A comparison of two strategies to assess sexual violence: general vs. specialised trauma screening strategies in two high-risk substance use health disparity samples

ABSTRACT

Background/Objective: Using two different high-risk samples, the present study compared and contrasted two different strategies/questionnaire types for assessing a history of sexual violence: a general trauma screening vs. specialised behaviourally-specific questionnaires.

Methods: Sample 1 included 91 men and women seeking detoxification treatment services in a publicly funded, urban clinic who completed a trauma and substance use questionnaire battery during treatment. Sample 2 included 310 women at a rural college who completed a trauma and religious coping questionnaire battery for course credit. All participants completed both types of questionnaires: One general trauma screening questionnaire (i.e. the Life Events Checklist [LEC]) and two behaviourally-specific specialised questionnaires (i.e. the 2007 Sexual Experiences Survey [SES] and the Childhood Trauma Questionnaire [CTQ]).

Results: There were large differences in the cases identified by the behaviourally-specific questionnaires (SES and CTQ) compared to the general trauma screening questionnaire (the LEC) in both samples but few differences in the prevalence rates of sexual violence detected by each questionnaire type. In the detoxification sample, the differences were especially notable for men. Follow-up analyses indicated that degree of traumatisation impacted results likely by increasing participant’s willingness to endorse face-valid items on the LEC.

Conclusions: For men, the behaviourally-specific questionnaires (SES/CTQ) were necessary to identify cases. For those with more severe trauma histories, the LEC was equivalent to the SES/CTQ in identifying a similar number of sexual violence cases. Thus, clinicians and researchers should consider the population when selecting assessments to identify sexual violence history.

HIGHLIGHTS

• For men and rural college women, general trauma questionnaires are not as accurate as specialised sexual violence measures in detecting cases.

• While prevalence rates were often similar, which cases were detected by general trauma and specialised questionnaires were different.

• Individuals with greater trauma exposure were more likely to endorse face-valid sexual violence items on the general trauma questionnaires.

Antecedentes/Objetivo: Utilizando dos muestras diferentes de alto riesgo, el presente estudio comparó y contrastó dos diferentes estrategias/tipos de cuestionario para evaluar una historia de violencia sexual: una evaluación general de trauma vs. cuestionarios especializados de comportamiento específico.

Métodos: La muestra 1 incluyó a 91 hombres y mujeres que buscaban servicios de tratamiento de desintoxicación en una clínica urbana financiada con fondos públicos y que completaron una batería de cuestionarios sobre trauma y consumo de sustancias durante el tratamiento. La muestra 2 incluyó a 310 mujeres de una universidad rural que completaron una batería de cuestionarios sobre afrontamiento del trauma y religión para obtener créditos en el curso. Todas las participantes completaron ambos tipos de cuestionarios: Un cuestionario general de detección de traumas (es decir, la Lista de Chequeo de Eventos Vitales [LEC]) y dos cuestionarios especializados de comportamiento específico (es decir, la Encuesta de Experiencias Sexuales del 2007[SES por sus siglas en inglés] y el Cuestionario de Trauma Infantil [CTQ por sus siglas en inglés]).

Resultados: Hubo grandes diferencias en los casos identificados por los cuestionarios de comportamiento específico (SES y CTQ) en comparación con el cuestionario general de detección de traumas (el LEC) en ambas muestras, pero pocas diferencias en las tasas de prevalencia de violencia sexual detectadas por cada tipo de cuestionario. En la muestra de desintoxicación, las diferencias fueron especialmente notables en hombres. Los análisis de seguimiento indicaron que el grado de traumatización influyó en los resultados, probablemente al aumentar la disposición de los participantes a respaldar elementos aparentemente válidos en la LEC.

Conclusiones: Para los hombres, los cuestionarios de comportamiento específico (SES/CTQ) fueron necesarios para identificar los casos. En el caso de las personas con antecedentes traumáticos más graves, el LEC fue equivalente al SES/CTQ a la hora de identificar un número similar de casos de violencia sexual. Por lo tanto, los clínicos y los investigadores deben tener en cuenta la población a la hora de seleccionar las evaluaciones para identificar los antecedentes de violencia sexual.

KEYWORDS:

• Rape
• sexual assault
• trauma
• assessment
• substance use
• rural

PALABRAS CLAVE:

• Violación
• agresión sexual
• trauma
• evaluación
• consumo de sustancias
• rural

One in five American women experience rape (Black et al., 2011). Yet, the burden of sexual violence is not distributed evenly. Women (Black et al., 2011), individuals with substance use disorders (Kilpatrick et al., 1997), and those living in poverty are more likely to be victimised (Harrell et al., 2014). Approximately 20% of survivors also experience substance misuse (Zinzow et al., 2012). Rural individuals are less likely to have access to care of many types, but especially substance use disorders and violence related care (Banyard et al., 2019; Dotson et al., 2014; Edwards et al., 2023).

Yet, the measurement and clinical assessment of sexual violence is complex due to the stigma assigned to and often internalised by survivors of abuse and violence in Western culture (Kennedy & Prock, 2018; Mellen & Hatzenbuehler, 2023). There are many different ways to assess sexual violence history that differ in amount of time to complete, complexity, and precision (Probst et al., 2011). A brief, general screening questionnaire may be ideal in a primary care setting whereas a more detailed assessment may be more appropriate in psychotherapy. Another complicating factor is that research on the measurement of sexual violence suggests there may be differences between populations in questionnaire accuracy (Anderson & Delahanty, 2020). Yet, there is little empirical information available to guide providers or researchers in making decisions regarding optimal assessment strategies. The goal of this study was to compare two different strategies for assessing sexual victimisation: general trauma history screening vs. behaviourally-specific questionnaires in two different high-risk, health-disparity samples, a sample of impoverished substance use detoxification treatment seekers and rural college women.

1. The stigma of rape complicates assessment

The stigma is assigned to rape is such that many who experience rape feel shame related to the rape and have concerns about disclosing their rape (Mellen & Hatzenbuehler, 2023). This internalised stigma interferes with accurate assessment as those who experience rape do not wish to admit or disclose it (Andresen & Blais, 2019; Wilson & Miller, 2016). Thus, general, face-valid (e.g. obvious, using lay terms) questions assessing rape like, ‘have you ever been raped?’ are not sensitive enough and under-estimate cases (Koss, 1998). Indeed, most rape survivors do not call their experience rape or sexual assault (Donde et al., 2018) and would likely therefore respond ‘no’ to this type of item even when they have experienced this type of trauma. Instead, behaviourally-specific approaches are considered the gold-standard in sexual violence research, used in approximately 80% of sexual violence related studies (Fedina et al., 2018). The behaviourally-specific approach eschews using face-valid items that also contain emotionally-loaded words like ‘rape’ or ‘sexual assault’ that are interpreted variably (Hamby & Koss, 2003) and carry stigma. An example of a behaviourally-specific item would be, ‘Someone put their penis into my butt, or someone inserted fingers or objects without my consent by: … Threatening to physically harm me or someone close to me’. This item comes from the Sexual Experiences Survey-Short Form Victimisation which contains between 25 and 35 behaviourally specific items (depending on gender of the respondent). This example shows how anal rape would be measured; the physical threat item immediately follows other items which specify use of verbal bullying, verbal criticism, alcohol incapacitation, and immediately before an item specifying use of physical force showing how each possible form of rape is comprehensively evaluated (item shown as used in Koss et al., 2007).

Behaviourally-specific approaches identify 2.5× as many cases of rape as face-valid items (Wilson & Miller, 2016). Using clear, behaviourally-specific terms cut through cultural perceptions about what real ‘rape’ is (e.g. violent assaults by strangers: Adams-Clark & Chrisler, 2018), potential self-stigma associated with experiencing a sexual trauma Andresen & Blais, 2019) and provides an exemplar for respondents to compare their experience to when formulating a response. Memory retrieval and the importance of multiple, behaviourally specific items may be more especially important for capturing sexual violence experiences that involved substances as substance use can disrupt memory encoding. Further, those who experience substance-related sexual violence are less likely to respond yes to face-valid items. Capturing substance related experiences may be additionally important for solving health disparities as those living in poverty and rural populations who may have limited treatment access. Without accurate assessment, the burden of substance-related sexual violence may be underestimated and resources for prevention and treatment inadequately allocated.

2. The general trauma history screening approach

Researchers and clinicians in the field of trauma have developed a number of well-studied general trauma history screening measures with strong psychometric properties that assess a wide range of potentially traumatic experiences, of which sexual violence is just one of many traumas assessed (Norris & Hamblen, 2004). A recent review and analysis of eight general trauma screening questionnaires suggested significant heterogeneity in the content of trauma screening measures (Karstoft & Armour, 2022). Specifically, this review identified 44 different types of traumas that were assessed; however, whether two different general trauma questionnaires measured the same trauma type was minimal, ranging from 22.7 to 63.6% of the 44 possible traumas. This suggests lack of consistency in which traumas are included, even for the most frequently included traumas – combat. Sexual assault was in the middle of this range suggesting only moderately consistent inclusion of this trauma type. Further, in these general screening questionnaires, sexual violence tends to be assessed using face-valid items like, have you experienced sexual assault? Thus, the evidence suggests general screening approaches will not be accurate due to inconsistent coverage of the trauma of sexual violence and inadequate specificity when using face-valid items.

3. Population specific considerations

Another issue potentially leading to discrepancies in findings is the population of study. The issue of under-counting because of face-valid items used on general trauma screening questionnaires may be especially concerning for populations that experience more intense stigma around rape, including men (Hlavka, 2017). Multiple studies suggest that men who have experienced sexual violence are even less likely to use terms like ‘rape’ than women (Artime et al., 2014). Men who experience sexual violence are less likely to receive care related to violence (Young et al., 2018). Thus, for men who have experienced substance-related sexual violence it is unlikely that they recognise it as such and seek care related to it; however, good assessment using behaviourally-specific questionnaires could highlight the impact of sexual violence related PTSD and be used to increase efficacy of substance use treatment when it is sought (Back et al., 2019). This disparity in care seeking may be heightened in low-income and rural populations where access to treatment is difficult.

Women in rural areas are more likely to experience sexual assault (Sinovich & Langton, 2014) and more likely to experience health disparities including in accessing mental health or substance use related care (Chan et al., 2016; Dotson et al., 2014). Further, some cultural values in rural areas promote more victim-blaming which heightens stigma (Annan, 2006) and may suppress responding on general trauma screening questionnaires. Substance use, particularly alcohol use is higher in some rural areas (Division of Population Health, 2022); yet, alcohol-related sexual violence is the least likely form to be recognised an captured using face-valid items (Walsh et al., 2016). Taken together this suggests behaviourally specific questionnaires may be especially important in rural settings to recognise cases of sexual violence, especially alcohol-related violence for rural women.

In contrast, research suggests that populations with high trauma exposure are more likely to respond affirmatively to face-valid questions such as those on general trauma history questionnaires. Considering the high prevalence of trauma and PTSD among those seeking treatment for substance use (Anderson, Hruska, et al., 2018), and the high trauma exposure of those living in poverty (Brattström et al., 2015), it is possible that impoverished substance use treatment seekers are more aware of the threat of sexual violence and would be more likely to respond accurately to face-valid measures than a less trauma exposed population.

4. The current study

Being able to prevent sexual violence first requires understanding the scope of the problem; yet assessing the scope of the problem is challenging due to the stigma of rape, stigma which is heightened in certain cultural contexts and for certain populations. Although screening for sexual violence is recommended by major healthcare organisations (e.g. American College of Surgeons, 2018; Veterans Health Administration, 2008), and is useful for substance use treatment planning (Goldberg et al., 2019); deciding which measure to use can be challenging. While we would not predict that general trauma screening would be as accurate as specialised behaviourally-specific assessment, understanding the degree to which general screening questionnaires do capture cases for certain populations is helpful for optimising screening and assessment strategies across settings. Optimising care via trauma screening programmes increases treatment efficacy and access for substance use and sexual violence (Babor et al., 2017; Kimerling et al., 2008); yet can be difficult to implement, especially in low resource settings such as publicly funded clinics and rural clinics (Chan et al., 2016; Fehr et al., 2020). Current research and theory suggest that behaviourally-specific questionnaires, like the Sexual Experiences Survey (SES) or items from the Childhood Trauma Questionnaire (CTQ), will identify many more cases than general trauma screening questionnaires due to increased specificity, omission of potentially emotionally-laden words like ‘rape’ or ‘sexual assault’ (Young & Maguire, 2003), and ability to overcome self-stigma in responding (Andresen & Blais, 2019). However, research also suggests that more highly trauma-exposed populations may not need such specific and potentially burdensome screening. The research thus offers contradictory recommendations, especially for clinicians in substance use treatment settings, where patients tend to be male with high trauma exposure (Clark et al., 2014).

Therefore, the goal of this secondary data analysis was to explore the use of a general trauma history vs. behaviourally-specific checklist approaches in two health-disparity populations both at high-risk for trauma and substance use and with low access to prevention and treatment. Sample 1 was a low-income, public detoxification clinic treatment seeking sample of mostly men. Sample 2 were women at a rural college in a state with high rates of binge drinking (Division of Population Health, 2022). Rural college women were chosen as a comparison sample given the wealth of research demonstrating how both trauma history approaches function in college women (Fedina et al., 2018; Norris & Hamblen, 2004); yet the lack of data on violence against rural college women specifically highlights the opportunity to better understand their experiences. We hypothesised that behaviourally-specific questionnaires, like the SES (adolescent/adult sexual violence) and the CTQ (childhood sexual abuse) would identify more cases than a general trauma history screening checklist in both samples (H1) but that the level of disagreement (in other words, discrepancy) would be smaller in the more highly trauma exposed detoxification sample (H2). Previous research has suggested rates of discrepancy between sexual violence measures agree that a case has a victimisation history to be around 18% (Anderson & Delahanty, 2020; Strang et al., 2013). We also explored whether substance use and degree of trauma exposure would be related to discrepancies.

5. Methods

5.1. Participants

5.1.1. Sample one – detoxification treatment seekers

Participants in the first sample were recruited through a community detoxification treatment facility in Great Lakes region, STATE MASKED, n = 91 between April 2016 and May 2017 that provides treatment regardless of income. Potential participants were approached in the treatment centre’s community room to participate in survey research and were compensated with snacks such as candy bars.

5.1.2. Sample two – rural college women

College women participants in the second sample were recruited via Psychology Department subject pools at two universities in rural states, one in the upper Midwest, MASKED UNVIERSITY 1 (82.6% of the sample) and one in the Mountain West, MASKED UNIVERSITY 2 (17.4% of the sample) as well as through flyers around MASKED UNVIERSITY 1’s campus and social media. Those who participated through psychology subject pools were compensated with course credit, and community members were entered into a drawing for a gift card. Demographic characteristics of both samples are summarised in Table 1.

Table 1. Sample demographics.

	Sample 1 – detoxification treatment seekers		Sample 2 – rural college women
	n	%	n	%
Gender identity
Cisgender women	38	41.8	310	100
Cisgender men	52	57.1	0	0
Genderfluid or nonbinary	0	0	4	1.3
Racial identity
White	68	74.7	253	81.6
Black	11	12.22	5	1.6
Other identities	1	1.1	7	2.3
Asian American	0	0	3	.9
Indigenous	0	0	10	3.2
Hispanic or Latina	0	0	17	5.5
Sexual identity
Heterosexual	–	–	252	81.3
Bisexual	–	–	37	11.9
Gay or lesbian	–	–	9	2.9
Queer	–	–	6	1.9
Another identity	–	–	6	1.9
Education
Less than high school	12	13.2	–
High school	37	40.7	–
Some college or greater	40	44.0	–
Currently in college	–		270	87.0
Median family income	–		$80,000–99,999
Age, range and M(SD)	20–65, 37.1(10.8)		18–31, 21.0(3.24)

5.2. Materials

5.2.1. Childhood Trauma Questionnaire

The Childhood Trauma Questionnaire (CTQ; Bernstein et al., 1994) is a 28-item scale measuring five domains of trauma that occurred before the age of 18: physical abuse, sexual abuse, emotional abuse, physical neglect, and emotional neglect. Statements are rated on a 5-point Likert-type scale from 1 (never true) to 5 (very often). CTQ scores have been found to be correlated with depressive symptoms, evidencing external validity (Cukor & McGinn, 2006). For this study, only the sexual abuse subscale was used. Three of the five childhood sexual abuse items are more behaviourally-specific, such as ‘Someone tried to touch me in a sexual way or tried to make me touch them,’ versus two face-valid items that use the term ‘molest’ and ‘sexually abused.’ Thus, the CTQ is a mix of both behaviourally-specific and face-valid items. In the detoxification treatment seeking group the vast majority of participants who endorsed behaviourally-specific items also endorsed the face-valid items (99.96%). In the rural college women group, 100% of participants who endorsed behaviourally-specific items also endorsed face-valid items. Thus, because almost all participants who experienced childhood sexual abuse endorsed both behaviourally-specific and face-valid items, we computed one overall dichotomous variable.

5.2.2. Sexual Experiences Survey–Short Form Victimisation

The Sexual Experiences Survey–Short Form Victimisation (SES-SFV; Koss et al., 2007) contains multiple behaviourally-specific items measuring sexual violence victimisation. Each item describes the sexual experience and lack of consent ‘someone had oral sex with me or made me have oral sex with them without my consent by’ in an item stem followed by the means this experienced was coerced in five sub-stems. Table 2 contains example items. Endorsement of any SES item indicates a history of sexual victimisation since age 14. In Sample 1 we included additional items beyond the traditional 35 items (7 stems x 5 sub-stems) to assess made-to-penetrate victimisation, a type of sexual violence more often experienced by men (Anderson et al., 2020). In the detoxification seeking group, participants responded yes/no regarding whether each item occurred since age 14. In the rural college women group, participants indicated how many times each item occurred in the last 12 months and since the age of 14 on a frequency scale of 0, 1, 2–5, 6–9, or 10+ times. Prior research has suggested good evidence of convergent validity of this questionnaire for men and women (Anderson, Cahill, et al., 2018; Johnson et al., 2017). Research suggests minimally adequate test-retest reliability for men (Anderson, Cahill, et al., 2018) and good test-retest reliability for college women (Johnson et al., 2017). Participants who experienced some type of victimisation on the SES were administered the Assault Characteristics Questionnaire (Littleton et al., 2009) to understand the characteristics of rape incidents. One item regarding the amount of alcohol or other substance used at the time of the rape was analysed in this study.

Table 2. Example items on SES, CTQ, and LEC.

SES	CTQ	LEC
A man put his penis into my vagina, or someone inserted fingers or objects without my consent by: Telling lies, threatening to end the relationship, threatening to spread rumours about me, making promises I knew were untrue, or continually verbally pressuring me after I said I didn’t want to. Showing displeasure, criticising my sexuality or attractiveness, getting angry but not using physical force, after I said I didn’t want to. Taking advantage of me when I was too drunk or out of it to stop what was happening. Threatening to physically harm me or someone close to me. Using force, for example holding me down with their body weight, pinning my arms, or having a weapon.	I was molested	Sexual assault Happened to me Witnessed it Learned about it Part of my job Not sure Doesn’t apply
Even though it didn’t happen, someone TRIED to have oral sex with me, or make me have oral sex with them without my consent by: Telling lies, threatening to end the relationship, … [list repeated from first item] … or having a weapon.	I was touched sexually	Other unwanted or uncomfortable sexual experience Happened to me Witnessed it Learned about it Part of my job Not sure Doesn’t apply
Someone fondled, kissed, or rubbed up against the private areas of my body (lips, breast/chest, crotch or butt) or removed some of my clothes without my consent (but did not attempt sexual penetration) by: Telling lies, threatening to end the relationship, … [list repeated from first item] … or having a weapon.	I was made to do sexual things	–

5.2.3. General Trauma Screening Questionnaire – Life Events Checklist

The Life Events Checklist for DSM-5 (LEC-5; Weathers et al., 2013) is a 17-item scale measuring exposure to events that are potentially traumatic, including sexual assault, natural disasters, and accidents. Participants indicate their level of exposure to each event, with response options including ‘happened to me,’ ‘witnessed it,’ ‘learned about it,’ ‘part of my job,’ ‘not sure,’ and ‘doesn’t apply.’ This scale has good evidence of validity; exposure to traumatic events as measured by the LEC-5 is positively associated with PTSD symptoms (Brewerton et al., 2020). Test-retest reliability appears to be stable over a period of 7 days, with fair to moderate Kappa values ranging from .23 (other serious accident) to .66 (combat and sexual assault: Gray et al., 2004). The LEC includes two items which assess sexual violence: the first which reads ‘sexual assault (rape, attempted rape, made to perform any type of sexual act through force or threat of harm)’ and ‘other unwanted or uncomfortable sexual experience.’

5.2.4. The ASSIST

The Alcohol, Smoking, and Substance Involvement Screening Test (ASSIST) is an eight-item measure designed to screen for the use of various substances and severity of symptoms (Humeniuk et al., 2010). The ASSIST was administered in the detoxification treatment seeking group to explore degree of substance use severity as a potential predictor of discrepancy. Participants indicated which substances they used in their lifetime, and symptoms (e.g. has anyone expressed concern about your substance use or have you tried to stop usage). This scale has a high internal consistency across substances (α = .77–.94; Humeniuk et al., 2010).

5.3. Procedures

For detoxification treatment seekers, questionnaires were administered in a set order with the CTQ being the first questionnaire, the LEC the second, and the SES being the last administered out of approximately ten study questionnaires. Data were collected in person via pencil and paper packets between July 2016 and May 2017 with informed consent being the first questionnaire in the packet. Participants completed questionnaire packets at large tables in groups that ranged in size from approximately one to ten participants. The overall parent study was designed to assess the relationship between substance use, trauma symptoms, and cognitive functioning. The study protocol was approved by a Great Lakes region Institutional Review Board (IRB-202011-062).

For rural college women, data were collected between November 2020 and April 2021 online via Qualtrics with the consent form as the landing page under the supervision of the MASKED (IRB: 16-317). Questionnaires were block randomised. The parent study was designed to assess the relationship between religious beliefs, rape, and mental health (masked citation).

5.4. Data cleaning, analytic plan, and A Priori power analysis

For detoxification treatment seekers, 39 women, 65 men, and 23 unidentified participants were removed from the detoxification sample because of missing data on the SES. For rural college women group, 9 women were removed due to missing data on the SES or LEC. We use the term discrepancy to refer to the lack of agreement in which cases were identified as positive for victimisation by the two measurement strategies. We used cross-tabulations and McNemar’s tests to examine prevalence rates and test for differences in the two prevalence rates. We also used McNemar’s to examine rates of case identification and discrepancies (yes/no) across the entire distribution (2 × 2) to examine whether the distribution of cases differed. We used McNemar’s tests considering McNemar is more appropriate than chi-square for dependent variables in related groups. For post-hoc follow-up tests we used ANOVA to examine how continuous variables may have varied across the four discrepancy pattern groups (both yes, both no, LEC yes/CTQ-SES no, LEC no/CTQ-SES yes). Prior research on differences in measurement strategy for sexual victimisation suggest medium effect sizes (phi = .37, Anderson & Delahanty, 2020), requiring a total sample size of 58.

6. Results

6.1. Sample 1: detoxification seekers

Of the 91 detoxification participants in the detoxification treatment seekers group, the SES/CTQ identified 58 cases of sexual victimisation (63.74%) while the LEC identified 47 cases (51.65%). Testing Hypothesis 1, overall prevalence rates were statistically equivalent, McNemar exact = .952, p = .329. Regarding Hypothesis 2, McNemar analyses of distribution (6.667) suggested that which cases were identified by each questionnaire was different, p = .007. In other words, while the questionnaires detected a similar number of cases, the questionnaires disagreed (statistically) as to whether an individual had a victimisation history for 20.88% of participants, consistent with prior college samples (Anderson & Delahanty, 2020).

Next, we repeated our initial analyses separately for men and women given prior research on gender differences (Anderson & Delahanty, 2020). There were 52 men in the detoxification treatment seekers group; for men, the SES/CTQ prevalence rate of 51.9% whereas the LEC rate was 36.5%. These were not different prevalence rates H1, ^menMcNemar = 1.100, p = .302, but there was discrepancy in which cases were identified, H2: McNemar = 4.900, p = .021. Specifically, for 37% of men, there was disagreement between the LEC and the SES/CTQ as to whether they had a victimisation history, see Table 3. This is larger than the rate of discrepancy for college men in Strang et al. (2013), χ²(1) = 10.15, p = .001. For the 38 women in the detoxification treatment seekers group, the SES/CTQ prevalence rate was 78.9% whereas the LEC rate was 71.1%. There was no difference in prevalence rates or discrepancy in cases identified, McNemar (.800), p = .375.

Table 3. Hypothesis 1: discrepancy between the LEC and the SES/CTQ.

	LEC – no	LEC – yes	Totals
A. LEC compared to SES only
Sample 1 (n = 91)
SES – no	31 (34.1%)	6 (6.6%)	37 (40.7%)
SES – yes	13 (14.3%)	41 (45.1%)	54 (59.3%)
Total	44 (48.4%)	47 (51.6%)	91 (100.0%)
Sample 2 (n = 310)
SES – no	89 (28.7%)	13 (4.2%)	102 (32.9%)
SES – yes	58 (18.7%)	150 (48.4%)	208 (67.1%)
Total	147 (47.5%)	163 (52.6%)
B. LEC compared to SES/CTQ
Sample 1 (n = 91)
CTQ or SES – no	31 (34.1%)	2 (2.2%)	33 (36.3%)
CTQ or SES – yes	13 (14.3%)	45 (49.5%)	58 (63.7%)
Total	44 (48.4%)	47 (51.6%)
Sample 2 (n = 310)
CTQ or SES – no	86 (27.7%)	9 (2.9%)	95 (30.6%)
CTQ or SES – yes	61 (19.7%)	154 (49.7%)	215 (69.4%)
Total	147 (67.4%)	163 (74.8%)
C. LEC compared to the SES/CTQ controlling for gender
Sample 1, Men (n = 52)
CTQ or SES – no	24 (46.2%)	1 (1.9%)	25 (48.1%)
CTQ or SES – yes	9 (17.3%)	18 (34.6%)	27 (51.9%)
Total	33 (63.5%)	19 (36.5%)
Sample 1, Women (n = 38)
CTQ or SES – no	7 (18.4%)	1 (2.6%)	8 (21.1%)
CTQ or SES – yes	4 (10.5%)	26 (68.4%)	30 (78.9%)
Total	11 (28.9%)	27 (71.1%)

6.2. Sample 2: college women

In the sample of 310 college women, the SES/CTQ identified 215 cases (69.35%) while the LEC identified 163 cases (52.5%). These prevalence rates differed, McNemar = 6.900, p < .01 (H1). The discrepancy rate in which cases were identified between questionnaires, was similarly significant, McNemar = 18.57, p < .0001, p < .001. In other words, the LEC and the SES/CTQ disagreed for 32.13% of the college women reporting sexual victimisation. This is larger than the percentage of discrepant cases for the detoxification treatment seekers group, χ²(1) = 4.264, p = .04 (H2), and larger than prior college samples, χ²(1) = 22.66, p < .0001.

6.3. Post-hoc follow-up analyses

We next conducted exploratory follow-up analyses to determine if there were differences in substance use or number of traumas across patterns of discrepancy (e.g. spread of cases identified) by group, see Table 4. Notably, because of methodological differences, not all analyses were conducted in both Samples. To estimate number of sexual violence incidents (e.g. only sexual violence trauma exposure) we created a continuous score by adding the number of SES/CTQ items endorsed. We created summed scores to estimate the number of trauma types (e.g. all possible trauma exposures) on the LEC.

Table 4. Post-hoc follow-up analyses: potential explanations for discrepancy.

Sample	Explanatory variable		Statistical Tests
	M	SD
	Age
Sample 1 – detoxification seekers
No victimisation (both no)	37.10	11.43	F(3) = .232, p = .874
Endorsement on the LEC only	37.50	14.85	Cohen’s d = .19
Endorsement on the SES/CTQ only	34.75	10.56
Endorsement on both LEC and SES/CTQ	37.73	10.52
Sample 2 – rural college women
No victimisation (both no)	20.48	2.92	F(3) = 1.13, p = .337
			Cohen’s d = .21
Endorsement on the LEC only	20.67	2.55
Endorsement on the SES/CTQ only	21.03	2.99
Endorsement on both LEC and SES/CTQ	21.27	3.511
ASSIST Score/Drinks at Victimisation
Sample 1 – detoxification seekers
No victimisation (both no)	57.80	29.68	F(3) = .538, p = .658
Endorsement on the LEC only	47.00	0.00	Cohen’s d = .15
Endorsement on the SES/CTQ only	62.25	11.24
Endorsement on both LEC and SES/CTQ	72.11	32.89
Sample 2 – rural college women
No victimisation (both no)	0	0	F(3) = 7.38, p < .001
Endorsement on the LEC only	0	0	Cohen’s d = .54
Endorsement on the SES/CTQ only	.63	1.73
Endorsement on both LEC and SES/CTQ	1.31	2.83
Number of Sexual Violence Incidents (SES)*
Sample 1 – women detoxification seekers
No victimisation (both no)	–	–	t(28) = .808, p = .426
Endorsement on the LEC only	–	–	Cohen’s d = .43
Endorsement on the SES/CTQ only	115.00	65.05
Endorsement on both LEC and SES/CTQ	88.11	24.59
Sample 1 – men detoxification seekers
No victimisation (both no)	–	–	t(25) = 5.32, p < .0001
Endorsement on the LEC only	–	–	Cohen’s d = 2.18
Endorsement on the SES/CTQ only	127.50	14.85
Endorsement on both LEC and SES/CTQ	92.50	18.36
Sample 2 – rural college women			t(213) = 3.114, p = .002
No victimisation (both no)	–	–	Cohen’s d = .47
Endorsement on the LEC only	–	–
Endorsement on the SES/CTQ only	9.74	21.27
Endorsement on both LEC and SES/CTQ	19.84	21.51
Number of Trauma Types Endorsed (LEC)
Sample 1 – women detoxification seekers
No victimisation (both no)	1.14^a	1.35	F(3) = 5.70, p = .003
Endorsement on the LEC only	4.00	0.00	Cohen’s d = .84
Endorsement on the SES/CTQ only	2.00^c	2.31	cb: t(28) = −2.26
Endorsement on both LEC and SES/CTQ	6.58^b	3.91	ab: t(31) = −3.58
Sample 1 – men detoxification seekers
No victimisation (both no)	4.25^a	3.48	F(3) = 2.83, p = .048
Endorsement on the LEC only	9.00	0.00	Cohen’s d = .48
Endorsement on the SES/CTQ only	4.44^c	3.40	bc: t(25) = −1.71
Endorsement on both LEC and SES/CTQ	7.17^b	3.81	ab: t(40) = −2.58
Sample 2 – rural college women
No victimisation (both no)	0.99	1.15	F(3) = 24.21, p = .048
Endorsement on the LEC only	3.22	1.56	Cohen’s d = .47
Endorsement on the SES/CTQ only	1.25	1.30
Endorsement on both LEC and SES/CTQ	3.69	1.81

Note: Group n’s are the same as those displayed in Table 3, analyses B for combined gender groups, analyses C for separate gender groups.

Age. The detoxification treatment seekers group was on average older than the rural college women group and had a greater range of ages represented. Because the stigma of responding ‘yes, I have been raped’ may decrease with age (Botta & Pingree, 1997), we tested whether differences in age could be related to discrepancies. Age was not significantly related to discrepancy in either sample, see Table 4.

Substance use symptoms. We examined substance use symptoms as a potential indicator of memory impairment that could hinder recall of substance use related sexual violence. For the detoxification treatment seekers group, there were no differences in substance use severity symptoms (as evaluated by ASSIST scores) across discrepancy groups. In the rural college women group, the number of alcohol beverages consumed at the time of the assault was used as an indicator of possible memory impairment. When compared across patterns of discrepancy, the LEC yes/CTQ-SES yes group reported more alcohol consumption than the no victimisation and LEC yes/ CTQ-SES no group.

Number of sexual assaults. In the detoxification treatment seekers group, there was no difference in the number of total incidents of sexual violence (as estimated by the SES) across patterns of responding for women but there was for men. For men, those who reported on only the SES/CTQ reported more incidents than those who reported on both types of questionnaires. For rural college women, those who reported victimisation on both types of questionnaires – the LEC and the SES/CTQ – reported approximately twice the number of sexual assaults.

Number of trauma types. In both samples, those who reported sexual victimisation on both types of questionnaires – the LEC and the SES/CTQ – reported more trauma types.

7. Discussion

The goal of this study was to compare and contrast two different approaches to identifying cases of sexual violence – general trauma history (e.g. LEC) vs. specialised, behaviourally-specific questionnaires (e.g. SES/CTQ) in two high-risk, health disparity samples. Findings support the use of multiple measurement strategies given the consistent significant differences in which sexual violence cases were identified by each approach.

Our hypothesis (H1) that the behaviourally-specific screening approach – SES/CTQ – would outperform the general trauma history screening approach – the LEC – was partially supported as the SES/CTQ outperformed the LEC in the rural college women group (but not the detoxification treatment seekers group). H2 was not supported and helps contextualises H1; even when prevalence rates were statistically similar across the two strategies (H1) there was discrepancy between the questionnaires in terms of which cases were identified as having a victimisation history (H2). This is consistent with prior research documenting there are often discrepancies between two measures designed to assess the same construct; however, in this case discrepancies were even greater likely because prior work has mostly compared measures that were both behaviourally-specific. Gender-focused analyses revealed that for men, additional cases were identified by the behaviourally-specific approach, SES/CTQ, but were missed by the general trauma screening checklist. H2 hypothesised that discrepancies would be lower in the lower in the detoxification treatment sample compared to prior research; however, we were wrong. Discrepancy rates were higher for detoxification treatment seeking men (37.0%) and rural college women (32.13%) than in prior research (18% in Anderson & Delahanty, 2020). This suggests that specialised, behaviourally-specific approaches will identify ≈1.7× more cases for men and rural college women.

We next explored possible explanations for discrepancies between the two measurement strategies. There were fairly consistent findings in both Samples that those who were trauma exposed were more likely to endorse face-valid items, like those on the LEC. Further, in the rural college women sample we also found that alcohol consumed at the assault was more likely to be captured by the SES/CTQ strategy. This suggests that the LEC may miss alcohol-related incidents, which comprise 80% of sexual violence for college students (Koss et al., 2022).

7.1. Implications for practice, research, and policy

This study recruited two infrequently sampled populations in sexual violence health disparity research – low-income men and rural college women. Our findings suggest that even the most widely used and psychometrically supported questionnaires may operate differently in different populations – and that general trauma screening questionnaires will fail to adequately capture the experiences of men who are sexually victimised and those who experience alcohol-involved victimisation. This study also documented higher discrepancy rates in this sample than in prior research which has relied on a mix of community and college samples from urban areas (Anderson & Delahanty, 2020; Strang et al., 2013) suggesting that the problems in violence measurement identified in earlier research may be magnified in health disparity samples.

A general trauma screening approach may be appropriate for particular settings – such as working with highly trauma exposed individuals and when the specific nature of the incident is less relevant. Indeed, general screening approaches may also be more acceptable to patients as sometimes high trauma exposed populations find behaviourally-specific questionnaires intrusive and upsetting (Siller et al., 2022). However, in instances where understanding the characteristics of sexual violence may be beneficial and health disparity populations are being included, the SES/CTQ approach may still be preferred in spite of the additional time to complete these more complex questionnaires.

Further, we encourage users to consider the specifics of the population being surveyed. For men, behaviourally-specific questionnaires are recommended, even for men with significant trauma histories. We were somewhat surprised to find such large differences for rural college women; however, follow-up analyses suggest this may be driven the lack of recognition of alcohol-related victimisation on the LEC. This is consistent with prior research suggesting that general trauma screening measures are in need of revision (Karstoft & Armour, 2022) as are behaviourally-specific measures like the SES-CTQ (Anderson, Cahill, et al., 2018) – to improve accuracy and inclusion of sexual violence as experienced by many different populations.

Our findings also suggest that research and interventions designed to reduce the stigma for all, including men and rural individuals, may be helpful at the population level. In terms of policy, ensuring that policies to support sexual violence research and related service provision be inclusive, especially of those with substance use disorders and of health disparity populations.

7.2. Limitations

Our research design allowed in-depth investigation of two popular behaviourally-specific strategies and only exploratory investigation of possible mechanisms for differences in measurement. Many participants in the detoxification sample did not complete demographic items or the SES suggesting a level of distrust or fatigue. Further, we did not recruit college men to be able to examine response patterns more thoroughly between gender groups. Within our college sample, some women were raised in more urban settings representing a mix of urban and rural cultural influences at the time of the study.

8. Conclusion

In summary, gender and trauma history impacted responding on both the general-screening questionnaire (the LEC) and the behaviourally-specific questionnaires (the SES/CTQ). Individuals with more severe trauma histories were more likely to endorse face-valid items on the general screening questionnaire and thus, produced fewer discrepancies when LEC responses were compared with the behaviourally-specific strategy (the SES/CTQ). However, men’s experiences would have been largely underreported had we relied only on the general-trauma screening strategy.

Acknowledgements

Thank you to the participants who gave their time for this study even while in crisis.

Disclosure statement

No potential conflict of interest was reported by the author(s).

Data availability statement

The data that support the findings of this study are available at https://osf.io/wknxu/ with password permission from author, RA, as participants did not consent to data being shared at the time of data collection.

Additional information

Funding

Dr. Anderson’s efforts were funded in part by a grant from the National Institute on Alcohol Abuse and Alcoholism (K01AA026643). The content is solely the responsibility of the authors and does not necessarily represent the official views of the funding agency.