individuals’ tendency to control and restrict women’s lives, behavior, clothing, spending, and social roles. It represents a perspective that justifies restricting women’s freedom, establishing dominance over them, and constantly monitoring their movements. This controlling attitude feeds on the tendency to view women as passive, obedient, and dependent on male approval. Views such as that it is “inappropriate” for women to go out late at night, that their work is “not naturally accepted”, or that their spending must be “approved by men” are part of this structure. This understanding is contrary to gender equality and denies women’s autonomy as individuals, limiting their role in society. This factor consists of 7–12 items and explains 13.1% of the variance, with a Cronbach’s $\alpha$ of 0.91, Gutman’s 0.91, and McDonald’s Omega $\omega$ of 0.94.

The Sahara Attitude Scale Toward Violence Against Women is a 5-point Likert-type scale (0 = Strongly Disagree, 4 = Strongly Agree). There are no reverse-scored items in the scale. The total scale score is obtained by summing the scores of all items, and the raw score ranges from 0 to 48 points. The raw score must be standardized to a range of 0 to 100 points according to the guidelines. An increase in the score indicates an increase in favorable attitudes toward violence against women.

Standardization has been applied to the scoring of this scale to enable easy and consistent comparisons between samples in studies using the scale and in future adaptation studies.

The following formula should be used to standardize the scores obtained from the scale to 100.

$$Standardizedscore=\left(\frac{RawScore}{ScaleRange(48)}\right)\times 100$$

For example, if a person scored 24 points on the scale, their standardized score can be calculated as follows:

$$Standardizedscore=\left(\frac{24}{48}\right)\times 100=50$$

In this case, the standardized score of a person who scored 24 points on the scale is 50 points. It is mandatory to use the standardized form of the scores obtained from the scale in the studies to be conducted.