Does God Comfort You When You Are Sad? Religious Diversity in Children’s Attribution of Positive and Negative Traits to God

: Children actively construct their understanding of God in early childhood, an understanding that incorporates affect-laden mental representations often referred to as God images. To explore religious variations in children’s association of positive and negative traits to God, 254 preschool-aged children from Protestant Christian, Catholic, Muslim


Introduction
A growing body of research has examined cultural and religious differences in children's concepts of God (e.g., De Roos et al. 2001a;Shtulman and Lindeman 2016;Nyhof and Johnson 2017).These studies have typically employed cognitive, explicit methods that assess children's beliefs about God's mental state capabilities (e.g., Burdett et al. 2020) or children's attribution of psychological, biological, and physical characteristics to God (e.g., Shtulman 2008).One limitation of these approaches is that they tend to focus on what might be considered 'cold' cognition, or the aspects of a God concept that do not incorporate affective and motivational processes (Zelazo and Müller 2011).As concepts of God are often profoundly relational even in young children, researchers have highlighted the importance of considering the relational nature of children's developing concept of God (Granqvist et al. 2007;King and Boyatzis 2015).Building on studies examining the positive and negative valence of God concepts in early childhood (De Roos et al. 2001a, 2001b;De Roos et al. 2004), the current study presents a descriptive account of religious diversity in children's concepts of God, specifically aspects of the God concept that incorporate affective and relational features.

Positive and Negative Views of God in Childhood
While developing an understanding of the nature of God, children are (often) in parallel coming to understand what it means to be in a relationship with God (e.g., Granqvist and Dickie 2006;Kirkpatrick and Shaver 1990;Rizzuto 1979).Studies examining this developmental process have explored the ways in which children associate positive and negative valence with God.To examine how young children think about God, Dickie et al.
Religions 2023, 14, 1181 2 of 15 (1997) established a measure for capturing how children understand God as an active agent with a 23-item questionnaire, including both positive (e.g., a loving God) and negative (e.g., an angry God) components.Researchers have leveraged this scale to study whether children perceive God through specific affectively and relationally valenced dimensions (e.g., a caring God and a punishing God) (De Roos et al. 2004).The findings indicated these measures reveal distinctive dimensions rather than opposite poles on one scale (De Roos et al. 2001a, 2001b).In other words, a child can understand God as an agent with both negative and positive relational components (e.g., God as both caring and punishing).
More specifically, De Roos et al. (2001a) identified five components to the God concept in 4-to 7-year-old Dutch children: caring God (e.g., God helps people), potency of God (e.g., God sees everything), punishing God (e.g., God is angry when you do something bad), God as a loving friend (e.g., God loves me), and God as parents (e.g., God looks as daddy does).The findings indicated religious group differences (state school, Catholic school, Dutch Reformed school, and Orthodox Reformed school) in some of these dimensions but not others.Dutch and Orthodox Reformed children had higher scores on God being caring and loving than other participants, and Orthodox Reformed children had higher scores than other children on God's potency.De Roos et al. (2003) highlighted how different schools associated with different Christian denominations differentially emphasize the extent to which God punishes wrongdoing.However, there were no religious differences in children's views of God as punishing, and mean values for scores on God as punishing were below those of the other, more positive characteristics of God (De Roos et al. 2001a).
Other studies have similarly indicated Christian children do not perceive God as punishing (De Roos 2006) and tend to focus on God's love rather than God's authority (Hertel and Donahue 1995;Nelsen et al. 1977;De Roos et al. 2001a;De Roos 2006).When children do report a view of God as punishing, this view has been related to parental child rearing practices, which predicted children's view of a punishing, strict, and powerful God (De Roos et al. 2004).In other words, the more parents practiced strict and assertive parenting styles, the more likely their children were to perceive God as angry, punishing, and powerful.
In summary, the few studies examining how young children feel about and relate to God reveal that children raised in religious homes develop a relationally dynamic understanding of God as primarily caring, but also occasionally punishing.However, these prior studies have been conducted mostly with children from Christian homes and backgrounds.Thus, the current study aims to explore religious diversity in how children feel about God.Research on other aspects of children's concepts of God has increasingly uncovered diversity in those concepts associated with religious affiliation.

Religious Diversity in Early Concepts of God
Within cognitive approaches, much of the research on the developing concept of God has examined how children incorporate and coordinate psychological, biological, and physical features into the God concept (e.g., Barrett et al. 2003;Knight et al. 2004;Kiessling and Perner 2014;Lane et al. 2010;Makris and Pnevmatikos 2007;Shtulman 2008).These programs of research have highlighted that children's concepts of God incorporate human-like properties to varying degrees based on religious context.
In general, studies have documented a shift over early childhood from children saying that God has characteristics indicative of having a body and being alive to being free of such constraints (e.g., Shtulman 2008).By the age of 5, children generally incorporate the understanding that God is omniscient into their concept of God (see Heiphetz et al. 2016).In addition, around the age of 5, children in Spain, Israel, and the UK acknowledged that God is not constrained by biological, life-cycle processes, such as birth and death (Burdett and Barrett 2016;Giménez-Dasí et al. 2005).In another example with 3-to 7-year-old children from varying religious traditions in the US, Muslim children were less likely to associate embodied biological (e.g., needs to eat food), physical (e.g., becomes wet in the rain), and psychological (e.g., forgetting) traits to God than Protestant, Catholic, and Non-Affiliated children (Richert et al. 2016).
Studies have additionally examined the utility of concepts of God, such as the ways in which children incorporate God into their understanding of cause and effect.For example, studies with children in the US have found that children are more likely to say an impossible event in a story could occur if there is a supernatural agent in the story (Corriveau et al. 2015;Woolley and Cox 2007;Woolley et al. 2011).In addition, children are more likely to say that God could perform impossible feats (e.g., be able to lift a house) (Boyatzis 2005;Burdett 2020).
This growing body of research has highlighted how the conceptual features embedded in concepts of God do (or do not) correspond with other concepts children are developing, and newer findings are demonstrating conditions under which children utilize their concept of God in reasoning about how and when different kinds of things can (or cannot happen).However, this body of research has yet to examine a critical function of the concept of God, that of forming an interpersonal bond and emotional connection with a divine being (Granqvist and Dickie 2006).
Some researchers have highlighted the ways in which a perceived relationship with God is similar to or meets similar needs as attachment relationships (e.g., Kirkpatrick and Shaver 1990;Granqvist and Kirkpatrick 2008;Granqvist et al. 2010).For example, Granqvist et al. (2007) studied the feelings of closeness to God in 5-to 7-year-old children from both non-religious and Christian homes.Children (mostly Christian children in these studies) indicated feeling closer to God in situations expected to activate the attachment system (death, illness, and pain) than more neutral situations (good and bad mood).Other researchers have highlighted how deeply relationality is connected with religious beliefs and spirituality (King et al. 2013;King and Boyatzis 2015;Mahoney 2010).Through both lenses, it is critical to understand how children's religious experiences become incorporated into their understanding of God's emotional and relational characteristics.A first step is to explore if children's religious affiliation and engagement with religion is related to their understanding of God as an emotional and relational being.

Current Study
The current study aims to provide descriptive data on religious diversity in children's concepts of God as emotional and relational.Although studies with adults have increasingly documented the importance of implicit representations of God utilizing measures, such as the Implicit Associations God Representation (IAGR) (Stulp et al. 2021), the current study aims to build on prior research examining children's explicit attributions of properties to God (e.g., Dickie et al. 1997).Because prior studies have documented the relation between engagement in religious practices and children's explicit attribution of body-dependent processes to God (Saide and Richert 2020;Saide and Richert 2022), the current study also includes a measure of religious engagement to explore relations with religious engagement.

Results
Children indicated their certainty that God loves them, comforts them when they are sad, helps people, scares them, punishes them, and is angry with them.Analyses (described in the Procedure Section below) indicated that the positively and negatively valenced items do not hold together as distinct factors; and each item was analyzed separately.
A univariate ANOVA with Tukey's post hoc tests also indicated a significant, large effect of religious affiliation on religious exposure, F(3,245) = 48.804,p < 0.001,  = 0.364.As depicted in Figure 1, religious exposure for the Muslim participants was significantly higher than religious exposure for Catholic (mean difference = 1.271; 95% CI: 0.454, 2.088; p < 0.001) and Non-Affiliated (mean difference = 3.527; 95% CI: 2.685, 4.369; p < 0.001) participants, but not for Protestant participants.Protestant participants also had a significantly higher religious exposure than Catholic (mean difference = 1.032; 95% CI: 0.232, 1.832; p = 0.005) and Non-Affiliated (mean difference = 3.289; 95% CI: 2.463, 4.113; p < 0.001) participants.Finally, Catholic participants had a significantly higher religious exposure than Non-Affiliated participants (mean difference = 2.256; 95% CI: 1.355, 3.156; p < 0.001).The overall ratings from children for each property are presented in Figure 2. Analyses exploring how each property was related to the others were conducted using Wilcoxon signed-rank tests with Bonferroni corrections for multiple tests (n = 13), which set the critical p-value at 0.004.Within-valence ratings indicated modest stability.Ratings for love and help did not differ from each other; ratings for comfort were significantly lower than ratings for both help, Z = 6.141, p < 0.001, r 2 = 0.15, and love, Z = 6.444, p < 0.001, r 2 = 0.17, but these differences were small.This was similar to the negatively valenced properties: ratings for angry were significantly higher than ratings for both scare, Z = 6.496, p < 0.001, r 2 = 0.17, and punish, Z = 5.977, p < 0.001, r 2 = 0.14; but again, the effect sizes were small, and scare and punish were not significantly different from each other.
Cross-valance analyses indicated large, significant differences between ratings for love and negatively valanced items: scare (Z = 11.530,p < 0.001, r 2 = 0.54), punish (Z = 11.291,p < 0.001, r 2 = 0.51), and anger (Z = 9.121, p < 0.001, r 2 = 0.34).There were also sig- The overall ratings from children for each property are presented in Figure 2. Analyses exploring how each property was related to the others were conducted using Wilcoxon signed-rank tests with Bonferroni corrections for multiple tests (n = 13), which set the critical p-value at 0.004.Within-valence ratings indicated modest stability.Ratings for love and help did not differ from each other; ratings for comfort were significantly lower than ratings for both help, Z = 6.141, p < 0.001, r 2 = 0.15, and love, Z = 6.444, p < 0.001, r 2 = 0.17, but these differences were small.This was similar to the negatively valenced properties: ratings for angry were significantly higher than ratings for both scare, Z = 6.496, p < 0.001, r 2 = 0.17, and punish, Z = 5.977, p < 0.001, r 2 = 0.14; but again, the effect sizes were small, and scare and punish were not significantly different from each other.
nificant differences between ratings for comfort and negatively valanced items: scare (Z = 9.696, p < 0.001, r 2 = 0.38), punish (Z = 8.617, p < 0.001, r 2 = 0.30), and anger (Z = 4.896, p < 0.001, r 2 = 0.10).In addition, ratings for help significantly differed from negatively valanced items: scare (Z = 11.273,p < 0.001, r 2 = 0.51), punish (Z = 10.921,p < 0.001, r 2 = 0.48), and anger (Z = 8.437, p < 0.001, r 2 = 0.29).In sum, children's ratings were the highest for love and help, followed by comfort and then anger.Children were least likely to say that God punishes and scares.Further analysis with Spearman correlations examined the relations among the emotional characteristics of God (see Table 1).These analyses indicated characteristics that De Roos et al. ( 2001b) associated with a caring God (love, comfort, and help) were strongly correlated with each other, and characteristics associated with a punishing God (scare, punish, and angry) were also correlated with each other, although less strongly.Additionally, in general, caring God characteristics were negatively correlated with punishing God characteristics, although only moderately or weakly.Of particular note, ratings of God as angry had a unique pattern of correlations, being only weakly but significantly, positively correlated with both positively and negatively valenced properties (although correlations between angry and scare were only trending toward significance).Further analysis with Spearman correlations examined the relations among the emotional characteristics of God (see Table 1).These analyses indicated characteristics that De Roos et al. ( 2001b) associated with a caring God (love, comfort, and help) were strongly correlated with each other, and characteristics associated with a punishing God (scare, punish, and angry) were also correlated with each other, although less strongly.Additionally, in general, caring God characteristics were negatively correlated with punishing God characteristics, although only moderately or weakly.Of particular note, ratings of God as angry had a unique pattern of correlations, being only weakly but significantly, positively correlated with both positively and negatively valenced properties (although correlations between angry and scare were only trending toward significance).Additional preliminary analyses tested Spearman correlations between each response and age and religious exposure (see Table 2).These analyses revealed older children are more certain that God does not scare or punish them and more certain that God loves, comforts, and helps them.In addition, more frequent religious exposure was related to an increased certainty that God loves, comforts, helps, and is angry, but not related to whether God scares or punishes.As there was a small correlation between age and religious exposure trending toward significance, r = 0.121, p = 0.056, we also tested the partial correlation between each variable and religious exposure while controlling for age, as well as between each variable and age while controlling for religious exposure (see Table 2).All significant correlations with religious exposure remained after partialling out the effects of age, but partialling out the effects of religious exposure removed the significant correlation between age and ratings about God's helping and punishing behaviors.
The next set of analyses utilized a univariate analysis of variance (ANOVA) with Tukey's post hoc analysis to explore religious group differences in attributions to God (see Table 3).Muslim children indicated strong certainty that God loves, comforts, and helps and does not scare or punish.In contrast to children in other groups, Muslim children were more likely to indicate moderate certainty that God becomes angry when someone does something wrong.Protestant and Catholic children indicated strong certainty that God loves them and helps people, and less certainty that God comforts them.Protestant and Catholic children also indicated moderate certainty that God does not scare or punish them, but were generally uncertain about whether God becomes angry.Non-Affiliated children were moderately certain that God loves and helps, were unsure of whether God comforts or becomes angry, and were moderately certain that God does not scare or punish.Given the significant correlations with age and religious affiliation, religious affiliation differences in age and religious exposure, and correlations among positive and negative properties, two multivariate analyses of covariance (one for positive valence and one for negative valence) tested for omnibus effects of religious affiliation, religious exposure, and age, as well as if religious group differences for each of the characteristics of God remained after controlling for religious exposure and age.

Discussion
The goal of the current study was to examine the extent to which concepts of God incorporate features of God as an emotional and relational being in early childhood.Research into the relationality of young children's concepts of God is critical, as children's interactions with God are profoundly relational (Granqvist et al. 2007;King and Boyatzis 2015) and may reflect how children utilize their existing relationships with parents and other adults to develop an understanding of what God is like (De Roos et al. 2001b;De Roos et al. 2004).As little is known empirically about how children come to conceptualize God's emotional and relational nature, much less about religious diversity in such aspects of the concept of God, the current study involved interviewing children from a variety of religious backgrounds about whether God has six properties: love, comfort, help, scare, punish, and angry.The analyses found relations between children's certainty that God has these properties and children's (a) age, (b) religious exposure, and (c) religious affiliation.In addition, the properties were related to each other in varying ways that suggest critical questions about the relationally dynamic nature of children's concept of God.

God as an Emotional and Relational Being
The findings suggest that, in general, the children viewed God as having positively valenced properties, although this view was neither exclusive nor ubiquitous.There was a broad agreement among the children that God helps and loves them and does not punish or scare them.The fact that children associate emotional and relational properties to God at all suggests theoretically intriguing questions about the cognitive and social processes involved in associating psycho-biological traits, such as emotions, to a disembodied agent.
The development of emotion recognition requires children to incorporate contextual information into their understanding of facial, vocal, and physical cues to predict or determine the emotion an individual is feeling at a particular point in time or in a particular setting (Denervaud et al. 2020).In other words, children's developing definitions of emotional states incorporate both things that are happening in the environment as well as things that are happening in a body.The findings in the current study indicate that, as children are increasingly likely to indicate that God does not have a body, they are also increasingly likely to indicate that God has certain emotional properties.This pattern is supported by prior research indicating decreasing anthropomorphism in children's concepts of God with age (e.g., Shtulman 2008) as well as religious differences in anthropomorphism in young children.More specifically, in prior studies, Muslim children have tended to have a less embodied and anthropomorphic concept of God (Giménez-Dasí et al. 2005;Lane et al. 2010).Yet, in the current study, Muslim children were as likely (if not more likely) than children from other religious traditions to indicate certainty that God loves, helps, comforts, and becomes angry.These findings suggest that, in certain contexts and interactions with abstract agents, children do not view embodiment as a necessary and sufficient characteristic of an agent for that agent to feel an emotion.Indeed, across religious traditions, and even for non-affiliated families, children view God as an agent with a rich, emotional life that interacts with the child's own emotional states (loving the child and comforting the child).
The exploration of some specific emotions reveals some intriguing possibilities.In particular, there was less agreement among children about whether or not God comforts them or becomes angry with them.Views of God's anger were especially variable among children.Furthermore, increased certainty that God does become angry was positively associated with an increased certainty that God has additional both positive and negative emotional properties.This particular pattern is different from other perceptions, which were more likely to cohere by valence.Any explanation of this pattern would be purely speculative at this point, as this study was exploratory; however, we propose two particular possibilities as avenues for further research.
One possible explanation is that the wording of the question about anger was somewhat different from the wording of the other questions.When asked about anger, children were asked "does God get angry," which requires children to reflect on and/or infer a general characteristic of God.In contrast, other questions were worded more with a focus on how God makes the child themselves feel (e.g., "does God love you," "does God scare you").This compromise in wording was negotiated with the relevant campus IRB for ethical concerns raised at the time about asking children if God becomes angry with them, but the general (rather than specific) nature of the question may explain the lower levels of certainty reported by children regardless of whether they indicated God becomes angry or not.That being said, children were more evenly split on their response to this question than any other question, especially about more negatively valenced characteristics.
Another, and not mutually exclusive, explanation is about how anger fits into children's developing working models of relational partners.It may be that anger is perceived differentially, and the purpose of anger may be different depending on the particular context.For example, God may be angry, but God can also be comforting and loving at the same time, because God is angry "out of love."Bowlby's (1973) attachment theory echoes this experience of anger: secure attachment working models endorse beliefs that the other party has good intentions and that negative behaviors (e.g., displaying anger) of the other partner can be reverted and are not long-lasting.In this perspective, anger might be perceived as beneficial for interpersonal and relational health between two parties, and may not necessarily always be regarded in a negative light.Empirical studies in adults also generally seem to be in agreement.Studies looking at Christian samples showed that secure attachment to God was inversely related to psychological distress and emotional problems (Ellison et al. 2014;Hiebler-Ragger et al. 2016).Furthermore, this finding regarding secure attachment to God having a negative relationship to psychological distress and emotional problems also holds true for a Muslim sample, which is especially interesting given that Allah is noted to be less personal than God in Christian samples (Miner et al. 2017).This finding also applied to a Jewish sample, a group in which the religious faith is more tied to communal interactions with God rather than one-on-one relational interactions with God (Pirutinsky et al. 2019).As such, future studies should further elucidate the connection of God's relational properties to children's attachment and investigate the influence of differential perceptions of the intention of an emotion in this kind of relationship.
This pattern of findings suggests future research should examine how each of these factors may have a unique developmental trajectory that varies by religious group influences and goals when it comes to children's formation of a relationship with God.For example, studies could investigate when and how context (e.g., during ritual participation and in the presence of religious imagery) plays a role in children's perception about God's relational properties.

Age and General Religious Exposure
Regarding age, older children were more certain than younger children that God loves, helps, and comforts them, and were also more certain that God does not scare or punish them.Interestingly, ratings about whether or not God becomes angry were unrelated to age.However, the finding that certainty about the other aspects of God were associated with age reveals several potential developmental possibilities.
One possibility is that age serves as a proxy for the general development of cognitive skills in this sample.The cognitive skills necessary to answer the specific questions asked of children involve the ability to engage in perspective-taking, switch between positively and negatively valenced concepts, and balance the potential cognitive dissonance that may arise when switching between these concepts.The developmental period represented in this sample (ages 3.5 to 7) is a time in which children are increasingly developing executive functioning skills that incorporate rule switching and cognitive flexibility in service of particular task-related or developmental goals (Doebel 2020).This age window also reflects the period in which children develop increased ability to place themselves into the mind of others, including religious entities (Heiphetz et al. 2016;Lane et al. 2010).As both the significant positive and negative correlations indicate increased certainty that God does or does not have these properties, respectively, the relations with age could indicate children's increasing ability to flexibly coordinate all of these properties into their concept of God.
Another possibility is that age serves as a proxy for children's experiences with God, which would indicate a critical role of religious experience and exposure in children's view of God as an emotional and relational being.Findings related to religious exposure partially support this interpretation.Religious exposure on its own was significantly related to children's certainty that God has positively valenced properties and anger.However, unlike age, religious exposure was not related to the other negatively valenced properties (i.e., scare and punish).Furthermore, partial correlations indicated the effects of age disappeared when partialling out the effects of religious exposure for help and punish, and effect sizes decreased for certainty about love, comfort, and scare.Finally, in the current sample, age and religious exposure had a small relation to each other, but the relation only trended toward being statistically significant.The differential patterns with age and general religious exposure suggest there is some overlap in what these variables are capturing, but also that each variable reflects unique developmental influences on children's views of God.
Thus, when considering religious experience in general, it is noteworthy that, when controlling for the effects of age, the correlations between religious exposure and perceptions of love, comfort, and help actually increased.This finding emphasizes the importance of considering the role of religious exposure in shaping perceptions of God.Prior studies have suggested parents' beliefs play a particularly important role in their children's certainty that God exists in cultural contexts with lower levels of community consensus about God (i.e., China; Cui et al. 2020) when compared with cultural contexts with widespread community consensus about God's existence (i.e., Iran; Davoodi et al. 2019).From this perspective, the findings in the current study related to differences by religious affiliation can be informative to understand the ways in which religious exposure relates to children's concepts of God.

Religious Affiliation
Preliminary, exploratory analyses of religious group differences indicated no group differences in views that God helps, scares, or punishes.In contrast, Protestant and Muslim children were more certain than Non-Affiliated children that God loves and comforts; and Muslim participants were more certain than Non-Affiliated participants that God becomes angry.Importantly, preliminary differences in perceptions of God between religious groups disappeared when controlling for differences in age and religious exposure, suggesting that these factors may be potential mechanisms of developmental change.In these models, age remained a significant predictor and the effects of religious affiliation and religious exposure disappeared, suggesting they account for overlapping variance.Although not directly measured, the implication of these findings is that children from different religious communities absorb different messages about God and what a relationship with God should be as they engage more frequently in religious instruction and activities.
Regarding group-level religious affiliation differences, early childhood provides a unique developmental phase in which children learn to distinguish in-group members from out-group members by familiarizing themselves with the complex beliefs and practices of their social groups (Legare and Nielsen 2015).Group effects may provide insight into how children's beliefs about God's relational characteristics are on their way to aligning with the beliefs of their religious community.Children may be motivated to align their beliefs with those of their religious community, as social cohesion and sharing a system of beliefs may function to foster the longevity of groups and allow their members to connect and cooperate with each other (Watson-Jones and Legare 2016).Future research is needed to shed light on how children come to understand the relational views of God within their social communities.For example, children's concept of God as an active agent in their everyday lives may stem from the caregiver's assumed responsibility of transmitting religious beliefs through direct socialization processes, such as by providing them with specific imagery, testimony, and biblical content (King and Boyatzis 2015).From this perspective, it is critical to consider family-level factors in addition to social-group-level factors.
Parent-child relationships are postulated to be one of the primary contexts for the development of individual differences in children's relational views of God.Prior research indicates that Christian mothers who have a concept of a loving God have children with a concept of a powerful, loving, and caring God (De Roos et al. 2004).Furthermore, when fathers and mothers stressed the importance of traditional goals for religious education, their 5-year-old children were more likely to perceive God as a positive religious agent that is powerful, helping, and loving (De Roos 2006).The proposed relationship between a parents' concept of God and children's concept of God's relational characteristics may be explained by the role many parents take on at home as their children's "religious anchor" (King and Boyatzis 2015).For example, through transmitting religious beliefs to their children through indirect or direct processes, such as sharing testimony, children's relational views of God may be more dependent on religious socialization in the home than in direct, formal religious instruction or observation (Harris et al. 2018).Thus, future studies should delineate the goals parents and communities have for children's religious involvement and developing relationship with God.

Limitations and Future Research
A key limitation of the current study is the methodological approach.In particular, the research method utilized a small set of structured questions to assess children's explicit attributions of traits to God.As such, the interpretation of the findings was limited by children's own working definitions of the states and traits they were asked about.In particular, the ability to adequately recognize emotions may be fundamental to infer another entity's emotions (Strand et al. 2016).Additionally, the process of emotion-label acquisition likely assists with categorizing emotions from more broad concepts based on valence (e.g., negative versus positive) to more narrow concepts based on distinctive properties (Ruba and Pollak 2020).Future studies should directly assess children's emotion recognition skills and their understanding of emotion terms; as such, these skills may be foundational to children's perception and certainty of God's emotional and relational nature.
Additionally, future studies should incorporate implicit narrative measures of children's understanding of the nature of God.Implicit measures, such as the IAGR, can provide "a richer and more personalized and contextualized picture of how people view and relate with God" (Stulp et al. 2021).Utilizing implicit and narrative measures will be critical for research aiming to understand not only how children view God, but also how children view God as a relational partner.
A related limitation of the current study is that it did not (even explicitly) directly examine children's relationship with God or what parents and communities encourage and value in that relationship.One intriguing finding from the current study is that Muslim children reported the greatest certainty that God both provides comfort but also can sometimes become angry.The ways in which these aspects of God are incorporated into a full understanding of what God is like and support a developing relationship with God are likely to differ from children in other religious traditions with differing attributions in their concept of God.It will be important for future research to consider questions such as what is emphasized for children at different ages in different communities and the differing expectations for children in differing communities about their understanding of God and how much agency they are expected to display in their developing relationship with God (King and Boyatzis 2015;Rogoff et al. 2015).
Finally, future research could additionally explore how relational interactions with an abstract being, such as God, is related to children's general understanding of emotions and emotions in social interactions.Although the relations with age suggest the need to examine age-related cognitive and socio-emotional factors in children's attributions of emotional valence to God, the reciprocal pathway is also critical to be understood.Prior research has begun to delineate the developmental affordances of children's engagement with other abstract beings in their fantasy play (Gilpin et al. 2015;Seja and Russ 1999) and imagination (Kushnir 2022;Thibodeau-Nielsen et al. 2021).It may be that the experience of engaging with an abstract being, such as God, can provide children the cognitive space to not only practice exploring their own emotional capacity, but also the emotional capacities of others.

Participants
Participants were 296 children (54.4% female) between the ages of 3.31 and 6.98 (M = 4.627, SD = 0.804).Thirty-two (10.8%) participants were excluded from analyses for varying reasons: child distracted/unable to finish interview (n = 16), assent/consent withdrawn (n = 5), did not meet eligibility requirements (e.g., language barriers or developmental delays; n = 7), or other (e.g., experimenter error; n = 4).Participants who indicated a Jewish religious affiliation were also excluded (n = 10), as that sample was too small for analyses of group-level comparisons.

Religious Exposure
Religious exposure was assessed as an average of four parent-report items (ranging from "never" [0] to "multiple times a day" [8]) (Saide and Richert 2020); parents reported how often the child (a) attends events sponsored by their religious organization, (b) par-ticipates in public religious practice, (c) participates in private religious practices, and (d) receives formal religious training (Cronbach's = 0.850; M = 3.322, SD = 2.158).

Procedure
The data analyzed in this manuscript were from the first wave of a six-wave longitudinal study into the development of religious cognition in early childhood (Shaman et al. 2016).Data collection occurred from March 2013 to June 2016.Child participants completed various tasks assessing their attribution of traits to God and humans and their understanding of and beliefs about prayer, as well as their theory of mind understanding and executive functioning.The data analyzed for the current study were not previously published.
The interview was broken into two sections: questions about God and questions about prayer.Half of the children received the questions about God first, and half of the children received the questions about prayer first.The questions about God's emotional and relational properties were included with the questions about God and always followed a set of questions asking about other characteristics of God, including biological (e.g., does God need to eat food), psychological (e.g., does God get bored), and physical (e.g., could you touch God).
For the measures specifically included in this analysis, children were asked: "Does God love you?", "Does God comfort you when you are sad?", "Does God help people?","Does God scare you?", "Does God punish you often?", and "Does God get angry when you do something bad?".After responding "yes" or "no" to each question, children were then asked, "Are you really sure or a little sure?".This method was chosen as it was in the same format as other questions participants were asked in the interview; thus, children were familiar with the structure and response options.Of note, the interview did not incorporate follow-up questions asking children for more reflection on how, when, or why they answered as they did.The constraints on what can be concluded from this method are discussed in the Limitations Section.
Children participated in the study either in a lab setting or in their homes (determined by parent preference).While children completed the interview (approx. 1 h), parents filled out a survey assessing demographic information about the child, their own religious beliefs and concepts of God, and their religious socialization practices with the child participant.Parents received USD 20 for their participation, and children chose a USD 1 toy.

Conclusions
This study described the ways in which children from four religious backgrounds (Protestant, Catholic, Muslim, and Non-Affiliated) view God as an emotional and relational being.To our knowledge, this is the first study that explores the relation between religious exposure, religious affiliation, and age with the concept of God's emotional and relational properties in early childhood.This study highlights that not only do children think about God's physical, biological, and cognitive capabilities, but also think about God's capabilities, especially in a relational context.

Institutional Review Board Statement:
The study was conducted in accordance with the Declaration of Helsinki, and approved by the Institutional Review Board (or Ethics Committee) of University of California, Riverside (HS-12-064, approved May 2012).
Informed Consent Statement: Informed consent was obtained from all subjects involved in the study.

Figure 1 .
Figure 1.Mean religious exposure by religious affiliation.

Figure 1 .
Figure 1.Mean religious exposure by religious affiliation.

Table 2 .
Spearman correlations between God concept items and age and religious exposure.

Table 3 .
Means (and standard deviations)for God concept items by religious group, with F-values and effect sizes.