Protecting the Innocence of Youth: Moral Sanctity Values Underlie Censorship From Young Children

Abstract

Three studies examined the relationship between people’s moral values (drawing on moral foundations theory) and their willingness to censor immoral acts from children. Results revealed that diverse moral values did not predict censorship judgments. It was not the case that participants who valued loyalty and authority, respectively, sought to censor depictions of disloyal and disobedient acts. Rather, censorship intentions were predicted by a single moral value—sanctity. The more people valued sanctity, the more willing they were to censor from children, regardless of the types of violations depicted (impurity, disloyalty, disobedience, etc.). Furthermore, people who valued sanctity objected to indecent exposure only to apparently innocent and pure children—those who were relatively young and who had not been previously exposed to immoral acts. These data suggest that sanctity, purity, and the preservation of innocence underlie intentions to censor from young children.

Keywords

censorship moral protection sanctity purity moral values

Society goes to great lengths to protect children, often shielding them from certain ideas and images. For example, parents limit the types of movies and television shows their children watch. The film rating system of the Motion Picture Association of America facilitates this. By attending to movie ratings, which typically warn of depictions of substance abuse, sex, and violence, parents can limit their children’s exposure to adult content. The present work examines the moral values underlying these special protections for children, specifically censorship of objectionable media.

We examined the moral values that underlie intentions to censor from children. We posit that caring for children is uniquely linked to protecting them from impurities (e.g., Curtis & Biran, 2001; Fessler, Eng, & Navarrete, 2005) and, relatedly, that children are seen as pure and innocent (Gino & Desai, 2012; James, Jenks, & Prout, 1998; Woodrow, 1999). Furthermore, we hypothesize that people who are especially sensitive to purity, those who value the moral concern of sanctity (Graham et al., 2011), are most willing to censor objectionable media from children.

Moral Values

Moral foundations theory (MFT; Graham et al., 2013; Graham, Haidt, & Nosek, 2009; Graham et al., 2011; Haidt & Graham, 2007) proposes that people’s moral values can be explained via five moral foundations: care, which praises empathy and condemns harm; fairness, which praises equality and condemns inequity; loyalty, which praises self-sacrifice for the group and condemns opposing one’s own group; authority, which praises respect of hierarchy and condemns rebelliousness; and sanctity, which praises purity and condemns indulgence in impure or disgusting acts. People’s moral values are, thus, defined by the weights they give to the five moral domains.

MFT helps explain meaningful individual differences in morality. Politically, liberals prioritize the care and fairness foundations, whereas conservatives give roughly equal weight to all five foundations (Graham et al., 2009; Graham et al., 2011; Haidt & Graham, 2007; Van Leeuwen & Park, 2009). MFT has also helped explain the link between lay beliefs and political affiliation (Van Leeuwen & Park, 2009), attitudes toward public policy issues (e.g., global warming; Dawson & Tyson, 2012), and proclivity for sports fandom (Winegard & Deaner, 2010).

MFT, thus, offers an established framework for assessing the values underlying intentions to censor. We tested whether endorsement of certain moral values as outlined by MFT is linked to censorship from young children.

Protection of Youthful Purity and Innocence

We sought to illuminate the moral values that underlie censorship from children, a form of moral protection. We propose that this protection is rooted in a fundamental motivation to care for and protect young children, particularly from pathogens. Children are perceived as physically pure and are vulnerable to illness, and so they demand physical protection. We propose furthermore than children are perceived as morally pure and vulnerable, thus demanding moral protection. Both motivations (i.e., to protect children from physical impurities and to protect children from moral impurities) are driven by a perception of children as pure and innocent (Gino & Desai, 2012; James et al., 1998; Woodrow, 1999), and we propose that media censorship is one manifestation of the desire to protect that.

Physical Purity, Innocence, and Protection

The notion that children are pure and need protecting may have evolutionary origins. One theory proposes a universal parenting motive that, when triggered, causes shifts in cognition and behavior that promote nurturing and caring for children (Kenrick, Griskevicius, Neuberg, & Schaller, 2010; Kenrick, Neuberg, Griskevicius, Becker, & Schaller, 2010). When people (including nonparents) view images of cute stimuli, thereby trigging parenting motives, they become more careful and vigilant in behavioral and cognitive tasks (Nittono, Fukushima, Yano, & Moriya, 2012; Sherman, Haidt, & Coan, 2009). Thus, parenting motives may trigger a cautiousness in thought and action that aids in protecting children from harm.

Crucially, the parenting motive seems also to include protection from pathogens, due to children being especially vulnerable to disease (Simon, Hollander, & McMichael, 2015). Indeed, women demonstrate heightened disgust sensitivity during pregnancy, which helps a mother to avoid pathogens, thereby protecting the vulnerable fetus (Fessler et al., 2005). Women also exhibit relatively high disgust sensitivity in general (e.g., Haidt, McCauley, & Rozin, 1994), and some have argued that this is driven by the need for women, who are often the primary child rearers, to protect their vulnerable children from pathogens (Curtis & Biran, 2001). Historical accounts corroborate a link between concerns for cleanliness and protection of the young. Namely, hygiene in the United States appears to have become moralized due to a desire to protect the health of young children (Sivulka, 2001). Along similar lines, Oaten, Stevenson, and Case (2009) described the moralization of disgust “as a means of managing disease-related threats, most notably relating to the care and management of those vulnerable to disease (typically infants and children)” (p. 315).

In short, parenting motives may promote not only protecting children from bodily injury but also keeping them from disgust-inducing sources of illness. We suspect that this integral disease-avoidance component separates protection of young children from other types of protection (e.g., protection of adult family members or friends).

Moral Purity, Innocence, and Protection

The desire to protect children from physical impurities may have in turn promoted a desire to protect them from moral impurities. Indeed, concerns about moral purity and sanctity are theorized as having origins in the avoidance of physical contamination (Haidt & Graham, 2007). Thus, purity concerns and experiences of disgust may have served initially to help avoid sources of disease. Over time, however, disgust evolved into a social emotion, linked to preserving purity of body, mind, and spirit. Similarly, disease protection mechanisms aimed at protecting children physically may have ultimately become linked to protecting children socially and morally.

Numerous findings suggest that parenting motives are linked to protection from social and moral impurities. During pregnancy, mothers experience both heightened disgust sensitivity (Fessler et al., 2005) and heightened ethnocentrism (Navarrete, Fessler, & Eng, 2007). Similarly, when parents’ caretaker roles are made salient, they become less trusting of strangers (Eibach & Mock, 2011). Thus, parenting motives trigger greater avoidance of potential contaminants in the social domain (e.g., foreigners).

Parenting concerns have also been linked to heightened concerns about moral behavior, purity behaviors in particular. When parents are reminded of their roles as caretakers, they more harshly judge moral purity transgressions (e.g., watching and becoming aroused by copulating animals; Eibach, Libby, & Ehrlinger, 2009). Furthermore, people reporting greater parental caretaking tendencies tend also to report greater condemnation of moral purity violations, such as incest or eating one’s dog (Buckels et al., 2015). Thus, when the motivation to care for children is highly active, condemnation of morally impure acts increases.

The above work suggests that people want to protect children from moral impurities. Evidence also suggests that people perceive children as morally pure and innocent (Gino & Desai, 2012; James et al., 1998; Woodrow, 1999). Moral purity refers to cleanliness, chastity, and innocence from wrongdoing (Gino & Desai, 2012; Haidt & Graham, 2007), and those who are perceived as morally pure are seen as nonthreatening and as needing protection and care (Goff, Jackson, Di Leone, Culotta, & DiTomasso, 2014). Thus, when people describe children as “pure” and “innocent,” they are communicating a judgment that children are morally clean and require protection. Given the apparent link between moral purity and protection of children, we hypothesized that moral purity values are uniquely linked to the desire to protect children from objectionable media.

The Sanctity–Censorship Hypothesis

We propose that the preservation of purity is central to the moral protections people extend toward children, including censorship. Parenting concerns trigger greater aversiveness to physical and social contaminants (Eibach & Mock, 2011; Fessler et al., 2005; Navarrete et al., 2007), perhaps to facilitate protecting children physically, socially, and morally. Moreover, parenting concerns have been uniquely linked in the moral domain to concerns about moral sanctity and purity (Buckels et al., 2015; Eibach et al., 2009). Therefore, we hypothesized that of the five moral foundations, endorsement of moral sanctity would be the value most linked to censorship from young children, because people who value sanctity should be most sensitive to violations of sanctity, namely, the degradation of innocent and pure children. Our sanctity–censorship hypothesis, thus, posits the following:

Hypothesis 1: The more people endorse moral sanctity, the more motivated they will be to censor objectionable content from children.

The sanctity–censorship hypothesis led us to an additional prediction that we tested. If people who care about censorship are motivated by a desire to protect youthful purity from contamination, then that concern should be mitigated toward youth who have lost that purity. In formulating this prediction, we drew on Rozin and Nemeroff (2002) who outlined the common patterns by which people perceive pure individuals to become contaminated, whether dealing with contamination by physical disease or moral degradation. They proposed that contamination is permanent and cannot be reversed, and that it is dose insensitive, so that contact with even the smallest amount of contaminant is sufficient to corrupt the whole. Therefore, we predicted that people who value sanctity would be highly protective of innocent children, but much less protective of children who have had any prior exposure to moral corruption and who have, thus, had their purity irreversibly degraded.

We, therefore, predicted that the sanctity–censorship link would be strongest for protection of innocent children and weakest or nonexistent for children whose innocence had already been tainted. We operationalized tainted innocence in two ways. First, we used age as a proxy for loss of innocence, given that children are perceived as less innocent and pure as they age (e.g., Goff et al., 2014). Thus, we predicted that the sanctity–censorship link would be weak when observing protection of older children (e.g., teenagers), who are no longer perceived as innocent. Second, we manipulated loss of innocence through prior exposure to immoral acts. We predicted that the sanctity–censorship link would be weakened when the children being protected have had prior exposure to impure or harmful acts. If a child’s innocence has already been tainted through exposure to moral corruption, then those who value moral sanctity should cease caring whether the child is exposed to further immoral acts (i.e., weakening the sanctity–censorship link).

Alternatives to the Sanctity–Censorship Hypothesis

We considered other, plausible links between moral values and censorship. Care values could be linked to censorship, insofar as censorship involves protection from psychological harm. Authority values could also be linked to censorship, insofar as censorship involves an authority figure (e.g., a parent) dictating which media children may consume. However, there are reasons to doubt these alternatives. Exposure to objectionable media may have harmful downstream effects (Anderson & Bushman, 2001; Bushman & Anderson, 2001), but such exposure on its own does not seem akin to the type of direct harm protected against by care values. Slapping a child is a prototypical act of harm (Graham et al., 2009); allowing a child to play violent video games seems more like an act of indulgence. Moreover, if a parent allows a child to consume objectionable media, the child is not less obedient. Thus, we suspected that censorship would be linked more to sanctity than to other moral values.

One seemingly viable alternative to the sanctity–censorship hypothesis is the personal values hypothesis, which posits the following:

Hypothesis 2: People censor acts that they themselves consider to be immoral.

Thus, when people decide to censor from children, they impose whatever constellation of moral concerns they value: People who value fairness censor depictions of unfair acts, people who value loyalty censor depictions of disloyal acts, and so forth. This is a plausible pattern, as endorsement of the various moral foundations ought to correlate with other measures of the same values (Graham et al., 2013; Graham et al., 2011). In addition, because children learn by observing others (e.g., Bandura, 1977), people may censor violations of important personal values to keep children from learning to engage in similar transgressions. People may, thus, censor transgressions of the moral values they endorse, essentially reflecting adherence to individual moral codes.

Although the personal values hypothesis would be consistent with prior work, we, nevertheless, favored the sanctity–censorship hypothesis. The personal values hypothesis assumes that the major motivation behind censorship is communication of moral values, and so it fails to capture a seemingly central driver of censorship—the desire to protect children. In the present work, we tested for evidence of both the sanctity hypothesis and the personal values hypothesis.

The Present Work

Across three studies, we tested whether endorsement of moral sanctity or of other moral values underlies intentions to censor. In each study, we measured participants’ endorsement of the five moral foundations and we measured attitudes toward censorship of diverse kinds of content. Our approach was to describe depictions of moral transgressions that spanned the five moral domains and then to measure attitudes regarding censoring those depictions from children.

Our assumption was that people censor depictions specifically of moral transgressions. Furthermore, we contend that the majority of acts that are censored from children are ones that are viewed at some level to be moral violations. In the present work, therefore, participants contemplated censoring depictions of a wide range of moral violations.

To be clear, our aim was not to test which types of moral violations people want to censor most. Such a test would require a wholly different approach than the one we used, because it would require taking diverse transgressions of equal magnitude and testing which ones people want to censor most. Instead, we assessed transgressions validated in previous work as conforming to the various moral foundations (e.g., Graham et al., 2009), which as an added benefit we considered to be typical of those depicted in movies and television, and these varied widely in severity. These transgressions included the killing of dozens of people (a relatively severe care violation) and the use of illicit drugs (a relatively mild sanctity violation). Given the range of severity across domains (e.g., care violations seemed the most severe), we expected that the willingness to censor would differ significantly across domains in our studies. However, our main interest was not in such mean-level differences in censorship between domains; instead, our main interest was in the moral values that underlie censorship across types of moral violations. Again, our hypothesis was that the more people valued sanctity, the more they would be motivated to censor diverse types of moral transgressions.

We tested for evidence of the sanctity–censorship hypothesis and the alternative personal values hypothesis across three studies. We tested whether sanctity values or other moral values are linked to intentions to censor various immoral acts (Study 1). We tested whether sanctity values predict censorship attitudes specifically when young (rather than adolescent or adult) children are involved and only for instances of parents (not the children themselves) causing exposure to objectionable content (Study 2). We also tested whether the relationship between moral sanctity values and censorship attitudes would decrease once children have had prior exposure to immoral acts (Study 3).

Study 1

Study 1 served as an initial test of the sanctity–censorship and personal values hypotheses. We developed and administered a measure of likelihood to censor from children, which required participants to indicate how likely they would be to censor a variety of movie scenes from a young child. Each movie scene was described as depicting a violation of one of the five moral foundations (e.g., Graham et al., 2009; Graham et al., 2011), with each foundation being represented in four different movie scenes. We drew the description of each movie scene (with slight modifications) from items used by Graham et al. (2009) to measure endorsement of the five moral domains, with the added exception that the wording was altered to include a protagonist in a movie scene (see Online Appendix A). This allowed us to test whether censorship intentions varied as a function of the types of moral violations being censored.

We measured each participant’s endorsement of the five moral domains to test for links between moral values and censorship intentions. We also explored whether other individual differences explained censorship intentions. We measured political orientation, given the link between conservatism and strict parenting styles (e.g., Altemeyer, 1981). We also measured age, sex, and parental status, as participants who are older, female, or parents might be more nurturing and protective of young children.

Method

Participants were 129 Amazon Mechanical Turk users in the United States (52% female; M_age = 36.4 years, SD_age = 11.3 years) who participated online for pay. Our main prediction was that endorsement of sanctity values alone would predict censorship intentions while controlling for endorsement of the other four moral foundations. An a priori power analysis indicated detection of such a relationship at .85 power, assuming a moderate effect size (f² = .15), would require 102 participants. We included additional participants to account for a conservatively high estimate of participant exclusion due to failed attention checks (34% for the target population; see Study 2 in Goodman, Cryder, & Cheema, 2013).

After providing informed consent, participants reported their censorship attitudes by indicating how likely they would be to censor each of 20 movie scenes from their own (hypothetical) 8-year-old child on a 1 (not at all likely) to 7 (very likely) scale (see Online Appendix A). Moral values were then assessed with the two subscales of the Moral Foundations Questionnaire (MFQ): the Moral Relevance Questionnaire (MRQ) and the Moral Judgments Questionnaire (MJQ; Graham et al., 2011). Although our measure of censorship attitudes was built from previous research on the MFQ (Graham et al., 2009), it is important to note that the two measures are conceptually distinct. Our measure of censorship attitudes asked participants to report whether they would show certain movie scenes to their child without reference to moral values, whereas the MFQ asks whether certain actions are important to morality and whether participants agree with certain moral judgments.

Both the MRQ and MJQ assess endorsement of the five moral domains with three items per domain, and both contain attention checks. We considered participants who failed at least one of the attention checks to be insufficiently attentive and were removed from analyses. Last, participants reported their political orientation (1 = very liberal, 7 = very conservative), age, sex (0 = male, 1 = female), and parental status (0 = not a parent, 1 = a parent).

Results

Fifteen participants were excluded for failing the attention checks, leaving 114 participants in the final analyses. We created censorship scores for each moral domain by averaging responses within that domain and a total censorship score by averaging responses to all 20 items. All censorship scores were reliable (α_care = .82, α_fairness = .88, α_loyalty = .71, α_authority = .83, α_sanctity = .77, α_total = .94). Participants’ moral values were calculated from their MRQ and MJQ responses. MRQ scores were consistently reliable (α_care = .70, α_fairness = .77, α_loyalty = .81, α_authority = .73, α_sanctity = .67) but MJQ scores were not (α_care = .55, α_fairness = .35, α_loyalty = .44, α_authority = .70, α_sanctity = .85). We, therefore, used only the MRQ data to represent participants’ endorsement of the five moral foundations.

Descriptive statistics

Mean censorship scores for each domain are displayed in Table 1. We did not test for differences between the various censorship scores, as we were interested in the individual differences (i.e., moral values) that underlie censorship intentions and not in whether certain categories of moral violations are censored more than others. Furthermore, it would be difficult to interpret those comparisons given that we did not standardize the severity of the violations across domains.

Table 1.

Mean Level and Standard Deviation of Censorship Attitudes.

Study	Condition	Domain of violations being depicted
Study	Condition	Care	Fairness	Loyalty	Authority	Sanctity	Total
Study 1		5.36 (1.47)	3.25 (1.78)	2.36 (1.23)	3.20 (1.66)	4.44 (1.44)	3.72 (1.32)
Study 2	Parental exposure-8	6.04 (1.09)	4.10 (1.44)	2.64 (1.13)	3.59 (1.39)	4.45 (1.10)	4.16 (1.01)
	Parental exposure-16	5.03 (1.55)	3.64 (1.65)	2.35 (1.15)	2.90 (1.46)	3.12 (1.20)	3.40 (1.18)
	Parental exposure-25	3.59 (2.10)	2.65 (1.53)	2.04 (1.02)	1.99 (1.17)	2.14 (1.21)	2.46 (1.23)
	Self-exposure	6.04 (1.07)	4.77 (1.27)	2.76 (0.92)	3.63 (1.13)	4.07 (1.37)	4.25 (0.87)
	Self-expression	6.21 (0.91)	5.12 (1.15)	3.00 (1.17)	3.74 (1.43)	4.52 (1.23)	4.52 (0.93)
Study 3	Control	5.97 (1.21)	4.34 (1.72)	2.54 (1.01)	3.48 (1.49)	4.45 (1.42)	4.15 (1.17)
	City control	5.91 (1.23)	4.31 (1.47)	2.66 (1.10)	3.55 (1.39)	4.45 (1.29)	4.17 (1.05)
	Impure exposure	5.62 (1.26)	3.81 (1.43)	2.47 (1.01)	3.22 (1.32)	3.99 (1.40)	3.83 (0.98)
	Harmful exposure	5.73 (1.25)	4.15 (1.57)	2.56 (1.15)	3.30 (1.47)	4.11 (1.15)	3.97 (1.09)

Note. Standard deviations are in parentheses. Items were answered on a 1 to 7 scale, with higher numbers in Study 1 indicating greater intentions to censor and higher numbers in Studies 2 and 3 indicating stronger negative reactions to the child seeing the violations.

Nevertheless, the mean levels of censorship in Table 1 are helpful insofar as they demonstrate that people were willing to censor in every domain. In addition, the variance in reported likelihood to censor was roughly equal across the domains. Table 2 displays the correlations between the different censorship domains, which were all significant, rs > .55, ps < .001.

Table 2.

Correlations Between Censorship of Different Domains From Study 1.

	Domain of violations being censored
	Care	Fairness	Loyalty	Authority	Sanctity
Care	—	.67	.55	.67	.76
Fairness		—	.74	.76	.69
Loyalty			—	.79	.61
Authority				—	.75
Sanctity					—

Note. All correlations p < .001.

Main analyses

To test whether certain moral values predicted willingness to censor, we performed separate multiple regression analyses for each of the five censorship domains and for total censorship (i.e., intentions to censor averaged across the five domains). The predictors in each regression model were participants’ moral values (i.e., their five MRQ scores). Results for each analysis are in Table 3.

Table 3.

Willingness to Censor as a Function or Moral Values in Study 1.

Moral values	Domain of violations being censored
Moral values	Care	Fairness	Loyalty	Authority	Sanctity	Total
Care	−.03	−.08	−.08	−.14	−.07	−.10
Fairness	.11	.01	−.08	−.02	−.06	.01
Loyalty	−.03	.30*	.22	.17	.13	.19
Authority	.06	.03	.23*	.22*	.11	.14
Sanctity	.47***	.27*	.23*	.35**	.29*	.37**

Note. Values are βs predicting willingness to censor. Each column represents a multiple regression analysis.

p < .05. **p < .01. ***p < .001.

Consistent with the sanctity–censorship hypothesis, endorsement of moral sanctity was linked significantly to willingness to censor in each of the five censorship domains and for total censorship. The more participants valued sanctity, the higher their intentions to censor, regardless of the types of acts being depicted. In contrast, only two censorship domains supported the personal values hypothesis: authority and sanctity, and the latter relationship could alternatively be explained by the sanctity hypothesis. The personal values hypothesis, thus, received little support.

Political orientation, parenthood, and other demographics

We analyzed the correlations between each demographic variable and overall willingness to censor. Political orientation was significantly correlated with overall willingness to censor, with more conservative political values being linked to greater willingness to censor, r = .35, p < .001. Age, r = .08, p = .42; sex, r = .04, p = .65; or parental status, r = .12, p = .19, were not significantly correlated with overall willingness to censor.

We then added the demographics variables as predictors in the same regression model of total censorship (i.e., containing the five moral values) used in the previous section. Results revealed that political orientation no longer predicted willingness to censor, β = .12, p = .25. Willingness to censor was also still not linked to age, β = –.04, p = .68, sex, β = .08, p = .33, or parental status, β = –.07, p = .42. However, sanctity values remained the strongest and only significant predictor of total censorship, β = .36, p = .003.

We suspected that political orientation correlated with willingness to censor due to the tendency for political conservatives to endorse sanctity values more so than political liberals (Graham et al., 2009). To test whether sanctity values or a different moral value mediated the relationship between political orientation and overall willingness to censor, we utilized the bootstrapping technique to measure multiple mediational pathways (Preacher & Hayes, 2008). Using 2,000 bootstrap samples, the bias corrected confidence intervals (CI) were calculated for each of the five moral foundations in mediating the relationship between political orientation and total censorship intentions. As expected, only sanctity values significantly mediated the relationship between political orientation and overall willingness to censor, 95% CI = [.05, .23].

Discussion

Sanctity values emerged as the strongest indicator of a person’s intentions to censor from a child, regardless of the types of violations being depicted. Political conservatives were more willing to censor than political liberals, and this relationship was fully mediated by conservatives’ greater endorsement of sanctity values (e.g., Graham et al., 2009; Helzer & Pizarro, 2011; Inbar, Pizarro, & Bloom, 2009). Although the sanctity–censorship hypothesis received strong support, there was little support for the personal values hypothesis. These findings suggest people censor not to communicate their personal moral views but instead due to moral sanctity concerns.

Study 2

Study 2 had three aims. First, we tested for replication of the main finding from Study 1 such that sanctity values predicted censorship intentions. There were two minor changes in that regard. The first involved switching the censorship questionnaire from first- to third-person perspective: Participants indicated how much it would bother them if someone else decided to show a child each of various movie scenes. The second involved measuring moral values before measuring censorship attitudes. In Study 1, censorship attitudes were measured first, and so it is possible that people who endorsed censorship felt obligated afterward to report valuing moral sanctity. By counterbalancing across studies, we could test whether there is a deeper link between sanctity values and censorship attitudes beyond such an ordering effect.

The second aim of Study 2 was to test whether the sanctity–censorship link generalized to protection of children of older ages. Across three conditions, we varied the age of the child described in the censorship questionnaire to be 8, 16, or 25 years old. Given the link between purity and youth, we predicted that sanctity concerns would be more predictive of attitudes regarding censorship from a young child than from an adolescent or adult.

The third aim was to test whether people’s strength of sanctity values predicted attitudes beyond prototypical censorship situations involving both a parent and child. We suspected that, in accordance with the sanctity hypothesis, people censor to avoid actively tarnishing children’s purity. Hence, when adults expose young children to inappropriate content, those who care about sanctity will object. In contrast, if children are exposed to immoral ideas through no fault of a parent or anyone else, or if young children produce immoral ideas on their own, people who value sanctity may not be similarly bothered. In these cases, there is no intentional purity-violating behavior on the part of a parent that is to be condemned (e.g., Cushman, 2008).

To test this idea, three versions of the censorship questionnaire were administered. In three parental exposure conditions (one for each of the three age conditions), the censorship questionnaire was similar to that in Study 1, with a hypothetical parent exposing a child to immoral acts. In a self-exposure condition, an 8-year-old child is described as having watched a movie scene depicting an immoral act albeit through no fault of a parent or anyone else. In a self-expression condition, an 8-year-old child is described as having written a story (absent any outside inspiration) depicting immoral acts. We hypothesized that sanctity values would predict how bothered participants would be in the parental exposure condition more so than in the self-exposure or self-expression conditions, in which no parent is actively corrupting the child’s innocence.

Method

Study 2 included 371 American participants: 164 undergraduates (60.5% female; M_age = 19.0 years, SD_age = 0.95 years) enrolled in an introductory psychology course who participated for course credit and 207 (53.6% female; M_age = 32.4 years, SD_age = 12.3 years) users of Amazon Mechanical Turk who participated online for pay. The study comprised a five-condition (parental exposure-8 vs. parental exposure-16 vs. parental exposure-25 vs. self-exposure vs. self-expression) between-subjects design. There were approximately 74 participants in each condition, which allowed for replication of the sanctity–censorship link from Study 1 (i.e., in the parental exposure-8 condition) at greater than 90% power.

Participants’ moral values were measured using the MRQ and MJQ, which contained the same attention checks as in Study 1. Participants completed the same censorship measure from Study 1, but the instructions varied by condition (see Online Appendix B). In the three parental exposure conditions, participants imagined that a parent intentionally showed each of various movie scenes to their child, who was described as 8, 16, or 25 years old. In the self-exposure condition, participants imagined an 8-year-old child had watched one of various movie scenes albeit through no fault of a parent or anyone else. In the self-expression condition, participants imagined an 8-year-old child wrote a story depicting one of various scenes, even though the child was not exposed to anything that would have inspired such a story. For each scene that the child watched or wrote, participants indicated how bothered they would be (1 = not bothered at all, 7 = very bothered). Last, participants completed measures for another line of work and a demographics questionnaire, including their political orientation (1 = very liberal, 7 = very conservative), age, sex, and, for online participants only, their parental status (0 = not a parent, 1 = a parent).

Results

Twenty-eight participants failed the attention checks, leaving 343 participants in the final analyses. Censorship and moral value scores were computed as in Study 1. Censorship scores were again reliable (α_care = .91, α_fairness = .86, α_loyalty = .64, α_authority = .81, α_sanctity = .77, α_total = .94). Moral value scores were again consistently reliable when using MRQ scores (α_care = .60, α_fairness = .70, α_loyalty = .67, α_authority = .58, α_sanctity = .72) but not MJQ scores (α_care = .30, α_fairness = .41, α_loyalty = .53, α_authority = .58, α_sanctity = .77). We, therefore, used MRQ scores to represent participants’ moral values. Moreover, because our sample was drawn from two populations, all analyses controlled for population main effects (undergraduates vs. online participants) and all two-way interactions involving population.¹

Main effects of experimental condition

Although peripheral to our hypotheses, initial analyses examined whether censorship scores varied across experimental conditions, independent of sanctity values or any other moral values. We conducted a one-way ANOVA on total censorship with condition as the independent variable, with planned contrasts comparing the parental exposure-8 condition with all other conditions and comparing the parental exposure-16 condition with the parental exposure-25 condition. As seen in Table 1, total censorship scores indicated that participants were significantly more bothered when the child was 8 years old than 16 years old, F(1, 337) = 17.21, p < .001, η²_p = .05, or 25 years old, F(1, 337) = 92.87, p < .001, η²_p = .22. In addition, participants were significantly more bothered when the child was 16 years old than 25 years old, F(1, 337) = 28.45, p < .001, η²_p = .08. When looking at the role of parental exposure, there was no significant difference between the parental exposure-8 and the self-exposure conditions, F(1, 337) = 0.32, p = .57, η²_p = .001; however, participants were significantly more bothered in the self-expression condition than in the parental exposure-8 condition, F(1, 337) = 4.01, p = .046, η²_p = .01. In summary, people were more bothered as the child was younger or had independently expressed the objectionable acts.

Conceptual replication of Study 1

We tested for replication of Study 1 by focusing only on participants in the parental exposure-8 condition, which consisted of 67 participants across the two samples. Separate regression analyses were conducted for each of the five censorship domains and for total censorship. Participants’ endorsement of the five moral domains served as the predictors. Again, consistent with the sanctity hypothesis, we found a significant positive relationship between sanctity values and attitudes toward censorship in all five moral domains and for total censorship (see Table 4). The data were less supportive of the personal values hypothesis. Only the care and sanctity domains supported the personal values hypothesis, and the latter link could be explained by the sanctity hypothesis.

Table 4.

Strength of Objections to Exposure as a Function of Moral Values in Study 2.

Condition	Moral value	Domain of depicted violation
Condition	Moral value	Care	Fairness	Loyalty	Authority	Sanctity	Total
Parental exposure-8	Care	.30*	.00	.02	−.05	.13	.08
	Fairness	−.22	−.03	−.08	−.20	−.23	−.18
	Loyalty	−.05	.07	.18	.04	−.04	.05
	Authority	.20	.07	.10	.27	.08	.18
	Sanctity	.37*	.36*	.47**	.45**	.55**	.53***
Parental exposure-16	Care	.29*	.27	−.01	.10	−.03	.17
	Fairness	.04	−.15	−.02	−.18	.14	−.05
	Loyalty	.15	.25	.25	.28	.00	.23
	Authority	−.11	.08	.18	.08	.11	.07
	Sanctity	.30	.20	.30	.46**	.30	.37*
Parental exposure-25	Care	−.03	−.06	−.23	−.06	.00	−.08
	Fairness	.09	.10	.03	−.01	−.05	.05
	Loyalty	.13	.07	.35*	.44*	.45*	.29
	Authority	.27	.36	.25	.00	.06	.23
	Sanctity	−.13	−.20	−.04	.07	−.10	−.10
Self-exposure	Care	.10	.11	.11	.17	.01	.13
	Fairness	.28*	.34*	−.14	−.12	.07	.13
	Loyalty	−.08	−.08	−.04	.14	−.22	−.09
	Authority	.40*	.23	.34*	.33*	.40*	.44**
	Sanctity	−.02	.04	.19	.17	.25	.16
Self-expression	Care	.27	.12	.15	−.04	−.05	.10
	Fairness	.01	−.03	−.18	−.11	−.08	−.11
	Loyalty	.07	.03	.24	.16	.13	.17
	Authority	.09	.26	.25	.29*	.22	.30*
	Sanctity	.18	.14	.12	.42**	.44**	.34*

Note. Values are βs predicting how bothered participants reported they would be, controlling for sample and sample by moral domain interactions. Each column within a condition represents a multiple regression analysis.

p < .05. **p < .01. ***p < .001.

Furthermore, as in Study 1, we tested the role of demographic factors on total censorship. Political conservatism, r(65) = .26, p = .03, and being older, r(65) = .25, p = .04, significantly correlated with total censorship, whereas sex did not, r(65) = .03, p = .81. In addition, for the 38 online participants in the parental exposure-8 condition, being a parent did not significantly correlate with total censorship, r(36) = .21, p = .20. If we included undergraduate participants under the assumption that none of them was a parent, the correlation between being a parent and total censorship was still nonsignificant, r(65) = .16, p = .21. However, given the low number of parents and small sample size, the nonsignificant effect for parenthood might be due to low statistical power. When we entered all demographic variables into the full regression model used above, only sanctity values, β = .46, p = .006, emerged as a significant predictor (all other ps > .08).

Following the procedure detailed in Study 1, we tested whether sanctity values or a different moral value mediated the relationships between either political orientation or age and total censorship. As expected, only sanctity values significantly mediated the relationships between political orientation, 95% CI = [.02, .24], and age, 95% CI = [.01, .04], with total censorship.

Age of the child

We tested whether the sanctity–censorship link is dependent on the age of the child being protected. We conducted the analysis described above but for participants in the adolescent and adult age conditions. In the parental exposure-16 condition, the sanctity–censorship link was significant only in one censorship domain (authority; see Table 4). Thus, the parental exposure-16 condition revealed weak evidence of a sanctity–censorship link. Meanwhile, in the parental exposure-25 condition, sanctity concerns were not linked to censorship attitudes for any of the five domains (see Table 4).

Further analyses tested whether the sanctity–censorship link differed between the age conditions. Given that the link between sanctity values and censorship attitudes were relatively constant across censorship domains, we simplified these analyses by collapsing the five censorship scores and focusing only on the total censorship scores.

A multiple regression analysis predicted total censorship scores as a function of age condition (parental exposure-8 vs. parental exposure-16 vs. parental exposure-25), participants’ five moral value scores, and the two-way interactions between age condition and moral value scores. Age condition was entered with two dummy-coded variables that compared each of the 16- and 25-year-old conditions with the 8-year-old condition. The results revealed that sanctity values interacted significantly with age condition when comparing the 8- and 25-year-old conditions, β = –.28, p = .007, but not when comparing the 8- and 16-year-old conditions, β = –.07, p = .50. Thus, the sanctity–censorship link differed significantly between the 8- and 25-year-old conditions but not between the 8- and 16-year-old conditions. As the child in question went from being 8 to 25 years in age, the strength of the sanctity–censorship link dropped significantly.

Role of parental exposure

We tested whether people who value moral sanctity object specifically when an adult exposes a young child to immoral acts, focusing on the parental exposure-8, self-exposure, and self-expression conditions. The analyses mirrored those done for the three age conditions. We tested for evidence of the sanctity–censorship link in the self-exposure and self-expression conditions, and then we compared these two conditions with the parental exposure-8 condition.

In the self-exposure condition, sanctity concerns did not significantly predict censorship attitudes in any of the domains (see Table 4). Thus, sanctity values do not predict negative attitudes toward cases in which children exposed themselves to objectionable content. However, authority concerns did significantly predict negative attitudes toward self-exposure in four of the censorship domains, a relationship we discuss further below.

In the self-expression condition, the link to sanctity values was significant in only two censorship domains (authority, sanctity; see also Table 4). There was, thus, little evidence that sanctity concerns were consistently linked to negative attitudes toward independent expression of immoral ideas in a young child.

We also tested for differences between the parental exposure-8, self-exposure, and self-expression conditions. Sanctity values interacted significantly with condition when comparing the parental exposure-8 condition against the self-exposure condition, β = –.34, p = .02, but not when comparing the parental exposure-8 condition against the self-expression condition, β = –.20, p = .13. Thus, the sanctity–censorship link was significantly stronger in the parental exposure-8 group than in the self-exposure group, but, although there was a trend such that the sanctity–censorship link appeared stronger in the parental exposure-8 than in the self-expression condition, this difference was not statistically significant.

Discussion

Once again, sanctity concerns were strongly linked to censorship attitudes. The more people endorsed moral sanctity, the more they were bothered by a parent showing an 8-year-old child objectionable movie scenes, regardless of the types of immoral acts being depicted.

Furthermore, sanctity concerns were relevant specifically when a young child was involved. In contrast, sanctity was only weakly linked to censorship attitudes when the protected child was 16 years old and not at all when the child was 25 years old. Indeed, the sanctity–censorship link was significantly stronger in the case of the 8-year-old child than it was for the 25-year-old child. Sanctity concerns are not linked to the sharing of objectionable media to everyone but specifically to the sharing of objectionable media with young children.

Sanctity concerns were also linked specifically to the prototypical censorship situation wherein parents choose whether to expose depictions of immoral acts to young children. When parents were described as choosing to expose a young child to objectionable media, people who valued moral sanctity objected. In contrast, sanctity concerns were not linked to objections when a young child watched objectionable movie scenes on his or her own. Indeed, the sanctity–censorship link was significantly stronger when exposure was at the hand of a parent than when it was due to self-exposure on the part of the child. Unexpectedly, authority concerns emerged as the main predictor of objections in the case of self-exposure, perhaps because the child who gained access to adult media was seen as undermining the parents’ authority. Moreover, when a young child independently expressed objectionable ideas, sanctity concerns were inconsistently linked to participants’ objections, predicting censorship attitudes in only two of the five domains. Thus, sanctity concerns seem linked most specifically to the notion that adults should not intentionally expose young children to objectionable media.

Ignoring the role that sanctity values play, participants were most bothered when the immoral acts were exposed to young children than to older children. In addition, participants were most bothered when a child self-expressed the immoral acts than when exposed to those same acts. This difference between exposure and expression suggests that people are most upset when a child is the architect of immoral acts than the observer of immoral acts. One possible explanation for this difference is that people may assume a more immoral character about the child when he creates the immoral content versus just being exposed to it. This is consistent with evidence that even thoughts and mental states are moralized and subject to judgment (e.g., Cohen, 2003; Cohen & Rozin, 2001).

Study 3

In Study 3, we tested whether people who value sanctity would no longer wish to censor from a child if that child has had prior exposure to immoral acts. If people who value sanctity censor to protect the purity and innocence of youth, then they may no longer censor for youth whose purity has already been tainted. Study 2 provides partial support for this hypothesis. Essentially, the age manipulation in Study 2 served as a proxy for loss of innocence. Adolescent and adult children have experienced more than young children and are, therefore, seen as less pure and innocent. And, as we observed, people who valued sanctity did not reliably express greater desires than others to censor from adolescent and adult children. Of course, as children age, many psychological changes take place beyond the loss of innocence (e.g., improved agency and ability to cope with stress). Therefore, in Study 3, we used descriptions of a child’s prior experience, rather than the child’s age, to manipulate that child’s innocence.

We hypothesized that people who value sanctity would endorse censorship from a young child, unless the young child’s innocence had previously been corrupted by exposure to immoral acts. We anticipated that this might be true only for certain kinds of immoral acts. Specifically, it might be the case that exposure to violations of moral sanctity (e.g., acts of sexual indecency) are especially potent in terms of tainting the innocence of youth. However, based on the findings from Studies 1 and 2 that showed that the strength of sanctity values predicts desire to censor immoral acts across all moral domains, not simply those in the sanctity domain, we hypothesized that exposure to moral sanctity violations and other violations could render a young child less innocent. Therefore, in the present study, we examined in one condition whether prior exposure to sanctity violations diminished the sanctity–censorship link and in another condition whether prior exposure to nonpurity violations (i.e., violations of moral care) also diminished the sanctity–censorship link.

Participants read about one of four children. In a control condition, participants simply imagined an 8-year-old child with no additional details about that child’s past. In an impure exposure condition, participants read about a child who grew up in an urban setting and had repeatedly witnessed sanctity violations, like drug-use and prostitution. Similarly, participants in a harmful exposure condition read about a child who grew up in an urban setting and had repeatedly witnessed care violations, like fights and other violent acts. We also included a fourth condition, a city control condition, in which an 8-year-old child was described as having grown up in an urban setting, but with no mention of exposure to immoral acts. We suspected that exposure to immoral acts, and not having grown up in an urban setting, would reduce the innocence of youth. The city control condition allowed us to test that hypothesis.

In sum, we anticipated that when a child has been repeatedly exposed to moral violations of any type, whether care or sanctity violations, people will perceive that child as being less pure and innocent. Thereby, we predicted that sanctity values would play less of a role in censorship from such children compared with children who had not repeatedly witnessed immoral acts.

Method

Participants were 267 (59% female; M_age = 34.5, SD_age = 12.0) users of Amazon Mechanical Turk who participated online for pay. The study comprised a four condition (control vs. city control vs. impure exposure vs. harmful exposure) between-subjects design. There were approximately 67 participants in each condition, which allowed for replication of the sanctity–censorship link (i.e., in the control condition) at greater than 90% power.

Participants’ moral values were measured using the MRQ and the MJQ. The censorship measure featured the same items from Studies 1 and 2, but the specific prompt varied by condition (see Online Appendix C). In all conditions, participants read about an 8-year-old boy named Jon whose parents intentionally showed him one of various movie scenes. In the city control, impure exposure, and harmful exposure conditions, participants also read that Jon lived with his parents in a large city. In the impure exposure and harmful exposure conditions, participants read that Jon had witnessed sanctity violations and care violations, respectively, on numerous occasions. Furthermore, in both of those conditions, participants read that Jon had shown no negative effects of this upbringing and was an otherwise normal 8-year-old. After completing the censorship measure, participants completed a demographics measure, including political orientation (1 = very liberal, 7 = very conservative), age, sex, and parenthood.

Results

Thirty participants failed the MFQ attention checks, leaving 237 participants in the final analyses. Censorship scores were created for each moral domain by averaging responses within that domain and for total censorship by averaging responses to all 20 items. All censorship scores were reliable (α_care = .84, α_fairness = .85, α_loyalty = .59, α_authority = .76, α_sanctity = .72, α_total = .92). Participants’ endorsements of the five moral domains were calculated from their MRQ and MJQ responses. MRQ scores were consistently reliable (α_care = .65, α_fairness = .75, α_loyalty = .78, α_authority = .64, α_sanctity = .76) but MJQ scores were again not (α_care = .46, α_fairness = .27, α_loyalty = .43, α_authority = .58, α_sanctity = .80). Therefore, subsequent analyses involving participants’ moral values only include MRQ scores.

Total censorship as a function of condition

Although peripheral to our study, we first conducted a one-way ANOVA with total censorship as the dependent variable and condition as the independent variable, with planned contrasts comparing the city control condition with each of the control, impure exposure, and harmful exposure conditions (see Table 1). There was no overall significant difference between conditions in terms of total censorship, F(3, 233) = 1.37, p = .25, η²_p = .02. There was no significant difference in total censorship attitudes between the city control condition (M = 4.17) and the control condition (M = 4.15), F(1, 233) = 0.01, p = .92, η²_p < .001, and the harmful exposure condition (M = 3.97), F(1, 233) = 1.09, p = .29, η²_p = .004. However, participants in the city control condition were marginally significantly more bothered than participants in the impure experiences condition (M = 3.82), F(1, 233) = 2.73, p = .082, η²_p = .01.

Replication of sanctity–censorship link

We next tested for replication of the sanctity–censorship link. This analysis focused only on participants in the control and the city control conditions (i.e., the two conditions in which we expected the young child’s innocence to be preserved and, therefore, worth protecting). Separate regression analyses were conducted for each of the two conditions. Furthermore, separate regression analyses were conducted for each of the five censorship domains and for total censorship, with participants’ endorsement of the five moral domains as the predictors. Consistent with Studies 1 and 2, there was a significant positive link between sanctity values and attitudes toward censorship in all five censorship domains and for total censorship for both the control and the city control conditions (see Table 5).

Table 5.

Strength of Objections to Exposure as a Function of Moral Values in Study 3.

Condition	Moral value	Domain of depicted violation
Condition	Moral value	Care	Fairness	Loyalty	Authority	Sanctity	Total
Control	Care	.29^†	.19	.10	.20	.03	.19
	Fairness	−.22	−.09	−.25^†	−.18	−.16	−.20
	Loyalty	.18	.26^†	.18	.11	.17	.22^†
	Authority	.07	.01	.25	.15	.08	.12
	Sanctity	.33*	.42**	.29*	.47***	.56***	.49***
City control	Care	.46*	.05	−.19	.02	.09	.11
	Fairness	−.05	.21	.05	−.12	.10	.05
	Loyalty	.09	−.02	−.02	−.09	−.01	−.02
	Authority	−.32^†	−.23	.08	.23	.04	−.05
	Sanctity	.44**	.47**	.57***	.54**	.59***	.64***
Impure exposure	Care	.25^†	.11	−.09	.08	−.05	.09
	Fairness	.16	.22	.03	−.04	.02	.11
	Loyalty	−.02	.07	.40**	.16	−.04	.12
	Authority	.05	−.04	.21	.23	.16	.15
	Sanctity	.19	−.04	.14	.25^†	.44**	.26^†
Harmful exposure	Care	.20	.08	−.05	.06	−.08	.06
	Fairness	.15	.13	.05	.01	.27^†	.15
	Loyalty	.04	.27^†	.21	.19	−.05	.17
	Authority	.06	.27	.35^†	.32*	.12	.28^†
	Sanctity	.12	−.00	.07	.28^†	.46**	.22

Note. Values are βs predicting how bothered participants reported they would be. Each column within a condition represents a multiple regression analysis.

†

p < .1. *p < .05. **p < .01. ***p < .001.

As in Studies 1 and 2, there was little support for the personal values hypothesis. In both conditions, only care and sanctity displayed at least marginally significant links consistent with the personal values hypothesis, and the latter relationship can also be explained by the sanctity–censorship hypothesis. The data, thus, support the primacy of sanctity values in predicting censorship attitudes.

As in Studies 1 and 2, we also tested the relationship between demographic variables and overall willingness to censor. For participants in the control condition, political conservatism, r(58) = .46, p < .001, parental status, r(58) = .32, p = .01, and being older, r(58) = .48, p < .001, were significantly correlated with greater overall attitudes endorsement of censorship, whereas sex was not, r(58) = .18, p = .17. Likewise, for participants in the city control condition, political conservatism, r(56) = .33, p = .01, parental status, r(57) = .47, p < .001, and being older, r(57) = .40, p = .002, were significantly correlated with greater overall endorsement of censorship, whereas sex was not, r(57) = .14, p = .28.

We then entered the demographic variables into full regression models with moral values as simultaneous predictors. For participants in the control condition, both political orientation, β = .20, p = .06, and age, β = .18, p = .096, emerged as marginally significant predictors of total censorship, such that more conservative and older participants tended to censor more than more liberal and younger participants. However, sanctity values were the strongest predictor of total censorship, β = .27, p = .06. For participants in the city control condition, only sanctity values emerged as a significant predictor of total censorship, β = .52, p = .001.

Following the procedure laid out in Study 1, we tested whether sanctity values or a different moral value mediated the relationship between the demographic variables (political orientation, age, and parental status) and overall willingness to censor. For participants in the control condition, only sanctity values significantly mediated the relationship between political orientation, 95% CI = [.05, .27], parental status, 95% CI = [.03, .68], and age, 95% CI = [.01, .03], and total censorship. A similar pattern held for participants in the city control condition, as only sanctity values significantly mediated the relationship between political orientation, 95% CI = [.07, .32], parental status, 95% CI = [.18, .95], and age, 95% CI = [.01, .03], and total censorship.

Role of life experiences

Next, we tested whether sanctity values continued to significantly predict censorship attitudes if the child had been exposed to immoral acts. To do so, we examined how moral values predicted censorship in the impure exposure and harmful exposure conditions separately. As shown in Table 5, sanctity values did not play as strong a role in either condition as it did in the control and city control conditions, significantly predicting censorship attitudes in only one of the five domains (sanctity) for both conditions.

To test whether the strength of the sanctity–censorship link differed between conditions, we conducted a multiple regression analysis predicting total censorship scores as a function of condition, the five moral value scores, and the two-way interactions between condition and moral values. Condition was entered with three dummy-coded variables that compared each of the control, impure exposure, and harmful exposure conditions with the city control condition. The reason that the city control condition was used as the point of comparison was because it had the fewest differences between it and each of the other conditions. The results revealed that sanctity values did not significantly interact with condition when comparing the control condition with the city control condition, β = –.05, p = .63. However, sanctity values did marginally significantly interact with immoral experiences when comparing the city control condition with the impure exposure condition, β = –.16, p = .064, and with the harmful exposure condition, β = –.20, p = .052. Therefore, sanctity played a marginally larger role in determining censorship attitudes in the city control condition than in either the impure experiences or the harmful experiences conditions.

Discussion

As in Studies 1 and 2, concerns for moral sanctity emerged as the strongest predictor of attitudes to censor objectionable material from 8-year-old children, and this effect was not diminished if the child had grown up in an urban setting. However, the link between sanctity values and censorship decreased if the child had repeatedly been exposed to immoral behavior. This reduction in the sanctity–censorship link occurred if the child had witnessed impure behavior or harmful behavior, suggesting that witnessing either form of immorality reduced the child’s innocence. People who valued moral sanctity more did not show increased desire to censor from children with prior exposure to immoral acts, presumably because there was less purity and innocence in the child to protect.

Thus, for people who valued moral sanctity, children previously exposed to harmful and impure acts in their neighborhood are no longer innocent and, thus, sanctity values no longer played a role in determining censorship attitudes. This finding is consistent with the proposed properties of purity whereby contamination of a pure entity is irreversible and dose insensitive (Rozin & Nemeroff, 2002). Living in a city is not by itself a contaminating act. But once a child living in that city has seen impure or harmful acts, the child’s purity is forever lost and people who highly value sanctity are no longer motivated to preserve it.

General Discussion

Concerns about sanctity, more so than other moral concerns, appear linked to censorship from young children. Censorship, and perhaps other special protections for children, may be driven by the desire to protect youthful innocence and purity. Indeed, across three studies, we found that the more people valued moral sanctity, the more they wanted to protect young children from exposure to immoral acts.

Other moral concerns were not consistently linked to censorship from children. It seems plausible that when people offer special moral protections to children, they seek specifically to protect children from exposure to acts that they themselves deem immoral, to model for children what is right and wrong (Bandura, 1977). For example, people who highly value loyalty might censor children from exposure to disloyal acts. The data, however, did not support this alternative to the sanctity–censorship hypothesis. Regardless of the objectionable content—whether containing disloyal acts, disobedient acts, or harmful acts—it was endorsement of moral sanctity that positively predicted willingness to censor it.

Concerns of moral sanctity are particularly linked to censorship from young children who are seen as innocent and pure. Sanctity values were strongly linked to censorship from an 8-year-old child. However, as the age of the child increased, this sanctity–censorship link weakened or disappeared entirely. In addition, prior exposure to immoral acts matters. If a young child was described as having repeatedly witnessed the immoral behavior of others, then the sanctity–censorship link again diminished. Presumably, prior experience of immoral acts taints a young child’s innocence so that people who value moral sanctity feel it no longer needs to be protected.

Whether a young child’s innocence is intentionally corrupted on the part of a parent also plays a role in censorship attitudes. Participants who endorsed moral sanctity objected when a parent intentionally showed an 8-year-old immoral acts. If the 8-year-old viewed such media without an adult’s consent or if the 8-year-old independently expressed objectionable ideas, then sanctity concerns were not consistently linked to objections. It is when a young child’s innocence is knowingly and intentionally corrupted that people who value moral sanctity object.

The present work is consistent with a general pattern emerging in the literature whereby caring for children is linked to concerns about purity and contamination. These purity concerns seem to be rooted in the fact that young children are vulnerable to disease (Simon et al., 2015) and so protecting them requires keeping them from pathogens and other contaminants (Curtis & Biran, 2001; Fessler et al., 2005). The desire to protect children from physical contaminants seems to have become inclusive over time of a desire to keep them from social and moral contaminants as well. For example, pregnant women have been found to be more ethnocentric and, thus, contamination averse in the social domain (Navarrete et al., 2007). Moreover, parenting motives have been linked to greater condemnation of morally impure acts (Buckels et al., 2015; Eibach et al., 2009). The present work demonstrates the reverse relationship—that people who endorse moral purity and sanctity values tend also to be motivated to protect children from moral degradation. Specifically, people who care about moral sanctity are simultaneously concerned with protecting the innocence of youth via censorship of objectionable media.

Censorship of Sanctity Violations Versus Other Violations

One question regarding our results is whether sanctity values should be more strongly linked to censorship of sanctity violations than to censorship of other types of violations. If censorship is partially explained by a concern with the child’s purity, then people who value sanctity could be especially concerned with censoring violations that are inherently impure. However, we do not expect that this should necessarily be the case. The chief concern for censorship is the innocence of the child and, as evidenced by Study 3, what violates that innocence is exposure to immoral acts broadly, not just sanctity violations. Thus, any immoral act should threaten a child’s innocence and should be targeted for censorship.

For censorship of sanctity violations to have an especially strong link to sanctity values, there would have to be some participants for whom sanctity values are high but other values (loyalty, authority, etc.) are low. But that is rarely the case. Sanctity values are generally one of the least endorsed of the moral values (e.g., Graham et al., 2011). If a person values sanctity, the person usually values just as much or more the other moral foundations, meaning that if a person values sanctity and thinks that impure behaviors are immoral, they also believe that other acts (disobedient acts, disloyal acts, etc.) are immoral. Therefore, someone who values sanctity will want to censor across domains, not just within sanctity.

Our data support the notion that people who value sanctity do not show an especially strong motivation to censor sanctity violations over other violations. In Study 1, the unique link between valuing sanctity and willingness to censor violations of sanctity (i.e., the sanctity–sanctity link) was directionally weaker than both the sanctity–care link and the sanctity–authority link. In Studies 2 and 3, the sanctity–sanctity link did tend to be the strongest of all links directionally, but that difference was statistically significant in only two cases out of a possible 16. This suggests that although sanctity values might occasionally be a stronger predictor of sanctity censorship than other types of censorship, this difference is inconsistent and not uniform. Therefore, there is minimal evidence to suggest that sanctity values should be especially attuned to censoring sanctity violations. Still, future work may delve further into this possibility.

Measurement Issues

One potential shortcoming of the present work is a perhaps tautological connection between one of the items used in the sanctity measure of the MRQ (“Whether or not someone violated standards of purity and decency”) and the censorship scale. That is, an alternative explanation for our findings is that much of the relationship between sanctity values and censorship could be due to conceptual and content overlap with that single item.

We tested whether this was the case. We recalculated participants’ MRQ sanctity scores using only the other two items and reran our main regression analyses. Sanctity values remained a significant predictor of four of the five censorship categories (with the exception of sanctity, β = .21, p = .12) and total censorship in Study 1, four of the five censorship categories (with the exception of harm, β = .25, p = .099) and total censorship for the parental exposure-8 condition in Study 2, and for five of the five censorship categories and total censorship in both the control and city control conditions in Study 3. Thus, only twice across 24 tests was sanctity not a significant predictor of censorship, and in both of those cases, the relationship was in the predicted direction and approaching conventional levels of significance. The sanctity–censorship link appears robust and not simply due to a confound with the one item.

Another potential concern with our findings is that we only used the MRQ subscale of the MFQ and not the MJQ subscale. As noted above, we did not use the MJQ subscale because of its weak internal reliability scores. One possible reason for the MJQ’s weak reliability is that participants may have perceived the relatively specific and concrete items in the MJQ as more distinct from each other than the relatively abstract items in the MRQ. But this is only speculative and does not explain why we found weak reliability compared with other researchers (e.g., Graham et al., 2011).

Nevertheless, it was the case that reliability scores for the sanctity foundation of the MJQ were sufficiently high, even though reliability scores for the other MJQ foundations were not. Given this sufficient reliability, we retested our hypotheses using the MJQ sanctity scores in place of the MRQ sanctity scores. Sanctity values significantly predicted censorship in all 24 cases where our hypothesis predicted it should: for all six tests (i.e., predicting each of the five censorship foundations as well as total censorship) in Study 1, Study 2 (i.e., in the parental exposure-8 condition), and the control and city control conditions in Study 3.

Limitations and Future Directions

The present work relied exclusively on hypotheticals. Therefore, much of the complexities present in actual children and in real adult–child relationships were absent from the present studies. Furthermore, we were able only to assess attitudes toward censorship instead of actual censorship behavior. Therefore, future work should examine whether the results generalize to more ecologically valid measures of censorship.

In addition, future studies could also move beyond the MFT framework we used to more directly examine the experience of censoring by asking people to express why they are motivated to censor: What are the momentary thoughts and feelings that motivate censoring from children? Such responses may illuminate the more proximal mechanisms by which sanctity values operate on censorship attitudes.

Future work may also explore how motivations to censor are linked to age and parenthood. We found mixed results in that regard. Age was not predictive of censorship in Study 1, but it was significantly, positively linked to censorship in Studies 2 and 3. Being a parent was not predictive of censorship in Study 1 or Study 2, but it was significantly linked to stronger procensorship attitudes in Study 3. Although these relationships were not of central interest in the present work, our results suggest there may be meaningful links between these variables and censorship that are worth investigating further.

One question that our data do not directly address but that they may nevertheless speak to is what motivates sanctity values in the first place: Why does sanctity have any moral weight at all? Researchers have proposed several possibilities. Both disgust sensitivity and religiosity correlate with moral sanctity values (Graham et al., 2011), suggesting that sanctity is moralized because of the value some people place on physical and spiritual purity and cleanliness. Alternatively, sanctity values may become moralized because of the perceived harm caused by disgusting and impure acts (Schein, Ritter, & Gray, 2016). Our data suggest that these concerns of purity and protection from harm may in some cases be in the service of vulnerable others more than the self. Young children in particular appear to be a highly vulnerable group whom those valuing sanctity seek to protect. Future research may shed light on motivations underlying moralization of sanctity, and to what degree protection of pure and vulnerable others is central among them.

In Study 3, we found general support for our hypothesis that the sanctity–censorship link was reduced when the young child in question had already been exposed in the past to impure or harmful acts. However, we did unexpectedly find that the sanctity–censorship link remained significant when the immoral acts being represented were violations of sanctity. One possible reason for this is that when the child’s purity is no longer an issue, people’s objections are more focused on the transgression being represented than on the effect the representation has on the already-tainted child. Thus, participants may simply be indicating how bothersome the depicted moral transgression itself is. If this is true, then in the impure exposure and harmful exposure conditions, we should see that participants object to any representation that violates a moral foundation they value (e.g., sanctity values uniquely predicting sanctity censorship and loyalty values uniquely predicting loyalty censorship). Indeed, we do see some evidence of this pattern (see Table 5). In the impure exposure and harmful exposure conditions, all these links have positive βs that are .13 or greater, and half of these links are significant (ps < .05) or approaching significant (ps < .10). In contrast, in the control and city control conditions, a minority of these links are statistically significant, and in two cases, the βs even show a nonsignificant negative relationship. Of course, these data are only weakly supportive of the notion that people are shifting their focus toward the acts within the representations themselves in the impure exposure and harmful exposure conditions. Thus, further research is needed to understand this unexpected pattern of results.

As an extension of the present research, future work should also examine the manner in which the perceived innocence of children elicits special treatment. Censorship is only one case. Children are also protected from blame (e.g., Cipriani, 2009), victimization (e.g., Norman-Eady, 2000), and even certain truths (e.g., that Santa Claus does not exist). Future work may examine how these practices are related to children’s perceived innocence and moral development and to the valuation of moral sanctity.

Conclusion

The present work revealed that moral protections for children may be driven by moral sanctity concerns more than other moral concerns. We found that people who value moral sanctity tended also to value protecting young children from viewing immoral acts. Showing immoral acts to young children may violate ideals of moral sanctity, given that such acts involve tarnishing young children’s purity and innocence (Woodrow, 1999). This pattern exposes one motivation underlying the special moral treatment of children.

Footnotes

Declaration of Conflicting Interests

The author(s) declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.

Funding

The author(s) received no financial support for the research, authorship, and/or publication of this article.

Notes

Supplemental Material

Supplementary material is available online with this article.

References

Altemeyer

R. A.

(1981). Right-wing authoritarianism. Winnipeg, Canada: University of Manitoba Press.

Anderson

C. A.

Bushman

B. J.

(2001). Effects of violent video games on aggressive behavior, aggressive cognition, aggressive affect, physiological arousal, and prosocial behavior: A meta-analytic review of the scientific literature. Psychological Science, 12, 353-359.

Bandura

(1977). Social learning theory. Englewood Cliffs, NJ: Prentice Hall.

Buckels

E. E.

Beall

A. T.

Hofer

M. K.

Lin

E. Y.

Zhou

Schaller

(2015). Individual differences in activation of the parental care motivational system: Assessment, prediction, and implications. Journal of Personality and Social Psychology, 108, 497-514.

Bushman

B. J.

Anderson

C. A.

(2001). Media violence and the American public: Scientific facts versus media misinformation. American Psychologist, 56, 477-489.

Cipriani

(2009). Children’s rights and the minimum age of criminal responsibility. Abingdon, UK: Ashgate.

Cohen

A. B.

(2003). Religion, likelihood of action, and the morality of mentality. International Journal for the Psychology of Religion, 13, 273-285.

Cohen

A. B.

Rozin

(2001). Religion and the morality of mentality. Journal of Personality and Social Psychology, 81, 697-710.

Curtis

Biran

(2001). Dirt, disgust, and disease: Is hygiene in our genes? Perspectives in Biology and Medicine, 44, 17-31.

10.

Cushman

F. A.

(2008). Crime and punishment: Distinguishing the roles of causal and intentional analyses in moral judgment. Cognition, 108, 353-380.

11.

Dawson

S. L.

Tyson

G. A.

(2012). Will morality or political ideology determine attitudes to climate change? Australian Community Psychologist, 24(2), 8-25.

12.

Eibach

R. P.

Libby

L. K.

Ehrlinger

(2009). Priming family values: How being a parent affects moral evaluations of harmless but offensive acts. Journal of Experimental Social Psychology, 45, 1160-1163.

13.

Eibach

R. P.

Mock

S. E.

(2011). The vigilant parent: Parental role salience affects parents’ risk perceptions, risk-aversion, and trust in strangers. Journal of Experimental Social Psychology, 47, 694-697.

14.

Fessler

D. T.

Eng

S. J.

Navarrete

C. D.

(2005). Elevated disgust sensitivity in the first trimester of pregnancy: Evidence supporting the compensatory prophylaxis hypothesis. Evolution & Human Behavior, 26, 344-351.

15.

Gino

Desai

S. D.

(2012). Memory lane and morality: How childhood memories promote prosocial behavior. Journal of Personality and Social Psychology, 102, 743-758.

16.

Goff

Jackson

Di Leone

Culotta

DiTomasso

(2014). The essence of innocence: Consequences of dehumanizing Black children. Journal of Personality and Social Psychology, 106, 526-545.

17.

Goodman

J. K.

Cryder

C. E.

Cheema

(2013). Data collection in a flat world: The strengths and weaknesses of Mechanical Turk samples. Journal of Behavioral Decision Making, 26, 213-224.

18.

Graham

Haidt

Koleva

Motyl

Iyer

Wojcik

Ditto

(2013). Moral foundations theory: The pragmatic validity of moral pluralism. Advances in Experimental Social Psychology, 47, 55-130.

19.

Graham

Haidt

Nosek

B. A.

(2009). Liberals and conservatives rely on different sets of moral foundations. Journal of Personality and Social Psychology, 96, 1029-1046.

20.

Graham

Nosek

B. A.

Haidt

Iyer

Koleva

Ditto

P. H.

(2011). Mapping the moral domain. Journal of Personality and Social Psychology, 101, 366-385.

21.

Haidt

Graham

(2007). When morality opposes justice: Conservatives have moral intuitions that liberals may not recognize. Social Justice Research, 20, 98-116.

22.

Haidt

McCauley

Rozin

(1994). Individual differences in sensitivity to disgust: A scale sampling seven domains of disgust elicitors. Personality and Individual Differences, 16, 701-713.

23.

Helzer

E. G.

Pizarro

D. A.

(2011). Dirty liberals! Reminders of physical cleanliness influence moral political attitudes. Psychological Science, 22, 517-522.

24.

Inbar

Pizarro

D. A.

Bloom

(2009). Conservatives are more easily disgusted than liberals. Cognition and Emotion, 23, 714-725.

25.

James

Jenks

Prout

(1998). Theorizing childhood. Oxford, UK: Blackwell.

26.

Kenrick

D. T.

Griskevicius

Neuberg

S. L.

Schaller

(2010). Renovating the pyramid of needs: Contemporary extensions built upon ancient foundations. Perspectives on Psychological Science, 5, 292-314.

27.

Kenrick

D. T.

Neuberg

S. L.

Griskevicius

Becker

D. V.

Schaller

(2010). Goal-driven cognition and functional behavior: The fundamental-motives framework. Current Directions in Psychological Science, 19, 63-67.

28.

Navarrete

C. D.

Fessler

D. M.

Eng

S. J.

(2007). Elevated ethnocentrism in the first trimester of pregnancy. Evolution & Human Behavior, 28, 60-65.

29.

Nittono

Fukushima

Yano

Moriya

(2012). The power of Kawaii: Viewing cute images promotes a careful behavior and narrows attentional focus. PLoS ONE, 7(9), e46362. doi:10.1371/journal.pone.0046362

30.

Norman-Eady

(2000, November, 22). Sentences for child abuse. Retrieved from http://cga.ct.gov/2000/rpt/2000-r-1064.htm

31.

Oaten

Stevenson

R. J.

Case

T. I.

(2009). Disgust as a disease-avoidance mechanism. Psychological Bulletin, 135, 303-321.

32.

Preacher

K. J.

Hayes

A. F.

(2008). Asymptotic and resampling strategies for assessing and comparing indirect effects in multiple mediator models. Behavior Research Methods, 40, 879-891.

33.

Rozin

Nemeroff

(2002). Sympathetic magical thinking: The contagion and similarity “heuristics.” In Gilovich

Griffin

Kahneman

(Eds.). Heuristics and biases: The psychology of intuitive judgment (pp. 201-216). Cambridge, UK: Cambridge University Press.

34.

Schein

Ritter

R. S.

Gray

(2016). Harm mediates the disgust-immorality link. Emotion, 16, 862-876.

35.

Sherman

G. D.

Haidt

Coan

J. A.

(2009). Viewing cute images increases behavioral carefulness. Emotion, 9, 282-286.

36.

Simon

A. K.

Hollander

G. A.

McMichael

(2015). Evolution of the immune system in humans from infancy to old age. Proceedings of the Royal Society B: Biological Sciences, 282(1821), 20143085.

37.

Sivulka

(2001). Stronger than dirt: A cultural history of advertising personal hygiene in America, 1875 to 1940. New York, NY: Humanity Books.

38.

Van Leeuwen

Park

J. H

. (2009). Perceptions of social dangers, moral foundations, and political orientation. Personality and Individual Differences, 47, 169-173.

39.

Winegard

Deaner

R. O.

(2010). The evolutionary significance of Red Sox Nation: Sport fandom as a by-product of coalitional psychology. Evolutionary Psychology, 8, 432-446.

40.

Woodrow

(1999). Revisiting images of the child in early childhood education: Reflections and considerations. Australian Journal of Early Childhood, 24, 7-13.

Supplementary Material

Please find the following supplemental material available below.

For Open Access articles published under a Creative Commons License, all supplemental material carries the same license as the article it is associated with.

For non-Open Access articles published, all supplemental material carries a non-exclusive license, and permission requests for re-use of supplemental material or any part of supplemental material shall be sent directly to the copyright owner as specified in the copyright notice associated with the article.

0.00 MB

0.02 MB