Introduction

Social status has been extensively linked to stress and health outcomes: Having higher rank within a social hierarchy is linked to reduced stress exposure and increased control over the stressors experienced (Dohrenwend, 1973; Turner & Avison, 2003); better health behaviors (Nandi et al., 2014); better physical and mental health (Adler & Ostrove, 1999; Sapolsky, 2004, 2005); heightened well-being (Tan et al., 2020); and improved life expectancy (Gallo & Matthews, 2003; Imami et al., 2022). These differences occur in all societies, even those with access to universal healthcare (Adler et al., 1993). The subjective perception of one’s own social status has also been linked to reduced stress and improved health, controlling for objective indicators like income or education (Adler et al., 1994).

The association of perceived status with stress and health – even among groups with access to universal healthcare – points to the critical role of the psychology of status in determining stress and health outcomes, rather than objective aspects of status per se. Seminal work by Sapolsky (1990, 1991) indicated that the physiological and health benefits of higher rank – in this case, endocrine stress responses in wild olive baboons – were actually correlates of a style of rank attainment, marked by social skill, affiliation, and having social control and outlets for frustration. This work also delineated other psychological facets of social hierarchies – specifically, the stability of a hierarchy and other social contextual factors – that contribute to inconsistent status-health relationships in primate hierarchies (Sapolsky, 2004, 2005). However, research on human hierarchies has generally lagged in examining traits that may influence or underly the link between social rank and outcomes related to stress and health.

Examining dominance and prestige, two routes by which social rank can be earned in human groups (Henrich & Gil-White, 2001), may provide insight on the evolved links between status and health in human hierarchies, due to how these routes differ in the reliance on more forceful versus more pro-social tendencies to earn social status. Dominance is defined by the use of force, fear, or intimidation to take high-status positions. Prestige relies on the free conferral of status to individuals deemed competent, cooperative, and who foster goodwill among group members. These two routes are not mutually exclusive – both dominance and prestige have been linked to earned rank in human groups (Cheng et al., 2021; McClanahan et al., 2021). Dominance and prestige likely evolved as complementary, alternative means of differentiating rank that reflect humans’ reliance on both conflict-determined (dominance) and conferred (prestige) status (Cheng et al., 2013; Van Vugt & Smith, 2019). Two corresponding personality types – trait dominance and trait prestige – relate to the preference for the use of dominance or prestige to earn status within a group (Cheng et al., 2010).

One might expect trait dominance to be associated with increased stress and worse health based on several key features of the dominance construct. For example, trait dominance has been associated with heightened behavioral responses indicative of being more sensitive to socially-threatening stimuli (Case & Maner, 2014; Mead & Maner, 2012). Trait dominance is also associated with hostility – a characteristic negative and oppositional attitude toward others with a tendency to inflict harm (Smith et al., 2004) – and hostility is a key affective state that predicts poor cardiovascular health (Miller et al., 1996; Siegman et al., 2000). Hence, having higher trait dominance is associated with heightened sensitivity to threat and may result in increased conflict with competitors and colleagues.

Several physiological aspects of trait dominance may also impact stress and health outcomes. Trait dominance is associated with heightened and unhabituating cardiovascular responses to stressors and other social interactions (Lee & Hughes, 2014; Newton & Bane, 2001). Trait dominance, when coupled with exogenously elevated testosterone levels, is also linked with heightened cortisol and negative affect responses to acute stress (Knight et al., 2017). This evidence of rapid and robust responses to stress suggests a physiological profile that evolved to provide relatively quick and sustained physiological responses to support trait dominant behaviors (i.e., related to hostility and aggression). However, over the course of a lifetime of increased exposure and heightened reactivity to stressors, this physiology that supports dominance behavior could contribute to allostatic load and longer-term health consequences.

An individual who is higher in trait prestige may instead rely on sociability and building social relationships to earn status (Cheng et al., 2010), and such psychosocial processes are linked to better health outcomes (Cacioppo & Hawkley, 2003; Cohen, 2004; Hawkley et al., 2008; Holt-Lunstad, 2021; Holt-Lunstad et al., 2017; Uchino et al., 1996). Since prestige-based status acquisition relies on free conferral from group members, trait prestige is less likely to be associated with the contentious, stressful contests associated with dominance. Further, prior evidence indicates that prestige (and not dominance) is associated with higher subjective well-being and that experimentally-induced downward comparisons of one’s prestige increased subjective well-being compared to upward comparisons in an online task (Anderson et al., 2012)Footnote 1.

However, a strong theoretical connection between trait prestige and health is hampered by a lack of work examining physiological correlates of prestige. In one experiment, trait prestige, on its own or in an interaction with exogenous testosterone, did not predict cortisol responses to social evaluative stress in men, although trait prestige did correlate with reduced negative affect in this experimentFootnote 2 (Knight et al., 2017). Considering other physiological systems, some theorizing suggests the evolution of the parasympathetic nervous system may have been critical for the development of socially-binding behaviors inherent to human prestige hierarchies, such as social engagement and emotional and behavioral regulation (Lenfesty & Morgan, 2019; Porges, 2001, 2007). Robust parasympathetic nervous system activity has, in turn, been associated with extensive health benefits (Kemp & Quintana, 2013; O’Connor et al., 2021; Thayer et al., 2012; Williams et al., 2019). Further, anthropological work indicates that maintaining a prestigious position within a hunter-gatherer group (e.g., holding niche knowledge on hunting grounds) confers health benefits (Sugiyama & Sugiyama, 2003). In combination with evidence of the pro-sociality of trait prestige, this work hints at a possible relationship between trait prestige and better health outcomes.

In sum, prior research suggests social status predicts stress and health outcomes, but the routes by which status is earned could be linked to stress and health outcomes as well. As a first step in examining these routes and their potential relation to stress and health outcomes, the present study examined the differential associations of trait dominance and trait prestige with an array of self-reported stress and health outcomes across one exploratory and two confirmatory samples. By examining status-stress and status-health associations across the multiple routes by which social status can be earned, this work can inform and improve theoretical frameworks linking social status to stress and health. Any evidence that integrates individual differences and social context factors may stimulate the critical work needed to understand and perhaps improve individual health outcomes across societal hierarchies.

Method

This study’s approach consisted of a cross-sectional, self-report design across three studies. Initially, an exploratory study (Study 1) was conducted that included examination of relationships between trait dominance and prestige and outcomes related to stress and health. Study 1 also consisted of exploratory follow-up analyses exploring other dominance- and status-relevant self-report measures as potentially confounding covariates. This exploratory phase was followed by a pre-registered study (Study 2) that attempted to replicate the exploratory phase findings related to stress and health outcomes and included a pre-registered exploration of trait self-enhancement as a possible explanation for trait prestige’s links with stress and health. A third pre-registered study (Study 3) then attempted to replicate and extend the findings to attitudes and behaviors relevant to a global pandemic. Since the main goal of this report is to examine stress and health relationships with trait dominance and prestige, the principal results reported are the meta-analyses across all three studies rather than individual, study-specific results in a fashion that corresponds to this study timeline. Equivalence testing is used to provide insights on study-by-study similarities and differences and individual study results are reported in supplemental tables.

Self-reported stress and health outcomes were examined due to their ease of collection and because prior work has demonstrated strong associations between self-reports and biomarkers of physical health across age groups (Bauldry et al., 2012; Idler & Benyamini, 1997; Lekander et al., 2004; Okun & George, 1984; Tomten & Høstmark, 2007; Undén et al., 2007; Wasylkiw & Fekken, 2002). The order of presentation of the online questionnaires was randomized. All data, materials, and pre-registrations are available on the Open Science Framework (https://osf.io/xc8h3/).

Participants

Participants consisted of university students from the human research pools of two universities who provided informed consent and then completed a series of online questionnaires for class credit. Across all studies, a total of N = 1032 participants completed the study. In Study 1, n = 40 participants completed a version of the survey that was missing all demographic questionnaires (i.e., age, sex, and race/ethnicity). An additional n = 14 had sporadically missing data across all three studies. Hence, this left a total sample size with usable data of N = 978 (67.1% women, 35.4% people of color; see Table 1 for demographics and descriptive statistics of all measures in each study).

Table 1 Descriptive statistics [mean (SD) or percent of study sample] of key variables
Full size table

Measures

Dominance and Prestige Scale

Trait dominance and prestige was determined from the Dominance and Prestige Scale (Cheng et al., 2013). This scale consists of 17 items that measure the extent to which an individual obtains social status via force, fear, and intimidation (e.g., “I try to control others rather than permit them to control me.” Cronbach’s α = 0.804) versus appearing competent, using adept social skills, or via respect (e.g., “Members of my peer group respect and admire me.” Cronbach’s α = 0.807) on a scale from 1 (not at all) to 7 (very much).

Self-reported Stress and Health Outcomes

Perceived Stress Scale

The Perceived Stress Scale was used to gauge participant’s stress experienced in the past month. Participants responded to ten questions (e.g., “In the last month, how often have you felt difficulties were piling up so high that you could not overcome them?”) on a 0 (never) to 4 (very often) scale, which was averaged into a single score (Cronbach’s α = 0.853).

SF-36

The SF36 is a standardized, self-report measure of physical and mental health. The measure’s 35 questions produce eight subscales (a 36th question relates to change in health and is not included in a subscale) that can be aggregated into a summary physical health (subscales: physical functioningFootnote 3, role limitations due to physical health, bodily pain, and general health) and mental health (subscales: vitality, social functioning, role limitations due to emotional problems, emotional well-being) component summary scores. The physical and mental health component summary scores (PCS and MCS, respectively) were calculated as follows: A factor analysis of the eight subscale scores with oblique (“promax”) rotationFootnote 4 produced estimated weights for two factors. The estimated weights for each factor were multiplied by the z-score for each of the eight subscales and summed to produce the PCS and MCS for each participant (Farivar et al., 2007). See Table S1 for factor loadings.

Health and Life Satisfaction Items from the Midlife in the United States Study

Several self-reported health and well-being items found in the Midlife in the United States (MIDUS) study were used in the present study (Radler & Barry, 2014; Turiano et al., 2012). These items included individual questions related to the participant’s past, present, and future health and life satisfaction on a 0 (worst possible health/life) to 10 (best possible health/life); how much control the participant feels they have over their health and their life on a 0 (no control at all) to 10 (very much control) scale; and how much thought and effort they put into their health and life on a 0 (no thought or effort) to 10 (very much thought and effort) scale. The responses to these sets of questions were averaged into separate indices of overall self-reported health (Cronbach’s α = 0.762) and life satisfaction (Cronbach’s α = 0.722).

The present study also included MIDUS’ self-reported comparison of the participants’ overall health, memory, vision, and hearing to other people their own age on a 1 (excellent) to 5 (poor) scale. Values were reverse coded so that higher values indicated better health compared to peers and were averaged into a single measure of relative health (Cronbach’s α = 0.719).

Participants were also asked to report their sense of control over their health across six items (e.g., “Keeping healthy depends on things that I can do.”) rated on a 1 (agree strongly) to 7 (disagree strongly) scale. These values were reverse-coded so that higher values indicated greater control over one’s health and averaged into a single measure (Cronbach’s α = 0.592).

Participants self-reported the extent to which they experienced a list of physical symptoms (e.g., headaches, backaches, pain or aches in the extremities, etc.) in the past 30 days on a 1 (not at all) to 6 (almost every day) scale. The average response across all items was calculated (Cronbach’s α = 0.813).

Finally, participants responded to four questions related to how often they experience sleep problems (e.g., “have trouble falling asleep”). These questions were answered on a 1 [never (0 times)] to 5 [almost always (4 or more times per week)] scale and were averaged into a single item (Cronbach’s α = 0.732).

Lifestyles and Habits Questionnaire – Brief

The Lifestyles and Habits Questionnaire - Brief (LHQB) is a survey developed to examine self-reported physical, mental, and behavioral health in a young adult population (Dinzeo et al., 2014). The questionnaire consists of 42 statements across 8 subscales (exercise and physical health; psychological health; substance use; nutritional health; environmental health; social health; accident prevention; sense of purpose), answered on a scale from 1 (strongly disagree) to 5 (strongly agree). For the purposes of reducing multiple tests in this study, each mean subscale score was averaged into a single measure of young-adult health (mean subscale scores, Cronbach’s α = 0.708; see Supplemental Materials for analyses of individual subscales).

Positive and Negative Affect

Affect has been extensively linked to health outcomes, with higher positive affect and lower negative affect each independently predicting better health outcomes (Blazer & Hybels, 2004; Ostir et al., 2000, 2001; Steptoe et al., 2009). Participants reported their current positive and negative affect via the Positive and Negative Affect Schedule - Extended form. The general negative (Cronbach’s α = 0.928) and general positive subscales (Cronbach’s α = 0.909) consist of ten items each on a 1 (very slightly or not at all) to 5 (extremely) scale. The hostility subscale, encompassing six items (Cronbach’s α = 0.857), was also explored because of its association with trait dominance and negative health outcomes.

COVID-19 Measures

Study 3 aimed to replicate the initial effects evident in Studies 1 and 2 and, in pre-registered exploratory analyses, extend the findings to health behaviors relevant to the COVID-19 pandemic. Whereas the self-reported stress and health measures mostly deal in generalities, the COVID-19 questions were specific to the participant’s behavior during the global pandemic (specifically, in Fall 2020). Hence, Study 3 provided an opportunity to examine the association of trait dominance and prestige with specific health behaviors and attitudes with known consequences for morbidity.

COVID-19 specific attitudes and behaviors were measured using the surveys developed for the Understanding COVID-19 in America study (Kapteyn et al., 2020). This set of questions consisted of a set of yes/no responses to a checklist of risky and preventative behaviors; ratings of how effective and how safe certain behaviors were (e.g., how effective is wearing a mask, how safe is attending a gathering of 100 people); whether an individual had been exposed to or developed COVID-19; and the self-reported percent likelihood that an individual felt they would develop COVID-19. Indices of effectiveness (Cronbach’s α = 0.799) and safeness (Cronbach’s α = 0.801) were calculated by averaging individual ratings. Per previous approaches (Kim & Crimmins, 2020), sets of the most preventative and riskiest behaviors within the set of full questions were examined. Pre-registered hypotheses included an expectation that trait prestige would relate to better health behaviors during the pandemic (e.g., stronger ratings of effectiveness of prophylactic measures) and an exploratory aim of examining the association of trait prestige and dominance with self-reported COVID19 morbidity without an a priori hypothesis of direction of effect.

Covariates

Principal Covariates

All models contained participant’s age, gender (0 = woman; 1 = man), and race/ethnicity (1 = person of color, 0 = white) as covariates. These variables were chosen due to their relevance to stress and health.

Exploratory Covariates

In specific follow-up models, sets of exploratory variables were added to control for and attempt to rule out confounding concepts or alternative explanations. These groups of conceptually related variables were chosen for reasons described below.

Objective and Subjective Indices of Socioeconomic Status

Prior work linking social status to stress and health has mostly relied on objective and subjective indices of socioeconomic status. Similar measures were included here to examine links between dominance, prestige, stress, and health, controlling for these previously known associations with socioeconomic status. To index objective socioeconomic status, participants self-reported their mother’s education (chosen to match prior work that has traditionally ignored paternal education) on a scale from 1 (some high school or less) to 7 (Ph.D. or professional degree) and their mother and father’s annual income on a roughly logarithmic scale. Participants also responded to two versions of the MacArthur ladder scale of subjective socioeconomic status (Adler et al., 1994). In one version, participants reported their position on the ladder in relation to the United States and, in the other, participants reported their position in relation to their community. Scores on each ladder scale were mean averaged into a single value.

Other Status-Relevant Traits

The Dominance and Prestige scale was the principal scale of interest due its ability to capture the two, evolved routes by which status can be earned in human groups. However, other scales exist that measure traits relevant to earning status. Two scales were included as covariates to explore the specificity of any stress and health outcomes to dominance and prestige. The Sense of Power (SOP) scale measures the feeling of possessing the ability to control outcomes in one’s social relationships (Anderson et al., 2012a). The scale consists of ten items (e.g., “In my relationships with others… I think I have a great deal of power.”) answered on a 1 (disagree strongly) to 7 (agree strongly) scale (Cronbach’s α = 0.836). The Personality Research Form (PRF) Dominance scale (Jackson, 1984) consists of sixteen true or false questions (Cronbach’s α = 0.807) that infer dominance by estimating one’s desire to hold high-ranking, powerful positions (e.g., “I would like to be a judge.”) and one’s ability to lead and sway others (e.g., “I am quite effective in getting others to agree with me.”).

Self-enhancement (Study 2)

The prestige route to status depends on conferral of respect from group members for an individual being competent and well-liked (Cheng et al., 2013). Individuals who are more motivated to seek positive appraisals from others (i.e., individuals high in trait prestige) may also respond to questionnaires in a desirable rather than accurate manner. Using data from Study 2, links between dominance, prestige, stress, and health were examined while covarying for self-reported, self-enhancing tendencies.

Self-enhancing tendencies were indexed in Study 2 via two measures: The Balanced Inventory of Desired Responding (BIDR; (Paulhus, 1984) and the Health Behavior Self-Enhancement scale (HBSE; (Christian, 2017). The BIDR consists of twenty questions on a scale from 1 (not true) to 7 (very true) that relate to some common self-focused attributes (e.g., “My first impressions of people usually turn out to be right.”). For the purposes of exploring self-enhancement within these models, the total score calculated from the mean averaged across all items was examined (Cronbach’s α = 0.810).

The HBSE consists of a list of six healthy and six unhealthy foods and supplements (e.g., “Vitamin C” and “fried foods”). Participants reported how often they consumed these items compared to an average person of their same age and gender on a scale from − 5 (very much less) to + 5 (very much more). The responses to unhealthy items were recoded so that higher values indicated much less usage of the item. Higher scores were then taken to indicate higher levels of self-enhancement related to these health behaviors. All responses were mean averaged to form a single scale (Cronbach’s α = 0.796).

Analyses

The main approach for analyzing the data in each study was to construct linear regression models with a given self-reported stress or health outcome regressed on trait dominance, trait prestige, and the covariates age, gender, and race/ethnicity. For example, a model examining the effects of dominance and prestige on perceived stress was constructed as follows:

$${PSS}_{i} = {b}_{0}+{b}_{1}{Dominance}_{i }+{b}_{2}{Prestige}_{i }+{b}_{3}{Age}_{i }+{b}_{4}{Gender}_{i }+{b}_{5}{Race}_{i }+ {e}_{i}$$
(1)

All variables (with the exception of gender and race/ethnicity) were z-score normalized within each study.

Meta-Analyses

Because the approach was equivalent across studies (i.e., using the same questionnaires), the principal results reported are meta-analyses of each individual regression model across the three studies. To examine the meta-effects, individual regression coefficients for the effects of trait dominance and trait prestige were extracted, converted to Pearson’s r correlations, and then to Fisher’s Z. Restricted maximum-likelihood meta-analyses with random effects by study were run for each set of Fisher Z coefficients linking trait dominance or trait prestige to a stress and health outcome. A 95% confidence interval (95%CI) that does not contain zero was considered evidence that a meta-effect was non-zero. Reported results were converted back to Pearson’s r for ease of interpretation.

Equivalence Testing

To examine the replicability of the results from the exploratory to the confirmatory phase, equivalence testing was performed for the associations between trait dominance or prestige and each of the stress or health outcomes using Anderson and Hauck’s (1983) approach for regression coefficients. This approach produces a single p-value, which indicates whether two regression coefficients can be considered statistically equivalent within some determined effect size (Counsell & Cribbie, 2015). Specifically, if a given association in Study 2 was found to be within a small-moderate effect size (δ = 0.5) of the same association found in Study 1, the effect was deemed equivalent. For Study 3, the equivalent testing comparison was made between effects found in Study 3 data and the effects evident from all available, prior data (i.e., the effects evident when using data collected in Study 1 and Study 2).

Exploratory Covariates

In exploratory follow-up analyses, covariates were added – that is, SES and other status-relevant traits in Study 1 and self-enhancement in Study 2 – to examine the effects of trait dominance and prestige, controlling for these other relevant constructs. Each set of covariates (i.e., SES, the status-relevant traits, or the self-enhancement variables) was added simultaneously for each given model. For example, the model exploring the association of trait dominance and prestige with perceived stress while controlling for other status-relevant traits was constructed as follows:

$${PSS}_{i} = {b}_{0}+{b}_{1}{Dominance}_{i }+{b}_{2}{Prestige}_{i }+{b}_{3}{Age}_{i }+{b}_{4}{Gender}_{i }+{b}_{5}{Race}_{i }+ {b}_{6}{SOP}_{i }+{b}_{7}{PRF}_{i }+ {e}_{i}$$
(2)

Results are reported with a focus on the difference in strength and significance of the result with and without controlling for the exploratory covariates.

Moderation by Gender

In post-hoc analyses, gender was explored as a moderator of the effects of dominance and prestige on each of the stress and health outcomes. Prior work indicates that the criteria for earning social status differ for men and women, with evidence suggesting men tend to rely on and be rewarded for behaviors related to dominance more so than women do (Buss et al., 2020; Hays, 2013). It is unclear if this gender difference impacts dominance and prestige’s associations with stress and health. Thus, moderation by gender was explored via meta-analyses with all available data to maximize power. Each outcome variable was regressed on the interactions of gender with trait dominance and prestige and the main effects of age and race/ethnicity as covariates. The meta-analyses were run as described previously but relied on extracting the coefficients representing the interaction terms.

Multiple Comparisons Correction

To correct for multiple comparisons, the Benjamini-Hochberg false discovery rate (FDR) correction was applied to each study’s set of principal results. Specifically, tests linking trait dominance and trait prestige to the array of stress- and health-relevant variables within a study were included in three sets of FDR correction. A conservative cutoff of q < 0.05 was taken as evidence of a non-zero effect when considering the multiple tests performed within each study. Because the principal analyses are focused on the meta-effects, the FDR correction results for each study are reported in the supplemental materials (Table S23).

Results

The results are organized by first reporting the meta-effects across the three studies. Next, the equivalence testing is reported, followed by the exploratory covariate analyses related to socioeconomic status and other status-relevant traits (Study 1), and then exploratory covariate analyses related to self-enhancement (Study 2). The COVID-specific outcomes are then reported (Study 3), followed by post-hoc exploratory analyses of moderation by gender.

Meta-Effects of Trait Dominance and Prestige

Self-Reported Stress

Based on meta-analyses of linear regression models that controlled for age, sex, and race/ethnicity, trait prestige was negatively associated with perceived stress (Pearson’s r = 0.299, 95%CI[-0.374, -0.221], p < 0.001) while trait dominance was positively associated with perceived stress (r = 0.126, [0.054, 0.196], p < 0.001; Fig. 1; see Table S2 for models from each individual study).

Self-Reported Health

SF-36

Examining the physical and mental health composites from the SF-36, trait prestige was linked with better physical (r = 0.232, [0.115, 0.342], p < 0.001) and mental health (r = 0.353, [0.297, 0.407], p < 0.001), controlling for sex, age, and race/ethnicity (Tables S3 and S4). In the same models, trait dominance was significantly linked with poorer physical health (r = -0.168, [-0.229, -0.107], p < 0.001) and poorer mental health (r = -0.074, [-0.136, -0.011], p = 0.021). Neither trait dominance (r = 0.037, [-0.100, 0.025], p = 0.243) nor prestige (r = 0.026, [-0.037, 0.089], p = 0.414) were associated with self-reported change in health (for better or worse) in the past year (Table S5).

MIDUS Questionnaires

The health and life questionnaires from the MIDUS study revealed similar patterns as the SF-36: Higher trait prestige was associated with better self-reported physical health (r = 0.350, [0.293, 0.404], p < 0.001), higher life satisfaction (r = 0.463, [0.412, 0.511], p < 0.001), better health compared to others (r = 0.292, [0.233, 0.348], p < 0.001), lower symptoms reported (r = -0.187, [-0.266, -0.107], p < 0.001), fewer sleep problems (r = -0.134, [-0.206, -0.059], p < 0.001), and greater reported control over one’s health (r = 0.365, [0.200, 0.511], p < 0.001; Tables S6-S11). Trait dominance was associated with poorer physical health (r = -0.109, [-0.170, -0.046], p < 0.001), poorer life satisfaction (r = -0.089, [-0.151, -0.027], p = 0.005), more symptoms reported (r = 0.166, [0.092, 0.237], p < 0.001), and less control over one’s health (r = -0.191, [-0.285, -0.094], p < 0.001); trait dominance was not associated comparative health (r = -0.044, [-0.106, 0.019], p = 0.174) or sleep problems (r = 0.027, [-0.048, 0.102], p = 0.474).

Fig. 1
figure 1

The effects of trait dominance and prestige on self-reported stress and health outcomes. Note: Effects plotted here are Fisher’s z; results reported in text are converted to Pearson’s r. PSS = Perceived Stress Scale; LHQB = Lifestyles and Habits Questionnaire – Brief

Full size image

LHQB

Similar patterns were evident when health was measured via the young-adult-focused health measures (Table S12). Specifically, after controlling for sex, age, and race/ethnicity, trait prestige was associated with better young-adult health (r = 0.478, [0.355, 0.584], p < 0.001) and trait dominance was associated with poorer young-adult health (r = 0.478, [0.355, 0.584], p < 0.001).

Affect

Trait prestige was associated with higher positive affect (r = 0.273, [0.123, 0.412], p < 0.001) and lower negative affect (r = -0.312, [-0.382, -0.238], p < 0.001), controlling for sex, age, and race/ethnicity (Tables S13-S14). In the same models, trait dominance was not significantly associated with positive affect (r = 0.046, [-0.018, 0.110], p = 0.157) and was associated with higher negative affect (r = 0.205, [0.143, 0.266], p < 0.001). Given prior associations of trait dominance with hostility and hostility with health, follow-up exploratory analyses focused on self-reported hostility. In these follow-up analyses, trait prestige was associated with lower hostility (r = -0.299, [-0.356, -0.240], p < 0.001) and trait dominance was associated with higher hostility (r = 0.315, [0.256, 0.371], p < 0.001).

Equivalence Testing

Given that data collection occurred across three studies (one exploratory and two pre-registered, confirmatory studies), it is important to determine whether the effects could be considered equivalent across studies. Further, despite robust meta-effects evident, some heterogeneity was also evident in the significance of the effects across the three studies (see supplemental tables).

Study 2 (First Confirmatory Study) vs. Study 1

Equivalence testing of the regression coefficients (Anderson & Hauck, 1983; Counsell & Cribbie, 2015) revealed that the effects evident in Study #2 were statistically equivalent for links between trait prestige and SF36 mental health and change in health; the MIDUS health, life, physical symptoms, and sleep questionnaires; the LHQB measures of young-adult health, and trait negative affect [i.e., just over half (54%) of effects tested; Table 2]. The remaining effects – perceived stress, SF36 physical health, and trait positive affect – could not be considered statistically equivalent, despite matching the direction and statistical significance of the exploratory effects. This degree of inequivalence might be reasonable given the difference in sample size: The Study 1 exploratory phase was a smaller sample and more susceptible to inflation of effect size. Indeed, Study 2’s effects were all smaller than Study 1, with the exception of trait positive affect, which was larger in Study 2. All tests associating trait dominance with self-reported stress and health outcomes were statistically equivalent between Studies 1 and 2.

Table 2 Equivalence testing results
Full size table

Study 3 (Second Confirmatory Study) vs. Studies 1 and 2

The effects evident in Study 3 were statistically equivalent to effects from models based on the entirety of data available from Studies 1 and 2, with the exceptions of trait prestige’s association with SF36 physical health (but see footnote about missing data in this study), the MIDUS sense of control over one’s health, and with the exceptions of trait dominance’s association with the MIDUS sense of control over one’s health (i.e., 73% of prestige effects and 91% of dominance effects tested were equivalent). In each of these exceptions, the effect evident in Study 3 was in the same direction as the effects evident in the combined data of Studies 1 and 2 and, for the exceptions related to trait prestige, were statistically significant in all studies.

In sum, meta-analyses revealed evidence of robust associations of trait dominance and prestige with self-reported stress and health. These meta-analyses revealed weaker effects for trait dominance (range of absolute values of significant Pearson’s r correlations = [0.074, 0.314]; see also Fig. 1 and supplemental materials for evidence of sporadic non-significant effects in smaller samples within a given study) compared to trait prestige (|r| = [0.134, 0.478]). Using equivalence testing, a majority of statistical effects (79%) were found to be equivalent between studies. Further, the associations of trait dominance and prestige with stress and health outcomes demonstrated a pattern of similar direction and statistical significance between the exploratory (Study 1) and confirmatory phases (Studies 2 and 3).

Exploratory Covariate Analyses

Objective and Subjective Socioeconomic Status (SES)

In Study 1, several covariates were explored to examine the specificity of the hypothesized associations of trait dominance and prestige with stress and health. Hence, models were re-run that included other, traditional measures of objective (mother’s education, family income) and subjective socioeconomic status (the MacArthur ladder scales for community and US). Including these measures did not meaningfully impact the results or interpretations of the initial findings of trait prestige’s positive associations with stress- and health-relevant outcomes (e.g., a median 3.3% change in effect size across all effects examined) and trait dominance being negatively associated (or unassociated in Study 1) with health-relevant outcomes (a median 1.0% change in effect size; Table S15).

Other Status-Relevant Traits

An initial check of correlations among the status-relevant trait surveys in Study 1 revealed evidence that the SOP scale was more strongly correlated to trait prestige (r = 0.633, p < 0.001) than dominance (r = 0.259, p < 0.001) and that the PRF Dominance scale was more strongly correlated with trait dominance (r = 0.539, p < 0.001) than prestige (r = 0.346, p < 0.001; see supplemental materials, Fig. S4). The strong prestige-SOP correlation is in line with prior work (Anderson et al., 2012).

The positive association between trait prestige and positive affect (B = 0.178, [0.002, 0.359], t(173) = 1.941, p = 0.054) and the negative association between trait dominance and negative affect (B = 0.133, [-0.024, 0.290], t(173) = 1.659, p = 0.099) were rendered non-significant by the inclusion of SOP and PRF in Study 1’s data. Otherwise, entering SOP and PRF as covariates did not substantially alter the interpretations of the associations of trait dominance (e.g., median 2.2% change in effect size) and prestige (median 1.8% change in effect size) with stress and health outcomes (Table S16).

Self-Enhancement

In pre-registered exploratory analyses, Study 2 focused on self-enhancement as a possible explanation for the patterns of results evident in Study 1. Hence, two measures of self-enhancement were included in models with trait dominance and prestige, as well as sex, age, and race/ethnicity. Models that included the self-enhancement measures as covariates revealed slightly weaker (e.g., approximately a 6% reduction in effect sizes) but still statistically significant relationships between trait prestige and lower stress, better health, higher positive affect, and lower negative affect (Table S17). This suggests that self-enhancement may only partially account for the relationship between trait prestige and self-reported health-relevant outcomes; direct associations between trait prestige and health are still evident after controlling for self-enhancement motives (see Supplemental Results for additional mediation analyses).

Study 3: Replication and Extension to COVID-relevant Health Behaviors

Study 3 aimed to replicate the initial effects evident in waves 1 and 2 and extend the findings to health behaviors relevant to the COVID-19 pandemic. Whereas the self-reported stress and health measures mostly deal in generalities (e.g., “I avoid binge drinking” in the LHQB or “In general, would you say your health is…” in the SF36), the COVID-19 questions were specific to the participant’s behavior during the global pandemic (specifically, Fall 2020). Hence, Study 3 provides an opportunity to examine cross-sectional associations of trait dominance and prestige with specific health behaviors and attitudes with known consequences for morbidity.

Full results for the COVID19-relevant behaviors are reported in the supplemental materials. Overall, trait dominance and prestige were mostly unrelated to self-reported behaviors and perceptions relevant to the global pandemic (Table S18-S19). Higher trait dominance was significantly associated with higher mean ratings of the safety of pandemic-relevant behaviors (B = 0.052, [0.006, 0.097], t(281) = 2.235, p = 0.026, pFDR = 0.105), such as the relative safety of attending a gathering of 100 or more people. This effect does not survive FDR correction, suggesting the effect may not be robust. Neither trait dominance nor prestige were associated with self-reported COVID19 exposure, worry about or the self-reported likelihood of contracting the disease, or self-reported COVID19 morbidity (i.e., either a positive test or a medical diagnosis with no test; Tables S20-S21).

Moderation by Gender

Although not pre-registered, gender was explored as a moderator of the effects of dominance and prestige on each of the stress and health outcomes using a meta-analytic approach to maximize power. Results indicate few, if any, robust moderating effects of gender (Table S22). There was weak meta-analytic evidence that gender moderated the association of trait dominance and prestige on the MIDUS measure related to sense of control over one’s health. In each case, the associations of trait dominance and prestige were somewhat blunted for women compared to men. However, these were the only two non-zero results out of twenty-eight meta-analyses tested. Hence, the overall interpretation of these results is a lack of robust moderation by gender on the associations between trait dominance and prestige with stress and health.

Supplemental Results

Analyses reported in the Supplemental Results include factor loadings for the SF-36 survey (Table S1); full mediation models for self-enhancement as mediator; alternative analyses with the original, pre-registered sample size in Study 2 (Fig. S1); analyses of individual facets of the LHQ-B (i.e., the young adult health questionnaire subscales; Fig. S2); p-values FDR-corrected for multiple comparisons within each study (Table S23); and tables for each regression model in each study.

Discussion

This report examined trait dominance and prestige – traits indicative of preference for two routes by which status is earned – and their association with self-reported stress and health. In wide ranging work, higher social status has been associated with reduced stress and better health (Adler et al., 1994; Adler & Ostrove, 1999; Dohrenwend, 1973; Gallo & Matthews, 2003; Sapolsky, 2004, 2005; Tan et al., 2020; Turner & Avison, 2003). The present results suggest these health benefits of social status may be more evident for individuals who rely on prestige to earn status, and not dominance. Indeed, relatively smaller, negative impacts on self-reported stress and health were consistently observed among individuals higher in trait dominance using a meta-analytic approach. These associations of trait dominance and prestige with self-reported stress and health were evident even when controlling for socioeconomic status indicators, other status-relevant traits, and one possible psychological explanation, self-enhancement motives.

Despite the strong associations of trait dominance and prestige with self-reported stress and health, the results linking trait dominance and prestige to specific health behaviors relevant to a global pandemic were not robust. As such, it remains unclear if trait dominance and prestige may impact health behaviors relevant to communicable disease. A limiting factor in this aspect of the present study was its cross-sectional design and reliance on self-reported prevalence. Future work could improve on these findings by examining medically-determined disease incidence in a sample over time.

This work follows from a broad literature linking personality traits to health outcomes (Smith, 2006), as well as work examining personality traits relevant to the concepts of dominance and prestige. For example, the so-called Type A and Type D personality types share a focus on aggressive tendencies and negative affect (respectively) with trait dominance and each has been linked to poor cardiovascular health (Denollet, 2005; Friedman & Rosenman, 1959). Later work showed Type A personality was primarily linked to negative health outcomes via its association with hostility (Miller et al., 1996; Smith et al., 2004). Another personality concept, the interpersonal circumplex, organizes personality along the orthogonal concepts of agency (desiring individuation, power) and affiliation or communion (desiring connectedness, friendliness; Gurtman, 2009). The concepts of agency and affiliation somewhat overlap with dominance and prestige, respectively, and have been associated with health outcomes (Gallo & Smith, 1998; Smith et al., 2010). Other traits that are relevant to trait dominance and prestige and that may explain stress and health outcomes include narcissism and self-esteem (Cheng et al., 2010; Johnson et al., 2012). However, Types A and D, the interpersonal circumplex, and other traits lack the explicit links to social status and social hierarchies evident in the concepts of trait dominance and prestige. Uniting relevant theory from the personality and social status fields may provide new insights on the links between the individual, their place in the social hierarchy, and health.

Possible Mechanisms

Physiological Mechanisms

Physiological research focused on dominance provides one possible explanation for the associations of trait dominance with poorer health. Trait dominance has been linked to heightened vascular constriction [indexed by total peripheral resistance (TPR)] in response to social threat, a response that does not tend to habituate to subsequent stress exposure (Lee & Hughes, 2014). Higher TPR increases blood pressure and is linked to risk of cardiovascular events and mortality in healthy and hypertensive populations (Fagard et al., 1996; Mensah et al., 1993). Within a laboratory setting, individuals who are high in trait dominance also demonstrated an exaggerated cardiovascular response to incongruencies between expected and actual status positions (Gramer & Schön, 2015). The links between trait dominance and physiology mirror evidence of heightened physiological responses among dominant, wild primates, who show heightened cardiovascular reactivity to sympathetic stimulation (Sapolsky & Share, 1994). Critically, these non-human primate associations between dominance and physiology may exist, in part, because of personality traits that underlie dominance rank in non-human primate hierarchies (Sapolsky, 1990, 1991). These physiological responses likely evolved to provide robust, sustained reactions to stress or threat. Such threat reactions provide metabolic energy needed to fight at the expense of increased allostatic wear and tear over longer term, repeated activations (McEwen & Gianaros, 2010).

The seeming benefits of trait prestige for stress and health may, in turn, be physiologically explained by the parasympathetic nervous system. In polyvagal theory (Porges, 2001, 2007), parasympathetic nervous system activity originating in the nucleus ambiguus, and perpetuated via the supradiaphragmatic vagus nerve, provides a physiological brake on sympathetic-driven “fight or flight” behaviors, which presumably reduces physiological wear and tear and allostatic load. Within polyvagal theory, parasympathetic activity also facilitates pro-social behaviors and social engagement (Porges, 2001, 2007). Later work adapted the polyvagal theoretical framework, arguing that parasympathetic-mediated behaviors were necessary for the evolution of human prestige hierarchies (Lenfesty & Morgan, 2019). Indeed, parasympathetic activity has been linked to behaviors relevant to prestige based hierarchies – including social engagement, emotional and behavioral regulation, and compassion, among other pro-social behaviors (Bornemann et al., 2016; Miller et al., 2016; Roos et al., 2017; Segerstrom et al., 2011; Smith et al., 2020) – but parasympathetic activity has not been linked explicitly to trait prestige. Such a relationship could mediate associations between trait prestige and health, given the extensive work linking parasympathetic activity to better health outcomes (Kemp & Quintana, 2013; O’Connor et al., 2021; Williams et al., 2019).

Social endocrine mechanisms could also be explored to better understand relationships between trait dominance, trait prestige, and stress and health. In particular, testosterone is theorized to underlie status-seeking motivations (Mazur & Booth, 1998) and is intricately and bidirectionally linked to cortisol and immune responses to stress (Muehlenbein & Bribiescas, 2005; Viau, 2002). However, testosterone is not robustly linked to self-report measures of dominance (Grebe et al., 2019; Sundin et al., 2021) and, instead, testosterone may be considered an implicit component of dominance that operates largely outside of conscious awareness (Akinola et al., 2016; Josephs et al., 2006; Knight et al., 2020; Schultheiss et al., 2005; Terburg et al., 2012) that modulates the associations of explicit, self-reported trait dominance with physiology and behavior (Carré et al., 2009, 2017; Knight et al., 2017; Slatcher et al., 2011). There are also inconsistencies in this implicit-explicit dominance interaction hypothesis, suggesting more work is necessary to develop this theoretical framework (Knight et al., 2022; Kutlikova et al., 2021).

Less work has examined trait prestige and testosterone, and initial evidence provides a mixed view. Trait prestige was associated with lower testosterone levels in men, but not women, in one sample, and trait dominance was unrelated to testosterone levels in men in the same study (Johnson et al., 2007). However, other work indicates that earned prestige (the conferred rank, which may be determined, in part, by trait prestige, the personality concept; (Cheng et al., 2021; McClanahan et al., 2021) in a large group was associated with increased testosterone levels across a two-month period (Cheng et al., 2018). In still other work, trait prestige was not found to robustly modulate testosterone’s causal effects on cortisol and affective responses to social evaluative stress in men (Knight et al., 2017). Hence, with somewhat limited data and mixed results, examining testosterone’s associations and interactions with trait dominance and prestige as a means of understanding how these traits relate to stress and health outcomes is suggestive but in need of further testing.

Psychological Mechanisms

The defined characteristics of trait dominance and prestige – that is, their relative reliance on more forceful versus more pro-social behaviors – provide an initial psychological understanding of the links between these traits and stress and health. Higher trait dominance was associated with higher stress, poorer health, and higher negative affect. This may be explained, in part, by trait dominance resulting in greater exposure to stressful or contentious social interactions, relying on relatively more anti-social behaviors, and being associated with greater threat sensitivity. In this proposed model, relying on dominance requires fighting one’s way to the top, being on guard for the need to defend one’s position, and possibly fighting off potential usurpers. Contrast this summary of trait dominance with the more pro-social tactics related to trait prestige, which include social engagement, building social relationships, and earning the liking and honor of group members. Prior work in a pre-industrial society suggests men with higher prestige also have higher reproductive success and gain more cooperation partners (von Rueden et al., 2011, 2019). These prestige-related psychosocial processes – that is, the pro-social tactics and interpersonal success – may be associated with objectively better lives and, in prior work, are associated with benefits for health and well-being (Cacioppo & Hawkley, 2003; Cohen, 2004; Hawkley et al., 2008; Holt-Lunstad, 2021; Holt-Lunstad et al., 2017; Uchino et al., 1996).

Exploratory analyses indicated that self-enhancement partly (but not fully) explained the association between trait prestige and stress and health outcomes. These results suggests that some (but not all) of the prestige-health association is likely due to trait prestige’s relationship with desires to maintain the good impressions of others, which could lead to responding in a desirable fashion on the stress and health questionnaires. To help delineate trait prestige’s direct and specific impact on health as opposed to indirect effects via desirable responding, future research could examine objective physiological indices of stress and health that are unlikely to be confounded with self-enhancement. Interventions that alter trait prestige levels could also be developed and implemented to produce experimental evidence of causal associations between prestige and stress and health outcomes, an approach used with other personality traits (Roberts et al., 2017).

Hierarchies also differ in their relative acceptance, encouragement, and rewarding of dominance and prestige behaviors (de Waal-Andrews et al., 2015). It is possible that the findings evident in this study are a result of the fit between the individual (trait dominance and/or prestige) and the social context (hierarchy style), in the same vein of an extensive literature on personality-environment fit (Mueller et al., 2019; Van Heck, 1997). No research to date has provided evidence of the modern US university system being more dominance versus prestige orientedFootnote 5. Evidence that indicates the university setting is more prestige oriented would suggest that trait prestige may reduce stress and improve health by fostering a good fit between an individual and his or her social setting. In other contexts – for example, within a military, athletic, or other dominance-oriented group – trait prestige may not be as beneficial, and so may not be related to reduced stress and better health outcomes. Trait dominance may instead predict personality-environment fit, which in turn may lead to trait dominance associating with better stress and health outcomes. This speculative psychological mechanism involving fit between an individual and his or her social setting could be studied in future work that measures or manipulates the dominance or prestige characteristics of a hierarchy (de Waal-Andrews et al., 2015) and examines downstream stress and health consequences.

Limitations and Future Directions

A primary limitation of the present work is the cross-sectional study design. While important for initial exploration and confirmation of hypothesized relationships, the cross-sectional design cannot eliminate the possibility that stress and health ‘outcomes’ precede the development of trait dominance and prestige characteristics. Major life events and health conditions are associated with changes in personality traits (Haehner et al., 2022; Jackson et al., 2017). For example, major stressors (e.g., in childhood) have been associated with the development of behaviors related to trait dominance, including heightened hostility and increased narcissistic rivalry (Clemens et al., 2022; Luecken, 2000), and chronic stress or health concerns may limit one’s ability to perform prosocial behaviors related to trait prestige. The complement to this model – that lower stress and better health precedes higher trait prestige and lower trait dominance – seems less probable: Lower stress and better health could permit or encourage use of pro-social tactics (e.g., because of an increased capacity to perform affiliative behaviors relative to an individual with higher stress, poorer health), but a healthy, unstressed state would seem unlikely to discourage dominant behaviors. The well established “tend and befriend” model of acute stress, in which stress is theorized to heighten human affiliative motivations (Taylor, 2006), further complicates any speculation that stress might precede higher dominant- and lower prestige-related traits. Future work that examines prospective, longitudinal associations of trait dominance and prestige with stress and health will be important for confirming causal direction.

Despite collecting three samples from two locations, the populations sampled share many demographic characteristics, perhaps especially consisting of young adults from large, public universities in the American West/Mountain West region. Such similarities make it difficult to determine the generalizability of the findings. Important demographic variables that are not strongly represented or were not selected for in the present sample must be specifically investigated, including variables related to race and ethnicity, sexual minority status, lower-income status, midlife and older adults, and the intersection among these and other variables (Crenshaw, 1989; Gkiouleka et al., 2018). Indeed, the associations between social status and health are likely most easily explained by rampant and unremitting inequality in our social systems. That a psychological concept (e.g., trait dominance or prestige) may explain some of the variability in stress and health outcomes does not negate our duty to examine and fix societal issues—issues that ultimately affect the stress and health of those already unfairly burdened.

As discussed previously, this initial exploration and confirmation of links between trait dominance, trait prestige, and stress and health should be followed up with an explicit focus on objective indices of stress and health. Such work could start with examinations of autonomic and endocrine activity, perhaps especially examining responses to manipulated status-relevant social stimuli, like a status-relevant threat. An experimental approach such as this, when coupled with longitudinal and ecological approaches linking trait dominance and prestige to longer term physiological indices, would produce a more comprehensive model of stress and health outcomes related to trait dominance and prestige.

Conclusion

These results point to intriguing associations between personality traits, social hierarchies, and health. Whereas trait dominance and prestige have each been linked with status attainment in human hierarchies (McClanahan et al., 2021; van Kleef & Cheng, 2020; Van Vugt & Smith, 2019) and social status has been linked with stress and health (Adler et al., 1994; Sapolsky, 2004, 2005), only trait prestige was associated with reduced stress and better health; trait dominance was associated with increased stress and poorer health outcomes. Hence, this work suggests the preferred route by which status is earned matters for understanding links between status, stress, and health. Future work that unpacks psychological and physiological pathways by which trait dominance and prestige may link to stress and health, with more diverse populations, will be important for determining the direct and sustained impact these personality types may have on health and well-being within our societal hierarchies.