Spirituality as compensation for low-quality social environments in childhood among young Kenyan men

Abstract

This study explores the role of spirituality as a coping mechanism for poor social conditions in childhood, asking whether spirituality moderates poor childhood social conditions and suicide ideation, self-rated health and collective self-esteem among young Kenyan men. Measured outcomes were worse among men who recalled fewer memories of relational warmth and safety in childhood, and better among men who reported higher spirituality. Consistent with the “religion as attachment” framework, spirituality significantly moderated associations between suicide ideation, self-rated health and childhood relational warmth and safety. Contrary to expectations, the association between low childhood warmth and safety and collective self-esteem was exacerbated, rather than compensated for, by higher spirituality. We consider whether “a safe harbor” may exist for people higher in spirituality to accept and critique social arrangements, and whether such a situation might illuminate another way spirituality compensates for poor social environments.

Introduction

The role of spirituality in social life has found renewed articulation as a component of relational attachment processes. Granqvist et al. (2010) proposed two pathways between social attachment histories and spiritual/God attachments. One pathway is labeled “the correspondence pathway,” as spiritual attachments correspond with social attachments in a positively correlated way. As social attachments strengthen, individuals perceive a deeper intimacy with God. We previously found that memories of social and emotional warmth and safety recalled from childhood – the requirements of secure attachment formation – predicted present spiritual intimacy and this pathway mediated associations between childhood memories and depression, hope and meaning in life as hypothesized (Goodman et al., 2020).

Granqvist et al. (2010) proposed a second pathway between attachment histories and spiritual attachments. The “compensation pathway” is a proposed explanation for religious conversions of individuals with insecure attachment histories and parental insensitivity. Children require warmth, responsiveness, proximity and safety in an adult attachment figure, typically a parent, in order to develop internal working models indicating their value and security. When such figures are unavailable, children may turn to substitute attachment figures to compensate for this lack of secure attachment base. A meta-analysis of childhood attachment and religious conversion using data from 11 studies across multiple countries found individuals with insecure attachments during childhood were significantly more likely to undergo rapid religious conversion (Granqvist & Kirkpatrick, 2004). If spirituality is a productive coping mechanism for those with challenging childhood social environments, spirituality will moderate the effect of childhood social environments on adult health outcomes associated with challenging childhood social environments.

Adult suicide ideation has previously been associated with challenging childhoods due to insecure social attachment in Kenya (Goodman et al., 2018). Empirical findings strongly support the Interpersonal Theory of Suicide, which posits suicide ideation begins when individuals experience thwarted belongingness and perceived burdensomeness (Van Orden et al., 2010). Thwarted belongingness occurs when individuals believe they do not belong socially, and experience the pain of social isolation. Perceived burdensomeness occurs when people believe they place a burden on others whom would benefit if the individual was no longer alive. These two constructs – thwarted belongingness and perceived burdensomeness – can combine with a sense of hopelessness that the situation will change, and individuals begin to ponder suicide. The experience of belonging, and being worthy of care from others, begins in early childhood and is marked by secure attachments with caregivers (Baumeister & Leary, 1995). Early attachment processes are extrapolated from caregiver relationships to peers, employers, teachers, friends, intimate partners and even pets (McConnell et al., 2011). Thwarted belonging, beginning in childhood, can lead to suicide behavior. The first question of the current study is whether perceived attachment with God can offset the pernicious effects of low levels of childhood warmth and security on suicide ideation in adulthood.

In addition to suicide ideation, self-rated health is another important marker of wellness that can be influenced by social belonging. In a study of 45 countries, Mansyur et al. (2008) found self-rated health was predicted by lower levels of individual social support and lower collective measures of social capital – the degree of trust within a community. A meta-analysis found that loneliness and social isolation predict mortality risk across the lifespan, but more so for individuals younger than 65 years of age (Holt-Lunstad et al., 2015). As belonging is a fundamental human need, one should expect to see worse health among those who feel a decreased sense of belonging. Self-rated health is a useful population health metric because summarizes a wide spectrum of cultural and biological information and predicts mortality and morbidity across cultures (Jylhä, 2009). Self-rated health is predicted by childhood social conditions that become repeating patterns across the life span, with adverse social exposures contributing to lower self-rated health (Bonomi et al., 2008; Goodman, Johnson, et al., 2017). If perceived attachment with God compensates for childhood threats to developing secure attachments, it should moderate associations between these conditions and self-rated health.

Similar to self-rated health and suicide ideation, self-esteem is also a reflection of one’s sense of belonging and social location with roots in childhood social environments (Robins & Trzesniewski, 2005). Self-esteem is a reflection of one’s perceived relational value to others (Leary, 2005). Children who receive positive affirmation from attachment figures, as opposed to neglect, disinterest, disapproval or rejection, are more likely to develop higher self-esteem and maintain higher levels of self-esteem across the lifespan (Robins & Trzesniewski, 2005). Self-esteem is important for its relationship to other psychological and behavioral characteristics. Individuals with lower self-esteem are more likely to experience loneliness and depression (Du et al., 2018; Orth & Robins, 2013). Low self-esteem also creates problems in intimate partnerships, as men with low self-esteem are more likely to resolve conflicts violently (Goodman, Serag, et al., 2017). Spirituality, compensating for low quality social relationships in childhood, should moderate associations between childhood social relationships and self-esteem in adulthood.

Collective self-esteem pertains to the valuation of one’s collective self – the network of relationships that provide a sense of identity – and one’s place within that collective. Personal self-esteem pertains to the valuation of one’s personal identity – as a separate construct from one’s social network, though a reflection of one’s social location. Research shows spirituality is associated with higher levels of both personal and collective self-esteem (e.g., Constantine et al., 2002; Joshanloo & Daemi, 2015). Empirical research on the role of spirituality to pathways between social attachment histories and self-esteem under the “religion as attachment” framework is new. In this study we selected a collective measure of self-esteem, as traditional Kenyan culture is more collectivist and primary notions of self tend to be social (Ma & Schoeneman, 1997). Research based in western cultures tend to adopt individual or personal self-esteem measures, and distinctions between these personal and collective self-esteem are not always clear (Oyserman et al., 2002).

Attachment theorist predict that attachment processes are universal to human experience. Ainsworth was working in Uganda when she made seminal observations about the importance of infant attachments with mothers (Mesman et al., 2016). Expressions of religiosity and spirituality are also found in most cultures across the world (WHOQoL SRPB Group, 2006). If spirituality is an intrinsic, natural and universal outgrowth of other attachment processes, relationships between spirituality and childhood attachment conditions anticipated in one culture should be found in another. While most of the theory and empirical data motivating the current study are derived from a Western setting, we hypothesize that dynamics in rural Kenya will support anticipated relationships.

Study aim

This study aims to determine whether spirituality moderates associations between recalled childhood attachment conditions and self-reported indicators of self-rated health, suicide ideation and self-esteem among Kenyan adult males (age 18–34).

Methods

Participants

Young men (aged 18–34 years) from 11 townships of North Igembe sub-county, Meru County, Kenya were surveyed using a structured survey with validated scales. Maua Methodist Hospital operates community clinics in the represented areas, and data to inform approaches to reduce suicide was collected.

Procedure

Male, Kenyan nursing students were to administer surveys and were paid to conduct interviews with selected respondents. The sampling frame included every other household along a predetermined route. Random number generators were used to determine direction for a uniform walk approach, which is a validated method of conducting household surveys when full housing registries are unavailable (Thompson, 2006). Selected households were assessed for the presence of an eligible male (aged 18–34 years). Where more than one eligible male was available, one male was randomly selected. A total of 772 houses were visited and found eligible for the larger study. Of these, 240 (31%) refused and 532 (69%) completed an interview. Interviews were conducted by the interviewer, who read each question aloud and recorded the response on an interview form. All data were entered twice into EpiInfo.

Survey items included in the current study comprised demographic characteristics as well as scales measuring early memories of relational warmth and safety, suicide ideation, self-rated health, self-esteem and spirituality. Interview questionnaires were created in English, translated into Kimeru and back translated for validation and refinement. Interviews were conducted in the local language, Kimeru. Data used in this study were collected to facilitate suicide prevention efforts, and are part of a larger dataset related to men’s mental and behavioral health.

Measures

All measures were selected based on prior validation across multiple cultures and languages.

Outcome variables

This study has three outcome variables – suicide ideation, self-rated health and self-esteem.

To measure suicide ideation, we used the Modified Scale for Suicide Ideation, an 18-item 4-point Likert type scale (MSSI; Miller et al., 1986). The MSSI provides a summative measure of desire, ideation, plans and preparation for suicide (Pettit et al., 2009) and has been validated across multiple cultures (Yang & Clum, 1994). Scale developers originally proposed three categories for severity of suicide ideation: low (range: 0–8), mild/moderate (range: 9–20), and severe (range: 21–54). In the current study to implement logistic regression models, we compare respondents in the “severe” category to those who scored in the lower two categories using a binary variable.

To measure self-rated health (SRH), we used the question “in general, how would you rate your health currently?” The response options included 4 levels – poor, fair, good and excellent. To utilize logistic regression analysis, we created a dichotomous variable from responses (poor/fair vs. good/excellent).

To measure self-esteem, we used the Collective Self-Esteem scale (CSE) (Luhtanen & Crocker, 1992). The CSE scale is a 16-item scale using 7-point Likert-type responses to assess an individual’s global positive evaluation of his or her collective/social identity (current study sample, α = 0.88). Scale items include statements like “overall, my social groups are considered good by others” and “I am a worthy member of the social groups I belong to.” Items are summed to calculate a scale score. The CSE scale has been validated across different countries, cultures, languages and income groups including Kenya (Kinoti et al., 2011). The CSE was developed to respond to a 4-factor structure (Membership esteem, Public collective self-esteem, Private collective self-esteem, and Importance to Identity), though our data did not display this structure in exploratory or confirmatory factor analysis.

Moderating variable

To create our moderating variable, we began with the Brief Multidimensional Measure for Religiousness and Spirituality (BMMRS; Fetzer Institute/National Institute on Aging Working Group 1999). During the time of the survey, there were heightened tensions related to religious identity stemming from recent terrorist activities by the Al Qaeda-linked terrorist group, Al Shabob. To avoid possible confusion, religious identification was removed from the BMMRS measure. We conducted exploratory factor analysis using iterative principal factor analysis with oblimin rotation to allow for oblique relations between factors (See Supplemental Table 1 for Factor Analysis results). Factor analysis yielded two factors with eigenvalues over 2, see Screeplot in Supplemental. We assessed the bivariate relationships between each of these two factors and the three outcomes, finding the second factor was significantly associated with the three study outcomes while the first factor was not. The retained factor, as shown in Supplemental Table 1, consisted of 10 items including statements such as “I feel God’s presence,” “I feel deep inner peace or harmony,” “I know that God forgives me,” and “I look to God for strength, support and guidance.” These 10-items are measured on a 6-point Likert-type response format, and had good internal reliability (α = 0.88). The characteristics of safety and security in one’s relationship to God is consistent with the “religion as attachment” framework provided by Granqvist et al. (2010) and reflected in this sub-factor. The unused factor measured constructs including religious observance, daily religious practice and obligations within religious networks, which reflect dimensions of religiosity not reflected in the central concerns of the “religion as attachment” framework. As such, we felt justified in not including it in the current study.

Exposure variables

Childhood social environment was measured using the early memories of warmth and safeness scale (EMWSS, Richter, Gilbert, & McEwan, 2009). The EMWSS is a 21-item recall measure assessing early feelings of secure attachment. Respondents are prompted to consider their earliest memories, and provide a rating (from 0 – never to 4 – most of the time) of how frequently they felt safe, secure, understood, comfortable, enjoyable, loved, cared for, and as though they belonged. The EMWSS can predict psychopathologies in English-speaking and Portuguese-speaking populations. In the present sample the scale showed high internal consistency (α = 0.97). Example items from the EMWSS include “I felt part of those around me,” and “I had feelings of connectedness” during childhood. The sense of belonging, receiving care and attention, safety and emotional warmth measured by the EMWSS are essential for developing secure attachments, so we used the EMWSS as a proxy for attachment security. While there are other scale measures that assess attachment styles directly, the EMWSS was chosen in collaboration with community partners as it measures behaviors that can more intuitively inform parenting behavioral interventions.

Control variables

To control for influence of sociodemographic characteristics, we included age, years of completed schooling, and household wealth in all regression analyses. The wealth index is a sum of the number of household assets in the following list: electricity, radio, television, telephone, refrigerator, land and number of rooms used for sleeping. The use of household wealth, including number of rooms for sleeping, is accepted as a way to develop a proxy measure of household wealth (Vyas & Kumaranayake, 2006). Age was included as a continuous measure of years reported. Education was measured as number of formal school years completed.

Statistical analysis

For descriptive analysis, we compared model variables across levels of the three outcomes – suicide ideation, self-rated health, and self-esteem. Suicide ideation and self-rated health are treated consistently as binary variables in all analyses – suicide ideation is present (1) or not (0). Self-rated health is good/excellent (1) or fair/poor (0). Self-esteem is treated as continuous. The correlation matrix for self-esteem, early memories of warmth and safety, spirituality and wealth is shown in Table 2.

Table 2.

Correlation Matrix of continuous measures (n = 541).

	1	2	3	4	5	6	7
1. SPIRITUALITY	1
2. SELF-RATED HEALTH	0.46***	1
3. COLLECTIVE SELF-ESTEEM	0.39***	0.36***	1
4. EMWSS	0.51***	0.41***	0.44***	1
5. WEALTH	0.25***	0.36***	0.19***	0.27***	1
6. AGE	0.09	−0.11	−0.07	0.01	−0.06	1
7. EDUCATION	0.09	0.2***	0.2***	0.16**	0.42***	−0.12	1

Correlation matrix of continuous model variables.

^*p < .05

^**p < .01

^***p < .001.

To assess effect moderation, we used fixed-effects multiple regression models to account for variation between geographic clusters, and determined associations between early memories of warmth and safety and the three outcome variables. We used a linear model to analyze self-esteem, and logistic models to analyze suicide ideation and self-rated health. We created interaction terms between spirituality and early memories of warmth and safety to determine possible moderation effects. All regression models control for respondent age, education and wealth.

Ethical consideration

The ethics committee at the Maua Methodist Hospital provided oversight and approval for the study prior to data collection. The Institutional Review Board at the University of Texas Medical Branch exempted ethical review of analysis and publication of deidentified data. All subjects provided informed consent prior to participating in the survey.

Results

Table 1 shows the bivariate analyses of model covariates against levels of the two binary outcome variables. Suicide ideation was present among 11.8% of respondents (sd: 0.32), and 77.8% of respondents had good/excellent self-rated health (sd: 0.42). Early memories of warmth and safety and spirituality were significantly lower among those who reported suicide ideation, and significantly higher among those who reported good/excellent self-rated health compared to poor/fair self-rated health.

Table 1.

Bivariate analysis of model variables and suicide ideation, general self-rated health and self-esteem (N = 541).

			No suicide ideation		Suicide ideation			Fair/poor SRH		Good/Excellent SRH
	Mean/%	SD	Mean/%	SD	Mean/%	SD	p	Mean/%	SD	Mean/%	SD	p
Suicide ideation	11.80%	0.32						35.65%	0.48	5.40%	0.23	<.001
Good/Excellent Self-rated health	77.80%	0.42	84.77%	0.37	34.90%	0.48	<.001
Collective self-esteem	4.79	1.12	4.89	1.14	4.04	0.49	<.001	4.22	0.8	4.96	1.15	<.001
EMWSS	4.94	1.01	5.05	0.92	4.12	1.2	<.001	4.26	1.21	5.13	0.85	<.001
Spirituality	5.32	0.66	5.39	0.61	4.78	0.73	<.001	4.79	0.8	5.47	0.53	<.001
Wealth	3.11	1.29	3.16	1.26	2.69	1.43	.02	2.28	1.43	3.34	1.14	<.001
Age	26	5	25.7	5.02	27.6	4.38	.005	26.51	4.61	25.82	5.08	.19
Education	8.97	3.77	9.09	3.64	8.02	4.53	.11	7.68	3.91	9.3	3.67	<.001

Table shows bivariate analysis of early memories of warmth and safety (EMWSS, standardized), spirituality (standardized), wealth index, age and education across levels of suicide ideation, self-rated health (SRH) and collective self-esteem (median split).

Table 2 shows the correlation matrix of the two spirituality sub-factors, self-rated health, early memories of warmth and safety, wealth, age and education. Spirituality is strongly positively correlated with overall health, self-esteem, early memories of warmth and safety, and wealth (r = 0.46, 0.39, 0.51, and 0.25, respectively; all p < .001). Self-rated health is significantly positively correlated with self-esteem, early memories of warmth and safety, wealth and education (r = 0.36, 0.41, 0.36, and 0.2, respectively; all p < .001). Self-esteem is significantly positively correlated with early memories of warmth and safety, wealth and education (r = 0.44, 0.19, and 0.2, respectively; all p < .001).

Table 3 presents coefficients predicted by fixed effects multiple logistic and linear regression models. Interaction terms of spirituality and EMWSS in all 3 models are significant, showing the presence of effect moderation.

Table 3.

Fixed effects logistic and linear regression of suicide ideation, GSRH, and self-esteem on model variables.

	Model 1: Suicide Ideation (logistic)			Model 2: GSRH (excellent/good vs fair/Model			3: Collective Self-esteem poor; logistic) (linear)
	OR	95% CI		OR	95% CI		Coef.	95% CI
EMWSS, std	0.54***	0.38	0.75	1.44**	1.1	1.88	0.34***	0.25	0.43
Spirituality, std	0.54***	0.37	0.79	2.37***	1.68	3.36	0.39***	0.28	0.5
EMWSS*Spirituality	0.75**	0.6	0.94	1.26*	1.01	1.56	0.17***	0.1	0.23
Wealth	1.13	0.87	1.47	1.59***	1.27	1.97	−0.01	−0.08	0.06
Education	0.95	0.87	1.04	1.05	0.98	1.13	0.03*	0	0.05
Age	1.11***	1.03	1.19	0.96	0.91	1.01	−0.02**	−0.03	0
LR Χ2		44.23			105.2			105.2
p-value		<.001			<.001			<.001

Model 1 shows fixed effects logistic regression odds ratios of suicide ideation (MSSI) on early memories of warmth and safety (EMWSS, standardized), spirituality (10 items from BMMRS), an interaction term (EMWSS x spirituality), wealth, age and years of schooling. Model 2 shows fixed effects logistic regression odds ratios of excellent/good general self-rated health (GSRH) vs fair/poor GSRH. Model 3 shows fixed effects linear regression coefficients of self-esteem on model variables.

^*indicates p < .01.

^**indicates p < .01

^***p < .001.

95% confidence intervals shown.

Table 3 shows the fixed effects regression models of suicide ideation, self-rated health (excellent/good vs. fair/poor), and self-esteem on EMWSS, spirituality, their interaction and additional covariates.

Suicide ideation

With each increased standard deviation of EMWSS, the odds of reporting suicide ideation decrease by 46% [OR = 0.54; 95%CI: 0.38–0.75]. Similarly, with each standard deviation increase in spirituality, the odds of reporting suicide ideation decrease by 46% [OR = 0.54; 95%CI: 0.37–0.79]. The interaction between EMWSS and spirituality show that the predicted odds of suicide ideation depend on the joint level of EMWSS and spirituality. As shown in Figure 1a, the slope between EMWSS and predicted probability of suicide ideation is lower among individuals who reported spirituality at or above the median, compared to individuals who report spirituality below the median. The level of suicide ideation is lower for individuals who report low EMWSS, but higher spirituality than for individuals who report low EMWSS and low spirituality.

Figure 1.

Interactions between early memories of warmth and safety (EMWSS), suicide ideation(1a), overall health (1b), and self-esteem (1c).

Excellent/good self-rated health

With each standard deviation increase in EMWSS, the odds of reporting good/excellent self-rated health increase by 44% [OR = 1.44; 95%CI: 1.1–1.88]. With each standard deviation increase in spirituality, the odds of reporting good/excellent self-rated health increase by 137% [OR = 2.37; 95% CI: 1.1–1.88]. The interaction term between EMWSS and spirituality is significant and positive, showing the predicted odds of reporting good/excellent self-rated health depend on the joint level of EMWSS and spirituality. Figure 1b shows the slope for EMWSS and self-rated health by level of spirituality. For individuals with spirituality less than the median, the slope between EMWSS and self-rated health was steeper. There was a larger difference in self-rated health among those with low EMWSS based on whether they reported lower or higher spirituality, but this difference was less for individuals with higher EMWSS.

Collective self-esteem

With each standard deviation increase in EMWSS, self-esteem increases by .34 standard deviations [B = 0.34; 95%CI: 0.25–0.43]. With each standard deviation increase in spirituality, self-esteem increases by 0.39 standard deviations [B = 0.349; 95%CI: 0.28–0.5]. The interaction between EMWSS and spirituality is positive and significant, showing the predicted level of self-esteem depends on the joint levels of EMWSS and spirituality. For individuals with EMWSS lower than −2 standard deviations, higher spirituality predicted lower self-esteem. For individuals with higher EMWSS, higher spirituality predicted higher self-esteem.

Discussion

The “religion as attachment” paradigm (Granqvist et al., 2010) asserts that attachment to divine figures relates to attachment to human figures through two different pathways. The corresponding pathway posits that individuals who are securely attached to human figures are more likely to develop secure attachments to a divine figure. The compensation pathway posits that individuals who are not securely attached to human figures may compensate for this attachment insecurity through developing a secure attachment with a divine figure. In this study, we aimed to determine whether spirituality moderated associations between social conditions in childhood and known outcomes predicted by insecure attachments – higher suicide ideation, lower health and lower self-esteem. That is, we asked whether spiritual attachments might compensate for the pernicious consequences to adult mental health of insecure childhood attachments.

We observed that individuals who recalled less emotional warmth and safety from childhood reported higher levels of suicide ideation, and lower self-rated health and collective self-esteem in adulthood. The associations between childhood emotional warmth, self-rated health and suicide ideation were statistically offset when individuals reported higher levels of spiritual attachment. The association between childhood emotional warmth and collective self-esteem, as moderated by spirituality, was not one of compensation but of exaggeration. Spirituality appeared to exacerbate pernicious associations between lower childhood emotional warmth (<−2 SD) and self-esteem in adulthood, while amplifying the association between higher collective self-esteem and more emotional warmth and safety during childhood.

Young men who felt more connected, loved, understood, and cherished in their earliest memories from childhood had significantly decreased odds of reporting suicide ideation, significantly higher odds of reporting good/excellent self-rated health and significantly higher self-esteem. Young men who felt less connected, loved, understood and cherished in childhood reported higher suicide ideation, and worse overall health, but not if they have higher levels of spirituality. Measured outcomes among young men who have high levels of spirituality improve if their childhood social environment improves. The best outcomes among men with high spirituality are found among men who also reported more affirming childhood memories.

Understanding the relationship between spirituality and attachment processes enables researchers to investigate how these ubiquitous dynamics interact to improve or threaten wellness and health. The “religion as attachment” framework provided by Granqvist et al. (2010) is helpful for understanding how spiritual relationships may extend social attachment relationships that are better understood by natural scientists. Consistent with the compensation pathway outlined in Granqvist et al. (2010), spirituality was observed in the present study as an effect moderator between recalled low quality social relationships in childhood and some health outcomes in young adulthood. However, spirituality also appeared to make the poor poorer and the rich richer in the association between childhood emotional warmth and collective self-esteem. While Granqvist et al. (2010) postulated that spirituality compensates for poor social relationships, they do not propose specific outcomes that may be moderated by increased spirituality.

Suicide ideation, driven in part by the sense of thwarted belonging, may be lower among people who believe God is present, forgiving, supportive, and offers a source of belonging beyond human relationships. As such, suicide ideation may be higher among individuals with insecure attachment histories, but spiritual attachments may weaken this association.

The self-rated health item registers multiple social, mental, and physical health characteristics into a summative statement regarding health. This summative statement is a good general predictor of mortality and morbidity across a population (Jylhä, 2009), but not a specific indicator of a specific health condition. As such, it is possible that differences in self-rated health by degree of spirituality or early memories of warmth and safety reflect social, mental or physical health characteristics. Childhood attachment security influences a range of social, mental and physical health outcomes – including romantic relationship stability, common mental disorders, self-harming behaviors and immunological function (Feeney & Noller, 1992; Maunder & Hunter, 2008; Mikulincer & Shaver, 2012). It is unclear which underlying health condition(s) is reflected in respondent’s self-rated health score, but it is noteworthy that spirituality modifies associations between childhood attachments conditions and overall health. Further research should clarify whether modification occurs through social, mental or physical health. Secondary analysis considered whether self-esteem explained the observed association between spirituality, early memories of warmth and safety and self-rated health. Self-esteem did not mediate observed primary associations.

Self-esteem is a reflection of one’s social acceptance and relational value, and an important indicator of other mental and social outcomes – e.g., life satisfaction, meaning in life, relational trust (Blattner et al., 2013; Dainton & Aylor, 2001; Diener & Diener, 2009). Research shows consistently that when children do not experience acceptance and validation, they often develop low personal self-esteem that can persist throughout the lifespan (Trzesniewski et al., 2003). We anticipated that individuals who experience less acceptance and validation during their childhood subsequently develop internal working models of God that provide acceptance and validation, improving collective self-esteem. Rather than compensate for the association between low childhood emotional warmth and safety and self-esteem, higher spirituality predicted worse collective self-esteem at the low end of childhood warmth and safety spectrum but higher self-esteem at the high end of the childhood warmth and safety spectrum. This is contrary to what we hypothesized from the compensation pathway postulated by the “religion as attachment” paradigm, and clearly warrants further study. One explanation for this apparent rejection of the compensation hypothesis, which we call “the safe harbor hypothesis”, is as follows.

A basic tenet of attachment theory for the past fifty years includes that people whose primary childhood socialization includes routine messages that they are not valued as part of their family will likely preserve internal working models of being less worthy throughout their lives (Bowlby, 1969). Acknowledging the reality that one is less valued within a social group, and that the social group is less valued at large, can be painful and prompt coping mechanisms including disengagement and denial (Leitner et al., 2014). Spirituality can promote acceptance of existential threats and reduce perceived risk of interpersonal rejection (Cicirelli, 2011; Laurin et al., 2014). As we used a collective measure to assess self-esteem, it is possible that respondents higher in spirituality were more able to recognize their childhood experience of low relational warmth and safety, and their relationships since then, as indicating they are not as highly valued by their social groups and their social groups are less valued by others. If this is the case, it may reveal one mechanism by which spirituality helps compensate for less secure attachment histories – that mechanism being by enabling a more honest acceptance of one’s less secure and less valued social location with potential to provide resistance with the picture in clearer view. Spirituality as “a safe harbor” for critiquing social arrangements would have broad implications, but requires empirical and theoretical development. To support the refinement and evaluation of the “spirituality as safe harbor” hypothesis, here are some testable hypotheses that would provide empirical support to the concept:

1. People higher in spirituality have lower threat vigilance, particularly toward social threats.

2. People higher in spirituality are more psychologically resilient to social rejection.

3. People with higher spirituality are more able to engage in critical group-reflection and are more aware of undesirable traits within membership groups.

4. Spirituality moderates associations between low collective self-esteem and personal self-esteem.

5. Spirituality moderates associations between social environments and collective self-esteem differently than between social environments and personal self-esteem – that is, personal self-esteem may be preserved in the presence of social hardship while collective self-esteem is reduced.

6. Spirituality increases during life transitions to mitigate the adversity of social role ambiguity on mental well-being.

Some of these tenets of the “safe harbor hypothesis” already find support in the literature, for example, transitioning young adults demonstrate higher spirituality (McNamara Barry, Nelson, Davarya, & Urry, 2010), though our proposal presents the function of spirituality as compensating for uncertain or insecure social attachments during these hypothesized conditions.

While requiring more research, “spirituality as a safe harbor” may enable young men higher in spirituality to advocate for more emotionally warm and secure environments for children as they understand the challenges imposed by environments with less warmth and security. It is imperative to note “spirituality” here does not indicate participation in religious groups or external religious behavior, but rather the inward sense of connection with a transcendent, personal, peaceful and forgiving God or God-like Being.

Higher spirituality predicted higher collective self-esteem overall, so it is not as though concern for self-esteem should contraindicate the role of spirituality. The question is about how spirituality interacts with childhood determinants, and in the case of collective self-esteem and childhood warmth and safety, the picture appears more complicated. Further research into personal self-esteem could clarify whether the hypothesized moderation between spirituality and childhood conditions is found there. The consistent moderation between spirituality and childhood conditions with respect to self-rated health and suicide ideation provides support for the hypothesis that spirituality can compensate for low quality childhood social environments.

Previous research has shown that college students from households that were emotionally cold were more likely to relate to God through a dismissive attachment style (McDonald et al., 2005). Understanding whether one’s image of God more substantially reflects one’s image of oneself or one’s image of other people, how God relations align with attachment styles, whether attachment styles are consistent in God- and human-relationships, and how various God attachment styles potentially impact health require further research. If more resilient individuals also report higher levels of spirituality, it is possible that spirituality reflects underlying resilience processes rather than promotes them directly. It is also possible that God beliefs and spiritual experiences buffer the pernicious effects of childhood exposures, compensating for poor social attachments as hypothesized.

Study data are cross-sectional and do not permit causal or even temporal inference. At most, findings and interpretation represent a “proof of concept” that require further investigation with multiple types of data and analysis. Qualitative research should explore how individuals understand the lingering effects of childhood social conditions, the role of God attachments and current mental and physical health. Quantitative research should explore longitudinal changes in spirituality and measured health outcomes, subsequent to baseline childhood exposures. Interventional research should consider how to develop spirituality within study subjects, and observe corresponding changes in suicide ideation, overall health and collective self-esteem. The hypothesis suggested by this study is that increasing spiritual attachment among young adults with less supportive childhood social environments will decrease suicide ideation, improve overall health and collective self-esteem.

In general, more research is required to understand relationships between attachment styles and conditions, spirituality and health outcomes. In considering whether, and how, God relationships can be understood as extensions of or compensations for human attachments, there are ongoing debates. One debate revolves around whether spirituality is better explained as corresponding to secure human attachments or developed as a substitute for insecure human attachments. Within the current data, individuals with better social conditions in childhood reported higher spirituality scores – supporting the view that spirituality develops as an extension of secure human attachments (Goodman et al., 2020). The present study, however, considers whether spirituality might compensate for suboptimal childhood social conditions – supporting the view that spirituality compensates for insecure human attachments. We found that, within the context of cross-sectional survey data of young men collected in semi-rural Kenya, spirituality statistically moderated anticipated associations between suboptimal social conditions and two indicators of adult health – self-rated health and suicide ideation.

From a public health perspective, the potential for spirituality to compensate for suboptimal childhood social attachments is important, though not as important as the overall place of spirituality in health. Overall, spirituality was protective against worse outcomes, and findings support the buffering mechanism proposed by the “religion as attachment” framework for suicide ideation and self-rated health. Future theoretical and empirical research should be conducted to explain why spirituality may have exaggerated the association between childhood warmth and safety and adult self-esteem.

Public health must also be concerned with how spirituality can be developed and utilized to promote population health, if indeed spirituality can successfully and predictably compensate for suboptimal human relationships. As mentioned, such confidence is beyond the scope of the present study and requires further investigation.

Another debate within the “religion as attachment” literature revolves around how internal working models function in the context of the compensation pathway. Within the correspondence pathway, working models governing impressions of a relational other are understood to transfer from visible humans to an invisible God (Miner, 2009). When experience of human attachments produce working models suggesting that others are untrustworthy, and one is unworthy, how do impressions regarding the trustworthiness of God derive? Within the development of spirituality as compensation for suboptimal human relationships, how does a person derive reliable models of God as a secure base such that health outcomes are transformed? It is notable that empirical research investigating “religion as attachment” is relatively new, and investigation into how secure God attachments may moderate health outcomes of insecure human attachments is even younger. Further investigation exploring whether spirituality moderates associations between suboptimal human relationships and health outcomes is needed, and should establish with greater clarity what health outcomes are improved by higher spirituality. More basic research is required to illuminate how and when God attachments emerge to compensate insecure human attachments, and whether God attachments provide “a safe harbor” to promote acceptance of uncomfortable social arrangements and how this “safe harbor” promotes resistance to these arrangements.

Limitations

Data were collected using self-report surveys of a cross-sectional, non-probabilistic sample of young Kenyan men. Self-report measures are open to various biases, including most importantly social desirability bias. We believe social desirability would lead respondents to under-report potentially harmful information, like suicide ideation, and over-report potentially socially beneficial information, like self-esteem and overall health. This differential misclassification would lead toward the null hypotheses and make observed associations conservative. Recall bias might lead to misrepresentation of childhood experiences. Where measured, respondents tend to report more favorable childhoods during adult recall than they experienced as children, leading again to more conservative estimates (Hardt & Rutter, 2004). Cross-sectional data, largely, prevent postulating temporal ordering or causal inferences. Some variables can be temporally ordered based on logical assumptions, e.g., childhood exposures came before adult outcomes, but it is not clear the extent to which adult conditions influence recall bias. The strength of this study is not in its causal claims, but in its development of new, testable hypotheses in a globally understudied population. As social attachment and religious expression are global phenomenon, meaningful connections between them should be applicable and informed by a variety of cultural settings. The non-probabilistic sample was necessary as there are no comprehensive lists of households in the study setting. We used a uniform walk method which has been validated elsewhere as a valid method of data collection. Ideally data would be taken from a random sample, but this is unfortunately not possible in community settings where there is no list of households. Finally, data were collected from young men. It is not possible to generalize findings beyond young men in a semi-rural setting in sub-Saharan Africa, but it should be clear what the findings of this study are. We are promoting testable hypotheses rather than settled causal conclusions about the study population, and these testable hypotheses may pertain to women, older subjects and populations outside of semi-rural Kenya.

Conclusions

Spirituality is a recognized source of resilience across many populations. New theoretical models link the development of spirituality to attachment processes through two routes – correspondence and compensation. This study explored whether spirituality may compensate for suboptimal social conditions in childhood with respect to adult health outcomes – suicide ideation, self-rated health and collective self-esteem. We found that spirituality statistically moderates expected associations between childhood social conditions and health outcomes among young Kenyan males. For suicide ideation and self-rated health, higher spirituality reduced disparities connected with fewer memories of emotional warmth and safety from childhood. For collective self-esteem, higher spirituality appeared to exaggerate the association, leaving people with higher spirituality and fewer memories of warmth and safety with even lower collective self-esteem. We provide a theory for how this unexpected association may yet be consistent, and provide greater depth to, the “religion as attachment” framework. Further research is required to understand when and how spirituality compensates for harmful consequences of challenging childhood social environments, and how spirituality might be promoted within public health approaches to population health.

Notes on contributors

Michael Goodman is a theologically-trained public health professional. He received a Master’s of Divinity from Emory University, a Master and Doctorate in Public Health from the University of Texas School of Public Health. He currently works as an assistant professor in the Division of Infectious Disease at the University of Texas Medical Branch, and has been active in Kenya on a variety of research projects related to social epidemiology since 2012.

Lauren Raimer-Goodman is a community-based pediatrician at the University of Texas Medical Branch. Trained at Baylor College of Medicine, she has been active in research and clinical service in sub-Saharan Africa since 2009.

Stanley Gitari holds a Masters in Public Health from the University of Ghana, School of Public Health and a Bachelor’s in Nursing from Oklahoma City University. He has performed community health service in Meru County, Kenya for over three decades. He is the country director of Sodzo Kenya.

Sarah Seidel is the direct of research at Sodzo International, and holds a doctorate degree in public health. Dr. Seidel has worked in Kenya on various research projects related to community resilience and child outcomes since 2014.

Data availability statement

The data described in this article are openly available in the Open Science Framework at https://doi.org/10.17605/OSF.IO/TPA6U.

Interview Questionnaire may be accessed: https://osf.io/f5bhs