Comparing Attitudes toward Sexual Consent between Japan and Canada

Abstract

Japanese and Canadian laws regarding sexual assault vary in the degree to which they incorporate the concept of sexual consent, with Japanese law being less consent-oriented than Canadian law. Although the Japanese law has incorporated the concept of sexual consent in the 2023 amendment, the public understanding of the concept is still limited. Reflecting such difference, it could be expected that the general public in both countries also differ in their perceptions and attitudes regarding punishment of sexual crime and sexual consent. The present study aimed to test these expectations and further examine the mediational mechanism that explains the national difference between Japan and Canada. The data from 1125 Japanese and 1125 Canadian respondents showed that Japanese respondents were less likely to perceive the imposition of punishment on an alleged perpetrator described in scenarios as appropriate. In contrast, the difference in the perceived victim’s consent was significant only in three out of seven scenarios. The relation between nations (Japan vs. Canada) and perceived appropriateness of punishment was mediated by the perceived victim’s consent.

1. Introduction

Social and academic interest in sexual consent has continued to grow in recent years. Not only Western societies such as the U.S., U.K., Sweden, and Germany, but also non-Western societies such as Korea and Taiwan have introduced legislation that criminalizes nonconsensual sex. Despite this increased emphasis on consent, the extent to which each society’s legal system criminalizes and punishes nonconsensual sex varies widely [1].

Previous research has argued that public attitudes toward crime often shape the criminal justice climate and culture of the legal system [2], and the law, in turn, shapes public attitudes by conveying a particular message to the public [3]. Given this interaction between law and public attitudes, societies with different degrees of the criminalization of nonconsensual sex are likely to have different public attitudes.

In addition, sexual behaviors are deeply cultural phenomena, shaped by the interaction of various forces such as social, cultural, and institutional forces. Emphasizing this, Gagnon and Simon [4] have argued in their sexual script theory that sexual behaviors are not innate but learned in the process of psychological development. Following this theory, it is expected that people in different societies with different institutions are likely to have different attitudes toward sex and sexuality.

However, the vast majority of studies on sexual consent have been conducted in Western societies, including Canada [5,6,7,8,9], with those conducted in non-Western societies, including Japan [10,11], remaining elusive. Comparative studies regarding sexual crime are even rarer, with very few exceptions [12,13,14].

This situation is problematic because it would lead to an overgeneralization of findings from Western societies to other non-Western societies. Underscoring this concern, researchers in the field of psychology have emphasized that findings in this area are biased toward those found in WEIRD (Western, Educated, Industrialized, Rich, and Democratic) societies, and have suggested a reorganization of research [15].

In light of these, the present study aims to fill this research gap by comparing two contrasting societies: Japan and Canada. The reason for choosing these two countries as comparative targets is twofold. First, Japan and Canada belong to Eastern and Western societies, respectively, so comparing these societies would help alleviate the problem of a narrow focus on Western societies. Second, as detailed below, sexual violence and consent have been the subject of widespread interest in each society. These two reasons only justify the inclusion of these two societies, not the exclusion of other societies with similar characteristics. Nevertheless, given the paucity of comparative research in this area of study, setting too high a standard for selection would unnecessarily hinder the development of research [16].

This paper attempts to answer two main research questions: (1) do Japanese people differ from Canadians in their perceptions of the punishment of sexual crimes and sexual consent? And (2) assuming that there are national differences in perceptions, are these differences mediated by perceived consent (i.e., the perception that the victim consented to the sexual conduct)?

1.1. Difference in Legal Context between Japan and Canada

1.1.1. Japan

Although Japan is often praised for its low general and sexual crime rates, its law on sexual assault remained unchanged from its enactment in 1907 until 2004. However, in response to a series of high-profile sexual crimes, the 2004 revision increased the minimum sentence for sexual assault from three to five years and introduced several new types of sexual crime. A more substantial reform was introduced in 2017 and 2023.

The 2017 amendment replaced the old crime of “rape” and its subtype with the newly established crime types of “forcible sexual intercourse” and its subtype (referred to in Japanese law as “constructive forcible sexual intercourse”), in which not only penile–vaginal penetration, but also penile penetration into the mouth and anus are equally punishable (Japanese Penal Code, Article 177). In addition, while the crime of “rape” required a formal complaint by the victim to initiate and maintain litigation, the new crime type eliminated this requirement.

Although these reforms were primarily intended to improve victim protection by criminalizing acts other than vaginal penetration and reducing the burden on the victim to file a formal complaint, critics continue to argue that the law does not provide adequate protection for victims [17]. Their central criticism concerns sexual consent. In particular, the “forcible sexual intercourse” and its subtype require sexual intercourse to be committed by “assault or intimidation” (Article 177) or “taking advantage of their unconscious state or inability to resist, or by causing the person to lose consciousness or have the inability to resist” (Article 178(2)). However, the absence of “assault or intimidation” or “inability to resist” does not mean the presence of consent. It has therefore been argued that the absence of the victim’s consent, not the presence of “assault or intimidation” or “inability to resist”, must be the standard for imposing punishment [17].

In response to criticism from scholars and victim advocates, the 2023 amendment took a step forward by introducing the concept of sexual consent into the law. The committee organized within the Ministry of Justice examined the current legal situation and finally published a reform proposal in February 2023. This proposal lists eight acts under which the crime of “nonconsensual sex” is to be established, including “assault and intimidation”, “ingestion of alcohol or drugs”, and “making the victim worry about the disadvantages to which he or she is subjected by virtue of his or her influence based on his or her position in economic or social relations”. Then, it recommends criminalizing sexual intercourse committed under the circumstances in which any of the eight acts are committed, thereby “making it difficult for the victim to form, express, or fulfill the intention not to consent”, or by taking advantage of such a situation [18]. This reform proposal was submitted to and approved by the Cabinet in March 2023, enacted in June, and promulgated in July [19,20].

Despite the legislation on sexual consent, public understanding of the concept remains limited. In a recent survey conducted by the Japanese Organization for International Cooperation in Family Planning in October 2023, 45.3% of men and 39.8% of women and non-binary individuals responded negatively when asked if they truly comprehended sexual consent [21].

1.1.2. Canada

Over the past fifty years, Canada has experienced a rapid evolution of its legislative framework regarding sexual assault. These changes have resulted in, among other things, the redefinition of the concept of consent, the introduction of new restrictions on the use of a victim’s sexual history as a defense, and the end to the exclusion of marital rape [22].

In 1982, the House of Commons passed Bill C-127 to amend the Criminal Code with respect to sexual offenses and other offenses against the person. This bill gave birth in 1983 to the Canadian Sexual Assault Act. The offenses of rape, attempted rape, and indecent assault were replaced in the Criminal Code by the offenses of sexual assault. In addition, under this Act, a spouse can be accused of sexual assault against the other spouse, whereas there was previously impunity for the “rape” committed between spouses. Both women and men can be recognized as victims.

According to the Criminal Code, the offense of sexual assault is an assault defined as willful touching of a sexual nature committed by a person who knows that the victim did not consent to the touching or who is willfully blind or reckless in this regard. It has three degrees of seriousness: sexual assault, level 1 (Canadian Criminal Code, RSC 1985 c C-46, s. 271), sexual assault with a weapon, level 2 (s. 272), and aggravated sexual assault, level 3 (s. 273).

In 1992, the notion of consent to sexual activity was clarified to ensure fairer treatment of victims and in 1999, it was decided by the Supreme Court of Canada that consent to sexual activity must be voluntary and communicated, thereby not allowing for the existence of implied consent to a sexual assault [23].

Section 265 (3) (RSC 1985 c C-46) of the Canadian Criminal Code says that there is no consent if the plaintiff had to submit or was unable to resist owing to (a) the application of force to the complainant or to a person other than the complainant; (b) threats or fear of the application of force to the complainant or to a person other than the complainant; (c) fraud; or (d) the exercise of authority.

The definition of consent to sexual activity is set out in section 273.1, and more specifically, situations that do not constitute consent in law in subsection 273.1 (2) of the Canadian Criminal Code (RSC 1985 c C-46). Consent is defined as the voluntary agreement of the complainant to engage in the sexual activity in question. Conduct short of a voluntary agreement to engage in sexual activity does not constitute consent as a matter of law (273.1 (1)). The subsection 273.1 (2) sets out specific situations where there is no consent in law; consent is not obtained where “the agreement is expressed by the words or conduct of a person other than the complainant; where the complainant is incapable of consenting to the activity; where the accused induces the complainant to engage in the activity by abusing a position of trust, power or authority; the complainant expresses, by words or conduct, a lack of agreement to engage in the activity, or where the complainant, having consented to engage in sexual activity, expresses, by words or conduct, a lack of agreement to continue to engage in the activity’’ [24].

1.1.3. Law and Social Norms

Previous studies have argued that there is an interactive relationship between the law and public attitudes. For example, Lappi-Seppälä [2] has presented a model that assumes that the penal climate is shaped by public opinion, as well as by political, economic, and criminal situations and the practices of the mass media and politicians. Conversely, Friedman [3] has argued that laws influence public attitudes by conveying a particular message to the public. For example, a law that criminalizes burglary and imposes a penalty for the act sends a message to the public not to burglarize. Consistent with these arguments, several studies have shown international differences in attitudes and judgments about punishment and crime [25,26,27,28,29].

Given this interactive nature, we expect that the general public in Japan and Canada, which contrasted sharply in the extent to which they criminalize nonconsensual sex, differ in the extent to which they perceive the imposition of punishment for a given act of nonconsensual sex to be appropriate.

1.2. Mediational Mechanism

Considering the sharp contrast between the legal systems in Japan and Canada, it seems somewhat obvious that the general public in both countries holds different perceptions. Therefore, the present study focuses primarily on the mediation mechanism by which the national difference in the perceived appropriateness of punishment by country is mediated and explained by a consent-related variable: the perceived victim’s consent. The reason we primarily address the perceived appropriateness of punishment by making it the dependent variable is that the imposition of punishment is the ultimate goal of all criminal justice proceedings and that it has serious consequences for both the defendant and the victim.

In addition, we focused on the consent-related variable as a mediator because of the importance of the concept of sexual consent in both countries. As described above, Canadian law requires the absence of consent to establish sexual assault, thereby giving consent a critical role in legal adjudication. In Japan, although the law has incorporated the concept of sexual consent in the 2023 amendment, public understanding of the concept is still limited [21]. Given this, it is expected that people’s perceptions of victim’s consent are related to their perceptions of punishment.

1.3. Hypotheses

Based on the above arguments, we aimed to test three hypotheses, two regarding national differences in perceptions of punishment and consent, and one regarding the mediating mechanism. Specifically, we hypothesized that Japanese respondents, compared to their Canadian counterparts, would

• H1: be more likely to perceive the imposition of punishment as appropriate.
• H2: perceive the level of victim’s consent as higher.
• In addition, we hypothesized that the relationship between country and perceived appropriateness of punishment would be mediated by perceived victim’s consent (H3).

2. Materials and Methods

2.1. Participants and Procedures

We conducted a web-based survey in both Japan and Canada. Respondents in both countries were recruited by the same company. A web-based survey was developed by the authors and posted on the company’s websites. Potential respondents who saw this posting and were willing to participate were asked to read the ethical terms presented on the first page of the questionnaire and then proceeded to complete the questionnaire. Upon the completion of the questionnaire, respondents received web points from the company. Data were collected in September and October 2021, prior to the 2023 amendment that incorporated the concept of sexual consent.

To maximize representativeness, the Canadian sample was first allocated according to the Canadian age and gender distribution based on the census, and the Japanese sample was allocated according to the Canadian sample. This strategy was used because the number of potential respondents in Japan was much larger, and therefore the Japanese sample was easier to collect in a nuanced manner. Since the age range of potential Canadian respondents was limited to between 20 and 69 years, people in this age group were targeted.

Although it is generally necessary to estimate the statistical power to determine an adequate sample size [30], the present study is one of the first to make an international comparison of differences in attitudes and perceptions regarding sexual consent. Given this unique feature, it seemed more appropriate to collect data that seemed sufficient to represent a national sample, rather than just a sample sufficient to detect a predetermined effect size. Therefore, we collected as many respondents as our budget allowed. Consequently, data from 1125 Japanese people (569 women, 556 men, M_age = 44.2, SD_age = 14.1) and 1125 Canadian people (569 women, 556 men, M_age = 44.1, SD_age = 14.3) were used in the analysis.

2.2. Development of Scenarios

The scenarios used in the present study were mainly adopted from a previous study [31], which presented more than 100 scenarios related to sexual assault and others. After translating the scenarios from English to Japanese and removing some scenarios that were not directly related to sexual assault (e.g., ones related to consent in civil law), two pilot studies were conducted with Japanese respondents. In the first pilot study, respondents (n = 17) were presented with a pool of 84 scenarios and asked to rate the perceived appropriateness of the punishment. They were also asked to identify scenarios that were difficult to understand or unrealistic. After the survey, forty-eight items that were listed as “difficult to understand” or unrealistic by at least one respondent were removed (e.g., a scenario describing the situation in which a woman has a serious illness and needs a transplant, and a man offers her an organ in exchange for sex). In addition, three items with a mean of five (i.e., those answered with “should be punished” by all 17 respondents) were dropped because they did not appear to be useful for examining individual differences. After these exclusions, three scenarios that conceptually overlapped with others were also dropped. In the second pilot study (n = 289), respondents were presented with the remaining 30 scenarios and asked to rate the perceived appropriateness of the punishment. We then referred to the scores of each variable, and selected items with as different scores as possible and with the least conceptual overlap with each other. Eventually, nine items were finally selected.

2.3. Variables

2.3.1. Perceived Appropriateness of Punishment

After reading the scenarios, respondents were instructed to “Please rate the extent to which you think A (male) should be punished in each situation.” They were then asked to rate their opinion on a 5-point Likert scale ranging from Should not be punished at all (1) to Should absolutely be punished (5). All English items and their corresponding Japanese and French items are shown in Table S1 in Supplementary Materials.

2.3.2. Perceived Victim’s Consent

Respondents were instructed to “Please rate the extent to which you think B (female) has consented to sexual intercourse.” Similarly, a 5-point Likert scale ranging from Not consented at all (1) to Completely consented (5) was used to rate their opinion. All English items and their corresponding Japanese and French items are shown in Table S2 in Supplementary Materials.

2.3.3. Demographic Variables

The respondent’s gender (man/woman) and age were asked for.

2.4. Translation

A back-translation technique [32] was used to ensure linguistic equivalence between Japanese, English, and French. Since the scenarios were originally developed in English, the scenarios were first translated into Japanese by native Japanese speakers and then back-translated into English by a native English speaker, and these two versions were compared and checked for linguistic invariance, i.e., whether items in different languages give respondents similar impressions.

2.5. Analytical Strategy

First, sample demographics and descriptive statistics were calculated to obtain an overview of the data. Then, exploratory factor analyses (EFAs) were conducted and measures were constructed to measure each variable. In an international comparison, simply constructing a composite variable and testing for difference is not appropriate due to the complexities caused by the translation of the measure [32] and the meanings imposed by languages [33]. With this in mind, we used an invariance testing methodology. This involved constructing several models with different constraints and comparing their differences in goodness-of-fit indices to reveal similarities and differences across groups. A conventional threshold of CFI = 0.900, RMSEA = 0.100 [34], and ΔCFI < 0.010 [33] was used as criteria for the fit indices and their comparison. Finally, a mediation analysis with multiple mediators was conducted to test the hypothesized model. In this analysis, instead of relying on the “causal steps approach” [35], the “product of coefficients approach” [36] was used to allow for a mediation analysis with multiple mediators. To test for mediation effects, the bootstrap method was used to estimate 95% confidence intervals (5000 samplings). If the 95% confidence intervals did not include 0, the indirect effect was considered significant. This analysis was performed using the package lavaan ver. 0.6.12 [37] with R ver. 4.2.1.

3. Results

3.1. Sample Characteristics and Descriptive Statistics

Before conducting a factor analysis and invariance tests, we first calculated the sample statistics of the samples. As shown in

Table 1. Sample characteristics.

Variable	Japan (n = 1125)		Canada (n = 1125)
	n	%	n	%
Age
20s	228	20.3%	228	20.3%
30s	241	21.4%	241	21.4%
40s	215	19.1%	215	19.1%
50s	232	20.6%	232	20.6%
60s	209	18.6%	209	18.6%
Gender
Woman	569	50.6%	569	50.6%
Man	556	49.4%	556	49.4%
Other	0	0.0%	0	0.0%
Language
Japanese	1000	100.0%	0	0.0%
English	0	0.0%	1019	90.6%
French	0	0.0%	106	9.4%

, the sample distributions of gender and age were virtually identical.

3.2. Exploratory Factor Analysis

Since the variables used in the present study had not been tested for their factor structure, EFAs were conducted separately in Japan and Canada using the maximum likelihood method. Regarding the variables related to scenarios, based on the eigenvalues of perceived appropriateness of punishment (Japan: 3.68, 1.81, 0.72, 0.64, 0.58...; Canada: 3.59, 1.88, 0.75, 0.75, 0.57, 0.54...) and perceived sexual consent (Japan: 4.06, 1.74, 0.80, 0.57, 0.46...; Canada: 4.23, 1.71, 0.74, 0.56, 0.44...) and the ease of interpretation, one-factor models were adopted. Two items with factor loadings below the conventional threshold of <0.40 (Matsunaga, 2010) were excluded. The final factor structures with the remaining seven scenarios for both variables are presented in

Table 2. Factor loadings of items measuring perceived appropriateness of punishment and perceived victim’s consent.

		Punishment		Consent
		Japan	Canada	Japan	Canada
pun4 con4	A proposed intercourse; B accepted on condition that A used a condom. A said “OK” and then penetrated B without a condom. B told A to withdraw. A refused until he ejaculated.	0.79	0.74	0.64	0.73
pun1 con1	A overpowered B physically, held her down despite B’s attempts to resist, and threatened her with additional force if she continued to resist.	0.76	0.62	0.82	0.76
pun6 con6	A and B had dated, but B had rebuffed A’s advances. On this occasion, B was severely intoxicated, although she had not passed out. In response to A’s advances, B said nothing when A removed her clothes, but was clearly too weak to resist.	0.71	0.73	0.79	0.79
pun5 con5	A, a supervisor at a large corporation, told B that he would engineer an undeserved promotion for which B had applied if she had sexual relations with him.	0.70	0.70	0.62	0.67
pun8 con8	A and B commenced what they thought would be a normal sexual interaction. B unexpectedly found intercourse painful on this occasion, and asked A to stop. A did not withdraw until he ejaculated.	0.57	0.66	0.59	0.67
pun7 con7	A was infatuated with B, but B had rebuffed him. A told B that he would commit suicide unless B had sexual relations with him.	0.54	0.48	0.75	0.67
pun3 con3	A, whose identical twin was married to B, slipped into B’s bed while she was half asleep. B believed that A was her husband.	0.54	0.53	0.68	0.70

Notes: Maximum likelihood method.

. All Cronbach’s αs were sufficiently high (αs > 0.82), indicating the reliability of the measures.

3.3. Invariance Testing

Based on these factor structures, we then proceeded to test the invariance of perceived punishment appropriateness and perceived victim’s consent. The preliminary estimation showed that the fit indices (CFI = 0.843, RMSEA = 0.114) were not acceptable. Therefore, we referred to modification indices, which suggested that the introduction of covariates between the following pairs would improve the fits: pun1–pun5, pun5–pun6, con1–con5, and con4–con6. Since these pairs of items tapped into similar content, we introduced the covariates between these items. After the introduction, the fits were improved to reach the cutoffs (CFI = 0.905, RMSEA = 0.091). Therefore, we adopted this as the baseline model (M1: configural invariance model). In addition, we constructed and estimated the models’ constraining loadings (M2: metric invariance model) and loadings and intercepts (M3: scalar invariance model) to be equal across countries. The results (

Table 3. Fit indices for measuring invariance of perceived appropriateness of punishment between Japan and Canada.

	Fit Indices					Comparative Fit Indices
	χ2	df	CFI	RMSEA	BIC	Target	ΔCFI	ΔRMSEA	ΔBIC
M1	1483.378	144	0.905	0.091	80,025.695
M2	1560.954	156	0.900	0.089	80,010.646	M1	−0.005	−0.002	−15.049
M3a	1747.191	168	0.888	0.091	80,104.259	M2	−0.012	0.002	93.613
M3b	1687.440	167	0.892	0.090	80,052.227	M2	−0.008	0.001	41.581

M1: configural invariance model; M2: metric invariance model; M3a: scalar invariance model; M3b: partial scalar invariance model (pun5 is freely estimated). Bold values indicate those with the final model.

) showed that ΔCFI (ΔCFI = 0.012) slightly exceeded the cutoff (ΔCFI < 0.010) between M2 and M3a. Therefore, the partial scalar invariance model was constructed. Specifically, we referred to the result of the univariate score tests. Since it showed that the largest χ² was with pun5, the model without equal constraint was constructed for this item (M3b). After this modification, the ΔCFI between M2 and M3b was 0.003. Therefore, M3b was adopted as the final model.

The relationships between perceived punishment appropriateness and perceived victim’s consent were −0.345 (p < 0.001) in Japan and −0.321 in Canada (p < 0.001). We also examined the difference in the latent mean (

Table 4. Latent mean difference between Japan and Canada.

	Coefficient	S.E.	z	p	Raw Mean (SD)
					Japan	Canada
Perceived appropriateness of punishment	−0.243	0.032	−7.542	<0.001 **	3.73 (0.69)	3.97 (0.71)
Perceived victim’s consent	0.079	0.040	1.997	0.046 *	2.21 (0.76)	2.18 (0.90)

Notes: Japan = 0, Canada = 1. ** p < 0.01, * p < 0.05.

). This revealed that Japanese respondents were more likely to perceive the punishment as less appropriate (z = −7.542, p < 0.001) and to perceive the victim as more consenting (z = 1.997, p = 0.046).

3.4. Mediation Analysis

Based on the model identified above, we conducted a mediation analysis using the model presented in Figure 1 to examine H3. The model controlled for respondent gender and age. Although the CFI of the model was slightly below the cutoff (CFI = 0.885), this was due to the constraint that the loadings and intercepts are equal, and the invariance test showed that this model was acceptable considering the differences in the fits of the models; we adopted this as the final model. Consistent with the above analysis, Japanese respondents were more likely to perceive the victim as consenting (B = −0.079, 95%CI [−0.154, −0.005]), and this perception reduced the perceived appropriateness of the punishment (B = −0.411, 95%CI [−0.473, −0.353]). Detailed results are presented in Table S3 in Supplementary Materials.

We then estimated the indirect effect using the bootstrap method (resampling: 5000). This showed that the indirect effect of country on the perceived appropriateness of punishment via perceived victim’s consent was significant (B = −0.091, 95%CI [−0.115, −0.068]), meaning that the national difference in perceived appropriateness of punishment is partially explained by perceived victim’s consent.

4. Discussion

4.1. Overview of the Results

Reflecting social and legal changes, empirical and theoretical studies of sexual consent are growing rapidly. However, most studies have been conducted in Western societies, and those conducted in other societies remain rare. Aiming to reduce this imbalance, the present study tested the invariance of perceived punishment appropriateness and victim’s consent. The results revealed interesting similarities and differences between the two countries. In terms of similarities, configural and metric invariances were confirmed, meaning that the structure and factor loadings of the items measuring perceptions do not differ. In contrast, scalar invariance was not confirmed, mainly due to pun5, which measures withdrawal of consent. In addition, there were significant differences in both perceptions, suggesting that Japanese respondents were generally less likely to perceive punishment as appropriate for the acts described in the scenarios and more likely to perceive the victim as having consented to the sexual acts. Moreover, the mediation analysis indicated that the national difference in perceived punishment appropriateness was partially explained by perceived victim’s consent.

4.2. Contribution of the Present Study

The contribution of the present study is twofold. Not only did it show that there are interesting international differences in judgments of sexual consent, but it also identified a mediating mechanism that explains these differences. Regarding the first contribution, in recognition of the research imbalance in the current literature, the present study focused on two countries, one in the West and one in the East. Previous non-empirical studies have argued that there is an interaction effect between the law and public attitudes [2,3]. Consistent with this argument, some studies found international differences in judgments about punishment and crime (e.g., [25]). However, international comparisons focusing specifically on sexual crimes remain rare. In this context, the present study contributes to the existing literature by showing that public judgments differ between two countries whose legal provisions on sexual assault differ significantly in terms of the scope of criminalization and the timing of criminalization.

With respect to the latter contribution, the present study found that perceived victim’s consent partially explains the international difference in perceived appropriateness of punishment. Since lack of consent is a prerequisite for establishing a crime in Canada and now in Japan, the finding that the degree of perceived appropriateness of punishment depends on perceived victim’s consent may seem somewhat obvious. However, the contribution of the previous study is that it showed that the national difference in perceived appropriateness of punishment is partially explained by perceived victim’s consent. Existing studies of rape in Japan have indicated that misconceptions about consent are common in Japan [10,38]. Finding that consent plays a role in determining punishment in such a society would shed a different light on sexual violence in Japan.

The result of the invariance testing that the caused scalar variance was pun5 also provides insight. This item tapped into a withdrawal of consent in which a victim once consented to sex but then withdrew her consent. This item had the largest difference in perceived appropriateness of punishment between societies (d = 0.42), suggesting that Japanese respondents (M = 3.36) were much less likely to perceive this act as punishable than Canadian respondents (M = 3.78). Thus, while many Japanese agree that sex without “consent” is criminal, their understanding of the concept may not be the same as that of Canadians (or Westerners in general), where consent is always revocable.

4.3. Limitations

Although the present study provides some insightful results, it also suffers from several limitations. First, while the present study controlled for respondent age and gender in our mediation model, no other variables were controlled. As discussed above, sexual behavior is a deeply cultural phenomenon [4]. Therefore, other demographic variables, such as ethnicity and respondents’ prior sexual experience, may be related to attitudes about sexuality. In addition, because sexual violence is, at least formally, a highly blameworthy act that often receives a punitive response [39], respondents may have given a more socially desirable response than they implicitly believe. Possible effects of other demographic variables and the present focus on explicit attitudes need to be considered when interpreting the present results.

The second limitation concerns sampling. The present study used a web-based survey, and the Japanese respondents were matched with their Canadian counterparts. While this matching resulted in an equal distribution of age and gender, it made the Japanese sample less representative than the Canadian sample. In addition, since the respondents were those who registered with the web survey company, the samples may have different characteristics from the general population. Taking these into account, the representativeness of the present samples, especially the Japanese sample, and the replicability of the present findings need to be investigated in future studies.

4.4. Future Research Orientation

Based on the present findings, several avenues for future research can be envisaged. First, the variables that contribute to the international differences in perceived appropriateness of punishment need to be further explored. While the present study showed that perceived victim’s consent partially mediated and explained the international difference, the explained variance was R² = 0.323, suggesting that approximately 77% of the variance was not explained by our variable. One promising variable may be exposure to sexual content. Several studies have shown that media exposure or use is associated with attitudes and behaviors related to sexual consent and violence [40,41,42,43]. In particular, the ubiquity of sexual content, which is sometimes cited as a uniqueness of Japanese society, at least compared to Western societies [44], may play a role in shaping attitudes and judgments. Moreover, although the present study did not include questions about respondents’ sexual history or gender identity, it is possible that these variables might contribute to the formation of attitudes. Future research exploring such possibilities is warranted.

Second, a longitudinal analysis based on the present study is a promising direction. As mentioned in the introduction, Japanese law on sexual assault is in a state of flux. In addition, social movements such as the #MeToo movement and the so-called “flower demo” in Japan, which were widespread at the time of data collection in 2021, may have influenced people’s attitudes [45]. Moreover, people in societies may have different attitudes than in 2021, when freedom of movement and interaction were restricted and anxieties were intensified due to COVID-19. In such a circumstance, collecting panel data before and after reforms and how public attitudes and judgments change as legal provisions are revised is highly relevant to the literature in the sense that it provides more detailed evidence for the causal relationship between the law and public attitudes.