Sexual assault victims face a penalty for adjacent consent

Significance

Sexual assault is pervasively underreported, and victims are often blamed and stigmatized. We highlight an important but overlooked factor that may contribute to these outcomes: adjacent consent. Even when a rape is objected to verbally and physically, and accomplished only through violent force, respondents see the victim as less virtuous and deserving of support if she previously consented to something sexual with her perpetrator (e.g., kissing, foreplay, sex on a prior occasion). Our findings are striking in an age of “no means no” and “affirmative consent”—and notably hold among even young and politically progressive respondents who do not otherwise disapprove of casual sexual activity. These results thus shed light on the psychological barriers to justice for sexual assault victims.

Abstract

Across 11 experimental studies (n = 12,257), we show that female victims of sexual assault are blamed more and seen as less morally virtuous if their assault follows voluntary sexual intimacy, a factor we term “adjacent consent”. Moreover, we illuminate a psychological mechanism contributing to this penalty: When a woman who provided no consent whatsoever is assaulted, people tend to see her as more moral than if she were not victimized (the “Virtuous Victim Effect”)—yet people do not extend the same moral elevation to victims who consented to sex-adjacent activity before they were assaulted. Adjacent consent plays a unique role in undermining the moral elevation of rape victims; respondents continue to elevate victims when, in the absence of adjacent consent, we introduce other information that makes the perpetrator seem less abhorrent or that makes the victim seem promiscuous, reckless, or sexually interested in her perpetrator. Furthermore, adjacent consent disqualifies rape victims from moral elevation in the eyes of a wide swath of respondents—including political liberals and undergraduates who, when no assault occurs, have no moral objection to (or even applaud) the victim’s voluntary sexual intimacy. Our results thus illuminate how sexual assault victims may be penalized for adjacent consent by even progressive or “sex-positive” communities. Finally, we identify a potential real-world consequence of adjacent consent: using field data from over 180,000 students across 33 U.S. universities, we find evidence that victims are less likely to report their sexual assaults in cases involving adjacent consent.

Sexual assault is pervasively underreported, and victims are often blamed and morally denigrated (1, 2). Many factors contribute to the blaming of sexual assault victims, who are disproportionately female, such as whether the victim consumed alcohol, whether she was assaulted by a stranger or an acquaintance, and the ideological and moral values of those assigning blame (including their endorsement of “just-world” beliefs, according to which bad things happen only to bad people) (3–11).

Here, we investigate an important but overlooked additional factor: whether the victim voluntarily consented to sexual intimacy with the perpetrator prior to the assault. We term this factor “adjacent consent”, to capture the idea that the victim has consented to something sexual, but what happens to them (e.g., sexual intercourse) extends beyond what they consented to (e.g., kissing, foreplay, sexual intercourse on a prior occasion).

Although the stereotypical rape happens out of the blue (e.g., a random attack in a dark alley) (12, 13), sexual assaults often occur “in a dating or ‘hook-up’ context” (2, 14), with some studies of college campuses suggesting that a majority of sexual assaults occur “during hookups” (15, 16). A large 2019 survey of students from U.S. universities found that many sexual assault victims identified “somebody I was involved with or intimate with at the time” as their assailant (17). Moreover, many respondents cited “the event happened in a context that began consensually” as a reason for not reporting their assaults (17), suggesting that victims who have provided adjacent consent may expect less support from their communities.

Prior work on victim-blaming suggests that rape victims may be devalued when it is unclear to outside observers whether the sex was truly against the victim’s will (13, 14). Importantly, however, the present research investigates clearcut cases that, even when they begin with voluntary sexual intimacy, end with unambiguous violations of consent, achieved through violent force. This approach enables us to isolate the negative effect of adjacent consent on support for rape victims, apart from any uncertainty about whether the sex that ultimately occurred was genuinely unwanted.

Across eleven preregistered experimental studies (total n = 12,257), we show that people regard female victims of sexual assault as less virtuous, more blameworthy, and less deserving of support when their assault follows consensual sex-adjacent activity. Moreover, we show that this penalty for adjacent consent does not merely reflect that some people hold “sex-negative” attitudes, and thus generally disapprove of women—whether they are victims or nonvictims—engaging in casual sexual activity. Rather, we illuminate a distinct psychological mechanism underlying the penalty for adjacent consent: Adjacent consent renders rape victims less likely to receive the moral affirmation that people typically grant to victims of wrongdoing.

Previous research highlights that, in general, people often react positively to victimized individuals and groups, and tend to see victims of wrongdoing as more morally virtuous than nonvictims who have behaved identically (18, 19). In one prior study, participants read a story in which Sarah let Gabrielle use her tablet. For some participants, the scenario ended there; others were told that Gabrielle subsequently stole Sarah’s tablet. Participants rated Sarah as more moral and trustworthy when she was a victim than when she was not. This tendency to morally elevate victims is termed the “Virtuous Victim Effect” (VVE), and has been documented across a variety of moral transgressions (e.g., theft, verbal abuse, medical misconduct), including rape (18).

Crucially, the VVE was not observed when Sarah lost her tablet in an earthquake; she received moral elevation only when she was the victim of another’s immoral conduct. And the elevation of victims was specifically moral; Sarah did not receive the same boost when respondents evaluated how intelligent, funny, or athletic she was. Moreover, people did not expect Sarah to behave more morally when she was a victim; they simply attributed more positive moral character to her. Accordingly, the VVE has been hypothesized to reflect an implicit recognition that, in light of having been wronged, victims have higher moral standing [i.e., they are owed more by their communities (20)] (18). The phenomenon thus recalls the moral theory articulated by philosopher Jean Hampton: When a perpetrator wrongs a victim, it casts doubt on the victim’s moral status by suggesting that “the perpetrator is superior to or higher in value than” her (21). The community must then “deny the wrongdoer’s false claim to superiority and … assert the victim’s equal value” (22). In other words, wrongdoing demands a community response that uplifts the victim and affirms her moral worth. The psychological tendency to morally elevate victims may contribute to this affirming response.

While it could seem irrational to regard a person as more virtuous simply because they were victimized, the VVE plausibly serves a useful psychological function: motivating community members to take justice-restorative actions. Indeed, experimentally manipulating a victim’s perceived moral character increases people’s willingness to help her (18, 23) and sanction the perpetrator (18, 24). And when participants are disincentivized from helping and punishing, the VVE has been shown to disappear (18).

Here, we show that when a woman who provided no consent whatsoever is sexually assaulted, people see her as more moral than if she were not victimized—that is, they show the VVE. Critically, however, people do not extend the same moral elevation to victims when their assault follows a consensual sexual encounter. Even when confronted with a case of clear, unambiguous, forcible rape, respondents withhold moral elevation from victims who willingly participated in sex-adjacent activity.

Strikingly, we observe this pattern across a wide swath of respondents, including political liberals and undergraduates who, when no assault occurs, have no moral objection to (or even affirmatively applaud) the relevant acts of voluntary intimacy. Our results thus suggest that sexual assault victims who provided adjacent consent may be penalized (i.e., deemed less virtuous than victims who provided no consent whatsoever) even by progressive or “sex-positive” communities.

We also show that adjacent consent plays a unique role in undermining the moral elevation of victims: respondents continue to elevate victims when, in the absence of adjacent consent, we introduce other information that makes the perpetrator seem less abhorrent or that makes the victim seem promiscuous, reckless, or sexually interested in her perpetrator.

Finally, our paper illuminates a potential real-world consequence of adjacent consent. Through preregistered analyses of the aforementioned 2019 survey (25), we provide evidence that victims who were “involved or intimate” with their perpetrator (which we treat as a proxy for adjacent consent) are less likely to report their assaults.

Experimental Paradigm

In all eleven experimental studies, participants read a vignette about a female target, evaluated her moral character, and made other judgments. Our studies drew on a 2×2, between-subjects design. The first factor (Adjacent Consent vs. Complete Nonconsent) manipulated whether the target initially participated in a consensual sexual encounter. The second factor (Victim vs. Nonvictim) manipulated whether the target was ultimately sexually assaulted.

Our vignettes drew on stimuli from refs. 9 and 18. The majority of our studies featured a party vignette that began by introducing the target, Alicia, with some incidental background information: Alicia is a politically moderate college student. Last Saturday night, Alicia and a few of her close friends went to a party at a campus fraternity. At the party, Alicia chatted with a few different classmates.

In the Complete Nonconsent conditions, Alicia meets a man named Michael and agrees to see a movie poster in his bedroom. The vignette continues: Toward the end of the night, Alicia talked with an acquaintance named Michael who lived in the frat. During their conversation, Alicia did not flirt with Michael, and he had no idea whether she was attracted to him. At some point, Michael and Alicia were talking about movies. Michael then offered to show Alicia a movie poster in his bedroom, which was just upstairs, and Alicia agreed. Then, in the nonvictim condition, the vignette ends: Once in his room, Michael showed Alicia the poster. In the victim condition, however, Michael assaults Alicia: Immediately after they entered his room, however, Michael closed the door behind him and then pushed himself on top of Alicia. Alicia tried pushing Michael off her but wasn’t strong enough. When she tried to scream, Michael covered her mouth, and shortly after he forced sexual intercourse with her.

In the Adjacent Consent conditions, by contrast, Alicia flirts with Michael, and then voluntarily “hooks up” with him in his bedroom, before asking to stop. After Alicia is introduced, the vignette continues: Toward the end of the night, Alicia talked with an acquaintance named Michael who lived in the frat. During their conversation, Alicia flirted with Michael, and he got the sense that she was attracted to him. At some point, Michael and Alicia were talking about movies. Later on, Michael suggested they go to his bedroom, which was just upstairs, and Alicia agreed. In Michael’s room with the door closed, they began hooking up. Then, in the nonvictim condition, the vignette ends: After a while, however, Alicia told Michael that she wanted to stop, so they stopped. In the victim condition, however, Michael assaults Alicia: After a while, however, Alicia told Michael that she wanted to stop, but Michael kept going. Alicia tried pushing Michael off her but wasn’t strong enough. When she tried to scream, Michael covered her mouth, and shortly after he forced sexual intercourse with her.

As outcome variables (all measured on 1 to 9 Likert scales), participants in all conditions rated the target’s morality (“How moral of a person is Alicia?”) and trustworthiness (“How trustworthy of a person is Alicia?”); following our preregistrations, we average these two items to form a composite measure of moral character (a robustness check in SI Appendix, section 3.1 reveals similar results for the individual morality and trustworthiness items). In the victim conditions only, participants made a variety of additional judgments that pertain to community responses to sexual assault—such as their sympathy for and blame of the victim, support for helping the victim and punishing the perpetrator, and attributions of rape and consent (with the included measures varying across studies).

Results

Victims Are Judged Less Positively Following Adjacent Consent.

We begin by asking whether victims who provide adjacent consent seem less virtuous, and receive more blame and less support. To investigate, we set aside the nonvictim conditions and examine, within the victim conditions, the simple effects of Adjacent Consent (vs. Complete Nonconsent) on our outcome measures.

We pool data across six studies—Study 1 and five studies that we later discuss in detail (Studies 7 to 11)—that share key features. These studies all presented a convenience sample of U.S.-based participants from MTurk with the party vignette. Study 1 (n = 997) employed the basic 2×2 design described above; Studies 7 to 11 included this same 2×2 design, plus additional conditions (described later). Thus, Studies 1 and 7 to 11 share four common experimental conditions (albeit with some very minor variation in vignette wording across studies), allowing us to aggregate data from these conditions across studies (total n = 5,383). We use linear regressions in all analyses, and include dummies for each study in analyses of the aggregate dataset. In SI Appendix, section 3.2, we reproduce our key analyses within each individual study in our aggregate dataset.

As reported in Table 1, comparing the two victim conditions in our aggregate dataset reveals that the victim was seen as less morally virtuous when she provided adjacent consent before being raped than when she provided no consent whatsoever. Participants also blamed her more, felt less sympathy for her, and were less supportive of helping her and punishing the perpetrator. We also find that participants in both the Adjacent Consent and Complete Nonconsent conditions expressed high absolute agreement that a rape occurred, and low absolute agreement that consent to sex was present. Still, when the victim provided adjacent consent, participants agreed less that rape occurred, and more that consent to sex was present. Furthermore, SI Appendix, section 3.3 reports that, when the victim provided adjacent consent, participants believed that she would be less likely to be believed by her peers, and that her peers would face diminished social incentives to help her and punish the perpetrator.

Table 1.

Effects of adjacent consent on reactions to sexual assault

Sample	Dependent variable (Measured on 1 to 9 scales)	Complete nonconsent	Adjacent consent	Comparison
n = 2,673 (six studies)	Moral character composite How moral of a person is Alicia? How trustworthy of a person is Alicia?	6.91 (1.43)	6.08 (1.72)	b = −0.83 [−0.95, −0.71], t = −13.61, P < 0.001
	Blame composite (Correlation with moral character: B = −0.45, P < 0.001) To what extent do you blame Alicia for this incident? How much do you think Alicia is responsible for this incident? How likely do you think it is that Alicia could have avoided this incident? How much do you think Alicia had control over this incident?	2.64 (1.56)	3.62 (2.03)	b = 0.98 [0.85, 1.12], t = 14.09, P < 0.001
n = 1,093 (two studies)	Sympathy composite (Correlation with moral character: B = 0.49, P < 0.001) How sympathetic do you feel toward Alicia? How bad do you feel for Alicia? How sorry for Alicia are you?	8.33 (1.21)	7.53 (1.89)	b = −0.80 [−0.99, −0.62], t = −8.44, P < 0.001
n = 2,174 (five studies)	Helping item 1 (Correlation with moral character: B = 0.37, P < 0.001) The university should provide Alicia with academic accommodations, such as extensions on deadlines, in order to help her cope with what happened with Michael.	7.71 (1.76)	7.04 (2.22)	b = −0.68 [−0.85, −0.51], t = −7.89, P < 0.001
	Helping item 2 (Correlation with moral character: B = 0.33, P < 0.001) Following what happened with Michael, Alicia's peers should go out of their way to support Alicia, even when doing so is costly or difficult for them.	7.45 (1.67)	7.04 (1.91)	b = −0.42 [−0.57, −0.27], t = −5.42, P < 0.001
	Punishment item 1 (Correlation with moral character: B = 0.31, P < 0.001) Michael should be expelled from the university and banned from its campus.	8.46 (1.21)	7.94 (1.81)	b = −0.53 [−0.65, −0.40], t = −7.95, P < 0.001
	Punishment item 2 (Correlation with moral character: B = 0.31, P < 0.001) Going to prison would be an appropriate consequence for Michael's behavior.	8.11 (1.50)	7.43 (1.98)	b = −0.68 [−0.83, −0.53], t = −9.03, P < 0.001
n = 1,580 (four studies)	Rape attribution item 1 (Correlation with moral character: B = 0.37, P < 0.001) Alicia is a victim of rape.	8.74 (0.79)	8.34 (1.44)	b = −0.40 [−0.52, −0.29], t = −6.94, P < 0.001
	Rape attribution item 2 (Correlation with moral character: B = 0.32, P < 0.001) Michael is a rapist.	8.67 (0.94)	8.19 (1.61)	b = −0.48 [−0.61, −0.35], t = −7.25, P < 0.001
	Consent attribution item 1 (Correlation with moral character: B = −0.33, P < 0.001) Alicia consented to sexual intercourse with Michael.	1.37 (1.24)	2.10 (1.99)	b = 0.74 [0.58, 0.90], t = 8.89, P < 0.001
	Consent attribution item 2 (Correlation with moral character: B = −0.31, P < 0.001) Michael believed that Alicia consented to sexual intercourse with him.	2.12 (1.83)	3.36 (2.55)	b = 1.24 [1.02, 1.45], t = 11.07, P < 0.001

When sexual assault victims provide adjacent consent, participants see them as less morally virtuous, blame them more, feel less sympathy for them, view helping them and punishing the perpetrator as less merited, and differentially endorse statements about whether rape occurred and consent was granted. Drawing from the victim conditions of Studies 1 and 7 to 11, we show simple effects of Adjacent Consent (vs. Complete Nonconsent) on each of these dependent variables. We also show that, within the victim conditions, moral character ratings correlate significantly with all other dependent variables, suggesting that perceptions of a victim’s virtue may predict how people will respond to her sexual assault more generally.

Together, these findings suggest that victims of forcible rape who provided adjacent consent—even when judged, in absolute terms, as clearly not having consented to sex—are liable to receive less supportive responses from their communities.

Next, we unpack why rape victims are penalized for adjacent consent. One possibility is that people disapprove of the sexual activity the victim consented to prior to the assault, and the penalty merely reflects this general sex-negativity. If so, the relevant sexual activity should trigger a similar penalty even in the nonvictim conditions, where no assault occurs. Alternatively, however, adjacent consent might alter the psychological response that people have to victimization—in particular, by diminishing their tendency to affirm and uplift victims.

To discriminate between these possibilities, we must consider how participants evaluated the target in both the victim and nonvictim conditions. We thus turn to focusing on a dependent variable that we could measure even when the target was not assaulted: perceptions of the target’s moral character. Whereas judgments like blame of the victim or punishment of the perpetrator make sense only when a sexual assault has occurred, the target’s moral character can be evaluated even in the nonvictim conditions. Thus, analyses of moral character can illuminate whether adjacent consent is penalized merely because people hold general sex-negative attitudes, or because adjacent consent serves to modulate how people respond to a target becoming a victim. Moreover, prior research shows that moral character attributions are fundamental to our impressions of others (26–28), and play a key role in guiding behavior, both toward people in general (29) and victims of wrongdoing in particular (18, 23, 30). Indeed, within our victim conditions, moral character ratings correlate significantly with all the other outcome measures (Table 1), suggesting that perceptions of a victim’s virtue may provide a helpful window into how her community will respond to her sexual assault more generally.

Adjacent Consent Undermines the Moral Elevation of Victims.

When we examine moral character ratings across all four conditions of the aggregate dataset (and thus consider both the victim and nonvictim conditions), we find that victims are not penalized for adjacent consent merely because people hold general sex-negative attitudes.

To be sure, we do find evidence for sex-negative attitudes. When no sexual assault occurs (i.e., in the nonvictim conditions), participants rate the target as significantly less virtuous in Adjacent Consent (where she hooks up with an acquaintance at a party before asking him to stop) than in Complete Nonconsent (where she simply attends the party and chats with the acquaintance) (Table 2, Row 1, Column 2 for statistics). In other words, when convenience samples evaluate our party vignette, sex-adjacent activity carries a moral penalty even in the absence of victimization. This baseline moral penalty for adjacent consent may reflect the well-documented disapproval, especially among conservatives (31–33), of “unchaste” sexual activity (e.g., casual sex, hookup culture) (34, 35) or of women who “tease” men by consenting to some sexual activity before asking to stop (10, 36).

Table 2.

Results across five experimental contexts

	Simple effects of adjacent consent (vs. complete nonconsent)		Simple effects of victim status (vs. nonvictim status)
	Victim conditions	Nonvictim conditions	Complete nonconsent conditions	Adjacent consent conditions	Interaction between adjacent consent & victim status
	Column 1	Column 2	Column 3	Column 4	Column 5
Party vignette, Convenience Sample Aggregate analysis of Studies 1, 7 to 11 (total n = 5,383)	No consent: 6.91 (1.43), Adj. consent: 6.08 (1.72), b = −0.83 [−0.95, −0.71], t = −13.61, P < 0.001	No consent: 6.33 (1.36), Adj. consent: 5.99 (1.65), b = −0.33 [−0.45, −0.22], t = −5.74, P < 0.001	Nonvictim: 6.33 (1.36), Victim: 6.91 (1.43), b = 0.58 [0.48, 0.69], t = 10.80, P < 0.001	Nonvictim: 5.99 (1.65), Victim: 6.08 (1.72), b = 0.08 [−0.05, 0.21], t = 1.26, P = 0.207	b = −0.50 [−0.67, −0.33], t = −5.92, P < 0.001
Apartment vignette, Convenience Sample Study 2 (n = 803)	No consent: 7.07 (1.44), Adj. consent: 6.43 (1.78), b = −0.64 [−0.96, −0.32], t = −3.93, P < 0.001	No consent: 6.44 (1.38), Adj. consent: 6.23 (1.71), b = −0.21 [−0.51, 0.10], t = −1.34, P = 0.181	Nonvictim: 6.44 (1.38), Victim: 7.07 (1.44), b = 0.63 [0.36, 0.91], t = 4.51, P < 0.001	Nonvictim: 6.23 (1.71), Victim: 6.43 (1.78), b = 0.20 [−0.14, 0.55], t = 1.16, P = 0.246	b = −0.43 [−0.87, 0.01], t = −1.93, P = 0.054
Apartment vignette + Temporally distant adjacent consent, Convenience Sample Study 3 (n = 792)	No consent: 7.39 (1.41), Adj. consent: 6.93 (1.66), b = −0.46 [−0.76, −0.16], t = −2.98, P = 0.003	No consent: 6.58 (1.35), Adj. consent: 6.57 (1.57), b = −0.02 [−0.31, 0.27], t = −0.12, P = 0.908	Nonvictim: 6.58 (1.35), Victim: 7.39 (1.41), b = 0.81 [0.54, 1.08], t = 5.84, P < 0.001	Nonvictim: 6.57 (1.57), Victim: 6.93 (1.66), b = 0.36 [0.04, 0.69], t = 2.23, P = 0.026	b = −0.44 [−0.86, −0.02], t = -2.07, P = 0.038
Party Vignette + Disclaimer, Convenience Sample Study 4 (n = 1,193)	No consent: 6.89 (1.46), Adj. consent: 6.45 (1.74), b = −0.43 [−0.69, −0.18], t = −3.33, P = 0.001	No consent: 6.41 (1.36), Adj. consent: 6.33 (1.65), b = −0.08 [−0.33, 0.16], t = −0.65, P = 0.516	Nonvictim: 6.41 (1.36), Victim: 6.89 (1.46), b = 0.47 [0.25, 0.70], t = 4.09, P < 0.001	Nonvictim: 6.33 (1.65), Victim: 6.45 (1.74), b = 0.12 [−0.16, 0.39], t = 0.85, P = 0.398	b = −0.35 [−0.71, 0.00], t = −1.96, P = 0.050
Party vignette, Undergraduate Sample Study 5 (n = 806)	No consent: 7.04 (1.54), Adj. consent: 6.75 (1.54), b = −0.29 [−0.60, 0.01], t = −1.90, P = 0.058	No consent: 6.11 (1.23), Adj. consent: 6.49 (1.57), b = 0.38 [0.11, 0.66], t = 2.75, P = 0.006	Nonvictim: 6.11 (1.23), Victim: 7.04 (1.54), b = 0.94 [0.67, 1.21], t = 6.80, P < 0.001	Nonvictim: 6.49 (1.57), Victim: 6.75 (1.54), b = 0.26 [−0.04, 0.57], t = 1.68, P = 0.093	b = −0.68 [−1.09, −0.27], t = −3.26, P = 0.001

Across five experimental contexts, we report effects on perceived moral character of adjacent consent, victim status, and their interaction. We find that adjacent consent consistently diminishes perceived victim virtue (column 1), despite not consistently triggering a baseline moral penalty in judgments of nonvictims (column 2). This reflects that participants reliably elevate rape victims who provided no consent whatsoever (column 3), but do not extend the same moral elevation to those whose rape follows consensual sexual intimacy (columns 4 and 5).

Critically, however, we find that adjacent consent also alters the way that people respond to victimization. When the target has not consented to anything sexual (i.e., within the Complete Nonconsent conditions), participants rate her as meaningfully more virtuous when she becomes a victim than when she does not (i.e., we observe a VVE) (Fig. 1A; see Table 2, Row 1, Column 3 for statistics). Yet when the target has voluntarily participated in a “hook up” before asking to stop (i.e., within the Adjacent Consent conditions), becoming a victim does not increase her perceived moral character (Fig. 1B; statistics in Table 2, Row 1, Column 4). In other words, adjacent consent undermines the VVE, resulting in a significant negative interaction between adjacent consent and victimization (Table 2, Row 1, Column 5). These results thus suggest that rape victims face a penalty for adjacent consent not just because people inherently disapprove of the sex-adjacent activity that was consented to, but also because voluntary intimacy renders them less eligible for the moral elevation that is typically bestowed on victims.

Fig. 1.

Adjacent consent undermines the moral elevation of victims. (A) When a woman who has provided no consent whatsoever is sexually assaulted, participants see her as more morally virtuous than if she were not victimized (the Virtuous Victim Effect, illustrated by the blue bars shifting rightward relative to the orange). (B) When assault follows voluntary sexual intimacy, however, participants do not morally elevate the victim (illustrated by the lack of shift in the blue bars, which overlap with the orange). We plot moral character ratings as a function of victim (vs. nonvictim) status, within the Complete Nonconsent and Adjacent Consent conditions, across six studies presenting the party vignette to convenience samples (Studies 1 and 7 to 11).

Next, we highlight an important implication of this latter mechanism: Victims can be penalized for adjacent consent even in contexts where people do not inherently disapprove of the sex-adjacent activity they consented to. In Studies 2 to 5, adjacent consent carried no significant baseline penalty—but nonetheless undermined the VVE.

In Study 2 (n = 803), we applied our basic 2×2 design to a different vignette featuring different contextual details. In this apartment vignette, the characters are older, the events occur not at a party but in the target’s apartment, and the target has a different relationship to the perpetrator—the pair have been on two dates in Adjacent Consent, and are neighbors in Complete Nonconsent. In the Adjacent Consent/Victim condition, like in our party vignette, the perpetrator forcibly rapes the target after she voluntarily hooks up with him and then asks him to stop.

In Study 3 (n = 792), we adapted the apartment vignette to change the timescale over which adjacent consent occurred. Here, in the Adjacent Consent/Victim condition, the pair had been dating a few months, and were regularly having sex. But on the day of the assault, the target did not consent to sex; when the perpetrator tried to initiate a sexual encounter, she immediately told him no. Thus, the adjacent consent (i.e., the prior voluntary sexual engagement with the perpetrator) is more temporally removed from the assault.

In Study 4 (n = 1,193), we again used the party vignette, but adapted it so that, in the Adjacent Consent conditions, the target communicates upfront that she does not wish to have sex: Alicia informs Michael, after the two begin kissing but before she agrees to go to his bedroom, that she does not want to go “all the way” tonight. Furthermore, when Alicia asks to stop, the vignette specifies that they had not progressed to the point of sex (to convey that Alicia’s upfront disclaimer was not contradicted by her subsequent actions). (For full vignette texts for Studies 2 to 4, see SI Appendix, section 6.)

In Study 5 (total n = 806), we used the original party vignette, but recruited undergraduates rather than adults on MTurk—first from the University of Michigan (n = 494), and then, after the university’s semester ended (and following our preregistration), from Prolific (n = 312 self-reported undergraduates).

The results of Studies 2 to 5 are plotted in Fig. 2 and are reported in Table 2 (alongside an aggregate analysis of Studies 1 and 7 to 11, for comparison). Within the victim conditions of each of these studies, we find negative effects of adjacent consent on perceived morality (Table 2, Column 1). Thus, victims of forcible rape are consistently seen as less virtuous if their rape followed sexual intimacy with their perpetrator, even if they were dating the perpetrator (Studies 2 and 3), provided zero sexual consent on the day they were raped (Study 3), or provided an explicit upfront disclaimer suggesting an unwillingness to have sex (Study 4).

Fig. 2.

Adjacent consent can undermine the moral elevation of victims even in contexts where it carries no baseline moral penalty. Across five distinct experimental contexts, adjacent consent consistently inhibited the moral elevation of victims—regardless of whether it carried a significant baseline moral penalty (A), carried no significant baseline penalty (B–D), or was evaluated positively at baseline (E). We plot the target’s perceived moral character across conditions, in (i) an aggregate analysis of Studies 1 and 7 to 11, (ii) Study 2, (iii) Study 3, (iv) Study 4, and (v) Study 5. Error bars are 95% CIs.

Yet in each of Studies 2 to 5, we observe no baseline moral penalty of adjacent consent (Table 2, Column 2). Within the nonvictim conditions of each of Studies 2 to 4, we find no significant effect of adjacent consent on perceived morality. Thus, convenience samples show no significant negative reaction to a woman, who is not assaulted, engaging sexually with a man she is dating, or with an acquaintance at a party following the disclaimer that she did not wish to go all the way. Strikingly, in the nonvictim conditions of Study 5, we find a significant positive effect of adjacent consent, highlighting that undergraduates actually applaud a woman who is not assaulted for participating in a casual sexual encounter. This result could reflect affirmative approval of the target’s sexual activity (i.e., “sex positivity”), or her assertion of sexual boundaries (saying she wanted to stop) following this sexual activity.

Critically, however, in each of Studies 2 to 5, we find evidence that adjacent consent undermined the VVE. In each study, we observed a significant VVE within Complete Nonconsent, reflecting that participants morally elevated victims who had never consented to anything sexual with their perpetrator (Table 2, Column 3). Yet participants did not extend the same elevation to victims who had provided adjacent consent (Table 2, Columns 4 and 5). In Studies 2 and 4, we observe no significant VVE within Adjacent Consent, and a marginally significant negative interaction between adjacent consent and victim status. In Study 3, we do observe a significant VVE within Adjacent Consent, but also observe a significant negative interaction between adjacent consent and victim status. In Study 5, we observe a marginally significant VVE within Adjacent Consent, and likewise observe a significant negative interaction between adjacent consent and victim status.

Together, Studies 2 to 5 suggest that adjacent consent can undermine the elevation of victims even in contexts where people do not, at baseline, disapprove of the relevant sex-adjacent activity—and even among a population that rewards that activity at baseline. They thus speak to the robustness of this undermining effect: We find evidence that adjacent consent inhibits moral elevation in several situations where one might have expected victims, having done nothing preassault that garners disapproval, to receive the typical VVE. At the same time, however, we note that Studies 2 to 4 (i.e., the experimental contexts that did not feature the standard party vignette) produced somewhat weaker evidence for the negative interaction between adjacent consent and victim status (Table 2, Column 5)—suggesting that adjacent consent may inhibit the elevation of victims more strongly in situations that resemble the party vignette.

We also note that the victim conditions of Studies 2 to 5 additionally measured blame; support for helping the victim and punishing the perpetrator; agreement that the victim was raped; agreement that the victim consented to sex; and (in Studies 2 to 4) sympathy toward the victim. These measures provide further evidence that adjacent consent is penalized; see SI Appendix, sections 4.2–4.5 for details.

Next, we provide further evidence that adjacent consent can undermine the moral elevation of victims even in the eyes of swaths of participants who do not penalize adjacent consent at baseline. Here, our evidence comes from analyses, described in detail in SI Appendix, section 3.4, investigating how our results are moderated by individual-difference variables, when aggregating data from all MTurk studies presented thus far (i.e., Studies 1 to 4 and 7 to 11).

These analyses reveal that the baseline moral penalty of adjacent consent is driven by conservatives and older individuals, as well as individuals with higher endorsement of “just-world” beliefs (8). Among liberals, younger participants, and individuals with lower just-world endorsement, no negative baseline attitudes toward adjacent consent are observed. (A multivariate analysis reveals that age and political ideology each predict baseline judgments independently, but just-world beliefs—which correlate positively with conservativism—do not.)

By contrast, adjacent consent undermines the VVE across a wide range of demographic subgroups, including the youngest and most politically liberal participants, as well as males and females, individuals with and without college degrees, higher- and lower-income individuals, different racial groups, and individuals with higher and lower endorsement of just-world beliefs. Thus, while negative baseline attitudes toward adjacent consent are driven by specific subgroups of individuals, the power of adjacent consent to erode the moral elevation of victims appears more universal. Illustrating these patterns, Fig. 3 plots results by respondent age (Fig. 3A), political ideology (Fig. 3B), and gender (as an example of an individual-difference variable that does not moderate results; Fig. 3C).

Fig. 3.

Adjacent consent robustly undermines the moral elevation of victims—even among subgroups of participants that show no baseline disapproval of sex-adjacent activity. We observe a baseline moral penalty for adjacent consent (within the nonvictim conditions, where no assault occurred) specifically among political conservatives and older individuals. Yet adjacent consent inhibits both conservatives and liberals from morally elevating victims—and this pattern also holds within every age cohort, and across other demographic groups (e.g., gender). We plot perceived character across conditions and age (A), dichotomized political ideology (B), and gender (C). Error bars are 95% CIs.

These results, taken together, provide strong evidence that adjacent consent can impede the moral elevation of sexual assault victims—even in contexts, or among populations, where people hold no inherent moral objection to the victim’s preassault activity. In this way, these results illuminate how even progressive or sex-positive communities may see rape victims as less virtuous when their assaults follow voluntary sexual intimacy.

The Unique Role of Adjacent Consent.

Next, we ask whether it is adjacent consent that is responsible for inhibiting the moral elevation of victims, or whether our results are merely driven by distinct factors associated with adjacent consent. In particular, when sexual assault follows voluntary sexual intimacy, observers might i) see the perpetrator as relatively less abhorrent (e.g., because his assault seems less premeditated, or potentially driven by confusion about the victim’s willingness to have sex), ii) see the victim as sexually interested in the perpetrator (such that his assault seems less unwanted), iii) see the victim as promiscuous, and/or iv) see the victim as having recklessly put herself at risk of being assaulted. Does adjacent consent undermine the VVE merely because of these four factors? We investigate this question across six studies, each employing the party vignette, conducted on MTurk, and described and analyzed in detail in SI Appendix, sections 4.6–4.10. We consistently find evidence that adjacent consent, apart from these factors, plays a unique role in undermining the VVE (Fig. 4).

Fig. 4.

Adjacent consent, apart from associated factors, plays a unique role in undermining the moral elevation of victims. Studies 6 to 11 provide evidence that adjacent consent does not undermine the VVE merely because adjacent consent makes the perpetrator seem less immoral (A), reflects sexual interest in the perpetrator (B), makes the victim seem promiscuous (C and D), involves the assumption of risk (E), or involves the combination of these factors (F). We plot perceived moral character across conditions for each study. Error bars are 95% CIs.

Perpetrator immorality.

In Study 6 (n = 794), we find evidence that adjacent consent does not undermine the VVE merely because it makes the perpetrator seem less immoral. Study 6 modified our basic 2×2 design to contrast our standard Adjacent Consent condition with a modified version of our Complete Nonconsent condition, designed to make the perpetrator seem less immoral for a reason unrelated to adjacent consent. Specifically, in the modified Complete Nonconsent condition, the perpetrator’s transgression was less severe: Rather than forcing sexual intercourse, he forcibly kissed the victim and grabbed her breasts. Participants now rated the perpetrator as slightly more immoral in Adjacent Consent (where the assault was more severe) than in this modified Complete Nonconsent condition (where the assault was less severe). Yet the Complete Nonconsent + Less Severe Assault condition nonetheless elicited a significant VVE (which was marginally significantly larger than the (nonsignificant) VVE in Adjacent Consent; see Fig. 4A). Study 6 thus suggests that adjacent consent, but not anything that makes the perpetrator seem less abhorrent, inhibits the moral elevation of victims.

Sexual interest in the perpetrator.

In Study 7 (n = 1,192), we find evidence that adjacent consent does not undermine the VVE merely because the victim is seen as sexually or romantically interested in the perpetrator (which could make his assault seem less unwanted). Study 7 tested whether uncommunicated sexual interest, in the absence of adjacent consent, undermines the VVE. We added a new condition to our basic 2×2 design, resulting in a 3×2 design. This condition was identical to the standard Complete Nonconsent condition, except that the vignette specified, before Alicia agreed to look at Michael’s movie poster, that she was interested in “hooking up” with him (but did not flirt with him, such that he had “no idea whether she was attracted to him”). This condition did trigger a baseline moral penalty, akin to the baseline penalty observed for Adjacent Consent. But it nonetheless elicited a significant VVE (which was no smaller than the VVE in the standard Complete Nonconsent condition, and significantly larger than the (nonsignificant) VVE in Adjacent Consent; see Fig. 4B)—suggesting that adjacent consent, but not uncommunicated sexual interest, inhibits the elevation of victims.

Perceived promiscuity.

In Studies 8 (n = 1,200) and 9 (n = 1,188), we find evidence that adjacent consent does not undermine the VVE merely because the victim is seen as a promiscuous person. Studies 8 and 9 mirrored the structure of Study 7, but their new conditions sought to increase the target’s perceived promiscuity, in the absence of adjacent consent, by describing her sexual history with men other than the perpetrator. The new conditions were identical to the standard Complete Nonconsent condition, except that they specified that Alicia i) hooked up with a different man on a prior occasion (Study 8), or ii) “regularly hooks up with different guys on campus” (Study 9). These conditions both triggered a baseline moral penalty, but nonetheless elicited significant VVEs (which were no smaller than the VVEs in the standard Complete Nonconsent conditions and significantly larger than the (nonsignificant) VVEs in the Adjacent Consent conditions; see Fig. 4 C and D)—suggesting that adjacent consent granted to the perpetrator, rather than any sexual activity that could be perceived as promiscuous, inhibits the elevation of victims.

Assuming the risk of victimization.

In Study 10 (n = 1,193), we find evidence that adjacent consent does not undermine the VVE merely because the victim is viewed as knowingly increasing her risk of being assaulted. Study 10 mirrored the structure of Studies 7 to 9, but in its new condition, the target puts herself at risk for assault, without providing adjacent consent. This condition was identical to the standard Complete Nonconsent condition, but specified that Alicia chose to attend the party despite knowing that the fraternity had a reputation for throwing parties where sexual assaults occur. This condition did trigger a baseline moral penalty, but nonetheless elicited a significant VVE (which was no smaller than the VVE in the standard Complete Nonconsent condition and marginally significantly larger than the (nonsignificant) VVE in Adjacent Consent; see Fig. 4E)—suggesting that adjacent consent, rather than any voluntary action that increases the risk of assault, inhibits the elevation of victims.

Combining factors.

Studies 6 to 10 each examined one factor associated with adjacent consent, and suggested that none of these factors is individually sufficient to undermine the VVE. But adjacent consent simultaneously i) makes the perpetrator seem less immoral, while also making the victim seem ii) more sexually interested in the perpetrator, iii) more promiscuous, and iv) more reckless. In Study 11 (n = 2,099), we thus examine these factors in conjunction. Study 11 mirrored the structure of Studies 7 to 10, but added two new conditions that built on the standard Complete Nonconsent condition. The first simply added the perpetrator factor: It sought to make the perpetrator seem less immoral in the absence of adjacent consent, by using the Complete Nonconsent + Less Severe Assault condition from Study 6. The second new condition added the perpetrator factor plus the three victim factors: It combined the Complete Nonconsent + Less Severe Assault condition from Study 6 with the uncommunicated sexual interest information from Study 7, sexual history information from Study 8, and assumption of risk information from Study 10.

In this latter condition, where the victim was sexually interested in the perpetrator, promiscuous, and reckless, a baseline moral penalty was observed. Yet both new conditions nonetheless elicited significant VVEs (which were no smaller than the VVE in the standard Complete Nonconsent condition, albeit not significantly larger than the (nonsignificant) VVE in Adjacent Consent; see Fig. 4F). Study 11 thus provides further evidence that adjacent consent does not inhibit the moral elevation of victims merely because it introduces the four factors we have examined.

Exploring the Consequences of Adjacent Consent for Reporting Sexual Assault.

Together, Studies 1 to 11 provide evidence that adjacent consent robustly and uniquely undermines the psychological tendency to morally elevate victims, contributing to a penalty for rape victims who engaged in voluntary sexual intimacy with their perpetrators.

In our final study, we consider a potential real-world consequence of these findings. If victims who have provided adjacent consent anticipate receiving less support (and more shame and blame) from their communities, they may be more reticent to report their assaults. Thus, we hypothesized that adjacent consent might be negatively associated with reporting behavior.

To investigate, we analyzed data from the 2019 Association of American Universities Campus Climate Survey, in which over 180,000 students across 33 U.S. universities were asked about experiences with sexual assault (17). For each incident of sexual assault a respondent described in the survey, respondents were asked whether the perpetrator was someone they “were involved or intimate with at the time”, and whether they reported the incident to any programs at their university. We treat the “involved” measure as a proxy for adjacent consent, because it suggests that the victim may have voluntarily engaged in sexual intimacy with the perpetrator, either during or prior to the incident. Thus, our analyses predict reporting to at least one university program as a function of the involved measure, taking incidents of sexual assault as observations and clustering SE on respondent. Before performing any analyses relevant to this hypothesis, we preregistered an analysis plan, detailed in SI Appendix, section 5.2.

The AAU dataset also includes measures of respondent characteristics (e.g., age, gender) and incident characteristics (e.g., which unwanted physical acts occurred) that constitute potential confounds (given that many characteristics differed across incidents with vs. without adjacent consent). In a model that attempts to account for such confounds by controlling for a preregistered set of 25 respondent- or incident-level covariates, adjacent consent negatively predicts reporting: b = −0.032 [−0.047, −0.018], t = 4.46, P < 0.001, n = 17,551 incidents across 16,697 respondents.

In SI Appendix, section 5.5, we report additional preregistered analyses that i) incorporate a larger set of covariates using a specification curve analysis (37), resulting in a significant negative coefficient for adjacent consent across each of 5,000 model specifications and ii) do not control for covariates, resulting in a positive coefficient for adjacent consent. These analyses reveal that the association between adjacent consent and reporting flips from negative to positive if we do not control for respondent and incident characteristics. In SI Appendix, section 5.6, we report post hoc unregistered analyses exploring which covariates are most responsible for this sign reversal. We find that the most responsible variables are indicators for certain unwanted physical acts (penetration and oral sex, which were more common during incidents involving adjacent consent and positively predict reporting) and for alcohol use before the incident (which was less common in incidents involving adjacent consent and negatively predicts reporting).

Thus, when controlling for respondent and incident characteristics, we find that victims who were involved or intimate with their perpetrator are roughly 3 percentage points less likely to report incidents of sexual assault to university programs—a difference of roughly 25% of the overall rate of reporting, which was 11.85% across the incidents we consider in our primary analyses. This association may in part reflect that such victims anticipate receiving less support and affirmation, as Studies 1 to 11 suggest.

Of course, other mechanisms may also be operative. For example, in cases involving adjacent consent, victims may blame themselves more, diminishing their propensity to report. Additionally, victims who were involved with their perpetrator may have been more motivated to protect them from punishment. With respect to this second possibility, however, in SI Appendix, section 5.5 we provide evidence (from preregistered secondary analyses) that the negative association between adjacent consent and reporting is observed even among respondents who reported never having been in a partnered relationship since enrolling at their university (and thus may have been less motivated to protect their perpetrators in cases involving adjacent consent). Finally, we note that SI Appendix, section 5.5 reports preregistered secondary analyses showing that our results are also robust to analyzing a broader set of incident types (including those involving incapacitation or coercion through threats or promises).

Discussion

We have shown that adjacent consent erodes support for sexual assault victims and have identified a psychological mechanism underlying this effect: Adjacent consent undermines the Virtuous Victim Effect. We find that a rape victim who is blindsided by something completely different from what she was expecting (e.g., sex instead of viewing a movie poster) is seen as a victim in a deep sense that motivates people to affirm her moral character. Yet a rape victim who voluntarily entered the arena of sexual activity with her perpetrator—even as participants provide strong explicit agreement that she was raped, and that her perpetrator deserves severe punishment (including imprisonment)—is viewed as a different kind of victim, one who is less deserving of moral affirmation. Importantly, voluntary sexual intimacy undermines the VVE across a variety of contexts and populations, including in situations where the victim has done nothing, preassault, that garners disapproval.

We have also illuminated a potential real-world consequence of these findings. Examining field data on campus sexual assault, we find evidence that victims who were involved or intimate with their perpetrator are less likely to report their assaults. Importantly, this evidence is correlational, and does not establish a causal effect of adjacent consent on reporting. Moreover, if adjacent consent does reduce reporting, this may reflect mechanisms other than the expectation of less community support (e.g., increased self-blame, or desire to protect the perpetrator). A mechanism grounded in reduced expectations of community support is, however, consistent with the results of our experimental studies—and with the responses of many survey respondents who cited “the event happened in a context that began consensually” as a reason for not reporting (17). Furthermore, reduced expectations of community support might plausibly give rise to other mechanisms: If victims expect less support in cases involving adjacent consent and “internalize” this expectation, they might experience more self-blame, further diminishing their propensity to report.

Our results are striking in light of salient cultural shifts: universities promoting “affirmative consent”; the #MeToo movement demanding greater consequences for sexual misconduct; and growing embrace of the idea that “no means no” (38–40). From these trends, one might expect people, especially liberals and young adults, to see adjacent consent as in no way diminishing an individual’s claim to victimhood—yet our studies suggest otherwise.

Our results do, however, accord with lines of reasoning the law has at times endorsed. Consider a 1985 legal case involving a California victim who provided and then withdrew consent to sex (41). The court determined that this victim had suffered a less “outrageous” affront to her “womanhood” than the prototypical rape victim: “If she withdraws consent during the act of sexual intercourse and the male forcibly continues…the sense of outrage to her person and feelings could hardly be the same magnitude as that resulting from an initial nonconsensual violation” (41). While later cases decried the patriarchal logic of this case, calling it “archaic” (42), the California court is far from alone in voicing the intuition that “[s]exual contact that begins with consent and ends as unwelcome is simply in a different category than rape” (43).

Our findings offer theoretical insights that complicate existing narratives of when and why communities sometimes fail to support victims. A vast psychological literature has highlighted that rape victims—particularly victims of acquaintance rape—are often blamed for their assaults (6, 12). The “most commonly cited” (6) explanation for such victim blaming is the “just-world hypothesis”, which contends that people are driven to believe that bad things happen only to bad people (8, 44). When somebody is victimized, the theory suggests, this drive triggers a motivated search for reasons the victim is deviant, responsible, and deserving of her injury.

Our results challenge this account. We find no evidence that learning that a target was assaulted prompts a motivated search for reasons to morally denigrate her, relative to if she was not victimized. In fact, for targets who provided no consent whatsoever, victimization leads to moral elevation. And for targets who have consented to sex-adjacent activity, victimization triggers neither denigration nor elevation. Instead, our results point to other reasons that people may sometimes respond less supportively to victims (including in cases of acquaintance rape, which frequently involve adjacent consent). Beyond a desire to see the world as just, operative psychological forces may include i) general moral objections to the victim’s consensual sexual activity (which we observe among conservatives and older individuals), and ii) a moral psychology whereby adjacent consent inhibits the tendency to elevate victims (which we observe across a wide swath of respondents).

As they illuminate these psychological forces, our results naturally raise the question of why adjacent consent undermines the moral elevation of victims. Studies 6 to 11 provide evidence that this effect does not merely reflect that adjacent consent makes the perpetrator seem less immoral or makes the victim appear sexually interested in her perpetrator, promiscuous, or to have assumed the risk of victimization: actions other than adjacent consent that introduce these factors do not trigger the same undermining. Yet our results do not provide a positive account of why adjacent consent plays a unique role in inhibiting the VVE.

Prior evidence suggests that the VVE functions to motivate people to help victims and punish perpetrators. The present work thus implies that, in sexual assault cases involving adjacent consent, communities may hold weaker norms for these justice-restorative actions. But why might that be, from an ultimate perspective? Future research should investigate this important open question, which raises the further question of what functions justice-restorative actions serve for communities—and why these functions might be seen as less important or achievable in sexual assault cases involving adjacent consent.

Uncovering the answers to these questions may help shed light on another question our results leave open: What forms of adjacent consent will—and will not—undermine the VVE? For example, would more minimal consensual acts, such as flirting or kissing, be sufficient to inhibit the elevation of rape victims? And for a consensual act to inhibit moral elevation, must it simply be similar to sex? Or might it be important that the act is seen as a steppingstone to sex, such that consenting to the act implies that sex is on the table?

With respect to this latter possibility, Study 4 suggests that adjacent consent can undermine the elevation of rape victims even when the victim tells the perpetrator, before agreeing to go his bedroom, that she does not wish to “go all the way tonight”—an upfront disclaimer that explicitly negates the idea that hooking up puts sex on the table. Yet considering the entrenched trope that a woman who goes to a man’s bedroom for a casual hookup is “asking for it” (36), conventional cultural scripts might serve to override even this explicit disclaimer, preserving observers’ perception that the victim’s preassault conduct implies a potential interest in sex.

Interestingly, the law expressly endorses the notion that cultural norms should shape the interpretation of conduct that extends beyond what was explicitly permitted. Consider a medical patient, for example, who consents to a three-inch incision during surgery, yet while under anesthesia the doctor determines that a five-inch incision is required. Legally, in most jurisdictions, the extra two inches would not violate the patient’s consent because the conduct is “not substantially different” (45) from what the patient agreed to. By contrast, performing surgery on a patient’s left ear after consent to operate on the right ear would violate consent, because the deviation is substantial (46). Tellingly, although the law cannot precisely define what constitutes a substantial deviation, the exercise of drawing this line is informed by “reasonable appearances and prevailing customs” (45). In light of such reasoning, one might expect social conventions to similarly inform lay intuitions surrounding when adjacent consent undermines a victim’s eligibility for moral elevation. Future research should investigate this idea, including the possibility that adjacent consent undermines the elevation of rape victims only when the consensual sex-adjacent activity is by convention understood to imply an interest in sex. And correspondingly, future work should investigate perceptions of adjacent consent in scenarios involving other gender identities (47), and in the eyes of populations from other countries and cultures, where different social scripts may be operative.

Finally, future research should investigate whether adjacent consent can undermine support for victims outside the domain of sexual assault. Consistent with this possibility, research on consent in other domains [e.g., medical procedures, police searches, and the like (48, 49)] has similarly highlighted the surprisingly low thresholds that people can have for seeing consent as morally meaningful—including in contexts where they arguably should not, according to certain normative theories. Indeed, people often perceive valid consent in situations where deception (48) or social pressure (50) would seem to compromise voluntary decision-making, even when they object morally to the way consent was obtained. For example, despite thinking it is wrong for a doctor to lie to a patient about a prospective medical procedure, many people believe that if the patient agrees to the procedure because of the doctor’s misrepresentation, the procedure is validly consented to (and, correspondingly, does not constitute medical battery) (48). In light of such findings—which suggest that people may assign moral meaning to seemingly minimal forms of consent in many domains—future work should explore whether and when adjacent consent might inhibit the elevation of victims of nonsexual transgressions.

In sum, we have illuminated an important factor that can erode support for victims of sexual assault: adjacent consent. When a woman voluntarily agrees to sex-adjacent activity, it can undermine the moral affirmation she will receive if assaulted—even if the assault is objected to both verbally and physically, and accomplished only through forcible overpowering. Ample evidence suggests that rape victims are often reticent to report their assaults, in part because they fear not receiving support from their communities (51). Our research illuminates why, in cases involving adjacent consent, that fear may not be unfounded.

Materials and Methods

Experimental Studies.

In each of Studies 1 to 11, approved by the Harvard University IRB (Protocol #: IRB21-0625) and the University of Michigan IRB (Protocol #: HUM00182995), participants provided informed consent, answered questions about a vignette, and completed demographic and survey questions. These studies were each preregistered. We adhered strictly to preregistered sample sizes, and closely to preregistered analysis plans; see SI Appendix, section 2 for a discussion of some deviations. Table 3 provides a design overview of each study. See SI Appendix, sections 1 and 6 for full design details, including information about dependent measures not analyzed in the main text, and SI Appendix, sections 3 and 4 for supplementary analyses that draw on these measures.

Table 3.

Overview of experimental studies

				Sample Demographics
Experiment	Vignette	Conditions (between subjects)	Dependent measures analyzed in main text	Mean age (years)	Gender composition
1 (n = 997, Mturk)	Party	Complete Nonconsent vs. Adjacent Consent X Victim vs. Nonvictim	Moral character, Blame, Sympathy	39.63	55% female, 45% male
2 (n = 803, Mturk)	Apartment		Moral character	41.48	54% female, 44% male, 2% other genders
3 (n = 792, Mturk)	Apartment + Temporally distant adjacent consent			41.71	48% female, 51% male, 2% other genders
4 (n = 1193, Mturk)	Party + Disclaimer			40.44	54% female, 45% male, 1% other genders
5 (n = 806, Undergraduates from UMichigan + Prolific)	Party			20.96	60% female, 39% male, 1% other genders
6 (n = 794, Mturk)		Adjacent Consent vs. Complete Nonconsent + Less severe assault X Victim vs. Nonvictim		42.57	55% female, 44% male, 1% other genders
7 (n = 1,192, Mturk)		Complete Nonconsent vs. Adjacent Consent vs. Complete Nonconsent + Sexual interest X Victim vs. Nonvictim	Moral character, Blame, Helping, Punishment, Rape attribution, Consent attribution	42.58	49% female, 50% male, 1% other genders
8 (n = 1,200, Mturk)		Complete Nonconsent vs. Adjacent Consent vs. Complete Nonconsent + Sexual history X Victim vs. Nonvictim		41.90	52% female, 47% male, 1% other genders
9 (n = 1,188, Mturk)				41.35	49% female, 49% male, 1% other genders
10 (n = 1,193, Mturk)		Complete Nonconsent vs. Adjacent Consent vs. Complete Nonconsent + Assumption of risk X Victim vs. Nonvictim		42.18	56% female, 43% male, 1% other genders
11 (n = 2,099, Mturk)		Complete Nonconsent vs. Adjacent Consent vs. Complete Nonconsent + Less severe assault vs. Complete Nonconsent + Less severe assault + Sexual interest + Sexual History + Assumption of risk X Victim vs. Nonvictim	Moral character, Blame, Sympathy, Helping, Punishment	43.93	54% female, 44% male, 1% other genders

Analyses of Field Data About the Reporting of Sexual Assault.

These analyses, approved by the Harvard University IRB (Protocol #: IRB23-0065), drew on the 2019 Association of American Universities (AAU) Campus Climate Survey. We applied for access to the data (URL to apply: https://www.icpsr.umich.edu/web/ICPSR/studies/37662), merged the incident-level dataset with items from the respondent-level dataset, and then preregistered analyses predicting the reporting of incidents as a function of our proxy for adjacent consent and covariates. See SI Appendix, section 5 for more details.

Supplementary Material

Appendix 01 (PDF)

Data, Materials, and Software Availability

Experimental data have been deposited in OSF (https://osf.io/ncr9k/) (52). Field data cannot be shared. Our analyses of sexual assault reporting data draw on the 2019 Association of American Universities (AAU) Climate Survey on Sexual Assault and Sexual Misconduct. The survey explored student experiences of sexual assault, sexual misconduct, and the campus climate surrounding sexual assault and misconduct. Gaining access to the data required obtaining a data-use agreement and IRB approval. Because the dataset is protected and requires an application to access, we are unable to make the raw data underlying our AAU analyses publicly available. However, we have shared survey instruments and variable descriptions that the AAU has made publicly available, as well as our analysis scripts. Additionally, we have provided in our main text and SI Appendix a URL to the website where one can apply for the data.