Attachment Theory and Concepts of God

We replicated and extended Buri and Mueller’s study, which tested whether one’s God concept is primarily self- or parent-referenced. Following their findings, we predicted that God concepts would primarily be self-referenced. However, we also predicted, in line with attachment theory, that God concepts would be influenced to some degree by attachment to mother. We found our predicted result in a sample of university students (N = 223). However, when we examined the question by gender, we found significantly different patterns. Concepts of God as Loving, Controlling, and Distant were self-referenced in women but were not so in men. In men, Loving God was predicted primarily by attachment to mother, Controlling God was referenced to attachment to both parents, and Distant God was related to a combination of viewing self as distant and experiencing parental attachment difficulties, primarily with father.

The psychological source of our concept of God has been a topic of discussion ever since Freud (1910Freud ( /1989Freud ( , 1913Freud ( /1952) speculated on the issue. Over the years, research has examined various sources, for example, parents, the self, and the larger social context, such as faith groups . In this study, we have chosen to focus on whether facets of our concept of God are referenced primarily to our parents or to our view of self. We also planned to explore whether there are differences influenced by gender. Buri and Mueller (1993) tested the prediction of psychoanalytic theory that the formation of one's God concept is largely a function of the experience of one's parents. They pitted the parent-referenced hypothesis against a rival hypothesis that one's concept of God as loving is largely selfreferenced. Contrary to the result predicted by psychoanalytic theory, Buri and Mueller found that a loving concept of God is referenced primarily to participants' self-esteem rather than their experience of their parents' nurturance or authority.
In the present study, we have chosen to replicate and extend Buri and Mueller's (1993) basic question, albeit from a different but related theoretical perspective, that of attachment theory. Attachment theory suggests the quality of the early experience of one's primary caretakers affects one's expectations in relationships. Also, rather than self-esteem, we chose to focus on self-concept, which would help us to answer whether one's concept of self is similar to one's concept of God. Like Buri and Mueller, we predicted that we would find that the concept of God as loving would be predicted primarily by one's concept of self as loving; however, we also predicted some influence from maternal attachment. We also predicted similar results for the concept of God as controlling and as distant, that is, that the corresponding concept of self (as loving, controlling, or distant) would be the primary predictor of the God concept, even after controlling for one's attachment to each parent. Also, because research Spilka, Addison, & Rosensohn, 1975), including recent attachment research (Dickie, Ajega, Kobylak, & Nixon, 2006;Reinert & Edwards, 2012), has shown that the sources of men's and women's God concepts cannot be assumed to be identical, we decided to examine the God concepts in men and women separately to explore potential gender differences. Freud's (1910Freud's ( /1989Freud's ( , 1913Freud's ( /1952 claim that God is a psychological projection of one's father sparked both theoretical 560547S GOXXX10.1177/2158244014560547SAGE OpenReinert and Edwards and empirical exploration of the psychological origins of the concept of God over the years. Rizzuto (1979) extended Freud's view and suggested that the influence of both parents is the key source. A number of studies have investigated this question (for review, see Francis, 1997). Early studies testing the Freudian father hypothesis have produced mixed results (Beit-Hallahmi & Argyle, 1975;Spilka et al., 1975). Some studies suggest that the influence is due to the mother (Dickie et al., 2006;Tamayo & Desjardines, 1976), or to the combination of both parents (Birky & Ball, 1988;Tamayo & Desjardines, 1976).

Psychological Origin of God Concepts
Alternatives to the projection theory have also been proposed. For example, Benson and Spilka (1973), as well as others (e.g., Roberts, 1989;Spilka et al., 1975), suggested that self-esteem may play an important role in the formation of God concepts. Buri and Mueller (1993) designed a study to determine which had more influence on the nurturing concept of God, parents or self-esteem, and concluded that the loving God concept was best predicted by self-referencing, as measured by self-esteem, rather than parent-referencing, as measured by respondents' experience of parental nurturance and authority.

Attachment Theory
Another theory that attempts to explain the psychological origin of one's concept or image of God is attachment theory (Kirkpatrick, 1992;Kirkpatrick & Shaver, 1990). One key element of this theory that has made it appealing and has generated a rather large body of research is that attachment theory not only guides researchers in making a prediction about what they might find but also proposes a potential source of why those dynamics should be present.
Attachment theory (Bowlby, 1969) proposes that a powerful affectional bond forms between an infant and the infant's primary caretakers, usually the parents, during the early stages of life. The primary purpose of this bond is for maintaining security and safety for the vulnerable child. Bowlby (1979) theorized that the child forms internal working models (IWM) of both self and others that reflect the degree of security or insecurity experienced in the parentchild bond. He further theorized that the characteristics of those bonds and subsequent IWMs influence the quality of one's relationships with others and one's sense of self over the life span. Kirkpatrick (1992;Kirkpatrick & Shaver, 1990) proposed that attachment theory can help explain religious dynamics among those who relate to a personal God, and found that among those who were securely attached to their parents, there was a corresponding attachment to God. Granqvist (1998Granqvist ( , 2002Granqvist ( , 2010Granqvist & Hagekull, 1999) found similar results and proposed that the early parental attachment security provides the conditions for the child's later social learning to take place. The child's IWMs of self and others influence the development of the subsequent concept of God and the degree to which one describes God as loving or harsh, close or distant (Cassibba, Granqvist, Costantini, & Gatto, 2008).
Attachment theory has informed a rather large body of empirical research in the psychology of religion (see Granqvist, 2010;Kirkpatrick, 1992), including on the concept of God. For example, research has shown the connection between secure attachment to parents with a loving God image (Granqvist, Mikulincer, Gewirtz, & Shaver, 2012;Kirkpatrick, 1998;Kirkpatrick & Shaver, 1992;Reinert & Edwards, 2009). In contrast, an insecure interpersonal attachment to parents has been found to be associated with distant and controlling God images (Granqvist et al., 2012;Kirkpatrick & Shaver, 1992) and is inversely associated with the concept of God as loving (Granqvist et al., 2012;Reinert & Edwards, 2009). Dickie et al. (2006) found that although attachment to parents influenced how young adults perceived God, it was attachment to mother that was the primary influence on the development of their children's God concepts. Buri and Mueller (1993) found that participants' self-esteem was a stronger predictor of the concept of a loving God than was participants' experience of their parents as nurturing and the style of authority they exercised. However, a question left unanswered by self-esteem studies is whether one's concept of oneself, as opposed to one's self-esteem, is predictive of one's concept of God. As Dickie et al. (2006) noted, although self-concept and self-esteem are related, they are not the same. The former is how one describes oneself and the latter is how positively one feels about oneself. Also, it is not known whether having negative conceptions of self are predictive of viewing God similarly on those same negative dimensions, such as viewing oneself as controlling, or oneself as distant from others.

Self-Esteem and Self-Concept
Self-esteem and self-concept are certainly related concepts and those who have written about the self or self-image have used a variety of definitions to describe various aspects of the self (Blyth & Traeger, 1983). Blyth and Traeger (1983) noted that the literature generally uses the term self-concept as a descriptive or a non-judgmental term to describe the self, whereas the term self-esteem is generally understood to be an evaluation of the self. Greenwald et al. (2002, p. 5) proposed that self-esteem be defined as "the association of the concept of self with a valence attribute" and that self-concept be defined as "the association of the concept of self with one or more (nonvalence) attribute concepts." One of the ways in which we wanted to extend Buri and Mueller (1993) was to propose that one's self-concept might be predictive of one's God concept. Buri and Mueller's choice of self-image limited the dimension of the self that they could focus on, namely, the valence (i.e., high vs. low self-esteem). With our proposed focus on self-concept, the description of self, we could investigate whether various concepts of self were related to a similar concept of God. For example, we could then ask whether one's concept of self as a person distant from others is predictive of whether one conceptualizes God also as distant.

God Concepts and Gender
Although some research has found little to no difference in God concepts between men and women, for example, Krejci (1998) found no differences except that males view God as more controlling than women, and Bassett and Williams (2003) found no gender differences, most studies have reported that God concepts differ in males and females. Likely, how God concepts are formed also differ in men and women . Ciarrocchi et al. (2002) summarized the general findings of the research, noting that women tend to see God as more loving, nurturing, and personal. In contrast, men tend to view God as more authoritarian, vindictive, malevolent, and irrelevant than women do. In addition, when influence of the self is taken into consideration, women tend to consistently have stronger relationships between their view of self and their image of God (Spilka et al., 1975). Typically for college-aged men, lower self-esteem is related to a more wrathful concept of God. For women, high self-esteem is related to loving, kindly, and close concept of God . When researchers explore the relative influence that same sex and opposite sex parents have on their young adult children's God concepts, they find consistently different patterns in the young men and women (Dickie et al., 2006;Reinert & Edwards, 2012). Spilka et al. (1975) commented that the gender differences that they discovered leave them " . . . questioning the appropriateness of combining the sexes in work of this type" (p. 157).
Because previous studies have frequently found differences in God concepts associated with gender differences, we decided to explore this dimension. However, given the somewhat disparate results of previous research and a lack of coherent theoretical rationale, we felt it would be premature to attempt to articulate a prediction of what we would find. Nevertheless, we decided it would be worthwhile to explore this area, and to see whether our findings might ultimately contribute to a deeper understanding of the formation of God concepts.

Research Hypotheses
From an attachment theory perspective (Bowlby, 1969(Bowlby, , 1979, when infants and young children experience consistent loving care from their primary caregivers, usually mother or mother-figure and father or father-figure, they develop an IWM of both self as lovable and capable of loving, and others as caring and loving. This IWM helps shape the person's expectations of others in relationship, as well as influences his or her concepts of others. So, based on attachment theory and past research (Dickie et al., 2006;Reinert & Edwards, 2012), we not only expected that participants' attachment to mother (see Dickie et al., 2006) or mother-figure would have some influence on their concept of God as loving, but also expected that their concept of self as loving would be the stronger predictor (Buri & Mueller, 1993). We expected the same pattern would model the concept of God as controlling. That is, we expected that there would be some influence from attachment to mother, but that the stronger predictor of God as controlling would be the concept of self as controlling. Similarly, we expected some influence by attachment to mother, but that God as distant would be predicted more strongly by one's concept of self as distant from others.
We expected that there would be gender differences, given the findings of past research (Dickie et al., 2006;Reinert & Edwards, 2012;Spilka et al., 1975). However, we made no formal predictions because the sources of the differences and the exact characteristics of those differences are not clear, either theoretically or empirically.

Method
We recruited 256 students from a mid-sized Midwestern liberal arts university but 33 failed to answer a significant number of items and were excluded from the study. Therefore, the final set of participants were 223 students, 82 males and 141 females, with an average age of 19.2 (SD = 2.77). Most were Caucasian (84%), and approximately 6% were African American, 4% Hispanic, and 2% Asian or Pacific Islander. The largest religious preference was Protestant (43%) and Catholic (22%), with 13% endorsing Other, and 22% no affliliation.

Procedures
We recruited general psychology students who were required to participate in one of several research projects for class credit, or to complete an alternate activity. Trained research assistants and graduate assistants administered the research materials outside class time and in large group settings.

Measures
Concept of God scales. We selected the Loving and Controlling God Scales (Benson & Spilka, 1973) to measure concepts of God. Each scale consists of five bipolar adjectives, for example, "accepting-rejecting," on which respondents rate the degree to which the adjective describes their view of God. We also chose the Distant God scale, an added scale created specifically for their attachment research by Kirkpatrick and Shaver (1990) using two of the Benson and Spilka filler items and three new items. Each scale was scored by summing responses, with higher scores indicating stronger preference for the named description. Cronbach's alpha for each of the scales in this study was, Loving God, .89; Controlling God, .74; and Distant God, .91.
Concept of self scales. To measure concepts of self that would correspond to the God concepts, namely, Loving, Controlling, and Distant, we adapted the above described concept of God scales (Benson & Spilka, 1973;Kirkpatrick & Shaver, 1990), by giving the instruction that participants should choose the adjective and rating that best describes their view of themselves. The resulting five-item scales reflected a rating of oneself as Loving, Controlling, and Distant and were scored in the same manner as the concept of God scales. Cronbach's alphas for each of the scales were, Self as Loving, .71; as Controlling, .66; and as Distant, .69.
Attachment to parents scales. We selected the Experiences in Close Relationships-Relationship Structures questionnaire (ECR-RS; Fraley, Heffernan, Vicary, & Brumbaugh, 2011) to measure attachment to mother and attachment to father. This 10-item Likert-type scale with 7 response options from strongly disagree to strongly agree was developed from the Experiences in Close Relationships-Revised Scale (Fraley, Waller, & Brennan, 2000) to form a brief and psychometrically sound scale that can be used to tap attachment to a precisely specified referent, such as mother, father, friend, or romantic partner. Fraley et al. (2011) demonstrated that the scale has two factors, Avoidance (6 items) and Anxiety (3 items), with the 10th item loading on both factors. Higher scores indicate greater avoidance or anxiety in the attachment relationship. For this project, we decided to use the full 10-item scale to measure the global attachment difficulties with each parent. Cronbach's alphas for the Avoidance and Anxiety subscales and the full Attachment to Mother Scale were .89, .86 and .89, respectively. For the subscales Avoidance and Anxiety, and the scale Attachment to Father, the reliability coefficients were .88, .93, and .91, respectively.

Results
Before examining our central research questions, we tested for gender differences among all the study's variables and found no significant differences using MANOVA, Wilks's lambda, (8, 214) = .938, p = .085. However, because the MANOVA approached significance, we report the variables that univariate contrasts show differences: Self as Loving and as Distant, p < .01, and God as Loving, Distant, and Controlling, p < .05 (see Table 1 for the means and standard deviations in males and females). We also found that the Pearson two-tailed correlations for this sample (N = 223) were in the expected range and similar to those reported in other studies (e.g., Reinert & Edwards, 2009Spilka et al., 1975). Loving God was related to Self as Loving (r = .30, p < .01) and inversely related to both Attachment to Father difficulties (r = −.18, p < .01) and Attachment to Mother difficulties (r = −.24, p < .01). Controlling God was related to Self as Controlling (r = .31, p < .01) and Attachment to Mother difficulties (r = .15, p < .05). However, Controlling God was not related to Attachment to Father (r = .11, p = ns). Distant God was related to Self as Distant (r = .44, p < .01), Attachment to Father difficulties (r = .18, p < .01), and Attachment to Mother difficulties (r = .27, p < .01).
Next, we attempted to replicate the findings of Buri and Mueller (1993) by testing each hypothesis using our entire sample (N = 223). In these analyses, we used a hierarchical multiple regression technique, with the concept of God, for example, Loving God, as the dependent variable. We entered the independent variable Self as Loving in Step 1, then in Step 2, we entered Attachment to Mother and Attachment to Father. Next, we reversed the order, entering the attachment variables in Step 1, and the concept of Self as Loving in Step 2. We tested the concepts Controlling God and Distant God using the same strategy. We found, as we had predicted, that the concepts of God as Loving, Controlling, and Distant are each predicted by the concept of self, respectively, as Loving, Controlling, and Distant, and also that Attachment to Mother contributed to the model for each concept of God. The strongest predictor in the model was the self-concept, but Attachment to Mother significantly added to the models. (Tables summarizing these analyses are available from the  first author.) Because past research Reinert & Edwards, 2012;Spilka et al., 1975) found gender differences among concepts of God and various self-and parent variables, we next calculated the Pierson correlations among the study variables for each gender. Table 1 presents the correlation matrix for males and females, along with means and standard deviations for the study variables in each gender. It is interesting to note that the concepts of God were significantly related to the reported attachment to each parent among the men, but the same pattern did not emerge among the women. In addition, the pattern of relationships among the concepts of God and the concepts of self seems to differ in women and men.
To test our first research question in males (n = 82), we used a multiple regression technique with Loving God as the dependent variable. In the first step of the model, we entered Self as Loving. It was not a significant predictor. In the second step, we entered the Attachment to Mother and to Father variables. Results (see Table 2) show that the model in Step 2 produced a significant R 2 change of .177, p < .001. We reversed the order of entry to evaluate the relative contributions of the independent variables. In Step 1 of this model, Attachment to Mother and to Father produced an R 2 change of .181, p < .001 and entering Self as Loving in Step 2 failed to account for any additional variance. Thus, we concluded that among the men, attachment to parents, in particular Attachment to Mother, was the significant predictor of the concept of God as Loving. Contrary to our prediction, the concept of Self as Loving failed to be a predictor of the Loving God concept.
We used the same modeling strategy in the females (n = 141). Loving God was significantly predicted by the concept of Self as Loving in Step 1, with a significant R 2 change of .147, p < .001. When we entered Attachment to Mother and to Father in Step 2, no additional variance was accounted for. When we reversed the order of entry, the finding that the concept of Self as Loving was predictive of the concept of a Loving God was confirmed. Although our prediction that the concept of Self as Loving would relate to the concept of God as Loving was supported, we did not find in these women the influence of Attachment to Mother that we expected. Table 3 shows the results of our analysis for the concept of God as Controlling. Similar to the above results in males, the model including Attachment to Mother and to Father together produced a significant R 2 change of .08, p < .05, and the concept of Self as Controlling failed to contribute to the model. In women, however, the concept of Self as Controlling was the predictor, with a significant R 2 change of .207, p < .001, and the attachment to parents failed to contribute to the model.
Finally, Table 4 presents the findings of our analysis for the concept of God as Distant. In males, results suggest that the combination of both the concept of Self as Distant and parental attachment difficulties, particularly difficulties in Attachment to Father, predict the concept of God as Distant. In women, the concept of Self as Distant is the strongest predictor, but difficulties in Attachment to Mother also contribute to the model. Note. Intercorrelations for female participants (n = 141) are presented above the diagonal and scale means and standard deviations for females are to the right of the matrix. Intercorrelations for male participants (n = 82) are presented below the diagonal and scale means and standard deviations for males are below the matrix. a Higher scores on the Experience in Close Relationships-Relationship Structures (ECR-RS) 10-item scale indicates higher levels of anxiety and avoidance in the attachment to the named person. *p < .05, two-tailed. **p < .01, two-tailed. Loving God in males (n = 82) Loving God in females (n = 141) Step

Discussion
We had expected, based on attachment theory, that participants' concept of God as Loving would be predicted both by their concept of themselves as loving, and also that it would be influenced by attachment to their parents, particularly their mothers. Attachment theory suggests that the relationship bonding between parent and child in the early stages of a child's life produces in the child an IWM of both self and others. In a secure attachment, the child develops an IWM of self being lovable and capable of love, expecting that others will manifest those caring and nurturing qualities. Kirkpatrick (1992) proposed that in religions that espouse a personal God, God can be conceptualized as a similar attachment figure, and therefore, similar attachment dynamics will operate in that relationship, as in any other. We decided to replicate the broad lines of Buri and Mueller's (1993) study in which they found that a nurturing concept of God is more strongly self-referenced, as measured by self-esteem, than parent-referenced. We decided to test the same question, only using participants' concepts of themselves rather than selfesteem. This allowed us to extend the question beyond the positive, nurturing concept of God, to more negative concepts of God, such as God as Controlling, and God as Distant. We asked participants to rate both God and self on the same set of adjectives, which would then provide an index of both God and self as Loving, as Controlling, and as Distant. In addition, we measured participants' level of attachment difficulties to each parent.
Step We predicted that participants' concept of self would be predictive of their concept of God, as Buri and Mueller (1993) had shown, but we also expected that the level of attachment to mother would also have an impact on the God concept. It is interesting to note that when we analyzed the data of the entire sample, we found that each God concept could be modeled as we had predicted, that one's concept of God would primarily be self-referenced but also influenced by attachment to mother. However, when we examined the data by gender, the results revealed significantly different patterns in males and females.
Broadly speaking, we found that God concepts for women were predicted most strongly by their concept of themselves. Thus, Loving God was best predicted by seeing themselves as loving, Controlling God by seeing controlling characteristics in themselves, and Distant God from their experience of themselves as distancing. Only in Distant God was the influence of their attachment difficulties with mother an additional contributing element to predicting the God concept. In this study, we found that women were much more self-referenced than parent-referenced.
In men, the findings were quite different. In men, the concept of God as Loving was parent-referenced, particularly due to attachment to mother. God as Distant was both selfand parent-referenced, particularly in experiencing father as distant. God as Controlling was primarily parent-referenced, with a combination of both father and mother attachment difficulties producing a significant but not particularly strong influence.
Clearly, there are gender differences between men and women in the formation of their concepts of God, but our study's design does not help us to tease out the reasons for such differences. We can only speculate on the source of our findings. One possibility is that some of the differences may be due to developmental issues in the formation of the selfconcept. Typically, women arrive at a stable sense of their self-identity at a younger age than do men. Because the average age of participants was 19, it is possible that the young women in this study, compared with the men, have a more developed, achieved sense of self at this point in their life. That is, the men may lagging behind developmentally, and may yet be relying more on their external relationships, that is, on their relationship with parents, to define themselves, whereas the women may have a more internalized sense of themselves. Future research on older adults may resolve the question of whether our findings reflect differential developmental issues.
An alternate possibility could be that our findings reflect differences that Cross and Madson (1997b) discussed when they surveyed the extensive literature on gender differences. They suggested that some of these differences may flow from different models of the self that men and women construct in the context of a particular culture. Men in the United States, Cross and Madson (1997b) suggested, have a selfconstrual that is more independent oriented, whereas women have a self-construal that is more interdependent. Both men and women desire interpersonal closeness, but may have different approaches to achieving that end. For example, women may tend to invest more in close relationships within an intimate group, whereas men may tend to seek connections in the larger social sphere (Baumeister & Sommer, 1997). Women, noted Cross and Madson (1997a), " . . . tend to incorporate close relationships into their own sense of self" (p. 52).
Whatever the explanation for the differences, our findings suggest that, psychologically, religious dynamics seem to be influenced significantly by gender dynamics. Going forward, we recommend that researchers studying God concepts take into consideration the potential role that gender differences may play. In this instance, when we analyzed data for the entire sample, it seemed that there was a rather straightforward model for predicting God concepts, namely, that the concepts are primarily self-referenced with some influence by attachment to mother. However, when examined by gender, it was clear that men are much more parent-referenced in their formation of God concepts, whereas women are more self-referenced.
Given that our focus on gender differences was exploratory, we are cautious in proposing implications based on those findings. However, counselors and spiritual guides may well be advised to recognize that concepts of the self and attachment history with parents may be related to how a person will tend to conceptualize God. This study, as well as related research, supports the notion that concepts of God and images of God are influenced by early attachments to key figures in one's early life, as well as by how one eventually comes to view oneself. Unfortunately, those who might most benefit from experiencing comfort in relating to a caring God may be the very persons who struggle with a harsher or more distant concept of God.
The study was limited by the usual limitations of paper and pencil self-report research. Also, because this research was conducted among emerging adults, and may reflect just this age group, we recommend that future research sample a wider age range to examine whether the same patterns are operative in the larger population. Also, these findings may not be generalizable beyond this study because it was based on a sample of convenience who responded to course requirements. In addition, it is unknown whether the results were skewed because a number of participants failed to respond to a substantial number of items and were thus eliminated from the study.

Conclusion
In this study, we set out to replicate and extend Buri and Mueller's (1993) study, which found that a nurturing God concept is predicted more by one's sense of self than by the influence of one's parents. We tested three God concepts to determine whether one's concept of self in the areas of loving, controlling, and distance would predict the corresponding God concept. We found that women tended to be more selfreferenced, whereas men relied more heavily on their experience of their attachment to parents in forming their God concepts. Because there is a growing body of results suggesting gender differences, we suggest that this area may be an increasingly fruitful area for further research that could reveal how men and women differentially approach their understanding of faith and relationship with God.

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

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