Ambivalent trust and ontological representations of death as latent factors of religiosity

Abstract The studies on the relationship between religiosity and the management of trust in postmodern society are an emerging field of research. The psychological contribute of the attachment theory shows how early relationships with the parental figures determine the adult attachment styles and the related dispositions to trust and religiosity. Lack of trust is a critical aspect of the insecure attachment styles: the avoidant and the anxious. This study in focused on these two traits, considering their level of trust compared to religiosity and representations of death. The Interpersonal Trust Survey, The World Health Organization Quality Of Life-Spirituality, Religiousness and Personal Beliefs, the Testoni Death Representations Scale and the Experiences in Close Relationships-Revised were utilized. The results illustrated different expressions of trust in avoidant and anxious styles.


ABOUT THE AUTHORS
The group is composed by psychologists (Ines Testoni, Davide Bisceglie, Valentina Pergher), a methodologist (Lucia Ronconi), and a physician (Enrico Facco). These specialists are interested in improving the studies inherent to the constructs, which constellate the area of spirituality and the representation of death. In particular, the group is interested in the issues inherent to hypnosis, near death experiences, transcendence, non ordinary mental experiences, considering how they influence the representations of death and the everyday life. In this study, we wanted to explore the connections among spirituality and faith in significant relationships. The next step of the collaboration regards the possible influence of trust in psychotherapy and hypnosis.

PUBLIC INTEREST STATEMENT
Literature has already widely highlighted the importance of religiosity for the management of everyday life. However, there are few studies focused on the relationships between faith and trust, despite the second one can be considered as a particular form of the first. In psychological and social contexts, trust has several meanings, however this concept typically refers to a situation characterized by the relationships where a trustor abandons control over the actions performed by the partner, accepting uncertainty. In this research, we wanted to consider how spirituality influences this kind of relationship and in particular we wanted to consider if the ontological representations of death assume a specific role in such an influence. A specific attention we dedicated to anxious and avoidant people because they are commonly considered to be lacking in trust.
of religiosity (Robertson, 2015a(Robertson, , 2015b. During the last two decades, important studies paid considerable attention to the issue of trust. In the area of interpersonal relationships and sociology, different definitions of this construct have been advanced. All of them share the assumption that it is inherent to the perceived benevolence to others (Holmes & Rempel, 1989), from which the improvement of communication, reciprocity, cooperation and cohesion derive (Coleman, 1990;Fukuyama, 1995;Putnam, 2001;Smith & Kulynych, 2002). In this perspective, trust entails the general beliefs and attitudes about the degree to which people are likely to be reliable, cooperative or helpful in daily-life contexts (Rotter, 1967), thanks to which a particular interdependence between the actors (trusters) and their partners (trustees) takes form (Holmes & Rempel, 1989). From a structural point of view, trust is substantially based on a system of faith, which permits to reduce uncertainty (Rempel, Holmes, & Zanna, 1985). In fact, ideological and religious factors intervene in the construction of such interdependence among individuals, through the transmission of moral artefacts and values (Erikson, 1968;Schwartz & Boehnke, 2004;Schwartz, Bukowski, & Aoki, 2006). The relationship between religiosity and trust is really cogent (Welch, Sikkink, Sartain, & Bond, 2004) because on one hand the latter facilitates the in-group communality promoted by the convictions arisen from the former and on the other hand lack of trust is a component of the inter-group conflicts (Frey & Tropp, 2006).
In the viewpoint of the dynamic psychology, also personal factors intervene in the disposition to trust. In particular, the attachment theory offers an important contribute showing how close relationships are mainly characterized by selective trust. These other-regarding preferences among kin promote cooperation, reciprocal altruism, direct and indirect reciprocity in long-term mutual benefit (Bowlby, 1969). From family early relationships, especially the parental ones, different working models derive which typify the development of specific attachment styles in childhood (secure, avoidant, anxious-resistant). These factors affect the way persons manage relationships with others, presenting different levels of trust, which tend to be consistent over time (Codato, Damian, Testoni, & Ronconi, 2013;Codato, Shaver, Testoni, & Ronconi, 2011;Hazan & Shaver, 1994;Mikulincer & Shaver, 2007). Following this perspective, a wide area of studies concerned with the transfer of such psychological models towards the religious dimension is developing. Desrosiers, Kelley, and Miller (2011) utilize the concept of "relational spirituality", to indicate how religiosity is characterized by a trustful and personal relationship with God or universe for guidance, Who/which are perceived as ever-present in daily life experiences and as benchmark in driving moral behaviour (Norenzayan, 2013). This sensation promotes the practice of forgiveness, and confidence towards other people. Some authors regained this idea, suggesting that attachment to caregivers is re-enacted in the attachment to God or to a universal being (Beck, 2006;McDonald, Beck, Allison, & Norsworthy, 2005;TenElshof & Furrow, 2000). In particular, Kirkpatrick andShaver (1990, 1992) described the similitude between relationships with parents and relationships with God. Within Abrahamic religions, God is seen as a paternal leader, guiding and protecting His offspring from evil and death. In a similar direction, further research explored in parallel, the correlations between religiosity, attachment styles and fear of death (Kirkpatrick, 2005). In particular, the mortality salience hypothesis of the Terror Management Theory (TMT) analysed how the fear of death influences romantic relationships, showing how the attachment styles affect the representation of symbolic immortality and the faith in an afterlife (i.e. Florian, Mikulincer, & Hirschberger, 2002;Hart, Shaver, & Goldenberg, 2005;Mikulincer, Florian, & Hirschberger, 2004). Furthermore, TMT illustrated how the cultural and religious defences against death anxiety influence the moral and altruistic behaviour (Jonas & Fischer, 2006;Testoni, Falletti, Visintin, Ronconi, & Zamperini, 2016). In this area, cultural and religious variables runs in parallel with close relationships and self-esteem, which all together constitute a Tripartite Security System, aimed to manage the terror deriving from the awareness of mortality (Hart et al., 2005). Other studies analysed how the representations of death and the religiosity modify the trust in the sense of existence, the reasons for living, well-being and coping with severe sicknesses (Ronconi, Testoni, & Zamperini, 2009;Testoni, Visintin, Capozza, Carlucci, & Shams, 2016).
This growing area of research shows that an implicit red thread binds the affective needs for protection from death with religiousness and management of trust. The present research moves in such a direction, focusing on trust, representations of death, religiosity-spirituality and the insecure attachment traits.

Aims and hypothesis
Since both avoidant (Collins, 1996;Feeney & Noller, 1990;Mikulincer, 1998) and anxious (Rodriguez, Dibello, Øverup, & Neighbors, 2015) styles present a significant lack of trust (Collins & Read, 1990), which is a component of the anxiety-avoidance (Daly & Mallinckrodt, 2009;Holmes & Lyons-Ruth, 2006), we wanted to: • Analyse how avoidant and anxious attachment experiences are related with trust, defining the role of spirituality-religiosity and of the representations of death; • Consider the hypothesis that their relationships may be described by a multilevel model, where the attachment orientations were the independent variables, the trust was the dependent variable, while spirituality and representations of death were moderators.

Participants
The research involved 443 university students. Table 1 illustrates the characteristics of participants, who were equipped with a foolscap folio containing four tests, providing a brief description of the research purposes. The collected data were anonymous and each participant was asked to sign an informed consent form. The study followed the APA Ethical Principles of Psychologists and Code of Conduct and the principles of the Declaration of Helsinki; furthermore it obtained the approval by the ethics committee of University of Padova.

Measures
The research design selected the Italian version of the following standardized questionnaires.

The Interpersonal Trust Survey (ITS)
ITS (

Testoni Death Representation Scale (TDRS)
TDRS (Testoni, Ancona, & Ronconi, 2015) comprised six items which were expressed in a five-point Likert scale. The measure assesses whether respondents perceive death as an absolute annihilation (Annihilation) or as a passage (Passage). Research showed that that individuals representing death as an absolute annihilation tend to suffer more from hopelessness and lack of resilience than people who think that death is a passage.

Experiences in Close Relationships-Revised (ECR-R)
ECR-R (Fraley, Waller, & Brennan, 2000) is a self-report questionnaire, conceived as a revision of the previous ECR (Brennan, Clark, & Shaver, 1998). The items of this scale were selected applying the Item Response Theory in order to estimate different levels of anxiety and avoidance. Studies by Watt, McWilliams, and Campbell (2005) pointed out that these aspects are strictly related to attachment. The tool consists of 36 items expressed in a 7-point Likert scale.

Rationale for Analyses
The analysis was conducted in three phases.
In the second phase, we performed a confirmatory factor analysis to assess one factor model for WHOQOL-SRPB variables (Domain 6) and two factors model for ITS (MB and OB) using the LISREL Version 8.7 statistical package (Jöreskog & Sörbom, 2004). In this study, the goodness-of-fit was evaluated using multiple criteria: Chi-square and df ratio, Root-Mean-Square-Error of Approximation (RMSEA), Non-Normed Fit Index (NNFI) and Comparative Fit Index (CFI). The cut-off point for Chi-square and df ratio value was below 3, RMSEA value was below 0.08, SRMR value was below 0.05, CFI and NNFI values were above 0.95, while AIC value is smaller than AIC for comparison model (Schermelleh-Engel, Moosbrugger & Müller, 2003). We finally computed the global score for each factor.
In the third phase, a hierarchical regression model on global scores of MB and OB was estimated setting as predictor variables attachment scales in the first step; spirituality variables (global score Domain6 of WHOQOL-SRPB and TDRS) in the second step; and interaction between attachment scales and spirituality variables in the third step. The analyses of first and third phases were performed using SPSS 21 statistical software package. https://doi.org/10.1080/23311908.2018.1429349

First phase: Preliminary analyses
In general, values reflected a high degree of internal reliability of all questionnaires. The skewness and kurtosis coefficients ranged from −1 to 1. We verified the internal reliability for each questionnaire by calculating Cronbach's alpha coefficient, which ranged from 0.70 to 0.97, with the exception of MRC, OAI, MME, ORC, OSI and MAI. Table 2 displays the psychometric properties of all the variables.
The analysis correlation among all the variables disclosed many significant interconnections.
With respect to the variables of ITS: MSI, MRC MME correlated in a positive way with almost all the variables of WHOQOL-SRPB, with Passage and Anxiety, while correlated in a negative way with Annihilation and Avoidance. Regarding the variables of WHOQOL-SRPB: MEANING and AWE were positively correlated with almost all the variables of ITS and Passage, and negatively with MAI and Annihilation and Avoidance. About TDRS, the representation of death as a passage was positively correlated with the following variables of trust (ITS): MRC, MCE, MME; and of the following variables of spirituality (WHOQOL-SRPB): SRBP, CONNECT, MEANING, AWE, STRENGHT, HOPE and FAITH. The representation of death as annihilation was inversely correlated with MME, SRBP, CONNECT, MEANING, AWE, STRENGHT, HOPE, FAITH. With respect to the ECR-R, Anxiety and Avoidance were both inversely correlated to SRPB, WHOLE and PEACE. They proceeded in the opposite way with MME, which in turn was inversely correlated with avoidance and positively with Anxiety. Specifically, Avoidance was inversely correlated with MSI, OSI, OAI, OCE, OME, MEANING and AWE, while Anxiety was positively correlated with MRC, MAI, MME, CONNECT and inversely with HOPE. Table 3 presents the correlations among all the variables.

Second phase: Confirmatory factor analysis
We carried out two confirmatory factor analysis to evaluate the same structural properties as the original scales for WHOQOL-SRPB and ITS. The CFA of WHOQOL-SRPB included nine observed variables (SRPB, CONNECT, MEANING, AWE, WHOLE, STRENGTH, PEACE, HOPE, FAITH) and one latent factor (Domain6). First model does not lead to an adequate solution (cf. WHOQOL-SRPB Model A in Table 4). Modification indices suggested that five additional paths had to be introduced: correlation between errors of CONNECT and FAITH, CONNECT and STRENGTH, FAITH and STRENGTH, PEACE and WHOLE, PEACE and HOPE. After allowing for these five paths, the model had a better fit (WHOQOL-SRPB Model A* in Table 4).
The CFA of ITS included 10 observed variables (MSI, MRC, MAI, MCE, MME, OSI, ORC, OAI, OCE, OME) and 2 latent factors (MB, OB). First model does not lead to an adequate solution (cf. ITS Model B in Table 4). Modification indices suggested that four additional paths had to be introduced: correlation between errors of MBRC and MBCE, MBRC and MBAI, MBAI and MBME, OBCE and OBSI. After allowing these four paths, the model had a better fit (ITS Model B* in Table 4).

Third phase: Hierarchical regression model
Hierarchical multiple regression analysis was conducted to explain the trust behaviour of both MB and OB as function of attachment and spirituality, and to evaluate the role of moderator of the attachment. In the first step, the attachment scale (measuring avoidance and anxiety) was comprised in the model. In the second step, the variables of spirituality (Annihilation-Passage and Domain6) were included. In the third step, all the interactions between scales and variables of attachment and religiosity were processed.
As the Table 5 shows, the variables of the first step always explained a significant proportion of the variance (R 2 = 19% for MB and R 2 = 69% for OB). The variables included in the second step added always a significant share of variance (ΔR 2 = 7% for MB and ΔR 2 = 4% for OB). Interactions inserted in the third step added a significant share of variance only with MB (ΔR 2 = 4% for MB and ΔR 2 = 2% for OB). The final model explained 29% of the total variance of MB and 11% of the variance of the OB.  At the first step, both scales of attachment were significant for MB (β = −0.46, p < 0.001 for Avoidance and β = 0.28, p < 0.001 for Anxiety) and only Avoidance for OB (β = −0. 24, p < 0.001). Avoidance had a negative impact on the behaviours of trust, or rather more Avoidance less confidence for both OB and for MB. On the contrary, Anxiety had a positive impact on MB.
At the second step, all the variables of spirituality were significant for MB (β = 0.16, p = 0.003 for Death Annihilation, β = 0.11, p = 0.042 for Death Passage and β = 0.24, p < 0.001 for Domain6) and only Domain6 for OB (β = 0.21, p < 0.001). In fact, Domain6 had a positive impact for both OB and MB. Furthermore, both the representations of death (Annihilation and Passage) had a positive score on MB.
At the third step, the interaction between avoidance and Death Annihilation was significant (β = 0.19, p = 0.001 for MB and β = 0.13, p = 0.050 on OB) and the interaction between Avoidance and Domain6 was positive for MB as well (β = 0.15, p = 0.002).
Opposite effect of Annihilation on MB and OB for subjects with High and Low Avoidance scores emerged when a slope analysis was done, in order to examine the interactions between representations of death and attachment. On one hand, Annihilation high scores were associated with greater trust in higher Avoidance, on the other, Annihilation high scores were associated with lower trust with Low Avoidance (Charts 1 and 2). However, it was to underline the greater effect of Domain6 for High Avoidance compared to Low Avoidance. In fact, high scores of Domain6 were associated with increased trust especially for High Avoidance (Chart 3).

Chart 1. Death annihilation and avoidance interaction on MB.
Chart 2. Death annihilation and avoidance interaction on OB.

Discussion
The preliminary analysis of the first phase showed that trust and spirituality were closely related and that a strong spiritual dimension corresponded to a greater willingness to develop trustful relationships. Furthermore, we confirmed that the two representations of death behave in an opposite way. Death represented as a passage, which is the essential conviction of most religions, exhibited more intense spirituality and optimism associated with stronger trust and reciprocity. As hypothesized following the idea that religiousness offers the protective image of reliable divinities, the representation of death as a passage eases the management of trust. According to the literature, which considers that God may be a positive attachment figure, Who facilitates the basic feeling of security (Desrosiers et al., 2011;Granqvist & Kirkpatrick, 2008;Kirkpatrick, 2005), indeed we confirmed that positive religiousness supports trust capabilities. On the contrary, the representation of death as absolute annihilation was characterized by a lack of spirituality and by a lack of trust. The coherence of these results confirmed that the ontological representations of death affect religiosity and management of trust.
However, anxiety and avoidance, which evidenced similar tendencies in spiritual variables, being inversely correlated with personal beliefs, completeness and peace, functioned in an opposite way in almost all the variables of trust. In this area, avoidance significantly manifested the greatest lack of trust. On one hand, Anxiety positively correlated with trust, in particular with the capability to meet MME and with MRC and MAI, while Avoidance was negatively correlated with it and with all the dimensions of ITS (significantly with: MSI, MCE, MME, OSI, OAI, OCE, OME). This difference was particularly interesting because, Avoidance and Anxiety maintained a similar performance in almost all the variables of WHOQOL-SRPB. In particular, Anxiety was positively correlated only with CONNECT and negatively with SRPB, WHOLE, PEACE and HOPE, meanwhile Avoidance negatively with MSI, MEANING, AWE, WHOLE and PEACE. These results suggested that, under the relationship among religiosity, representation of death and attachment, a further latent dimension was operating, which could become visible through the hierarchical regression model.
The confirmatory factor analysis of the second phase permitted us to use the latent factor Domain6, in order to analyse the general spirituality indicated by WHOQOL-SRPB, and OB and MB as the fundamental indicators of trust. After this verification, the third phase confirmed the hypothesis that the relationships among religiosity, trust, representation of death and avoidant-anxious attachment experiences were related in accordance with a structural model. In fact, religiosity was linked to Anxiety and Avoidance with respect to trust, where the ontological representations of death assumed the role of moderator for Avoidance. In fact, the first step of the multiple hierarchical regression evidenced that both Anxiety and Avoidance were significant to respect with MB, where the first Chart 3. Domain6 and avoidance interaction on MB.
was positively and the second inversely related. Furthermore, Avoidance was negatively significant in OB as well. The anxious individuals resulted preoccupied to deserve the trust of others, and it is possible to hypothesize that their anxiety could result from the need for approval not yet sufficiently obtained in infancy. On the contrary, Avoidant people seemed not to be affected by any kind of trust, as the attachment theory literature illustrated (Ainsworth & Bowlby, 1965). This first result showed that the dimension of the personal behaviours aimed at promoting the personal reliability characterized the relational strategies of the anxious attachment style.
The second phase of the research confirmed the correlation between religiosity and trust, already discussed by literature, which analysed the representation of God as attachment figure and the role of religion as relationships' facilitator (Beck, 2006;McDonald et al., 2005;Norenzayan, 2013;TenElshof & Furrow, 2000). In this fundamental structure, it would be expected that only the representation of death as a passage should have intervened on trust in a similar way. In fact, since religions announce an afterlife, offering a consolatory effects, it would be possible to hypothesize that the representations of death as a passage was the dominant element of such a structure. Surprisingly, on the contrary, both the ontological representations influence trust and only MB positively without any effects on OB. It means that, any kind of representation of death positively influences the personal trust behaviour. We can therefore interpret this result considering the effect of the representation of death as an intimate motivation in the personal behaviour aimed at gaining trust, independently from the kind of representation.
However, the most important result of our research appears in the third step of the hierarchical regression, where the importance of the representation of death as annihilation emerged as significant in both MB and OB for Avoidance, and further resulting significant in MB with Domain6. In order to better clarify this result, we realized the slope analysis, which showed an opposite effect of Annihilation for High and Low Avoidance. On one hand, Annihilation high scores were associated with greater MB and OB in higher Avoidance; on the other, Annihilation high scores were associated with lower trust with Low avoidance (Charts 1 and 2). Besides, it emerged that high scores of Domain6 were associated with increased trust, especially for High Avoidance (Chart 3). This result suggested that the representation of death as annihilation moderates the effect of Avoidance on trust, where also spirituality intervenes in the case of MB. This last result may be explained as a personal strategy, useful to orientate the trust behaviour. In fact, following the TMT assumptions, death is the fundamental cause of any human paralysing anguish, which is managed by cultural worldviews indicating symbolic or literal immortality. In this sense, the representation of death as an absolute annihilation is more terrifying of the representation of death as a passage because the latter is the negation of the former and is a psychological remedy to the anxiety of any death related thought.
What is more, the fact that both Domain6 and Annihilation moderated in the same way MB in High Avoidance casts suspicion on a particular aspect of Avoidance: the same role of religiosity, which implied trust in a universal saviour principle, and of the absolute absence of such an entity. It means that, the more the person is avoidant, the more spirituality and its opposite (death as annihilation) moderate its own trust behaviour. Both the representations equally and ambivalently motivate to assume behaviours deserving trust and do not appear as excluding each other. As Kirkpatrick andShaver (1990, 1992) demonstrated, people with an avoidant God attachment have a distant and aloof relationship to God. We can add that, He is a presence Who produces an effect similar to the "annihilating nothing" one.

Limit of the research and future developments
The most important limit of this research was inherent to lack of a specific analysis of trust in God. In this research, it came out that, it is important to develop this dimension, especially in order to intercept the nature of the role of the representations of God and death in the relationships among attachment traits, trust and spirituality.