The Revised Sexual Experiences Survey Victimization Version (SES-V): Conceptualization, modifications, items and scoring

Abstract

The Sexual Experiences Survey [SES] is considered the gold standard measurement of non-consensual sexual experiences. This article introduces a new victimization version [SES-V] by a multidisciplinary collaboration, the first revision since 2007. The 2024 SES-V is designed to measure the construct of sexual exploitation since the 14th birthday. Notable revisions are adoption of a freely given permission standard for non-consent, introduction of new tactics and acts including made to perform or to penetrate another person’s body, tactics-first wording order, and emphasis on gender and sexual orientation inclusivity. The SES-V is modularized to allow whole or partial administration. Modules include Non-contact, Technology-facilitated, Illegal (largely penetrative), and Verbally pressured sexual exploitation. Tables provide item text, multiple scoring approaches, module follow-up, specific incident description and demographics. Future plans include developing a scoring algorithm based on weighting our hypothesized dimensions of sexual exploitation severity: invasiveness, pressure, and norm violation combined with frequency. This article is the first in a special issue on the SES-V. Subsequent articles focus on the taxonomies and literature that informed each module. The issue concludes with two empirical papers demonstrating the feasibility and validity of the SES-V: (1) psychometric comparison with the 2007 SES-SFV; and (2) prevalence data from a census-matched adult community sample.

The Revised Sexual Experiences Survey-Victimization Version (SES-V): Conceptualization, modifications, items and scoring

The Sexual Experiences Survey [SES] is considered the gold standard measure in assessing non-consensual sexual experiences and has been foundational in advancing research and policy related to sexual violence (Rutherford, 2017). The SES has been in continual use since the late 1970s, with three versions published in the literature (Koss & Oros, 1982; Koss et al., 1987; Koss et al., 2007). The SES was initially developed to document the prevalence of rape, both victimization and perpetration, including the substantial proportion of these experiences not captured in criminal justice data. Further, the SES sought to situate rape along a continuum of increasingly transgressive and invasive nonconsensual sex acts. The SES was innovative in its use of explicit, behaviorally specific terminology to assess non-consensual sexual experiences. Supporting the utility of the SES, a recent review found that up to 70% of research studies on the sexual exploitation of college women (a frequently utilized population) adopted one of the three existing SES versions (Fedina et al., 2018). The present article and the other papers in this special issue introduce the revised 2024 SES victimization version (SES-V). The revision team decided to pursue revision of the perpetration version independently so that conceptual and measurement issues could better be addressed than through simple wording reversal to make the items reference acts performed rather than experienced as in prior versions. This work is underway.

Many researchers who utilize the SES have modified it in various ways. For example, some researchers sought to assess only specific forms of sexual violence (e.g., completed rape), and thus reduced the SES to a subset of items relevant to their purpose. Others added items to assess categories of sexual exploitation more fully than the existing versions did (e.g., alcohol- and drug-facilitated sexual exploitation). A number of researchers modified or added items to the SES to be more inclusive of the experiences of diverse individuals, including cis-gender men, and sexual and gender minority individuals. Furthermore, many methodological papers were published pointing to changes that improved SES validity or sensitivity. The SES-V attempted to align measurement with advancements in the field. This article traces the progression of the revision process from conceptual stages to a consensus revised Sexual Experiences Survey-Victimization version (SES-V). The team intended to improve inclusivity of diverse persons, methods of scoring, applicability across disciplines (criminal justice, public health, and mental health) and utility for researchers, practitioners, advocates and policymakers.

This article delineates and contextualizes what the SES-V measures. We begin with a description of the domains of sexual exploitation as we define them (Figure 1). Later in the article, we specify the dimensions that are hypothesized to predict severity of SES-V scores (Figure 2). This article contains the text of the revised SES-V items (Table 1), and instructions for several scoring approaches (Table 2). Additionally, there are accompanying assessments that we believe increase the validity and utility of the SES-V items including: (1) six follow-up questions for each module (Table 3); (2) a “most serious” specific incident description (Table 4); and (3) a demographic assessment inclusive of race/ethnicity, gender identity, sexual orientation, social economic conditions during formative years, and disability status based on contemporary best practices (Table 5).

Figure 1.

Domains of the sexual exploitation construct.

Figure 2.

Three dimensional model of sexual exploitation.

Table 2.

Sexual Experiences Survey-V Scoring Approaches with Item Coding¹

Dichotomous Scoring by Module2
Module	Coding 0=no and 1=yes2
Non-contact sexual exploitation	Score 1 (yes) for responses > 0 on any items 1 through 5. Score 0 if no >0 responses disclosed on these items.
Technology-facilitated sexual exploitation	Score 1 (yes) for responses > 0 on any items 6 through 15; Score 0 if no >0 responses disclosed on these items.
Illegal sexual exploitation	Score 1 (yes) for responses > 0 on any items 16 through 26. Score 0 if no >0 responses disclosed on these items.
Verbally pressured sexual exploitation	Score 1 (yes) for responses > 0 on any items 27 through 37. Score 0 if no >0 responses disclosed on these items.
Dichotomous Scoring by Sexual Act2,3
Sexual Act	Coding 0=no and 1=yes
Penetrative contact	Score 1 (yes) for responses > 0 on any items 16 through 25 and 27 through 37, any option except a. Score 0 if no >0 responses disclosed on these items.
Attempted penetrative contact	Score 1 for responses > 0 on item 26 except item a. Score 0 if no >0 responses disclosed on this item.
Non-penetrative contact	Score 1 for response > 0 on items 16 through 37 option a only. Score 0 if no >0 responses disclosed on these items.
Non-contact	Score 1 for response > 0 on any items 1 through 5. Score 0 if no >0 responses disclosed on these items.
Cybersex	Score 1 for responses > 0 on any items 6 through 15. Score 0 if no >0 responses disclosed on these items.
Continuous Scoring4
Module	Coding=Sum of “how many times…”
Non-contact sexual exploitation	Sum number of times for items 1 through 5.
Technology-facilitated sexual exploitation	Sum the number of times for items 6 through 15.
Illegal/potentially illegal sexual exploitation	Sum the number of times for items 16 through 26 (include options a through f for all items).
Verbally pressured sexual exploitation	Sum the number of times for items 27 through 37 (include options a through f for all items).
SES-V Total Score	Sum of “how many times…” for each of the items 1–37, including all options a through f.
Ordinal Scoring5
Ranks in Imputed Order of Severity	Coding=ordinal
Penetrative sexual exploitation	Score 5 for any response > 0 on any items 16 through 25, options b, c, d, e, and f. The responses to any other items are ignored.
Non-penetrative sexual exploitation	Score 4 for any response > 0 on item 26 or any response >0 on items 16 through 25, option a only. No responses >0 on penetrative items. (16–25) Responses to items 1 through 15 and 27–37 are ignored.
Verbally pressured sexual exploitation	Score 3 for any response > 0 on items 27 through 37 with no >0 responses on items 16 through 26. Responses to items 1 through 15 are ignored.
Technology-facilitated sexual exploitation	Score 2 for any response > 0 on items 6 through 15. Responses on items 16 through 37 must be zero. Ignore responses on items 1–5.
Non-contact sexual exploitation	Score 1 for any response > 0 on items on items 1 through 5. No >0 responses items 6 through 37.
No sexual exploitation disclosed	Score 0 if no response >0 occurs on any option of any item.
Scoring Rape Prevalence Percentage Consistent with the FBI Definition6
Total rape	Total rape is the sum of percentages of respondents who meet scoring criteria for completed or attempted rape, regardless of the number of experiences.
Completed rape	Completed rape is the percentage of respondents with a >0 response on items 16 through 22 options b, c, or e. Ignore response to item 26.
Attempted rape	Attempted rape is the percentage of respondents with a response > 0 on item 26, options b, c, or e AND no >0 responses on items 16 through 22.

Notes:

¹There is no best approach to scoring the SES-V because the choice depends on study goals. Each scoring method has uses for which it is best matched as well as pros and cons. Further discussion is in the text.

²Using dichotomous scoring, respondents receive a yes or no score for each module regardless of the types or number of experiences. However, individuals often have experiences in more than one module. Therefore, dichotomous scoring by module generates percentages that exceed 100%. The same caveat results for dichotomous scoring by sexual act or by tactic (not illustrated; see Peterson, Koss et al., this issue.

³The response options within each item are a=non-penetrative sexual contact, option b=oral penetration, option c=vaginal/genital opening penetration, option d=made to penetrate another person’s vagina/genital opening, option e=anal penetration, and f= made to penetrate another person’s anus.

⁴Continuous scoring reflects the cumulative burden of sexual exploitation using frequency. However, used alone, continuous scoring equates the severity of each item. Specifically, one experience of sexualized comments contributes 1 point to the summary score, which is equal to the 1 point that would be counted for forcible penetration. Additionally, if multiple acts and/or tactics occurred in a single incident, continuous scoring may not accurately reflect the unique number of incidents of sexual exploitation experienced.

⁵Ordinal scoring is non-duplicative and prevalence percentages add to 100%. However, this scoring approach is based on assumptions about the relative severity of the acts that are contradicted by some empirical data and the scores violate the assumptions of various statistical analyses. The text elaborates the recommendation to avoid using ordinal scoring.

⁶Many users of prior versions of the SES were most interested in the total rape, completed rape, and attempted rape prevalence percentages.The SES has always based these categories on the Federal Bureau of Investigation definition of rape used in collection of the Uniform Crime Reports. These scoring instructions are grounded on the most current (2018) definition, which is: “Penetration, no matter how slight, of the vagina or anus with any body part or object, or oral penetration by a sex organ of another person, without the consent of the victim. Attempts or assaults to commit rape are also included in the statistics presented here; however, statutory rape and incest are excluded.” FBI — Rape, Attempted rape and completed rape are scored as mutually exclusive and summed to calculate total FBI-defined rape prevalence.

This article is the first of a special issue of The Journal of Sex Research on the SES-V. The other papers focus on the theoretical foundations, relevant literature, rationale for item choices and wording within each sexual exploitation module. The modules are named groupings of items including non-contact (Peterson, Littleton, et al. this issue), technology-facilitated (Kowalski & Thompson, this issue), illegal acts including both non-penetrative and penetrative (Peterson, Littleton et al., this issue), and verbally pressured sexual exploitation (Anderson, Peterson, Canan et al. this issue). Two empirical papers complete the issue. The first focuses on psychometrics (Anderson, Peterson, Koss et al.). The data compare the frequencies of disclosure on the existing 2007 SES to those found with the revised 2024 SES-V using two national samples of students. The second empirical paper is based on a nationally representative, census matched community sample (Peterson, Koss et al.). Both empirical papers report prevalence percentages using the SES-V scoring methods summarized in Table 2 and selected correlates derived from the accompanying assessments. Each individual article has domain-specific recommendations for future research to better understand SES-V performance and utility.

The SES-V Revision Process

The first author sought to compile a team that included some early and mid-career scholars who could foster SES research among future academic generations. The final 14-person SES Revision Collaboration team included the first author, two senior career scholars from the 2007 SES revision (Antonia Abbey and Jacquelyn White) and 11 new members identified via the assistance of a member selection group. This group included editors and associate editors of relevant journals who made recommendations based on their familiarity with active researchers using the SES, particularly in diverse populations including sexual and gender minorities. The member selection group also sought individuals who had published critical data pointing to content or measurement problems with the SES, as they were best positioned to suggest remediation. In contrast to previous versions, the revision team was multidisciplinary with appointments including nursing, social work, public health, psychology, public policy, and gender and sexuality studies.

The project began in 2021 and coincided with the COVID-19 pandemic. Thus, the work was conducted primarily through Zoom and e-mail. Smaller work groups were arranged by module based on expertise. These groups communicated among themselves outside of bi-weekly large-group discussions. They were accorded latitude to draft their own items based on theoretical and empirical sources including their own qualitative and quantitative work, both published and in process. The suggested items received feedback from the entire group through iterative drafts, but final decisions were made by the experts on each module team. The SES-V is a tribute to all members, whose names appear in the authorship, for the range of constructive and innovative changes that were advocated and subsequently made by consensus.

Defining Terms

The conceptualization of phenomena is integral to their measurement (Cook et al., 2017; DeVellis & Thorpe, 2022; Lord & Novick, 2008). A foundational question the team confronted was how to label the construct assessed with the SES-V. The intent of the original SES was to identify legally-defined rape acts that were going unmeasured in criminal justice surveys. However, from the start, the SES situated rape on a sexual victimization continuum with other unwanted, non-consensual experiences.

Existing Definitions

The challenge the team confronted is that different terms and definitions are used to refer to the scope of SES measurement. For example, the United Nations uses the term violence against women, which is defined as:

“Any act of gender-based violence that results in, or is likely to result in, physical, sexual, or mental harm or suffering to women, including threats of such acts, coercion or arbitrary deprivation of liberty, whether occurring in public or in private life” [https://www.who.int/health-topics/violence-against-women#tab=tab_1].

Many terms appear to be used interchangeably in both scholarly and public media such as sexual aggression, sexual victimization, sexual coercion, unwanted sexual contact, sexual assault, sexual violence, gender-based violence, and rape. For example, the World Health Organization [WHO] defines sexual violence as:

“Any sexual act, attempt to obtain a sexual act, unwanted sexual comments or advances, or acts to traffic or otherwise directed against a person’s sexuality using coercion, by any person regardless of their relationship to the victim, in any setting, including but not limited to home and work.” [Violence against women (who.int)].

The U.S. Bureau of Justice Statistics that conducts the United States National Crime Victimization Survey uses the term sexual assault, which includes:

“… a wide range of victimizations, separate from rape or attempted rape. Includes attacks or attempted attacks generally involving unwanted sexual contact between victim and offender, with or without force. Includes grabbing or fondling and verbal threats. Rape and sexual assault are combined into one victimization measure” [https://ncvs.bjs.ojp.gov/terms].

The National Crime Victimization Survey definition is problematic for using the term sexual assault to subsume rape together with a range of other non-consensual acts. Rape and sexual assault are defined differently yet they are often synonymous across U.S. state statutes, thus undercutting the legitimacy of sexual assault as an umbrella. We sought an overarching term to label many non-permissive sex acts. Any construct label had to meet a list of criteria that accomplish: (1) subsuming acts that occur in cyberspace or in the real world; (2) involving non-contact as well as contact acts; (3) including potentially illegal acts, or acts that currently are illegal or criminalized; (4) harmonizing crime, health and psychological perspectives; (5) exploring the conceptual space beyond seduction where words alone achieve sex acts that lack freely-given permission; and (6) eliciting disclosure from both those who received and those made to perform non-permissive sexual acts on another person’s body.

Sexual Exploitation

The revision team selected sexual exploitation as the construct label. This term already is used by several organizations including the United Nations Refugee Agency, the World Health Organization, and the United Nations Office on Drugs and Crime [https://www.unhcr.org/us/what-we-do/how-we-work/tackling-sexual-exploitation-abuse-and-harassment/what-sexual-exploitation; https://www.unodc.org/documents/data-and-analysis/tip/2021/GLOTiP_2020_15jan_web.pdf; https://www.who.int/docs/default-source/documents/ethics/sexual-exploitation-and-abuse-pamphlet-en.pdf?sfvrsn=409b4d89_]. Sexual exploitation is defined by the WHO in the context of those providing care through their humanitarian work as:

“Actual or attempted abuse of a position of vulnerability, power, or trust, for sexual purposes, including, but not limited to, profiting monetarily, socially or politically from the sexual exploitation of another”.

In the U.S., the word exploitation is seen in the name of the National Center for Missing and Exploited Children [https://www.missingkids.org/] where it references child abuse, child trafficking and sextortion. However, empirical evidence fails to support the restriction of these forms of sexual exploitation to children (Wolak et al., 2018). Our collaboration appreciated the potential of expanding the term sexual exploitation to describe the scope of experiences assessed by the SES-V. According to the Cambridge Dictionary (2022), exploitation refers to using someone unfairly to your own advantage [https://dictionary.cambridge.org/us/dictionary/english/exploitation].

The term sexual exploitation remediates several definitional challenges. It is versatile because an individual may be exploited or exploit others. The SES-V is intended to assess those who are exploited. Those who exploit are being addressed by a team currently developing the SES-P [Perpetration]. The revision team reached a consensus that the SES-V will be considered a measure that operationalizes sexual exploitation in the broadest sense that is practical for any individual questionnaire. Sexual exploitation as a construct avoids limitation to gender-based incidents, does not restrict incidents to intimate relationships and does not characterize all acts as violence. The term sexual exploitation implies transgression and is inclusive of multiple non-consensual acts. Although conveying wrongfulness, sexual exploitation responds to skepticism that all these acts involve violence as that word is commonly understood.

Figure 1 depicts the construct of sexual exploitation using circles to represent the domains that comprise it and text to specify the acts identified as residing within each domain. The domains are named non-contact, technology facilitated, illegal, and verbally pressured sexual exploitation. The purpose of Figure 1 is to visualize via a Venn diagram that sexual exploitation as measured by the SES-V is a departure from the continuum perspective used in all previous SES versions. Rather than being depicted as a spectrum, Figure 1 illustrates that sexual exploitation domains overlap because many non-permissive acts cross boundaries and resist arrangement as a hierarchy. Whether an act without permission is conceptualized as within a single domain or as residing where the circles of the Venn diagram overlap is depicted by the placement of the text. We anticipate concerns from a carceral perspective that the SES-V illegal acts domain represents net widening that would scoop more acts into the criminal justice system purview. The authors are united in opposing mass incarceration. We have not included acts into the illegal domain that are not already criminalized in some or most jurisdictions or condemned in national and global human rights and health proclamations. Furthermore, except for rape and commercial sexual exploitation, the acts in the illegal domain are not generally offenses with carceral consequences.

Some acts spill across domains to include elements of two. For example, non-contact sexual exploitation such as voyeurism and exposure are routinely classed as lower level (misdemeanor) sex crimes, whereas other forms of non-contact such as staring or leering at someone are not. The largest number of overlaps involve acts that are inconsistently viewed across jurisdictions as illegal or criminalized or are in flux. Prime examples are reproductive coercion and being made to penetrate acts on another person’s body. Reproductive coercion is currently illegal in several countries and criminalized as a civil offense in some U.S. states when it involves violation of agreement to have sex with a condom (stealthing), but not when it is expressed as pressure to have children. Misuse of authority overlaps three domains depending on sex act, context, and victim age. Under some circumstances it is a crime and in others is a civil rights violation under U.S. federal and state statutes. When a co-worker or employee with greater power is involved, it is a civil rights violation. Although historically most likely to involve verbal forms, misuse of authority has increasingly migrated to text, email and unwanted images transmitted through technology. Technology-based sexual exploitation is not depicted as clearly illegal because legal scholars and advocates are grappling with criminal justice responses to scenarios such as transmission of images that qualify as child pornography sent by one teen to another or videos of intimate acts are uploaded to public media without permission, commonly as revenge when a relationship ends.

Some forms of sexual exploitation are explicitly not assessed by the SES-V. It is not intended to assess child abuse that occurred before the 14^th birthday. The SES-V excludes measurement of genital mutilation, sexual exploitation of child soldiers, and child marriage. Honor killings are not included because, although they may be punishment for sexual behavior, murder is not a sexual act. Rape in conflict zones is excluded. We do not mean to say that persons raped during war are excluded from responding to the SES-V. We intended to convey that the items do not, and could not, reflect all the criteria needed to measure war crimes. The United Nations defines rape as a war crime as follows:

“Rape and other forms of sexual violence, when committed in the context of an armed conflict either international or non-international, constitute violations under international humanitarian law (IHL). Sexual violence constitutes a war crime and a crime against humanity when committed as part of a widespread or systematic attack directed against any civilian population. The Statute of the International Criminal Court includes rape and some other forms of sexual violence in the list of war crimes and in the list of acts that constitute crimes against humanity”. [Five things to know about sexual violence in conflict zones (icrc.org)]; [OHCHR | Rape as a grave and systematic human rights violation and gender-based violence against women].

Even though the SES-V could exchange the word “experiences” for “exploitation” and retain the same acronym, the name Sexual Experiences Survey was retained for the practical purpose of facilitating literature searches to avoid missing the extensive pre-revision literature. Experiences is also a neutral label for describing the measure to institutional review boards and respondents. Of course, the description of the survey would reveal that it measures sexual experiences lacking permission. The term sexual exploitation is not intended to be seen by respondents, nor are they expected to know what it means. Later in the paper we discuss our approach to assessing the extent to which victimized persons accept sexual exploitation as a potential label for what they perceive as their most serious non-permissive sexual experience.

Non-Consent

The language used in the SES to indicate the non-consensual nature of sexual acts has evolved over time. In its first iterations (Koss & Oros, 1982; Koss et al., 1987), the SES referred to sexual acts as occurring “when you didn’t want to,” but subsequent research highlighted a sometimes-meaningful distinction between wanting sex and consenting to sex (O’Sullivan & Allgier, 1998; Peterson & Muehlenhard, 2007). The 2007 version of the SES changed the phrase to “without my consent” (Koss et al., 2007). More recent research has highlighted that there is no single, agreed-upon definition or indicator of consent. Individuals interpret the word consent in highly variable ways such as if consenting to kissing means that one has also consented to penetrative sex or viewing marriage as consent even when sex is unwanted (e.g., Hamby & Koss, 2003). Extant prevention programs, which often define consent in ways that are not consistent with legal definitions, may inadvertently introduce variability in individual interpretations of the word consent (for a review of the contested meaning of consent, see Muehlenhard et al., 2017). The revision team reached consensus that sexual exploitation refers to sex acts imposed without freely given permission, a term that focus groups have demonstrated to be widely understood. Peterson, Littleton et al. (this issue) elaborate on the empirical foundation for selection of the word permission.

We learned from writing the items that the language to signify lack of permission is dependent on the domain. For example, technology-facilitated sexual exploitation frequently does not offer the opportunity for the recipient to object, meaning that the communication arrives spontaneously or is continued after a request to stop. Non-contact acts often lack an opportunity to agree or refuse because they occur in public spaces or by surprise. Verbally pressured sexual acts often occur when victims are worn down by persistent pressure and feel that it is useless to continue to resist although they do not freely agree to the sexual activity. In contrast, illegal sexual exploitation involving bodily contact references incidents that happen in person, thus permitting a broad range of behaviors to express non-permission. For these reasons, consent is queried with different words depending on the circumstances. Except for illegal acts, consent terminology is incorporated into the item itself. The non-contact and technology-facilitated items include phrases such as “after I asked them to stop,” or “without first asking if it was okay” (see Kowalski & Thompson, this issue; Peterson, Littleton et al., this issue). The verbal pressure items are introduced by the following sentence: “These questions ask about sexual experiences that happened without your freely given permission. Freely given permission is absent when you are pressured to comply” (Anderson, Peterson, Canan et al., this issue). For the illegal acts, respondents are presented with eight examples of explicit behaviors that can indicate lack of permission. The selection of examples was guided by research on how individuals refuse sex (Griner et al., 2021; Marcantonio & Jozkowski, 2020; Muehlenhard et al., 2017; Riemer et al., 2022). Several choices of inability to consent are intended to address victims who did nothing overtly because they were frozen by fear, taken by surprise or when the act(s) occurred within a power dynamic where they felt unable to refuse. The SES-V behaviorally specific scenarios for lack of permission are included in Table 1 just prior to the questions labeled as “contact”. Next, we highlight the changes in the content and format of the SES-V that differentiate it from prior versions.

Design Decisions

Measurement of sexual exploitation has been known to be sensitive to survey format, item wording, and scoring method (Cook et al., 2017). Some settled decisions are that behaviorally specific wording is superior to vague or abstract terms (Fisher et al., 2009) and that two stage scoring methods result in lower rates of detection than single stage scoring (Cook et al., 2011; Daigle et al., 2016; Krebs, 2014). All previous versions of the SES and the SES-V are behaviorally specific and use single stage scoring. Format and content modifications are discussed in the following paragraphs and scoring approaches are the focus of a later section.

Item Structure

Tactics first.

Prior versions of the SES first presented the sexual act that occurred, with a subsequent list of exploitative tactics that might be used to obtain that act. In other words, prior versions of the SES framed the questions around the non-permissive sexual act. However, some research has suggested that presenting the tactic first followed by the list of sexual acts results in higher rates of reported sexual exploitation. For example, using an experimental design, Abbey and colleagues (2005) found that a version of the 1987 SES that presented the tactic first resulted in an additional 13% of women reporting sexual exploitation over the standard version in which the sexual act was presented first (also see Schuster and colleagues, 2020). Framing items in terms of tactics resulted in nearly identical or slightly higher validity compared to the traditional act-first framing (Abbey et al., 2021; Anderson, Garcia et al., 2021). Given that sexual exploitation is generally considered under-reported, higher rates of reporting would imply that that measurement is eliciting a fuller scope (Abbey et al., 2005; Fisher, 2009). For these reasons, the SES-V departs from earlier versions and frames the questions in relation to the tactic—describing the tactic first before presenting the possible sexual acts that might result from that tactic.

Expanded response range.

Original versions of the SES asked respondents to indicate whether each type of sexual exploitation had happened using a yes/no format (Koss & Gidycz, 1985; Koss & Oros, 1982). Since 1987, respondents have indicated “how many times” each experience has occurred with response options of 0, 1, 2, and 3 or more times (Koss, Gidycz et al., 1987; Koss et al., 2007). Research on behavioral self-reports suggests that the range of response options may influence participants’ reports (Anderson & Cuccolo, 2021). For example, Schwarz and colleagues (1985) found that respondents seem to infer that the midpoint of the response range is the average or typical response, and, thus, when a scale is shorter, participants are less likely to endorse values on the high end. Respondents also tend to feel worse about their own behavior when the range is shorter, perhaps because they are more likely to view their responses as outside of the norm. Furthermore, Davis and colleagues (2014) demonstrated that regardless of the scoring approach used, all were improved by incorporating frequency data. Given these findings, a larger range of response options might help respondents feel more comfortable reporting, increase the variance, and improve the validity of disclosures. The SES-V, therefore, includes 11 response options ranging from 0 to 10 or more for each tactic by act inquiry. The SES-V frequency of acts approach differs from the NISVS, which instead asks “how many PEOPLE” have ever…” (Kresnow et al., 2021).

Recall period.

The 2007 version of the SES asked participants to report on experiences of sexual exploitation in the past twelve months and separately since their 14^th birthday. According to the instructions, respondents should exclude experiences in the prior twelve months from their recall of the entirety of their experiences. It is unlikely that participants always carefully read or understood those instructions. Thus, it was hard to ascertain whether the same experience/s were included by respondents in both recall periods (Hilton et al., 1998). Therefore, we eliminated the prior 12-month responses. Most researchers likely do not require both incidence (the past year) and prevalence (cumulative experiences since age 14) in the same study. Further, some researchers will need to use alternate time frames. For example, campus climate surveys might be restricted to the start of the school year or since the respondent arrived on campus. In longitudinal studies, researchers may want to ask about sexual exploitation since age 14 at baseline, but, thereafter, only in the period since the prior assessment (Humphrey & White, 2000; Swartout, Koss et al., 2015; and Thompson et al., 2013).

There are several strands of evidence that support maintaining age 14 as a recall boundary. The age of consent was recently changed from 13 to 16 in Japan. It is 16 in Canada and many U.S. states (18 states stipulate age 18); 15 years in France; and 14 in Germany and Italy [https://www.washingtonpost.com/world/2023/06/16/japan-age-of-sexual-consent-16/ ].The median age at first intercourse in 2019 was 16 years in Scandinavian countries [https://obgyn.onlinelibrary.wiley.com/doi/full/10.1111/aogs.13732], similar to the U.S, although the age has been rising in recent years [https://dhsprogram.com/data/Guide-to-DHS-Statistics/Age_at_First_Sexual_Intercourse.html]. Research by Finkelhor and colleagues (2014) document that, around age fifteen, a shift away from familial abuse to juvenile/peer perpetrators occurs (also see Smith et al., 2018). By age fifteen, peer perpetrators are twice as common as familial ones. Fourteen is the boundary for delineating commercial sex exploitation of children under federal law. Livingston et al. (2004) and Testa & Livingston (1999), observed that sexual exploitation after age 14 is more reflective of adult sexual experiences as opposed to the forms of child sexual abuse. Limiting the SES-V to experiences that occurred since the 14th birthday overlaps the transition to the age of consent, timing of first intercourse, and shift to peers as the predominant perpetrators. This brief review demonstrates that it would be impossible to select a recall boundary applicable across jurisdictions. The pragmatic rationale for retention of the 14^th birthday is to maintain continuity with prior versions of the SES to permit comparisons over time (e.g., Koss et al., 2022).

Gender identity and sexual orientation inclusivity.

The revision team was committed to improving inclusion of gender identity and sexual orientation [SOGI] throughout all components (e.g., Canan et al., 2020). The 2007 SES used gender-neutral pronouns, but anatomy and binary sexual identity remained confounded (e.g., men were instructed to skip the item inquiring about vaginal penetration). The SES-V avoids confounding genital anatomy with gender identity. For the vagina, we have adopted the preferred terminology identified in qualitative work with transgender and gender diverse individuals for surgically created genitalia, which is genital opening (Peitzmeier et al., under review). Thus, the relevant sex acts include both the term vagina and genital opening. It is possible that some respondents do not know what a genital opening is. However, among those for whom it is relevant, the terminology is more likely to make them feel included, and potentially more likely to disclose penetrative victimization because of language precision and preference.

The list of sex acts included in the prior SES versions is expanded by two additions. Past research has identified that some individuals, particularly cisgender men, have experiences of sexual exploitation wherein they are compelled or pressured to perform either non-penetrative and/or penetrative acts on another person’s body (DuMont et al., 2022; Littleton et al., 2020; Smith et al., 2022). The U.S. Centers for Disease Control and Prevention [CDC] assessed what they labeled made to penetrate experiences in the previous two National Intimate Partner and Sexual Violence Surveys (NISVS, Basile et al., 2022; Kresnow et al., 2021). The entire NISVS is gender-binary. Kresnow and colleagues (2021) indicate that made to perpetrate is asked only of males with the following text: “How many FEMALES have ever used physical force or threats of physical harm to make you put your penis in their vagina?”. A prominent scenario is exemplified by a person with a penis, who is incapacitated or asleep, and becomes conscious to discover someone having caused or taken advantage of an erection making the physical movements required to accomplish anal or vaginal/genital opening penetration without the victim’s initiation or opportunity to agree. However, restriction to men is inconsistent with our inclusivity goal, not only for gender identity but also from a sexual orientation perspective. The CDC item is insufficient to cover the scenarios under which individuals might be made to perform non-penetrative sex acts on or to penetrate someone else’s body (see Peterson, Littleton et al., this issue). Anderson, Peterson, Koss et al. (this issue) discuss the empirical process of developing and testing the item wording for assessing made to perform and made to penetrate. They compare the results of alternate wording. Data on the final wording for made to perform and made to penetrate [MTP] in a community sample is reported by Peterson, Koss and colleagues (this issue).

Format

The SES-V is conceptualized as measuring the construct of sexual exploitation. This conceptual approach and the existing literature on how investigators have modified prior versions of the SES suggested several format changes.

Modularization.

The SES-V construct of sexual exploitation is envisioned as consisting of four modules labeled non-contact, technology-facilitated, illegal, and verbally pressured. In the actual SES-V that respondents see, the module names are edited to reduce potential impact on disclosure. As seen in Table 1, the modules within the administration version of the survey are non-contact, cybersex, contact, and verbal pressure. The content of each module is elaborated more fully later. We speculated that earlier versions of the SES might not have provided sufficient breadth within the domains of sexual exploitation assessed, which drove researchers to create their own non-standard extensions. However, no standard survey could be long and broad enough to anticipate what a multitude of researchers require. Longer surveys encourage non-standard alterations to shorten it and shorter surveys invite investigators to add an idiosyncratic mix of items to augment their specific focus. The revision team decided that modularizing the content would permit adequate domain coverage and present a standardized method to reduce length. Investigators could use as many or as few modules as needed for their study goals. On average, each module covers 10 different tactics. When time permits, the team suggests administering the full SES-V in addition to the adjunct assessments discussed later in this article. Doing so increases the scoring and re-scoring options to achieve more nuanced results. When administered online, particularly by mobile phone, the SES-V benefits from the time-saving features of automated surveys including drop down menus, pop-up definitions, and branching. In our preliminary data, the full SES package including all adjunct assessments without using any branching was completed by a community sample in a median of 18.46 minutes (Peterson, Koss, Anderson et al., this issue). This time estimate is lengthened by outliers who stepped away from their computers and returned later to finish (the maximum time allowed was one attempt of 56 minutes).

The revision team strongly requests that investigators shorten the SES-V by administering those modules most relevant to the study goals or through elimination of module follow-up (Table 3), specific incident description (Table 4), or inclusive demographics (Table 5). With these choices, the time for a single module would be quite short. Many investigators use the SES-V solely for selecting rape victims for further study. Just eight items in the illegal acts module are needed to identify FBI-defined cases by following the scoring method found in Table 2. When the opposite is the case and investigators seek additional information, suggested locations for expanded inquiry are the module follow-up (Table 3) or the specific incident description (Table 4). Using the modules as written and modifying outside the SES-V items themselves contributes to the aggregation of data across studies so large data sets can be accumulated.

Disaggregation of acts and incidents.

The 2007 SES as well as other assessments of sexual exploitation have confounded the number of times a respondent discloses a >0 value on an item with the number of occasions on which a person has been victimized. Many SES-V items are compound and ask about six sex acts that could have resulted from the single tactic described under the item number. This structure does not permit determination of whether the sex acts endorsed represented separate victimizations, or single occasions where multiple acts were perpetrated by the same person(s). To disaggregate the number of act/tactic combinations from the number of occasions on which they occurred requires follow-up that has been absent outside criminal justice surveys. The inability to disaggregate into independent incidents violates the assumptions of many statistical analyses (Swartout, Koss et al., 2015; Swartout, Thompson, et al., 2015). The SES-V adopted two approaches to disaggregation that are discussed later in this article because they are placed in follow-up assessments, rather in the SES-V items themselves. Briefly, following each module (see Table 3), respondents are asked, “Sometimes, people report multiple experiences of __________________ [fill in module name] that happened together. Since your 14^th birthday, please estimate how many times more than one of the acts occurred on the same occasion” [fillable box]. The question wording was based on reviewing and revising an item used by Anderson and colleagues (2023). The item demonstrated construct and convergent validity with college students through cognitive interviewing and correlations with estimated frequency counts. The same respondents may be represented again in subsequent module follow-ups. When multiple tactics spanning different modules were involved, the same occasion will be re-referenced in each applicable module. Therefore, summing across modules would inflate the number of incidents. Cross-module incident data are obtained by the specific incident follow-up (see Table 4). Respondents are asked to focus on what they consider their most serious experience. Then they are requested to check all tactics that happened during the selected incident and in a second item, to check all sex acts that occurred. This approach does not produce an estimate of the total number of incidents but does provide more detail about the co-occurrence of tactics and sexual activities across all forms of sexual exploitation assessed by the SES-V. Peterson, Koss et al. (this issue) compare the victimization estimates at the item versus incident level.

SES-V Item Text by Module

Table 1 contains the SES-V item text, organized by modules, and formatted for paper and pencil administration. Most investigators today would opt for electronic administration on mobile phones, tablets, or PCs. In electronic format, the survey has a more streamlined appearance using drop down menus and hyperlinks to pop-up definitions for sex acts, permission, gender identities and sexual orientation. Researchers may also incorporate branching to shorten the survey, but we did not use it in the studies reported in this issue. SES-V materials are available through Open Science at [https://osf.io/hxpsk/]. Included are Microsoft Word and pdf files for the SES-V items alone, the SES-V combined with follow-up assessments, scoring approaches, and importable Qualtrics code. The following description of each module is in the order of presentation within the questionnaire. The labels in this article differ from how they appear on the survey contained in Table 1. The intent is to provide terminology for scholars in this publication and more neutral wording for the survey itself. Scoring approaches for the SES-V are found in Table 2 and are described in a subsequent section of the article.

Non-Contact Exploitation

Non-contact sexual exploitation references acts that involve no direct touching. A recent national survey in the United States found that 76% of women had experienced unsolicited catcalls, sexually explicit comments about their appearance, and repeated sexual advances (University of California San Diego Center on Gender Equity and Health, 2019 [https://stopstreetharassment.org/wp-content/uploads/2012/08/2019-MeToo-National-Sexual-Harassment-and-Assault-Report.pdf]. Additional sex acts where there is no body contact are voyeurism, exposure and/or masturbating in front of someone without their permission. Some non-contact acts may be considered minor crimes (misdemeanors). Despite the perception that these acts are tolerated or normative, they are associated with negative outcomes, particularly if repeated, including depression, anxiety, poor sleep quality, and substance use (DelGreco & Christensen, 2020; Reed et al., 2020). A possible explanation for the relative neglect of non-contact victimization is that most researchers are not simultaneously assessing sexual exploitation with and without physical contact. Researchers studying exclusively contact sexual exploitation can eliminate these items, numbered 1 through 5 in Table 1.

Technology-Facilitated Sexual Exploitation

Since the SES was revised in 2007, there has been a rapid evolution of digital technology (Auxier et al., 2021; Perrin, 2021). The United Nations defines technology-facilitated gender-based violence as:

“Any act that is committed, assisted, aggravated or amplified by the use of information communication technologies or other digital tools which results in or is likely to result in physical, sexual, psychological, social, political or economic harm or other infringements of rights and freedoms. These are forms of violence that are directed against women because they are women and/or that affect women disproportionately. It encompasses many forms, including intimate image abuse, doxing (the sharing of personal or identifying details), trolling (posting messages, images or videos and the creation of hashtags for the purpose of provoking or inciting violence against women and girls) and sharing of deep fake images” [https://www.unwomen.org/en/what-we-do/ending-violence-against-women/faqs/tech-facilitated-gender-based-violence ].

Although this definition is helpful in specifying what constitutes technology-facilitated sexual exploitation and maps well onto the SES-V items, the reader should notice that the United Nations focuses on women and girls. Also notable is classification of these acts as forms of violence. The SES-V is gender inclusive and uses the term sexual exploitation to subsume these acts rather than violence, which could be perceived as hyperbole when applied to settings where there is no face-to-face contact. It may surprise readers that there currently does not exist a gold standard measure to assess technology-facilitated sexual exploitation. Kowalski and Thompson (this issue) provide a two-factor structural model of technology-facilitated sexual exploitation including text abuse and image abuse. The items that operationalize their empirical foundation are numbers 6 through 15 in Table 1.

Artificial intelligence generated fake photos and manipulated pornographic videos are increasing at a rapid rate (e.g., placing the face of a famous performer into a pornographic video they did not make; or digitally undressing women who did not pose nude) [https://www.washingtonpost.com/technology/2023/11/05/ai-deepfake-porn-teens-women-impact/ ]. We believe that the items in Table 1 map onto these instances without referencing that the content was machine-generated. At the time of writing, artificial intelligence can create text and image content only in response to human instructions and the major AI models provide some level of content moderation. Regardless of origin however, what the victim sees is uninvited text or images containing sexual content depicting others or intimate images of themselves sent, forwarded, or uploaded without their permission. In the absence of empirical data, we can’t speculate on relative impact of actual versus machine-generated sexually exploitive content. It may be that personally directed, real content adds trust abuse to humiliation. However, fake images have a wider reach and are more difficult to track and remove from multiple sites. Two technology-facilitated questions are designed to capture potential grooming (i.e., “Someone tried to get me to talk about sex online when I did not want to” and “Someone asked me online for sexual information about myself when I did not want to tell”. Unfortunately, when a young recipient cannot identify the sender, they may receive communications from someone impersonating a younger person with the intent to create unlawful opportunities to interact with underage persons.

Illegal Sexual Exploitation

Illegal sexual exploitation consists of non-penetrative and penetrative sexual acts. These are items 16 through 26 in Table 1. Each of them begins with a tactic that could be used to obtain sex acts without permission. The tactics include being exploited when unable to give permission by someone who served strong drinks/drugs to hasten intoxication, surreptitiously administered drinks/drugs, or after voluntary consumption. Other forms of incapacitation include being unconsciousness or sleep. Additional tactics included in this module are threatened physical harm, actual physical violence, misuse of authority, violation of the conditions on which consent was contingent, and using force, fraud or coercion to make someone take money in exchange for sex. The sexual exploitation tactics could result in one or more of six non-permissive sexual acts represented by options a through f following item stem (see Table 1). They are briefly described below and elaborated by Peterson, Littleton and colleagues (this issue).

Non-penetrative acts.

Option (a) for items 16 through 37 (see Table 1) references non-penetrative sexual contact. These acts are defined for respondents as, “being touched in a sexual way, having sexual parts of your body (breasts/chest/butt, crotch/genitals) stroked or grabbed…also includes making you touch these parts of someone else’s body”. Non-penetrative contact acts may occur in private or public settings such as crowded clubs with dancing. The attempted rape item (#26) options (b) through (f) refer to acts that victims perceived as indicative of intention to penetrate, but for various reasons no penetration occurred. In the 2007 SES, multiple items were presented to measure attempted rape separately by each tactic. Responding to attempted rape items differs from all other SES-V items by diverging from observable behavior and defaulting to victims’ perceptions of the other person(s) intent. The highest discrepancy between SES self-report and narratives occurs for these items and the verbal pressure items (Testa et al., 2004). In addition, the ratio of attempts to total rapes has declined suggesting that more rapes proceed to completion today (Koss et al., 2022). In response to empirical data, the collaboration decided to reduce assessment of attempted rape to one item and to expand assessment of verbally pressured sexual exploitation, which can be phrased on the basis of observed behavior and not perceptions of other people’s motivations.

Penetrative acts.

Many investigators use the SES primarily to measure rape prevalence or to identify rape victims for further research. Additionally, a large literature exists where an individual’s history of rape is used in multi-factorial designs to predict health or justice outcomes, particularly the development of mental health conditions such as posttraumatic stress disorder, help seeking, disclosure, receiving social support, and reporting to criminal justice or campus authorities. Identification of rape victims is also essential to various research foci including study of alcohol effects and prevention evaluation. Delineating those respondents who have experienced rape is complicated because legal definitions differ among state statutes, federal law, and national and international human rights and health documents. In a later section, various approaches to scoring the SES-V are described including one based on the FBI definition. Peterson, Littleton, and colleagues (this issue) elaborate on definitions of illegal acts including rape in a lengthy section of their paper. The following is a summary restricted to defining rape.

The United Nations model rape law passed by the UN General Assembly on July 15, 2021 is an excellent touchstone [https://www.google.com/url?q=https://undocs.org/Home/Mobile?FinalSymbol%3DA%252FHRC%252F47%252F26%252FAdd.1%26Language%3DE%26DeviceType%3DDesktop%26LangRequested%3DFalse&sa=D&source=docs&ust=1684965242848599&usg=AOvVaw2McG8awhT3nTaGPfIafGjb]. The definition of rape in the model law is:

“A person (the perpetrator) commits rape when they:(a) engage in non-consensual vaginal, anal or oral penetration of a sexual nature,however slight, of the body of another person (the victim) by any bodily part or object; or(b) cause non-consensual vaginal, anal or oral penetration of a sexual nature,however slight, of the body of another person (the victim) by a third person; or(c) cause the victim to engage in the non-consensual vaginal, anal or oralpenetration of a sexual nature, however slight, of the body of the perpetrator or anotherperson”.

In contrast, the U.S. The Federal Bureau of Investigation [FBI] definition used to compile the United States Uniform Crime Reports is narrower. It reads: “Penetration, no matter how slight, of the vagina or anus with any body part or object, or oral penetration by a sex organ of another person, without the consent of the victim” [https://ucr.fbi.gov/crime-in-the-u.s/2013/crime-in-the-u.s.-2013/violent-crime/rape)]. The model rape law is notable for including points (b) and (c), which refer to MTP acts or scenarios involving more than one bad actor. The SES-V items that measure these non-permissive sex acts are found in Table 1, items 16 through 25, options (d) and (f) and in the module follow-up (Table 3) regarding acts that involved more than one person. Item 16 through 25 options (b), (c) and (e) apply to the UN’s first point.

Three forms of sexual exploitation that are generally illegal or criminalized are measured by single items. Single item assessment is not ideal for behaviorally specific coverage, interpretation of responses, unreliability and reduced variance compared to longer and specialized questionnaires. The rationale for assessing these exploitative acts with single items knowing that this was not optimal was to populate our construct of sexual exploitation. The alternative was to fully exclude growing forms of sexual exploitation that require an infeasible number of items to fully assess. We define these forms of sexual exploitation briefly and in each case refer readers to comprehensive assessments.

Reproductive sexual exploitation.

Grace and colleagues (2018) define reproductive coercion as:

“…behavior that interferes with the autonomous decision-making of a woman, with regards to reproductive health. It may take the form of birth control sabotage, pregnancy coercion, or controlling the outcome of a pregnancy” (p. 371).

The earliest studies defined reproductive coercion as withholding sexually infectious disease status, particularly HIV (Toska et al., 2015). More recent work includes studies on condom negotiations that often include scenarios where permission for penetrative sex was contingent on agreement to use them, but that agreement was violated (Chen et al., 2023). Condom removal after contingent consent (stealthing) is relatively common (Davis et al., 2023). For example, Bonar and colleagues (2021) found that 19% of women reported being a victim of stealthing. Stealthing is increasingly being codified as illegal (see Branigin, 2022; Hernandez, 2021; Papenfuss, 2022; Robinson, 2018; and Stonehouse, 2019). The SES-V includes a single item to assess reproductive coercion: “Someone withheld information about sexual infection status, lied about using birth control, or tampered with or removed a condom after agreeing to use one” (Table 1, item 23). Among other issues already noted, this type of multi-scenario item risks misinterpretation by the respondent that they must have had all the experiences. Available multi-item scales include the Reproductive Coercion Scale (McCauley et al., 2017), which consists of nine items assessing two underlying domains: pregnancy coercion and condom manipulation. It is notable that contemporary measurement of reproductive coercion seems not to include the original meaning of this term, which was withholding disease information.

Misuse of authority sexual exploitation.

Tanya Burke formed a metoo movement in 2006 to focus on Black women [Get To Know Us | Tarana Burke, Founder (metoomvmt.org)]. Around 2017 a similarly named #MeToo emerged associated with a different organization. It achieved massive media exposure of the pervasiveness of abuse of power and authority (O’Neil et al., 2018). Multiple recent cases in the popular media support that sexual exploitation due to misuse of authority occurs globally at the systemic level and within multiple institutional contexts (e.g., Diaz, 2022; Jaffe et al., 2020; Krahé et al., 2015; Maclellan & Macaskill, 2022; Raymond, 2022). Misuse of authority to obtain sex acts is assessed in the SES-V by item 24 (Table 1). Surveys that fully query sexual exploitation by persons in authority are almost as long as the SES-V. Cortina and Areguin (2021) provide an excellent literature synthesis and review of comprehensive measurement tools.

Commercial sexual exploitation.

The 2020 Global Report on Trafficking in Persons found that 50% of detected victims in 2018 were trafficked for sexual exploitation purposes as compared to forced labor (38%) and forced criminal activity (6%), with the remainder accounted for by coerced begging, forced marriages, organ removal, and other purposes [https://www.unodc.org/unodc/en/human-trafficking/faqs.html#h11]. The number of human trafficking victims identified in 2020 by the Council of Europe has nearly doubled in recent years according to a publication by the Group of Experts on Action against Trafficking in Human Beings released in 2020 [https://www.coe.int/en/web/anti-human-trafficking/greta].

The SES-V uses the U.S. federal definition of commercial sexual exploitation: “… a crime involving the exploitation of someone for the purpose of compelled labor or a commercial sex act through the use of force, fraud, or coercion.” [https://www.congress.gov/bill/106th-congress/house-bill/3244]. Commercial sexual exploitation is measured by item 25, Table 1. A caveat about this item is that federal law distinguishes commercial sexual exploitation for victims between ages 14 and 17 and imposes additional requirements that only apply after age 18. Because the SES-V bounds retrospective recall to age 14, the text will undercount the youngest victims of commercial sexual exploitation. After significant debate, it was decided that inclusion of a different recall period for one item (i.e., ages 14 through 17) would confuse respondents. If commercial sexual exploitation is the main study focus, a longer survey with a range of specific items covering both scenarios applying to younger and older respondents is needed. (For a review of the multi-item tools currently available, see Eickhoff et al, 2023; Gerasso et al, 2023; Macy et al., 2021).

Verbally Pressured Sexual Exploitation

The number of verbal coercion items was expanded from two in the 2007 SES to 11 in the SES-V. These acts are defined as sexual acts achieved predominantly by verbal means of creating pressure that interfere with an individual’s ability to freely give permission. This module assesses experiences in the zone of behavior that oversteps consensual seduction and borders illegal acts. Verbally pressured acts are generally not illegal, and many people erroneously regard them as normative. Anderson, Peterson, Canan et al., (this issue) summarize the seven-part taxonomy of verbally coerced sexual exploitation they derived from integration of empirical literature, both qualitative and quantitative. The resulting taxonomy consists of the following forms of verbal pressure: positive, neutral, substance-related, negative, threats to the relationship, social threats to reputation or exclusion, anger, reminders of past violence by the perpetrator, threats to critical resources, and threats by the perpetrator to harm themselves.

The verbally pressured sexual exploitation module is introduced by the following statement: “These questions ask about sexual experiences that happened without your freely given permission. Freely given permission is absent when you are pressured to comply”. These situations often occur when victims are worn down by persistent pressure and feel that it is useless to continue to resist although they do not freely agree to the sexual activity. The tactics stem of each individual item (Table 1, items 27 through 37 stem begins with, “Someone pressured me by…”). Research pre-dating the 2007 revision has consistently demonstrated that the items measuring verbally pressured strategies have the lowest agreement with ratings derived from narratives provided by the respondent. Concurrence is significantly below the items assessing rape (Testa et al., 2004). The SES-V verbally pressured sexual exploitation items are placed at the end of the assessment. The team expected that better discrimination could be achieved by differentiating the verbal pressure items from illegal acts.

Scoring

The SES-V itself does not contain attention checks. The rationale is that attention checks have been found to change participant behavior, not just in increasing attentiveness but also potentially increasing suspiciousness, which is not optimal for the SES-V items that query about experiences that people generally have difficulty disclosing (Hauser & Schwarz, 2015). To minimize reactivity, questions to assess inattentiveness were placed in follow-up items outside the SES-V itself. The empirical data presented by Peterson, Koss et al., (this issue) contain the results of attention checks including the number of respondents eliminated due to failing them. Investigators themselves should apply their preferred approach.

The scoring design for the SES-V as with all prior versions is a direct, one-stage approach. This method is the alternative to a gate approach where a small number of broad, vague, and presumably less offensive questions are used to determine those respondents that will be branched into the behaviorally specific items. It is only the latter items that determine whether a disclosure meets qualifications for FBI defined rape or the broader umbrella term of sexual assault used in the NCVS. The consensus is that two-stage methods under-identify victimization. Fisher (2009) found that 11 times more cases were detected in an experimental comparison of one stage behaviorally specific scoring with two stage scoring. Two stage scoring introduces additional error variance because both sensitivity and specificity of each stage (i.e., gate questions and behaviorally specific follow-up questions) require psychometric examination (Cook et al., 2011; Krebs, 2014). Specifically, there is movement of persons who reveal victimization on gate questions but deny it on behaviorally specific inquiry and the reverse demonstrated by respondents who deny the gate questions but reveal incidents in the behaviorally specific items. The SES-V scoring instructions presented in Table 2 all involve a single stage methodology.

Ordinal Scoring

In all prior versions of the SES an ordinal scoring approach was recommended to achieve non-duplicated prevalence percentages that represented the highest victimization severity respondents had experienced (Koss & Gidycz, 1985; Koss et al., 1987; Koss et al., 2007). SES items were arranged by the authors in a presumed order of severity. The presentation in Table 2 updates the legacy scoring of ordinal ranks to map onto the added content in the SES-V. These new ranks include one for technology-facilitated exploitation that is wholly new and the added acts and tactics that are subsumed within the illegal acts module. Although the new items represent forms of sexual exploitation that are in flux from a legal perspective, the collaboration has included them as illegal because they are in line with UN declarations and WHO model law, as well as being measured to a minimum degree by the National Intimate Partner and Sexual Violence Survey (NISVS; Kresnow et al., 2021). The ordinal scoring rules in Table 2 assign a non-empirically based severity rank to each module that is generally consistent with findings that illegal experiences are the least frequent (Leemis et al., 2022), and that the most violent tactics and penetrative acts tend to be associated with more severe health consequences (Ullman, 2007). Following the assumed order of increasing severity, ordinal rank scores are 0=no victimization, 1=non-contact, 2=technology-facilitated, 3=verbal pressure, 4=non-penetrative contact and 5=penetrative contact. The word rape is not used in these labels because if that data point is what investigators seek, the instructions for FBI-defined rape scoring in Table 2 should be used. Ordinal scoring creates mutually exclusive percentages that sum to 100% of the sample.

However, research has accumulated that demonstrates problems with ordinal scoring. Other papers in this issue focusing on single modules provide citations for studies that document measurable impacts on both mental and physical health for each of the module content areas. These data challenge the premise that the order of severity reflected in ordinal scoring is sufficient to represent the nuance in outcomes. Outcomes may vary from person to person depending on context and the past personal experiences of the respondents (Koss 1988; Kern et al., 2022). Ordinal scores also preclude determination of the prevalence of lower ranked victimization because responses to these items are ignored if the respondent has responded to higher ranked items. Research suggests that ordinal scores may be unreliable because the error variance associated with assessing the presence/absence of any one category is compounded (Anderson, Garcia et al. 2021). Finally, ordinal scoring does not take advantage of the frequency data the SES generates by use of a “how many times” response format. Davis and colleagues (2014) found that validity was improved by including frequency into any of the six scoring approaches they compared. We want to be clear in advocating against the use of ordinal scores in future research and encourage the use of alternatives that raise fewer cautions. These options are discussed next.

Dichotomous Scoring

Early versions of the SES used a yes/no format for item response. Although item responses were changed in 1987 to continuous (i.e., how many times?), dichotomous approaches can be used to score the data.

Dichotomous scoring by module.

This approach consists of assigning a score at the module level (yes=1 or no=0) representing whether the respondent has any >0 responses on one or more items within that module only. These scores yield a percentage of people that have experienced the type of exploitation captured by each module. Dichotomous module scores avoid reliance on potentially unwarranted assumptions of relative severity. The module scoring rules presented in Table 2 do not utilize the frequency data generated by the SES-V, which is true of most of the approaches in Table 2 and that issue will be discussed at a later point. Users should note that the percentage of >0 responses within a module cannot be added to percentages generated for the other modules to achieve a summary prevalence estimate. Doing so would duplicate many respondents because people will have responses in multiple modules, thus summing to a victimization percentage greater than 100%.

Dichotomous scoring by sexual act or tactic.

The same yes/no scoring approach described for the module level can be applied to the specific sexual acts endorsed (i.e., penetrative contact, attempted penetrative contact, non-penetrative contact, technology-facilitated, and non-contact). Dichotomous scoring may also be performed at the tactic level (e.g., using technology, exploiting incapacitation or impairment resulting from alcohol or drugs, employing physical force, threats of force or harm to self or others, violating reproductive consent, misusing of authority, commercializing sex by force, fraud or coercion, and forms of verbal pressure). Table 2 presents scoring by sexual act. Tactic scoring follows the same logic and is not included in the table. These scoring approaches are suggestions for understanding the nature of sexual exploitation in each sample. However, this scoring option is subject to the same cautions as module-based scoring previously discussed. Specifically, prevalence percentages of individuals reporting non-permissive sex acts or tactics cannot be summed due to participants who responded to multiple options. As with dichotomous scoring by module, these approaches do not utilize frequency data.

Continuous Scoring

The sum of >0 number of times for each item within modules or for the SES-V as a whole make use of frequency data (Davis et al., 2014). Continuous scoring may be useful to capture the total burden of sexual exploitation, and future investigators can experiment with ways to combine these frequencies with other scoring approaches. Davis and colleagues (2014) found that among six scoring models they compared (e.g., by acts, by tactics, by ordinal ranks, etc.), none were clearly superior in validity, and each was improved by weighting for frequency. Continuous scores are amenable to many analyses based on a general linear model. However, users must recognize that without weighting, continuous scoring equates all items in severity. For example, a person reporting sexual comments about their body that occurred ten times would get a higher score than a person reporting one anal penetration. Additional measurement concerns with continuous scoring are that it sums non-independent data points and is affected by outliers that contribute disproportionately because of the high frequencies of some forms of sexual exploitation, such as technology facilitated.

Federal Bureau of Investigation Rape Prevalence

Many investigators have used prior versions of the SES to select rape victims for further study or to project rape prevalence estimates following the federal legal definition. The definition used to compile the Uniform Crime Reports by the Federal Bureau of Investigation was provided earlier. Table 2 contains instructions to identify those respondents who have experienced rape using this definition, including estimation of prevalence for attempted rape, completed rape and total rape. By using only items that comport with the FBI definition, Koss and colleagues (1987) estimated a total rape prevalence estimate of 27.5% or 1 in 4 college women, a figure that was broadly disseminated and influential in public policy due in part to a basis in federal rape law. However, investigators often have a broader focus across forms of sexual exploitation and prefer definitions of rape typically used in disciplines outside of criminology.

Emerging Directions in Scoring

Table 2 represents scoring methods that are available now. The collaboration has identified some of the next steps in refining and increasing the predictive power of the SES-V.

Index scoring

Measures of behaviors such as the SES-V immediately confront the question of whether sexual exploitation is a latent construct (Diamantopoulos, & Siguaw, 2006). Latent constructs are phenomena that cannot be directly observed or quantified so proxy measurements must be used. For example, posttraumatic cognitions cannot be directly observed. However, sexual exploitation can be directly described in terms of observable acts and therefore is not a latent phenomenon. Prior versions of the SES-V have not assumed that sexual exploitation is a latent construct. However, much of the psychometric research on the SES is predicated on a latent construct model (Johnson et al., 2017). The implication of assuming that SES-V items are caused by a latent construct would be that those who are exploited possess some characteristic(s) that directly or indirectly caused their own victimization (DeVellis & Thorpe, 2022). This assumption is inconsistent with multiple publications that report little success in predicting victimization from any personal characteristic. For example, using 14 risk variables grouped into vulnerability-creating traumatic experiences, social-psychological characteristics, and vulnerability-enhancing situations, Koss and Dinero (1989) reported that statistical modeling could improve only 10% over chance and that most rape victims could not be differentiated from non-victims. Many studies that purport to distinguish one group from another at a practically meaningful level do so based on aftereffects as opposed to variables pre-dating sexual exploitation and fail to identify the amount of variance the analyses contribute above chance. Further, we express concern that this assumption is not only empirically but morally and ethically flawed.

We suggest, instead, that measures of sexual exploitation, including the SES-V, are better thought of as indexes of behavioral experiences because separate experiences of sexual exploitation may or may not be related. Further, there is no singular trait that an individual is born with that confers risk for sexual exploitation without outside influence. As noted repeatedly in rape risk reduction interventions such as E-AAA (Senn et al., 2015), or Flip the Script (Crann et al., 2022), no one is at risk of experiencing rape without the presence of an exploitative individual. Sexual exploitation as measured by the SES-V is the effect of behaviors imposed on the respondents. A collection of formative indicators (i.e., items), such as the SES-V, when combined constitute an index (Bollen, 1989; Bollen & Diamantopoulos, 2017; Diamantopoulos & Siguaw, 2006). Conceptualizing the SES-V as an index has several implications for its use and scoring. First, because each item is viewed as a discrete indicator of sexual exploitation, individual item-level incidence and prevalence data is informative. Although indexes typically have a standard scoring procedure that requires all indicators, working with fewer than the full set of indicators, such as modules in the SES-V case, could permit future researchers to focus on types of experiences that raise risk for that form of sexual exploitation (for an example, see Seitz et al., 2022).

Summary Scoring

Our team felt that representing sexual exploitation as unidimensional, as has been done in past versions of the SES, was too simplistic. We would ultimately like to achieve a summary score for the SES-V that is empirically based and maximizes the information that can be gleaned from responses. The SES-V follows Fisher (2009) in hypothesizing that three dimensions determine the severity of a given experience. The hypothesized dimensions with one exception are similar. Our hypothesized dimensions predictive of the severity of sexual exploitation are illustrated in Figure 2. The dimensions include Axis X—the degree of norm violation; Axis Y—the degree of pressure used to compel a sex act without permission; and Axis Z—the degree of bodily invasiveness (Kern et al., 2022). Norm violation represents the degree of condemnation that an act is morally wrong, is at odds with community values, violates social trust, and is potentially illegal. What is perceived as a norm violation varies across time and culture, with perceptions of whether rape can occur within marriage as an example. Someone may live in a culture that condones marital rape (low norm violation) or condemns it (high norm violation). Many people would perceive sexual comments made by strangers on the street as a less significant norm violation than telling lies to coerce someone into a sex act. Pressure references the tactics used to compel a person into sexual acts. Tactics involving physical force and threats of bodily harm are likely at the upper end of the dimension whereas the lower end is populated by acts initiated without offering an opportunity for permission such as unsolicited sexual comments or online interactions. Some victims’ perceptions may vary from general opinion, however. An individual may view unsolicited sexual comments as high pressure perhaps due to who is sending the comments or traumatic past experiences. Invasiveness reflects loss of autonomy, with penetrative sexual acts conceptualized at the upper end of this dimension. The many sexual exploitation experiences with no physical contact represent the lower end of invasiveness, although here too, some individuals may feel highly invaded by humiliating on-line content or another person’s proximity or threatening posture, even if no direct physical contact occurs.

In utilizing this three-dimensional framework, the team sought to ensure that the SES-V provided a theoretical basis for multivariate scoring of severity. Scaling studies are needed to test the degree to which each SES-V item is perceived as invasive, pressured and non-normative. We do not conceptualize these dimensions as orthogonal. In fact, each item would have a score on the X, Y, and Z axis and these would be expected to be correlated. These data will not be easy to collect or interpret. Our expansion of the construct of sexual exploitation to include non-contact and technology-facilitated experiences, behaviors that are nearly universal, reduces item discrimination values (Papp & McClelland 2020; Peterson, Koss et al., this issue). As of this writing, this type of theoretically focused research does not yet exist precluding us from suggesting how to achieve a summary score weighted by the dimensions of severity and by frequency.

Supplemental Assessments

The data reported in the two empirical articles of this issue used the SES-V items contained in Table 1 to test the SES-V under real world conditions (Anderson, Peterson, Koss et al., this issue; Peterson, Koss et al., this issue;). The protocol of these studies also included supplemental assessments including module follow-up (Table 3), specific incident characteristics (Table 4) and demographics (Table 5). These assessments are not formal components of the SES-V and therefore are ideal locations for investigators to add items that reflect perceived missing pieces in the SES-V, and to amplify with questions specifically tailored to study goals. Information obtained through the supplemental assessments enhance interpretation of SES-V responses, especially in counting incidents as opposed to acts and specifying the acts that occurred on a single occasion. The supplemental assessments are described here in greater depth and data derived from them are found in the reports of initial empirical data (this issue).

Module Follow-up

Table 3 presents six questions to follow-up each module. Respondents are directed to think of all the experiences they disclosed in the just-completed module. The follow-up asks for how many times more than one of the acts occurred on the same occasion. For example, an occasion could involve multiple sexual acts such as non-permissive touching, oral and vaginal/genital opening penetration. The perpetrator may have used several tactics for non-permissive acts including verbal pressure, exploiting incapacitation, and posting intimate pictures or videos. This item is intended to estimate the number of occasions as opposed to item by item frequency of each tactic/sexual act combination. With this information, the investigator can determine how many unique victimization incidents are represented by the item responses. The module follow-up also collects the ages when experiences occurred, how many involved more than one perpetrator, the perceived gender identity of the perpetrator and the relationship to the respondent. This information permits re-scoring of SES-V responses by incident and projection of risk ratios for different intersectional groups and relationships.

Specific Incident Characteristics

Table 4 presents a short set of items that are premised on the respondent selecting “the one you consider most serious” from all their disclosures on the SES-V (or on any subset of modules the investigator has selected). Those who responded “no” to all the items would skip this set of questions. Existing studies vary in the instructions for selecting the incident. Some common examples include the one considered “most upsetting,” “most recent,” “most clearly remembered,” or “the one you least want to talk about.” To our knowledge, no empirical data has identified a clearly superior choice, nor anecdotally does it seem to make an observable difference in results (Dean Kilpatrick, personal communication, December 21, 2022). Although we initially experimented with asking respondents to choose their “best remembered incident (Peterson, Koss et al., this issue), we settled on directing respondents to what they consider the “most serious” incident. The rationale for rejecting some other options is as follows. Best remembered is problematic for those whose memories are impaired by alcohol, drugs, sleep, or other reasons. Most upsetting primes for reporting greater emotional distress on impact outcomes. Most recent may be an insignificant experience compared to others the respondent had endured earlier.

The incident description consists of eight questions. They resemble some of the items in the module-follow-up, but in this case refer to one specific experience in the entire SES-V. Respondents are asked to check all tactics and all the sex acts occurred during this experience. They are also questioned about whether one or more perpetrators were involved. The final three questions address the victim’s labeling for the incident. Even among individuals who endorse experiences of FBI-defined rape, many fail to identify themselves as having been raped or having experienced a crime. This finding has been known since the first version of the SES. The original term, unacknowledged rape, is still used today and has generated a significant literature (Anderson, Tarasoff et al., 2021; Artime et al., 2014; Littleton et al., 2007). Research on acknowledgment frequently has used a binary item (is it rape yes/no; for an exception, see Koss, Gidycz & Wisniewski, 1987). Others have argued that the most important distinction is whether the experience is conceptualized as some type of illegal sexual act or crime versus a more normative experience (Littleton et al., 2017; Wilson & Miller, 2016). Recent research demonstrates that, among those who experienced completed rape, individuals choose a variety of labels such as bad sex, a seduction, a mistake, a miscommunication, or say they are not sure how to label the experience (Littleton et al., 2017, 2018; Peterson & Muehlenhard, 2011). The specific incident description presents a list of options based on prior research for how individuals could label their experience. Two versions of the item are asked. In the first, respondents are asked to select all responses that apply. In the second, they are asked to pick their primary choice. The options include typical sexual experience, hook-up, serious miscommunication, mistake on my part, mistake by the other person(s), sexual exploitation, sexual coercion, sexual harassment, rape, sexual assault, social media bullying, some kind of crime but not rape, and something else (fillable box).

Some researchers would prefer follow-up questions after every affirmative response. However, Peterson, Anderson, & Koss (this issue) found an average of 6.3 incidents limited to those respondents within the illegal exploitation module only. An incident report for each separate occasion would greatly extend administration time, increase respondent fatigue, and raise questions of burden versus benefit. Discontinuance becomes a bigger problem the longer a survey gets, especially after the respondent realizes that if they say zero to every item, they branch out of further questions and can finish faster.

Demographic Assessment

Table 5 is a demographic assessment. Virtually all investigators collect these data. They are essential to characterizing samples and required by the American Psychological Association style [https://apastyle.apa.org/jars/race-ethnicity-culture]. Demographic items are often written individually by each study team. Our objective with the items in Table 5 is to provide a standardized item set consistent with intersectionality theory (Crenshaw, 1989; 2013) to inclusively assess: age, sex assigned at birth, current gender identity, sexual orientation, race/ethnicity, socioeconomic status, and (dis)ability. These questions in Table 5 represent our team’s assembly of best practices. Each of the components is discussed in the following material and linked to its source.

Race/Ethnicity.

The Office of Management and Budget is at work revising the assessment of race/ethnicity for the U.S. Census [https://www.documentcloud.org/documents/23587529-initial-proposals-for-updating-ombs-race-and-ethnicity-statistical-standards#document/p18/a2199840]. An interagency technical working group with representatives from 13 statistical agencies picked up work of a previous project conducted between 2014 and 2017. Methodology included both focus groups and public comment. The group proposed the minimum categories that should be offered when more detailed data collection is not feasible or scientifically justified. For self-response data collection, the use the question, “What is your race or ethnicity?” The respondent is instructed to select all that apply from the following choices: White, Hispanic or Latino, Black or African-American, Asian, American Indian or Alaska Native, Middle Eastern or North African, and Native Hawaiian or Pacific Islander. This suggested approach is used word-for-word in Table 5.

Gender Identity and Sexual Orientation [SOGI]:

Young and Bond (2023) conducted a meta-analysis with the intent to describe best practices in measurement of SOGI. They observe, “There is a clear tension between the priority for more nuanced SOGI data and the need for actionable summaries that can advance understanding and advocacy around health disparities” (Young & Bond, 2023, no pagination). As with race/ethnicity, we have attempted to construct items that assess the essential elements based on a credible scientific source. As with a number of areas previously discussed, more extensive questioning is recommended when feasible. That said, Young and Bond (2023) “there is currently no gold standard for how to approach measurement” (2023, no pagination). We have included written the SOGI items to comply with the primary messages of their meta-analysis. These include separate assessment of sex assigned at birth (if this is relevant to the research goal), gender identity, and sexual orientation with specification of whether the question refers to the public or private realm. Given that many potential respondents do not understand SOGI terminology, they suggest definitions for all gender identity and sexual orientation items. Examination of Table 5 reveals that we have incorporated definitions into the items themselves. However, an alternative is to use pop-up definitions as was done in the Qualtrics platform for the SES-V to display definitions for lack of permission, and the sexual acts that may be compelled by exploitative tactics.

Socioeconomic Status.

The measurement of socioeconomic status is the MacArthur Scale of Subjective Social Status-Adult Version (https://sparqtools.org/mobility-measure/macarthur-scale-of-subjective-social-status-adult-version/).

Differently Abled:

The (dis)ability items are based on CDC guidance recommending the WG Short Form on Functioning available from the Washington Group on Disability Statistics (https://www.washingtongroup-disability.com/question-sets/wg-short-set-on-functioning-wg-ss/; also see Adler, 2000).

We do not claim that Table 5 represents a gold standard for assessing demographics. However, it draws together standardized measures and recommended best practices for each of the most frequently assessed personal characteristics. If all investigators collected these minimum demographics, aggregation of multiple samples would be facilitated. The ability to combine samples is essential for assessing sexual exploitation among gender and sexual minorities, individuals with disabilities, those born into poverty, racial and ethnic minoritized individuals, and intersectional identities. Even in large, representative samples, the numbers in these groups frequently fails to provide sufficient power to project stable prevalence and risk estimates for every cell. Existing large-scale studies are skewed to the majorities. The ability to combine studies would allow aggregation into a secondary analysis sample where marginalized and smaller populations were of sufficient size to achieve the power required to meaningfully include them in analyses. Data aggregation would also facilitate achieving results comparable to expensive longitudinal studies using imputed temporal sequences achieved through data from comparable samples collected in different eras.

Discussion

The SES-V 2024 is introduced in this article, the first in a special issue of the Journal of Sex Research. The issue also contains three articles that elaborate conceptual and empirical bases and item content for each module and two empirical papers. Each of these individual articles addresses limitations and envisions future research. Thus, the special issue taken as a whole presents a comprehensive evaluation of strengths, limitations, and research agendas. The SES-V is a free-standing assessment of sexual exploitation that contrasts with other well-known surveys where these items are presented in a larger reproductive health, intimate partner violence, or crime context. The revision sought to expand assessment to a broadly defined construct of sexual exploitation and to make methodological changes suggested by accumulated empirical data on victimization measurement. The notable revisions include the definition of non-consent as acts without freely given permission, a tactics-first as opposed to a sexual acts-first format, use of a single recall period of “since your 14^th birthday”, an extension of frequency assessment to 10 or more times, and a modularized organization that permits part or whole administration. The sexual acts assessed are augmented by inclusion of incidents without prior permission where individuals were made to perform non-penetrative and penetrative sex acts on someone else’s body. The tactics involving alcohol and drugs are expanded to include both surreptitious administration, voluntary consumption to the point of incapacitation, and pressure to consume more than intended. Verbal pressure strategies to achieve sex acts are significantly expanded to better interrogate the conceptual space between seduction and illegal acts. New items were added to assess several areas of contemporary interest including technology-facilitated, reproductive coercion, misuse of authority and commercial sexual exploitation. Three forms of sexual exploitation are measured by single items (reproductive coercion, misuse of authority, and commercial sexual exploitation) when multiple items would be a stronger approach. These items were added to populate the construct of sexual exploitation that the SES-V operationalizes. Despite the larger scope defined, some forms of sexual exploitation are excluded from the SES-V such as child abuse, where alternate measures exist.

Each of the revision decisions raises empirically testable questions. There are several structural features of the SES-V that are worthy of study. For example, given that respondents use the context of the overall survey to understand individual items (Schwarz, 1999), the new modularization of the SES-V raises questions regarding the similarity of responses when only one module versus all modules are administered. Likewise, how do disclosure rates on the SES-V compare to questionnaires where sexual exploitation is embedded in a larger scope such as all forms of intimate violence (e.g., NISVS) or a range of crimes (e.g., NCVS)? Furthermore, what is the difference between asking respondents how many people (NISVS) versus how many times (SES-V)? Frequency data measured by number of times improves the performance of multiple SES scoring approaches (Davis et al., 2014). Therefore, this difference in response format has validity implications, particularly in how responses are sorted into incidents to project prevalence and score severity.

The impact of data collection method on disclosure of sexual exploitation could also be better understood, especially in this changing technological era. Our U.S. national surveys are primarily administered by live interviewers (mostly women) by telephone using computer assisted administration. The SES-V can be administered electronically or by paper and pencil methods. Both provide greater privacy than speaking live. The comparative participation rate on the SES-V when sent by URL to a sample identified by a reproducible method versus the random-digit dialing or community recruited volunteers is unknown. Participation rates are a field-wide problem. The most recent NISVS resulted in such a low response rate that investigators express more confidence in 2016/17 data (Basile et al., 2022; Kresnow et al., 2021).

Scoring methods are another fruitful area for future study. Many criminology surveys still use a two-stage classification (e.g., the NCVS) where a disclosure on a few vague and non-behaviorally specific gate questions are used to trigger an incident follow-up. Coding of victimization status only occurs at the second stage (i.e., incident report). This approach is known to result in underestimation of victimization (Berzosky et al., 2021; also see Krebs et al., 2007; 2016). In contrast, the NISVS, like the SES-V, is scored in one stage (Kresnow et al., 2021). Experimental studies must contrast these victimization identification methods and quantify each in terms of sensitivity, specificity, reliability and validity. The impact on public policy is muted when federal sources and independent estimates differ, especially when the discrepancies can be substantially explained by methodology features that are too nuanced for media dissemination and are of little interest to the public.

A broader agenda of studies can be envisioned to examine the psychometric functioning of the SES-V in a wide range of populations and to better understand why the SES-V operates the way it does. We encourage researchers to utilize the COSMIN checklist to guide their work (Mokkink et al., 2016). Validity research might elicit narratives that are then independently coded to evaluate whether they fit with the intended meaning of the items (e.g., Testa et al., 2004; Littleton et al., 2019). Further studies are needed to compare the present SES-V wording to other investigator’s alternatives that may achieve higher comprehensibility of the items, disclosure rates, greater predictive or concurrent validity, and/or improved inter-rater agreement between self-report and qualitative narratives. Also, of interest is how the SES-V items function in a variety of populations. Very little research has examined how interpretations may vary across racial, cultural or gender-defined intersections. For example, it is possible that there are unique types of sexual exploitation experienced by marginalized groups that are not captured by the SES-V. For example, does the single item on commercial sex map well onto the experiences of indigenous women and recent immigrants, both of which groups are more at risk than other women (Hodgins et al., 2022)? In the United States, racist cultural norms and stereotypes have been used for centuries to justify sexual violence toward African American and Indigenous women and they continue to affect others’ perceptions of the seriousness of what occurred and survivors’ mental health (Lewis et al., 2016; Ward et al., 2023). For example, the Jezebel stereotype of African American women as hypersexualized, animalistic, and manipulative has been used to justify White men’s rape of African American women (West, 1995). Concerns about high rates of African American men’s imprisonment can also discourage African American women from disclosing sexual exploitation by African American men (Gomez, 2023). Distrust of scientific research may affect how African Americans and other members of marginalized groups answer SES-V questions and further community-based participatory research is needed to ensure that the questions and survey format are acceptable (Abbey et al., 2010; White et al., 2013).

Gender comparisons have been an emphasis in some past measurement research (e.g., Anderson & Delahanty, 2020; Anderson, Carstees et al., 2021). Prior work has demonstrated that gender differences can be minimized or amplified depending on the items used to measure sexual exploitation, such as whether they are questioned with gender binary framing (Anderson et al., 2020; Cecil & Matson (2006). Although Peterson, Koss and colleagues (this issue) provide data that suggest the SES-V functions well for cisgender men and women, that is an initial analysis to be built upon. We have already collected and analyzed a national data set on 550 LGBT persons. This work will contribute to evaluation of the extent to which the SES-V is psychometrically equivalent across gender groups and identifies high risk sub-groups for focused interventions.

Racial comparisons reveal that risk for severe forms of sexual exploitation, such as rape, do in fact, discriminate by race (Basile et al., 2022). Indigenous people and multiracial individuals are identified as the most likely of all racial groups to experience contact sexual exploitation (Basile et al., 2022; Rosay, 2016). Hispanic and Asian women may experience less risk for sexual exploitation than White individuals, at least according to the instrumentation used by the CDC NISVS in 2016/17 data (Basile et al., 2022). Distinct from differential risk is acceptability. Acceptability has been primarily examined in relation to interventions, and it encompasses whether individuals find a given approach appealing and appropriate as well as their other cognitive-emotional reactions to it (Sekhon et al., 2017). Acceptability in this case may be thought of more as fit of the measure to the population. For the SES-V, issues such as how the sexual acts are defined and labeled, the language used to describe non-consent, how substance use and incapacitation are described, and even the order of the items might influence acceptability. Thus, understanding whether the SES-V content is acceptable across groups is key for health equity research, and is an important empirical question to be tested (White at al., 2013).

Our team has attempted to measure a broad range of sexual exploitation, and modularized it to provide options for shortening through mapping items to study goals. Yet, we predict that short forms of the full SES-V will be developed. A typical approach might be to identify those items that contribute unique information about sexual exploitation prevalence and demonstrate good reliability and predictive validity across many populations. The limitation of this approach is that when two items are highly correlated, for example misuse of authority and commercial sexual exploitation, only one will make the cut for a shortened version. The end result is a summary score that achieves comparable predictive power of the whole SES-V, but may have less policy impact because descriptive data linked to face valid items (those that measure what they look like they measure) is often more persuasive. A higher altitude view of the SES-V reveals that 14 people contributed expertise to it and the endeavor took more than two years. Our experience with thinking through conceptual issues, writing and iteratively tweaking item wording demonstrates that sexual exploitation questions are harder to produce than it may appear. We hope that sharing our process will discourage individual investigators from modifying wording of the SES-V. Journal editors have relayed to us a challenging problem with people submitting “modified SES” versions and presenting them as replacements. A single study is insufficient to justify that the entire research community adopt changes to a standardized measure. In the future, we hope that investigators will respond to the substantial agenda of SES-V research just proposed. Each contribution is important and as studies accumulate in the future, the authorship team plans to meet periodically to review recent literature, incorporate scientific advances, and release a new standardized version in a continuous cycle of quality improvement that is empirically informed.

The SES-V has several limitations the authorship acknowledges. The group was not fully inclusive. Steps that would have been advisable prior to revision, especially focus groups and cognitive testing were not systematically done. We sought to balance timely availability with feasibility. The ideal process is predicated on funding for activities such as focus group input to what the content should be, cognitive testing of draft items, comparison of alternative approaches to the format including context and response options, and refinement of scoring methods. Most researchers in this field are aware of the meager financial support that is available, especially given that the federal government spends its resources on their own large-scale surveys. Recent research has revealed that 0.3% of NIH dollars are spent on funding violence research by independent investigators (Williams et al., 2022).

Conclusions

The SES-V attempts to bring the SES legacy into the next generation of research through a concerted interdisciplinary effort to incorporate contemporary theory, scholarship, and methodological innovations. As an independent initiative, the authors could focus on the measurement of sexual exploitation without restraints by political considerations or broader aims that contextualize sexual exploitation (e.g., measurement of other crimes, injuries and reproductive health, or forms of intimate partner violence). Historians of social science view surveys as drivers of ontological politics (i.e., what exists politically or what is part of political reality; Rutherford, 2011; 2017; also see Pache, 2022). The SES has been singled out for its contribution to raising awareness, “structuring the experience it purports to measure,” and “creating a statistical reality that can be communicated easily and wields significant political leverage” (Rutherford, 2017, p. 102). The SES-V revision team embraced the opportunity to continue this legacy. We hope that our efforts encourage others to use this standard measure and contribute their data to Open Science repositories. The field could be synergized by coordinated efforts of individual investigators, using minimal resources to collaboratively accumulate sample sizes and conduct analyses with sufficient power to stand their ground against well-funded public sources.

Tables 1, 3 - 5

Table 1.

Sexual Experiences Survey Victimization Version (SES-V)

NIHMS2056704-supplement-Table_1.docx ^{(207.7KB, docx)}

Table 3.

Module Follow-up

NIHMS2056704-supplement-Table_3.docx ^{(42.2KB, docx)}

Table 4.

Specific Incident Description

NIHMS2056704-supplement-Table_4.docx ^{(29.1KB, docx)}

Table 5.

Demographic assessment

NIHMS2056704-supplement-Table_5.docx ^{(47.8KB, docx)}