The number of women without children is increasing in Western cultures (Kohli and Albertini 2009; Martinez et al. 2012) and in the United States (Dye 2008). In 2010, approximately 43 % of U.S. women aged 15–44 did not have children, and 6 % of these women had deliberately chosen to forgo motherhood (Martinez et al. 2012). This increase has been associated with greater rates of postponing or forgoing marriage, improved access to abortion and contraceptives, increased divorce rates, and delay of initial childbirth (Kohli and Albertini 2009; Lundquist et al. 2009). It has also corresponded with the financial, social, and political gains associated with the women’s liberation movement (Lundquist et al. 2009). Indeed, academic achievement (Abma and Martinez 2006; Lundquist et al. 2009) and greater opportunity for women to participate in the labor force (Gubernskaya 2010) are among the strongest predictors of women’s nonparental status.

The present study uses a number of terms to describe nonparents. The term childless describes individuals who do not have children, but the reasons for the status are unidentified or diverse (Houseknecht 1987). The childfree individual has made an intentional decision not to parent and includes “…those who expect to have no children in their lifetimes, and are either fecund (physically able to have a birth) or are surgically sterile for contraceptive reasons” (Martinez et al. 2012, p. 4; italics included in the original quote). The involuntarily childless individual may wish to parent but has not due to uncontrollable impediments (e.g., infertility, lack of a partner; Jeffries and Konnert 2002). Thus, distinctions between being childfree and involuntarily childless arise from differing attitudes toward and degree of control over one’s parental status.

Contemporary literature has demonstrated that childfree women are demographically distinct from involuntarily childless women. Childfree women typically achieve advanced education (Lundquist et al. 2009; Martinez et al. 2012), are dedicated to reaching their personal and career aspirations (Majumdar 2004), are more likely to hold full-time professional employment (Abma and Martinez 2006), and have high income (Abma and Martinez 2006; Lundquist et al. 2009). Additionally, women who are childfree are more likely to be older than are parents (Majumdar 2004; Martinez et al. 2012), as well as unmarried (Abma and Martinez 2006; Martinez et al. 2012), and non-religious (Abma and Martinez 2006).

Attitudes Towards Being Childless or Childfree

Early research on perceptions of parental status primarily focused on studying attitudes toward women and, secondarily, couples. Many early studies also failed to distinguish between women who were involuntarily childless and those who were childfree. Nonetheless, in early cross-cultural literature, childfree women evoked more negative perceptions than did mothers (Kopper and Smith 2001; La Mastro 2001; Lampman and Dowling-Guyer 1995) and involuntarily childless women (Kopper and Smith 2001; Lampman and Dowling-Guyer 1995). For example, Polit’s (1978) U.S. participants rated both childfree men and women as more socially distant, immature, in need of social support, and selfish than were parents; however, childfree women were perceived less favorably than were childfree men. Additionally, childfree women in Lampman and Dowling-Guyer’s (1995) study were rated by U.S. participants as less caring and ambitious than were involuntarily childless women. Moreover, Kopper and Smith (2001) found that U.S. childfree male–female couples evoked negative emotional reactions, but involuntarily childless couples evoked sympathy. Involuntarily childless women are also sometimes perceived more negatively than are mothers (Calhoun and Selby 1980; Callan 1985; Peterson 1983). Although far fewer studies have assessed perceptions of childfree men, research confirms that childfree men are also perceived more negatively than are fathers (Callan 1985; Calhoun and Selby 1980; Jamison et al. 1980; La Mastro 2001; Lampman and Dowling-Guyer 1995) and involuntarily childless men (Callan 1985; Kopper and Smith 2001; Lampman and Dowling-Guyer 1995).

Although some contemporary findings suggest that attitudes toward women without children may be improving in Western cultures (Gubernskaya 2010; Merz and Liefbroer 2012; Noordhuizen et al. 2010), other findings support the persistence of negative perceptions of women without children in the United States (Ashburn-Nardo 2016; Coffey 2007; Koropeckyj-Cox and Pendell 2007; Koropeckyj-Cox et al. 2007; Vinson et al. 2010) and other countries (Çopur and Koropeckyj-Cox 2010; Dolińska 2014; Kemkes 2008; Rowlands and Lee 2006). For example, Ashburn-Nardo (2016) found that childfree women and men were evaluated as less psychologically fulfilled than were parents, and they elicited feelings of anger, disgust, and disapproval from participants. Additionally, contemporary qualitative research provides a particularly rich picture of the experiences of childfree women who describe continued negative social reactions, stigmatization, and stereotyping based on their nonparental status (Doyle et al. 2012; Rich et al. 2011; Shaw 2011).

Research suggests that childfree women are well-aware of the negative sanctions against being childfree. Childfree women have described being stereotyped by others and feeling pressured to have a child from family and friends (Mueller and Yoder 1999) and report experiencing social reactions of criticism, pity, shock, and dismay (Mollen 2006) along with responses of disbelief, disregard, and deviance from others (Gillespie 2000). Negative perceptions of nonparents may also affect real-world outcomes. For example, nonparents are sometimes expected to work longer and more inconvenient hours and to forgo holidays and weekends to accommodate parenting coworkers (Mollen 2006; Picard 1997). In one study, single parents were offered jobs with more stable locations and higher merit-based stipends than single nonparents (Eby et al. 2004). Childfree women also describe being denied voluntary sterilization by medical professionals who insist that they will regret their decision (Gillespie 2000; Mollen 2006; Mueller and Yoder 1999). Participants in other studies were more likely to allocate medical resources to parents than to their nonparenting counterparts (Wiseman 2007, 2010).

Stereotypes, Emotional Responses, and Behaviors

Women without children are stereotyped as materialistic and immature (Callan 1985; Kemkes 2008), emotionally unstable and selfish (Peterson 1983), less likely to live happy and satisfied lives (Mueller and Yoder 1997), and maladjusted (Polit 1978). The Stereotype Content Model (SCM) additionally suggests that competence and warmth are principal and universal dimensions of social evaluations (Cuddy et al. 2009; Fiske et al. 2002b, 2007). Competence includes traits such as confidence, intelligence, and skillfulness, whereas warmth includes traits such as trustworthiness, friendliness, and sincerity (Fiske et al. 2002b). Thus, traits that are often attributed to nonparents may be subsumed under the competence and warmth dimensions. The SCM further predicts that perceptions of competence and warmth vary systematically with perceptions of a social group’s competition and status. Competition reflects the extent to which the goals of the outgroup align with the goals of the ingroup, whereas status reflects the outgroup’s ability to achieve their goals and compete adequately with the ingroup (Cuddy et al. 2007; Fiske et al. 2007). Outgroups sharing the ingroup’s goals without the ability to achieve those goals are perceived as noncompetitive and warm; outgroups who share the ingroup’s goals and can also achieve those goals are perceived as competitive and cold. Moreover, high status groups with the ability to achieve their goals are perceived as competent, whereas low status groups with little ability to achieve their goals are perceived as incompetent.

The SCM challenges conceptualizations of stereotypes as purely negative, proposing instead that groups can be perceived as purely negative or positive (i.e., unambivalent stereotypes), or perceived both positively and negatively (i.e., ambivalent stereotypes). Thus, SCM framework predicts four competence-warmth combinations accompanied by emotional responses from others (Fiske et al. 2002a, b). The first group is unambivalently and positively stereotyped as high in both competence and warmth (HC-HW). This is often a reference group, admired outgroup, and/or the ingroup and evokes emotions of pride and admiration due to their sanctioned high status and noncompetition with the ingroup (e.g., Americans). The second group is unambivalently and negatively stereotyped as low in both competence and warmth (LC-LW). This group evokes anger and disgust (i.e., contemptuous prejudice) due to their alleged attempts to exploit the ingroup and perceived responsibility for their negative outcomes (e.g., homeless individuals). The two remaining ambivalent combinations receive high ratings on one dimension and low ratings on the other (Fiske et al. 2002b). Low competence-high warmth (LC-HW) groups evoke pity and sympathy (i.e., paternalistic prejudice) due to perceived lack of responsibility for their low status and noncompetition (e.g., the elderly). Finally, high competence-low warmth (HC-LW) groups evoke resentment and envy (i.e., envious prejudice) due to their high, yet unsanctioned status and competitiveness (e.g., wealthy individuals).

The Behaviors from Intergroup Affect and Stereotypes (BIAS) map expands the SCM to include expected behaviors associated with competence-warmth combinations (Cuddy et al. 2007). Behaviors are classified on the active-passive and facilitative-harmful dimensions. The active-passive dimension predicts behavioral intensity, such that active behaviors are direct and intentional, whereas passive behaviors are indirect and less intentional. The facilitative-harmful dimension predicts behavioral valence, with groups being advantaged by facilitative behaviors and disadvantaged by harmful behaviors (Cuddy et al. 2007). The BIAS map predicts four combinations resulting from the active-passive and facilitative-harmful dimensions (Cuddy et al. 2007). Active facilitation acts for and overtly benefits a group (e.g., antidiscrimination laws), passive facilitation acts with and indirectly benefits a group (e.g., hiring illegal immigrants at reduced pay), active harm acts against and overtly disadvantages a group (e.g., hate crimes), and passive harm acts without and indirectly disadvantages a group (e.g., avoiding contact). According to the BIAS map, active behaviors are positively associated with warmth, and passive behaviors are positively associated with competence (Cuddy et al. 2007).

By combining the cognitive, emotional, and behavioral components of intergroup evaluations and the SCM, the BIAS map predicts that: (a) high competence-high warmth (HC-HW) groups evoke admiration and pride as well as active and passive facilitation, (b) low competence-low warmth (LC-LW) groups evoke contemptuous prejudice as well as active and passive harm, (c) low competence-high warmth (LC-HW) groups evoke paternalistic prejudice as well as active facilitation and passive harm, and (d) high competence-low warmth (HC-LW) groups evoke envious prejudice as well as passive facilitation and active harm (Cuddy et al. 2007). The BIAS map is also a mediation model, with stereotype content exerting an indirect influence on behaviors through at least one emotion in samples in the United States and Europe (Becker and Asbrock 2012; Cuddy et al. 2007; Echebarria-Echabe 2013; Levine and Schweitzer 2015). For example, Cuddy et al. (2007) found that warmth indirectly influenced active facilitation through admiration and pity, warmth influenced active harm through contempt, competence influenced passive facilitation through admiration, and competence influenced passive harm through pity.

In previous research, subgroups of women tend to be ambivalently stereotyped. Brambilla et al. (2011) found that most Italian lesbian subtypes were rated as HC-LW. The majority of Eckes (2002) German subgroups were stereotyped as either HC-LW (e.g., feminist) or LC-HW (e.g., wallflower). Although Eckes (2002) did not include parental statuses, the LC-HW “housewife” and HC-LW “career woman” tentatively resembled mothers and childless women, respectively. Using a U.S. sample, comparable findings were reported by Wade and Brewer (2006) for businesswomen (HC-LW) and homemakers (LC-HW). Although housewives do not necessarily equate with mothers, nor career women with childless or childfree women, these findings have SCM and BIAS map implications for parental statuses. Indeed, Cuddy et al.’s (2004) U.S. sample rated working mothers as LC-HW and working childless women as HC-LW, resulting in less hiring and promotion of mothers. In contrast, Australian participants in Masser et al.’s (2007) study rated a pregnant job candidate as warmer and more competent than her non-pregnant counterpart, but still viewed the pregnant woman as a less desirable candidate. Yet, Cuddy et al. (2004) did not distinguish between involuntarily childless and childfree women by describing the cause of childlessness and future intent to mother. Furthermore, no known study has examined BIAS map relations among stereotype content, emotions, and behaviors for women’s parental statuses.

Research demonstrates that when reasons for nonparental status are known, different perceptions emerge of involuntarily childless and childfree women (Kopper and Smith 2001; Lampman and Dowling-Guyer 1995). Further, childlessness is viewed as a negative outcome in most Western cultures. The SCM and BIAS map suggest that those lacking responsibility for their negative outcomes tend to be pitied as well as elicit active facilitation and passive harm. In contrast, individuals with some responsibility over their outcomes may elicit envy and contempt as well as passive facilitation and active harm. Thus, when fertility control and parenting decisions are inferred, women who are childfree and involuntarily childless may be stereotyped differently. Because they either already parent or desire to parent, mothers and involuntarily childless women may both be perceived as warmer than competent, eliciting paternalistic prejudice as well as active facilitation and passive harm. Conversely, because of their intentional decision not to mother, childfree women may be viewed as more competent than warm, eliciting envious prejudice as well as passive facilitation and active harm. Furthermore, perceived competence and warmth of women’s parental statuses may indirectly influence behaviors through emotions.

The Present Study

Using an SCM and BIAS map framework, I first assessed stereotype content (i.e., competence and warmth) for women’s parental statuses. Second, I examined relations between social structures (i.e., competition and status) and stereotype content for parental statuses. Third, I evaluated the emotions associated with stereotype content. A fourth aim was to assess associations between stereotypes and expected behaviors toward women’s parental statuses. Finally, I examined the mediating effect of emotions on stereotype → behavior relations. Thus I tested five hypotheses. (a) Women’s parental statuses would be ambivalently stereotyped, with mothers and involuntarily childless women rated as low in competence and high in warmth (LC-HW) and childfree-by-choice women rated as high in competence and low in warmth (HC-LW) (Hypothesis 1). (b) Social structures would predict stereotype content, such that warmth and competition would be negatively associated and competence and status would be positively associated across all parenting statuses (Hypothesis 2). (c) Emotions would vary across women’s parental statuses such that high warmth groups would evoke the most admiration, low warmth groups would evoke the most contempt, high competence groups would evoke the most envy, and low competence groups would evoke the most pity (Hypothesis 3). (d) Expected behaviors would vary toward women’s parental statuses, with warmth associated with active behaviors and competence associated with passive behaviors (Hypothesis 4a). Additionally, high warmth groups would evoke the most active facilitation, low warmth groups would evoke the most active harm, high competence groups would evoke the most passive facilitation, and low competence groups would evoke the most passive harm (Hypothesis 4b). (e) Finally, stereotype content would predict expected behaviors toward women’s parental statuses through emotions; thus, emotions would mediate stereotype → behavior relations (Hypothesis 5).

Method

Participants

Using the α = .05 criterion for achieving a 95 % chance of detecting a medium-sized effect by conducting a one-way ANOVA with three groups, a priori power analyses yielded a sample size estimate of 252 participants (Faul et al. 2007). Participants were 299 college students enrolled in psychology courses at Virginia Commonwealth University in Richmond, Virginia. The sample ranged in age from 18 to 50 years-old (M = 20.81, SD = 4.09, mdn = 20) and was predominantly female (70.9 %, n = 212), heterosexual (87.0 %, n = 260), and Christian (50.5 %, n = 151) or non-religious (28.4 %, n = 85). The most commonly reported race/ethnicity was White (45.2 %, n = 135), followed by Asian (18.1 %, n = 54), Black (17.4 %, n = 52), Hispanic/Latino (a) (8.7 %, n = 26), Multiracial (7.7 %, n = 23), and Other Race (3.0 %, n = 9). Political affiliations included Democratic (38.8 %, n = 116), Independent (15.7 %, n = 47), Republican (13.7 %, n = 41), Other Affiliation (25.5 %, n = 76), or No Affiliation (6.0 %, n = 18). The class rank was fairly evenly distributed across first-year students (33.4 %, n = 100), sophomores (20.7 %, n = 62), juniors (20.4 %, n = 61), and seniors (23.7 %, n = 71). Most participants were single (54.2 %, n = 162) or in a romantic relationship (29.1 % not cohabitating, n = 87; 12 % cohabitating, n = 36) and had no children (95.3 %, n = 285). The average size of participants’ family-of-origin was 4.34 people (SD = 1.89, range = 0–15). The majority of the sample intended to parent (73.2 %, n = 219), although some were uncertain about their parenting intentions (19.7 %, n = 59) or did not intend to parent (5.7 %, n = 17). Two (.7 %) participants with 20 % or more missing responses were excluded from analyses. There were no participant demographic differences among parental statuses groups, providing evidence of successful randomization of participants to conditions.

Procedure and Measures

Before commencing with data collection, approval was obtained from Virginia Commonwealth University’s Institutional Review Board. Participants were recruited from a research pool of undergraduate students enrolled in introductory psychology courses. Participants were first informed of the anonymous online study’s purpose and then provided their informed consent electronically. Next, participants were randomly assigned to rate cultural responses to one of three parental statuses (“mothers,” “women who are involuntarily childless,” “women who are permanently childfree-by-choice”) for stereotype content, social structures, elicited emotions, and expected behaviors. The study took approximately 15 min to complete, and participants received course credit for their participation. A demographic survey collected data on participants’ age, gender identity, year in school, political affiliation, racial identity, sexual orientation, relationship status, size of family-of-origin, current number of children, and intent to parent. Participants completed measures in the following order: stereotype content scales, social structures scales, elicited emotions scales, expected behaviors scales, and the demographic survey.

Competence and Warmth

Each participant rated a target group’s competence and warmth using the SCM’s competence and warmth scales (Fiske et al. 2002b). Six items each composed the competence (i.e., capable, competent, skillful, confident, intelligent, efficient) and warmth (i.e., trustworthy, warm, well-intentioned, sincere, friendly, good-natured) scales. To reduce socially desirable responding, examine cultural stereotypes, and remain consistent with previous uses of the instrument, each item prompted participants to rate a group using the following format: “As viewed by society, how [trait] are [one of the three parental statuses]?” Responses were provided on a 5-point Likert-type scale from 1 (not at all) to 5 (extremely), with higher averaged scores indicating more positive evaluations of the social group. Missing responses were extrapolated from the mean of non-missing responses. Previous studies provide reliability estimates ranging from α = .74 (de Paula Couto and Koller 2012) to α = .90 (Fiske et al. 2002b) for warmth, and α = .76 (de Paula Couto and Koller 2012) to .94 (Fiske et al. 2002b) for competence. The current study also found strong estimates of internal consistency for competence (α = .86) and warmth (α = .92).

Competition and Status

Also using SCM scales (Fiske et al. 2002b), participants rated one parental group on the social structures of competition and status. The status scale had two items assessing a group’s prestige and economic success, whereas the competition scale had two items assessing the degree to which a social group competes with the participants. Items in the status scale followed previously described SCM formatting and were “As viewed by society, how prestigious are the jobs of [one of the three parental statuses]?” and “As viewed by society, how economically successful are [one of the three parental statuses]?” However, the competition scale items deviated from this format and were: “Resources that go to [one of the three parental statuses] are likely to take away from the resources of people like me” and “If [one of the three parental statuses] get special breaks (such as preference in hiring), this is likely to make things more difficult for people like me.” Responses were again provided on a 5-point Likert-type scale from 1 (not at all) to 5 (extremely), with higher averaged scores indicating more of the construct being assessed. Previous studies’ reliability estimates ranged from α = .73 (Brambilla et al. 2011) to α = .79 (Cuddy et al. 2007) for competition and α = .75 (Brambilla et al. 2011) to α = .87 (Cuddy et al. 2007) for status. For the present study, alphas were .76 for competition and .87 for status.

Emotions

Participants rated one group’s parental status on four SCM emotions scales (Cuddy et al. 2007; Fiske et al. 2002a) with two items each: admiration (“admiring,” “proud”), contempt (“contemptuous,” “disgust”), envy (“envious,” “jealous”), and pity (“pity,” “sympathy”). Following Cuddy et al. (2007), the two items with the highest factor loadings in previous studies (Fiske et al. 1999, 2002b) were used in the current study. Emotion items followed standard SCM formatting. Ratings were made on a 5-point Likert-type scale from 1 (not at all) to 5 (extremely), with higher averaged scores indicated more of the assessed emotion. Previous reliability estimates are α = .80 for admiration, α = .60 for contempt, α = .82 for envy, and α = .71 for pity (Cuddy et al. 2007). Estimates of internal consistency using Cronbach’s alphas in the current study were .73 for admiration, .43 for contempt, .78 for envy, and .63 for pity. Because internal consistency for the contempt scale was in an unacceptable range (DeVellis 2012), analyses were conducted using each contempt item (contempt and disgust) separately. Both items were retained in analyses to maintain conceptual correspondence with original operationalization of the scale. The contempt scale is described as measuring both contempt and disgust as “…univalent, downward contrastive emotions…” (Cuddy et al. 2007, p. 634), with each leading to distinct behaviors (e.g., contempt leads to demeaning paternalism; disgust leads to attempts to eliminate, forcefully expel, and/or destroy a harmful stimulus). Thus, both contempt and disgust appear to be the primary constructs that this scale is intended to measure.

Expected Behaviors

Participants rated a group’s parental status on the four BIAS map behavior scales (Cuddy et al. 2007), each of which contained two items: active facilitation (“help,” “protect”), active harm (“fight,” “attack”), passive facilitation (“cooperate with,” “associate with”), and passive harm (“demean,” “exclude”). Items on the expected behaviors scales followed SCM formatting (e.g., “To what extent do people in society tend to [behavior] [one of the three parental statuses]?”) and used the same 5-point Likert-type scale from 1 (not at all) to 5 (extremely). Thus, participants reported how they expected others to behave in general toward women in one parental status. Previous estimates of reliability are α = .60 for active facilitation, α = .59 for active harm, α = .61 for passive facilitation, and α = .68 for passive harm (Cuddy et al. 2007). The current study estimated internal consistency with Cronbach’s alphas of .74 for active facilitation, .73 for active harm, .48 for passive facilitation, and .70 for passive harm. Because the estimate of internal consistency for the passive facilitation scale was in an unacceptable range (DeVellis 2012), analyses were conducted using each passive facilitation item (cooperate and associate) separately. Again, both items were retained in analyses to maintain conceptual correspondence with the original operationalization of the scale, which is described as “acting with” behaviors (Cuddy et al. 2007, p. 633). As with the contempt scale, there did not appear to be adequate theoretical justification for using one passive facilitation item over the other to represent the construct the scale is intended to measure.

Results

Preliminary Analyses

Because mine is the first use of these scales to assess perceptions of women’s parental status, separate item analyses on scales with low reliability (disgust/contempt and passive facilitation) were conducted to determine if these items actually represent conceptually distinct constructs, have differing levels of importance, and account for unique variance when assessing specific social groups based on women’s parental status rather than many social groups in aggregate. Analyses were also conducted to assess differences between female and male participants for the outcomes of interest. Participants identifying as “genderqueer” (n = 4, 1.3 %) or “other” gender (n = 1, .30 %) were excluded from these analyses due to insufficient sample sizes. According to an independent samples t-tests with a Bonferroni correction (p < .003), there were gender differences for passive harm, t (292) = 3.06, p = .002, d = .39; women (n = 212; M = 2.76, SD = .87) reported more passive harm than did men (n = 82; M = 2.41, SD = .94). Women’s and men’s ratings did not differ for warmth, competence, status, competition, admiration, contempt, envy, pity, active harm, active facilitation, association, and cooperation.

Regarding other demographic variables, there were no gender differences in age, size of family-of-origin, number of current children, class rank, political affiliation, racial identity, religious affiliation, sexual orientation, and parenting intentions. According to a chi-square test of independence, female and male participants (N = 281) differed for relationship status χ 2 (2) = 9.29, p = .01. Because of inadequate cell sizes (<5), participants who identified as married/partnered, separated/divorced/widowed, and “other” were excluded from these analyses. Women were more likely to be in a relationship (both cohabitating and not cohabitating), whereas men were more likely to be single. The effect size for this difference was small, Cramer’s V = .18.

Hypothesis Testing

Hypothesis 1

Hypothesis 1 predicted ambivalent stereotypes for parental statuses, with mothers and involuntarily childless women rated as LC-HW and childfree women rated as HC-LW. Two one-way analyses of variance (ANOVA) across the three parental statuses and three t-tests assessed Hypothesis 1, with a Bonferroni correction to account for increased Type I error rate associated with conducting multiple tests. Thus, the standard .05 alpha level was adjusted for five analyses, resulting in an alpha of .01. Competence and warmth met normality assumptions and had no outliers because no cases had standardized residuals greater than three standard deviations from the mean (i.e., ±3.29, Tabachnick and Fidell 2007). Competence met assumptions of homogeneity of error variances (p = .21), but warmth did not (p = .03). Thus, results from Welch’s ANOVA and Games-Howell post hoc tests, which are robust to violations of homogeneity of variances, are reported where appropriate.

Results of a one-way ANOVA (N = 297) indicated that parental statuses differed for competence, F (2, 294) = 13.62, MSE = .49, p < .001, ω2 = .08. In Tukey post hoc analyses, childfree women were rated as more competent than were mothers (p = .04, d = .36) and childless women (p < .001, d = .73; see Table 1). Mothers were also rated as more competent than were childless women (p = .02, d = .39). A second one-way ANOVA (N = 297) revealed that parental statuses also differed for warmth, Welch’s F (2, 191.11) = 54.23, MSE = .56, p < .001, ω2 = .24. In Games-Howell post hoc analyses, mothers were rated as warmer than were childfree women (p < .001, d = 1.38) and childless women (p < .001, d = 1.01), but childfree and childless women did not differ (p = .09, d = .31; see Table 1).

Table 1 Sample sizes, means, and standard deviations for all measures
Full size table

Paired sample t-tests assessed for ambivalent competence-warmth differences within each parental status. Competence and warmth were standardized for comparison, with standardized values reported here. Mothers (n = 102) were warmer (M = .77, SD = .73) than competent (M = −.05, SD = .90), t (101) = 11.14, SEM = .07, p < .001, d = 1.00; childless women (n = 96) were warmer (M = −.10, SD = .95) than competent (M = −.42, SD = 1.01), t (95) = 4.23, SEM = .08, p < .001, d = .33; and childfree women (n = 99) were more competent (M = .28, SD = .89) than warm (M = −.38, SD = .93), t (98) = −7.93, SEM = .08, p < .001, d = −.72. Thus, Hypothesis 1 was supported, because women in all parental statuses were ambivalently stereotyped. More specifically, mothers and childless women were stereotyped as warmer than competent, and childfree women were stereotyped as more competent than warm.

Hypothesis 2

Hypothesis 2 predicted that warmth and competition will be negatively associated, and competence and status will be positively associated. Pearson correlations assessed relations between stereotype content and social structures for Hypothesis 2. Status and competition met normality assumptions and had no outliers. Competition negatively correlated with warmth, and status positively correlated with competence (see Table 2). Correlations between competence and competition, and warmth and status were not significant. Thus, Hypothesis 2 was also supported, because competition was negatively associated with warmth, and status was positively associated with competence across all parental statuses.

Table 2 Correlations among all measures
Full size table

Hypothesis 3

Hypothesis 3 stated that HW groups (e.g., mothers and childless women) will evoke the most admiration; LW groups (e.g., childfree women) will evoke the most contempt; HC groups (e.g., childfree women) will evoke the most envy; and LC groups (e.g., mothers and childless women) will evoke the most pity. Five ANOVA’s with a Bonferroni adjusted alpha of .01 assessed Hypothesis 3. Due to the low reliability estimate for contempt, analyses were conducted using both contempt items (contempt and disgust) separately. All emotions met normality assumptions and had no outliers. Admiration and envy met assumptions of homogeneity of variances (p’s > .05), but pity (p = .03), the contempt item (p = .02), and the disgust item (p = .001) did not. Thus, results from Welch’s ANOVA and Games-Howell post hoc tests are reported for analyses involving pity, the contempt item, and the disgust item where appropriate. Demographic variables that accounted for significant variance in the outcome variables during preliminary analyses were controlled as covariates.

In a one-way ANOVA (N = 297), parental statuses differed for admiration, F (2, 294) = 57.49, MSE = .66, p < .001, ω2 = .28. In Tukey post hoc tests, mothers were admired more than were childfree (p < .001, d = 1.10) and childless women (p < .001, d = 1.48), but childless and childfree women did not differ (p = .053, d = .33; see Table 1). In a second ANOVA (N = 297) that controlled for political affiliation as a demographic covariate, parental statuses also differed for envy, F (2, 285) = 14.02, MSE = .74, p < .001, ω2 = .08. In Tukey post hoc tests, childfree women were envied more than were mothers (p = .02, d = .38) and childless women (p < .001, d = .74); mothers were also more envied than were childless women (p = .02, d = .38; see Table 1). In a third ANOVA (N = 297) that controlled for family-of-origin size as a demographic covariate, parental statuses differed for pity, F (2, 291) = 32.08, MSE = .80, p < .001, ω2 = .17. In Tukey post hoc analyses, childless women were pitied more than were mothers (p = .01, d = .42) and childfree women (p < .001, d = 1.11); mothers were also pitied more than were childfree women (p < .001, d = .77; see Table 1). Regarding analyses with separate contempt items, a one-way ANOVA (N = 297) revealed that parental statuses did not differ for contempt alone, Welch’s F (2, 294) = .90, MSE = .80, p = .41, ω2 = .00, but did differ for disgust alone, Welch’s F (2, 294) = 8.61, MSE = 1.15, p < .001, ω2 = .05 (see Table 1). Games-Howell tests revealed that childfree women elicited more disgust than did mothers (p < .001, d = .59), but not childless women (p = .07, d = .32); mothers and childless women also did not differ for disgust (p = .16, d = .26).

Hypothesis 3 was partially supported, because mothers (a HW, LC group) elicited the most admiration; childfree women (a LW, HC group) elicited the most disgust and envy; and childless women (a HW, LC group) elicited the most sympathy. Contrary to predictions, mothers elicited more admiration than childless women even though both groups were stereotyped as HW. Similarly, childless women elicited more pity than mothers even though both groups were stereotyped as LC. As predicted, childfree women elicited the most envy for their HC and the most disgust for their LW.

Hypothesis 4

Hypothesis 4a predicted that warmth will be associated with expected active behaviors and competence will be associated with expected passive behaviors. Hypothesis 4b further stated that HW groups (e.g., mothers and childless women) will evoke the most active facilitation; LW groups (e.g., childfree women) will evoke the most active harm; HC groups (e.g., childfree women) will evoke the most passive facilitation; and LC groups (e.g., mothers and childless women) will evoke the most passive harm. Five ANOVA’s with a Bonferroni adjusted alpha of .01 assessed Hypothesis 4. Due to the low reliability estimate of passive facilitation, analyses were conducted using both passive facilitation items (cooperate and associate) separately. All behavior tendencies met normality assumptions and had no outliers. Active facilitation, passive harm, and the single items cooperate and associate met assumptions of homogeneity of variances (ps > .05), but active harm (p = .004) did not. Thus, results from Welch’s ANOVA and Games-Howell post hoc tests are reported where appropriate. Demographic variables that accounted for significant variance in the outcome variables were controlled as covariates.

Warmth and competence both correlated with all expected behaviors: active facilitation, active harm, passive harm, the cooperation item, and the association item (see Table 2). In a one-way ANOVA (N = 299), parental statuses differed for active facilitation, F (2, 296) = 45.11, MSE = .62, p < .001, ω2 = .23. In Tukey tests, mothers elicited more active facilitation than did childless (p < .001, d = .81) and childfree women (p < .001, d = 1.35); childless women also elicited more active facilitation than did childfree women (p = .001, d = .52; see Table 1). In a second ANOVA (N = 299) that controlled for parenting intentions as a demographic covariate, parental statuses differed for active harm, F (2, 290) = 15.89, MSE = .68, p < .001, ω2 = .09. Tukey tests revealed that childfree women elicited more active harm than did mothers (p < .001, d = .73) and childless women (p < .001, d = .59); mothers and childless women did not differ for active harm (p = .69, d = .12; see Table 1). A third ANOVA (N = 299) indicated that parental statuses did not differ for passive harm, F (2, 296) = 4.35, MSE = .80, p = .01, ω2 = .02 (see Table 1). Regarding the separate passive facilitation items, a fourth one-way ANOVA (N = 299) revealed that parental statuses differed for cooperation, F (2, 296) = 8.65, MSE = 5.84, p < .01, ω2 = .05. Tukey tests revealed that mothers elicited more cooperation than did both childless (p = .02, d = .37) and childfree women (p < .001, d = .60); childless and childfree women did not differ for cooperation (p = .38, d = .19; see Table 1). A final one-way ANOVA (N = 299) revealed that parental statuses also differed for association, F (2, 296) = 6.39, MSE = 4.44, p = .002, ω2 = .78. Tukey tests revealed that mothers elicited more association than did both childless (p = .002, d = .51) and childfree women (p = .03, d = .37); childless and childfree women did not differ for association (p = .68, d = .12; see Table 1).

Hypothesis 4a was partially supported, because warmth and competence were associated with both passive and active behaviors. Hypothesis 4b was also only partially supported, because mothers (a HW, LC group) elicited the most active facilitation, cooperation, and association; childfree women (a LW, HC group) elicited the most active harm; and childless women (a HW, LC group) most often elicited behaviors similar to those elicited by childfree women despite being stereotyped similarly to mothers.

Hypothesis 5

Hypothesis 5 stated that stereotypes will have an indirect effect on behaviors through emotions. Mediation tests using ordinary least squares path analysis, 5,000 bootstraps, and 95 % biased-corrected confidence intervals (PROCESS, Hayes 2013) assessed the indirect effects of stereotype content on behaviors via emotions. According to this model, when a confidence interval does not include zero, a significant result is inferred. Because both competence and warmth significantly correlated with all behaviors (see Table 2), warmth was included as a covariate in competence → behavior analyses and competence was included as a covariate in warmth → behavior analyses. The parallel mediation model simultaneously assessed for specific indirect effects of all emotions on stereotype → behavior relations while accounting for associations between emotions (see Table 2). All variables met assumptions of normality, multicollinearity, and homoscedasticity and had no univariate or multivariate outliers.

The results of analyses examining the mediating effect of emotions on the relation between warmth and the five expected behaviors are presented in Fig. 1a Warmth indirectly affected active facilitation (95 % CI [.16, .42]) via admiration (95 % CI [.14, .30]) and pity (95 % CI [.04, .13]), but not contempt (95 % CI [−.00, .04]), disgust (95 % CI [−.01, .10]), or envy (95 % CI [−.02, .05]; see Fig. 1a). As warmth increased, so did admiration and pity, leading to more active facilitation. Comparing the size of the indirect effects of significant mediators revealed that admiration (.21) was a stronger mediator than was pity (.08). (b) Warmth also indirectly affected active harm (95 % CI [−.34, −.07]) via envy (95 % CI [−.08, −.00)] and disgust (95 % CI [−.26, −.10]), but not contempt (95 % CI [−.04, .00]), admiration (95 % CI [−.10, .03]), or pity (95 % CI [−.05, .02]; see Fig. 1b). As warmth decreased, envy and disgust increased, leading to greater active harm; however, disgust (−.18) was a stronger mediator than was envy (−.03). (c) Warmth indirectly affected passive harm (95 % CI [−.28, .01]) via pity (95 % CI [.02, .10]) and disgust (95 % CI [−.28, −.11]), but not contempt (95 % CI [−.05, .00]), admiration (95 % CI [−.11, .03]), or envy (95 % CI [−.05, .02]; see Fig. 1c). As warmth increased, pity increased and disgust decreased, leading to more passive harm; disgust (−.18) was a stronger mediator than was pity (.05). (d) Regarding analyses with passive facilitation behaviors, warmth indirectly affected cooperation (95 % CI [.05, .33]) via admiration (95 % CI [.03, .18]) and pity (95 % CI [.03, .12]), but not contempt (95 % CI [−.01, .03]), envy (95 % CI [−.04, .03]), or disgust (95 % CI [−.01, .11]; see Fig. 1d). As warmth increased, admiration and pity increased, leading to greater cooperation; admiration (.10) was a stronger mediator than was pity (.07). (e) Warmth also indirectly affected association (95 % CI [.10, .39]) via admiration (95 % CI [.02, .18]) and disgust (95 % CI [.02, .13]), but not contempt (95 % CI [−.03, .01]), envy (95 % CI [−.07, .00]), or pity (95 % CI [−.07, .01]; see Fig. 1e). As warmth increased, admiration increased and disgust decreased, leading to greater association; admiration (.09) was a stronger mediator than was disgust (.07).

Fig. 1
figure 1

Emotions as parallel mediators of the relation between warmth and expected behaviors, specifically a active facilitation, b active harm, c passive harm, d cooperation, and e association. The coefficient (β) is followed by standard error in parentheses. Arrows with solid lines indicate significant paths; dashed lines, nonsignificant paths. *p < .05. **p < .01

Full size image

The results of analyses examining the mediating effect of emotions on the relation between competence and the five expected behaviors are presented in Fig. 2a Competence indirectly affected active facilitation (95 % CI [−.26, .00]) via pity (95 % CI [−.13,-.03]), but not contempt (95 % CI [−.02, .01]), disgust (95 % CI [−.01, .03]), admiration (95 % CI [−.01, .11]), and envy (95 % CI [−.06, .02]; see Fig. 2a). As competence decreased, pity (−.07) increased, leading to less active harm. (b) Competence indirectly affected active harm (95 % CI [−.05, .21]) via envy (95 % CI [.004, .10]), but not contempt (95 % CI [−.01, .03]), admiration (95 % CI [−.04, .00]), disgust (95 % CI [−.07, .06]), and pity (95 % CI [−.02, .05]; see Fig. 2b). Increased competence led to greater envy (.04), which increased active harm. (c) Competence indirectly affected passive harm (95 % CI [−.21, .08]) via pity (95 % CI [−.10, −.01]), but not contempt (95 % CI [−.01, .03]), admiration (95 % CI [−.04, .00]), disgust (95 % CI [−.07, .06]), or envy (95 % CI [−.03, .06]; See Fig. 2c). As competence decreased, pity (−.05) increased, leading to greater passive harm. (d) Regarding passive facilitation behaviors, competence indirectly affected cooperation (95 % CI [−.04, .24]) via pity (95 % CI [−.13, −.03]), but not admiration (95 % CI [−.01, .05]), disgust (95 % CI [−.01, .03]), envy (95 % CI [−.05, .04]), or contempt (95 % CI [−.02, .01]; see Fig. 2d). As competence increased, pity (−.07) decreased, leading to reduced cooperation. (e) There was no indirect effect of competence on association (95 % CI [−.06, .23]) via admiration (95 % CI [−.00, .06]) disgust (95 % CI [−.03, .03]), contempt (95 % CI [−.01, .01]), envy (95 % CI [−.00, .08]), or pity (95 % CI [−.01, .06]; see Fig. 2e). In summary, Hypothesis 5 was generally supported, because at least one emotion mediated most stereotype → behavior relations.

Fig. 2
figure 2

Emotions as parallel mediators of the relation between competence and expected behaviors, specifically a active facilitation, b active harm, c passive harm, d cooperation, and e association. The coefficient (β) is followed by standard error in parentheses. Arrows with solid lines indicate significant paths; dashed lines, nonsignificant paths. *p < .05. **p < .01

Full size image

Discussion

In the present study, women of various parental statuses were ambivalently stereotyped, with mothers and childless women rated warmer than competent and with childfree women rated more competent than warm. As predicted, warm parental statuses were perceived as noncompetitive, and competent statuses were perceived as high status. Stereotype content of women’s parental statuses elicited specific emotional responses and expected behaviors from others. Mothers were the most admired group, eliciting facilitative behaviors; involuntarily childless women elicited pity; and childfree women elicited envy, disgust, and harm behaviors. Both warmth and competence predicted active and passive behaviors, but warmth was more predictive than was competence. Finally, stereotype content influenced most behaviors through at least one emotion, supporting the primacy of emotions over stereotype content in predicting behaviors.

Most of the study’s hypotheses were supported or partially supported by current findings. Ambivalent stereotyping of all parental statuses aligns with previous literature reporting that most subgroups of women are ambivalently stereotyped (Brambilla et al. 2011; Eckes 2002) and groups equivalent to mothers and involuntarily childless women are stereotyped similarly (Cuddy et al. 2004; Eckes 2002; Wade and Brewer 2006). Regarding relations between stereotypes and social structures, competitive parental statuses were perceived as cold and noncompetitive statuses were perceived as warm. Additionally, groups with high status were seen as competent, whereas those with lower status were seen as less competent. Previous SCM studies found similar results for status and competition relations across a variety of groups (Cuddy et al. 2007; Eckes 2002; Fiske et al. 2002b, 2007). Both competence and warmth promoted facilitation and inhibited harm in the current study, which contrasts with previous reports that warmth predicts active behaviors and competence predicts passive behaviors (Cuddy et al. 2007). In fact, warmth was generally a stronger predictor of behaviors than was competence, which is not without precedent in the literature (Becker and Asbrock 2012). Indeed, these results tentatively suggest that warmth may be a highly salient trait in perceptions of women’s parental status. This conclusion also aligns with previous findings that mothers and/or women who were perceived as future mothers were rated as warmer than were women without children or those who were not perceived as future mothers (Koropeckyj-Cox et al. 2007; La Mastro 2001).

Mothers, an ambivalently stereotyped low competence-high warmth group, elicited the most admiration and facilitation (i.e., helping, protecting, cooperating, associating), supporting previous research that high warmth groups receive admiration (Fiske et al. 2002a, b) and helping behaviors (Cuddy et al. 2007). However, mothers did not elicit any of the harm behaviors characteristically experienced by low competence groups (Cuddy et al. 2007). Perhaps mothers experience forms of passive harm other than the overt exclusion and demeaning measured by SCM scales, such as being unable to relate to others or being treated as insignificant. Although intending to parent does not indicate that the participants viewed mothers as the ingroup per se, mothers’ status as the most admired and helped group implies that participants may have viewed mothers as more similar to the ingroup than to the outgroup. Indeed, although most of the participants in my study did not currently have children, most intended to reproduce, which perhaps signifies that participants may have identified with parenthood as a potential future identity. Alternatively, participants may have viewed mothers as an admired outgroup, evoking unambivalently positive emotions despite ambivalent stereotyping.

The ambivalently stereotyped high competence-low warmth voluntarily childfree group elicited the most disgust, envy, and active harm (i.e., fighting, attacking). Because disgust contributes to contempt scales, this finding partially supports previous reports that low warmth groups who are responsible for their outcomes elicit contemptuous prejudice (Fiske et al. 2002a, b). However, disgust may be especially salient over contempt when considering women’s parental statuses. The envy and harm elicited by childfree women supports prior findings that high competence-low warmth groups evoke envious prejudice (Fiske et al. 2002a, b) and fighting and attacking behaviors (Cuddy et al. 2007). This also aligns with reports that childless women elicit harm behaviors, such as being denied voluntary sterilization by doctors (Gillespie 2000; Mollen 2006; Mueller and Yoder 1999) and given less desirable job expectations (Mollen 2006; Picard 1997). Yet, childfree women did not elicit the passive facilitation (i.e., cooperation, association) behaviors characteristic of high competence-low warmth groups in previous literature (Cuddy et al. 2007). In the current study, all helpful behaviors were associated with high warmth, and all harmful behaviors were associated with low warmth (similar to perceptions of obese individuals; Levine and Schweitzer 2015). Thus, passive-active distinctions found in studies assessing numerous and diverse social groups in aggregate may not be completely replicable when assessing specific social groups in isolation. Furthermore, permanently childfree-by-choice women may evoke ambivalent stereotypes (i.e., high competence-low warmth) and emotions (i.e., contempt/disgust and envy), but unambivalent harm behaviors.

Current findings regarding childfree women are also consistent with suggestions that people generally believe that raising children is vital to a fulfilling life experience (Eibach and Mock 2011a, b). However, Eibach and Mock (2011a, b) argue that parenthood is idealized or exaggerated as emotionally rewarding to rationalize the high economic, emotional, and psychosocial costs of actually raising children. Considering a childfree woman may prime the dissonance created by this paradox, increasing the likelihood of negative emotional and behavioral responses toward childfree women to resolve the dissonance. Eibach and Mock (2011a) also note that most people have some awareness of the financial costs of raising children. Thus, even nonparents might idealize parenthood and/or exaggerate its private rewards to justify an exploitative system in which all of society benefits from parents’ costly investment in children (Eibach and Mock 2011a, 2012). It is therefore conceivable that this dissonance might have occurred even in a sample of predominantly nonparents, prompting the observed disgust, envy, and harm behaviors toward childfree women.

That childfree women evoked disgust and harm behaviors might also be situated in the larger literature on disgust and moral cognition. Broadly, disgust is “…a basic negative emotion, the appraisal of which involves both a sense of offensiveness and revulsion accompanied by thoughts of contamination…[and] linked with a motivation to avoid and reject any perceived offensive entity” (La Rosa and Mir 2013, p. 223). Moral disgust, in particular, may be evoked by individuals, ideas, or behaviors that violate common moral and social norms, commit egregious social transgressions, and/or impose individual- or group-level costs or harm on society (Chapman and Anderson 2013; La Rosa and Mir 2013). Childfree women may have elicited disgust because they violate social and gender norms that prescribe caregiving and mothering for women. Furthermore, by not contributing to the next generation of laborers and tax-payers, childfree women may be perceived as imposing an unfair cost on society. Important to the study of childfree women, moral disgust can also be evoked by violations of bodily morals, defined as “…acts that offend categorical moral norms about what should or should not be done with the body and its products…” (Russell and Giner-Sorolla 2013, p. 330). As such, childfree women may be violating norms prescribing what they should be doing (e.g., reproducing) and producing (e.g., children) with their bodies, thus evoking an emotion akin to bodily moral disgust.

Indeed, current findings closely align with those of Ashburn-Nardo (2016), who assessed the association between moral outrage (i.e., feelings of anger, disapproval, disgust, and/or contempt that motivate retribution toward those who deliberately do something wrong or inflict harm) and responses toward childfree individuals. Childfree women and men in Ashburn-Nardo’s (2016) study were evaluated as less psychologically fulfilled than were parents, a relation that was explained (i.e., mediated) by increased moral outrage. Based on these results, Ashburn-Nardo (2016) concluded that parenthood remains a moral imperative in the United States and that violations of this prescription by childfree individuals elicit outrage. Findings of my study tentatively support this assertion.

Nonetheless, childfree women’s intentional decision to forgo the high costs associated with parenthood may have also evoked the envy observed in my study. Quintanilla and de López (2013) suggest that envy is an inherently social emotion involved in comparisons of the self to others. Envy occurs when one feels inferiority, resentment, or longing regarding the desired, but unequal possessions or success of another individual and the wish that this individual would lose their advantage (Quintanilla and de López 2013). Thus, envy toward childfree women may also be evoked by their typically high educational, occupational, and economic success (Abma and Martinez 2006; Lundquist et al. 2009; Martinez et al. 2012). According to Quintanilla and de López (2013), envy can also embody a combination of positive (e.g., admiration, praise) and negative (e.g., ill-will, hostility) components, which is reflected in the ambivalent stereotyping of childfree women as competent (i.e., positive) but cold (i.e., negative).

Findings were more mixed for involuntarily childless women, who were ambivalently stereotyped as warmer than competent as well as pitied. This aligns with previous findings that low competence-high warmth groups with little perceived control over their outcomes elicit paternalistic prejudice (Fiske et al. 2002a, b) and sympathy (Levine and Schweitzer 2015). This further aligns with reports in the childlessness literature that involuntarily childless women evoke reactions of sympathy (Kopper and Smith 2001) and are sometimes perceived more negatively than are mothers (Calhoun and Selby 1980; Callan 1985; Peterson 1983). However, in other domains, the involuntarily childless vacillated between being viewed similarly to mothers (e.g., LC-HW, eliciting little active harm), childfree women (e.g., eliciting little admiration, cooperation, and association), and somewhere in between the two (e.g., evoking moderate contempt, disgust, and active facilitation). Thus, involuntarily childless women may possess a contradictory mixture of characteristics common to both mothers (e.g., desire to have children, high warmth) and childfree women (e.g., not a mother, career-focused).

Regarding the indirect effects of stereotype content on behaviors, at least one emotion mediated nearly all stereotype → behavior relations. Positive emotions of admiration and pity mediated the warmth → active facilitation relation, supporting previous work (Becker and Asbrock 2012; Cuddy et al. 2007; Echebarria-Echabe 2013; Levine and Schweitzer 2015). Warmth had an indirect effect on active harm through envy and disgust, both supporting and contradicting findings by Cuddy et al. (2007). Additionally, admiration and pity mediated warmth’s effect on cooperation, whereas admiration and disgust mediated warmth’s effect on association. Finally, pity mediated warmth’s effect on passive harm. Equivalent relations between these and similar variables have been reported previously (Echebarria-Echabe 2013; Levine and Schweitzer 2015).

Several emotions mediated competence → behavior relations. Competence had an indirect effect on active facilitation through pity, which conflicts with several previous findings (Echebarria-Echabe 2013; Levine and Schweitzer 2015). In the current study, envy mediated the competence → active harm relation, aligning with Cuddy et al. (2007). However, significant competence → active harm relations contradict the primacy of competence in predicting passive behaviors and warmth in predicting active behaviors (Becker and Asbrock 2012; Cuddy et al. 2007). Regarding relations between competence and passive behaviors, pity mediated the influence of competence on passive harm, which mirrors past findings for this and similar variables (Cuddy et al. 2007; Echebarria-Echabe 2013; Levine and Schweitzer 2015) and contradicts others (Becker and Asbrock 2012). Additionally, the influence of competence on cooperation was mediated by pity, but no emotions mediated the relation between competence and association.

These findings do not align with the results of previous studies using comparable variables. For example, Cuddy et al. (2007) found that admiration and envy mediated the relation between competence and passive facilitation, whereas Echebarria-Echabe (2013) reported that respect and indifference mediated the influence of competence on cooperation. Furthermore, Becker and Asbrock (2012) found that competence and passive facilitation increased together, but this relation was unmediated by emotions. Thus, mediation findings that differ from previous research tentatively suggest that results differ when assessing specific groups of women based on parental status. Indeed, cooperation (e.g., working with others toward a common goal) and association (e.g., joining oneself with a partner, ally, friend, or companion) may carry different meanings in the context of women’s parental status.

Current findings regarding the mediating effects of emotions that differ from previous studies may also be due to methodological differences between studies, such as sample demographics and target groups. For example, the current study used an undergraduate sample in the United States, whereas other studies used a variety of undergraduate and community-based samples in the United States and other countries (Becker and Asbrock 2012; Cuddy et al. 2007; Echebarria-Echabe 2013; Levine and Schweitzer 2015). Similarly, the target group (s) in each study differed from the current study. For instance, Cuddy et al. (2007) studied ratings of up to 20 U.S. social groups in one study and of one fictitious ethnic group in another. Furthermore, Echebarria-Echabe (2013) assessed perceptions of countries, Levine and Schweitzer (2015) examined ratings of obese individuals, and Becker and Asbrock (2012) investigated perceptions of “yuppies” (young urban professionals) and elderly individuals. Thus, differences between current and previous findings may be due to the sampling technique and groups under study. In these varying circumstances, some emotions may be more salient than are others in mediating relations between stereotype content and behavioral outcomes. Yet, despite methodological differences between the current study and past studies, only a few studies had been conducted that could inform hypotheses in the current study.

Strengths of the Study

The current study has several strengths. Conceptually, it provides a multifaceted understanding of social evaluations of women’s parental status by assessing the cognitive, affective, and behavioral components of the SCM and BIAS map models. Not only does my study support the persistence of negative perceptions of women without children in U.S. emerging adults, but it also provides a tentative explanatory mechanism (i.e., emotions) for the influence of stereotypes on behavioral outcomes. By using Hayes (2013) PROCESS mediation model, refined and complex statistical analyses controlled for stereotype content as covariates and accounted for simultaneous emotions as mediators. In addition to being a powerful test of mediation, the PROCESS model detects and quantifies indirect effects without increasing the number of inferential tests (thereby reducing Type I error). Methodologically, random assignment of participants to conditions likely reduced the influence of individual differences on results, increasing my study’s internal validity. Additionally, participants reported on cultural perceptions of parental statuses instead of their own, which likely reduced social desirability bias when combined with anonymous online data collection.

Limitations and Future Research Directions

Several limitations should also be considered. Despite using a powerful multiple mediation model, causal relations among stereotype content, emotions, and behaviors cannot be conclusively established without further experimental studies. Also, reliability estimates in the current study for the contempt and passive facilitation scales were unacceptable; for pity, undesirable (DeVellis 2012). Several potential explanations for these low reliability estimates might be offered. First, original research using these scales examined responses to a variety of aggregated social groups rather than a few groups in isolation (Cuddy et al. 2007). In other words, the original SCM and BIAS map scales were developed and validated for use with a large number of targets. Thus, when examining just a few groups of women who vary by parental status, individual items in these scales (e.g., contempt and disgust) may not represent the singular construct (e.g., contempt) observed for aggregate groups. Indeed, disgust may be more salient than contempt in distinguishing women based on parental status. Second, some research suggests that low Cronbach’s alphas are often expected for scales with few items (Cortina 1993). Yet, conducting separate analyses on the items composing the contempt (i.e., disgust, contempt) and passive facilitation (i.e., cooperate, associate) scales conceivably changed the constructs being assessed, which limits the interpretability of findings involving these scales and renders a reliability estimate for these single items impossible. Thus, findings associated with these items must be interpreted with caution and carefully replicated in future studies.

Furthermore, the convenience sample of college students limits the generalizability of results to older, more homogeneous populations that may hold different or more unfavorable perceptions of women without children. Indeed, cultural proscriptions against pregnancy in young adults may promote decidedly negative attitudes toward pregnancy and parenthood in this age group (Frost et al. 2012; Miller 2011; Vasilenko et al. 2012). However, because these proscriptions might be expected to produce favorable perceptions of women without children, current findings demonstrate the pervasiveness of negative perceptions of childfree women.

Research should continue to explore and clarify stereotype content, emotional reactions, and expected behaviors based on women’s parental status. In particular, these studies should assess the specific negative outcomes that may be occurring in social, medical, and occupational settings due to relations among these variables. Future studies could benefit from more heterogeneous samples that include older, less politically liberal participants with varied sexualities, religions, educations, and gender identities. Moreover, scales validated to assess perceptions of childless women should be created and used in future studies. Finally, research should further explore the mechanisms underlying relations among stereotype content, emotions, and behaviors, such as how the saliency of warmth and competence influences perceptions.

Practice Implications

Previous research suggests that women without children experience negative interpersonal outcomes, such as stereotyping and stigmatization by others (Doyle et al. 2012; Gillespie 2000; Mollen 2006; Mueller and Yoder 1999; Rich et al. 2011; Shaw 2011). Studies also suggest that nonparents experience disparities in hiring and stipend decisions (Eby et al. 2004), employee scheduling (Mollen 2006; Picard 1997), the accessibility of voluntary sterilizations (Gillespie 2000; Mollen 2006; Mueller and Yoder 1999), and the allocation of medical resources (Wiseman 2007, 2010). Results of the current study corroborate these findings by demonstrating that stereotype content and emotional responses influence harmful behaviors toward childfree women. If the number of childfree women in the United States and other Western countries continues to grow, it will become increasingly important to establish how stereotype content and emotional responses influence those with the power to make the interpersonal, occupational, and/or medical decisions that affect childfree women. It will also be important to determine if one type of response (e.g., stereotype content vs. emotional response) influences some types of behavioral outcomes more than others (e.g., medical vs. occupational). Additionally, policymakers, business owners, and administrators may need to consider how institutional policies could subtly influence or be influenced by the stereotyping and negative emotional responses toward women without children that ultimately affect others’ behaviors.

Given the current finding that women without children elicited stereotyping, negative emotional reactions, and (in the case of childfree women) harm behaviors, mental health services might address the negative intrapsychic and interpersonal consequences of these social reactions. Results of my study support previous suggestions that mental health professionals should increase their awareness of cultural and personal biases regarding parenting, consider these biases in treatment planning and diagnosis, challenge and revise these biases, and validate the choice to be childfree with clients (Gold and Wilson 2002; Mollen 2006). For example, Mollen (2006) suggests that mental health professionals acknowledge that childfree populations are marginalized by traditional developmental theories that assume that parenting is a necessary component of adult growth. Furthermore, a number of authors advise that mental health professionals should empower women without children to use stigma management techniques in social situations (Mollen 2006; Park 2002). Other authors promote specific models and modalities for treating women or couples without children. For instance, Pelton and Hertlein (2011) outline a staged intervention model for mental health professionals working with married, heterosexual, childfree couples. Furthermore, Motherwell and Prudent (1998) discuss the use of group psychotherapy for treating the presenting concerns of both childless and childfree women. Finally, specific psychoeducational interventions might also be designed to reduce the stereotyping and negative emotional responses by others toward childless and childfree women.

Conclusions

Research across several decades suggests that women without children are perceived more negatively than are mothers (Dolińska 2014; Ganong et al. 1990; Jamison et al. 1980; Kopper and Smith 2001; Polit 1978; Vinson et al. 2010), a finding that is supported and extended in the current study using a contemporary sample of U.S. emerging adults. In my study, childfree women were perceived more negatively than both involuntarily childless women and mothers. In social cognition research, the SCM and BIAS map propose that a variety of social groups are subject to univalent and ambivalent stereotyping based on perceptions of competence and warmth and that these stereotypes are accompanied by specific emotional and behavioral reactions (Cuddy et al. 2007; Fiske et al. 2002a, b, 2007). In the first known study to assess perceptions of women’s parental status using both the SCM and BIAS map, women of various parental statuses were ambivalently stereotyped, with mothers and involuntarily childless women rated warmer than competent and with permanently childfree-by-choice women rated more competent than warm. These findings also extend previous work by demonstrating that when reasons for childlessness are inferred, involuntarily childless women are stereotyped more similarly to mothers than are childfree women. As predicted by previous research (Cuddy et al. 2007; Fiske et al. 2007), warm parental statuses were also perceived as noncompetitive, and competent statuses were perceived as high status.

Stereotyping of women’s parental statuses also elicited specific emotional responses and expected behaviors from others. Mothers were the most admired group, eliciting facilitative behaviors, whereas involuntarily childless women elicited pity. One of the most novel findings of my study was that childfree women elicited envy, disgust, and harm behaviors. These results may tentatively explain reports that parental status influences actual occupational (Mollen 2006; Picard 1997), medical (Gillespie 2000; Mueller and Yoder 1999; Wiseman 2010), and social (Doyle et al. 2012; Rich et al. 2011; Shaw 2011) outcomes for women without children. Indeed, both warmth and competence predicted active and passive behaviors, but warmth was more predictive than competence, suggesting that warmth may play a particularly salient role in perceptions of women’s parental status.

Furthermore, my study found that stereotype content influenced nearly all behaviors through at least one emotion, supporting previous reports of the primacy of emotions over stereotype content in predicting behaviors (Becker and Asbrock 2012; Cuddy et al. 2007; Echebarria-Echabe 2013; Levine and Schweitzer 2015). Despite several limitations in the current study, it is an important step in understanding the mechanisms underlying negative perceptions of women without children. As such, future studies should continue to examine the relations among stereotyping, emotional responses, and expected behaviors toward women based on parental status. Indeed, understanding associations among these variables and their impact on outcomes for women will become an increasingly important body of work as the number of women without children will likely continue to rise in developed countries.