"Are you afraid to die?" Religion and death attitudes in an adolescent sample.

The purpose of this study was to provide an in-depth examination of the relationship between religion and death attitudes. First, we investigated whether the results found in previous research on adults could be generalized to adolescents. Second, we implemented a person-centered approach in studying this relationship. Questionnaires assessing death attitudes and religious attitudes were completed by 213 adolescents in a secularized country. Results from hierarchical regression analyses were in line with previous findings on adults, confirming our hypothesis that the link between religious attitudes and death attitudes is important during diverse stages of the life span. Furthermore, cluster analysis was used to investigate naturally occurring profiles of death attitudes in order to offer a qualitative refinement as compared to the commonly used variable-centered approach. A meaningful 3-cluster solution (i.e., Natural Process, Acceptance, and Anxiety) was retained and each cluster was characterized by its own unique scores on religious attitudes. Limitations concerning the generalization of the conclusions are discussed.

**********

Coping with the finitude of life might be viewed as a life long task for all human beings. The idea of mortality and the realization of the finitude of life often elicit negative feelings in individuals, such as anxiety, despair, sadness, and uncertainty. For instance, existential psychology indicated that each individual is anxious about his or her death (Yalom, 1980). Tillich (1952) also noted that death anxiety is a basic, universal, and inescapable feeling. Frankl (1958), however, emphasized a more positive attitude toward the existential theme of death and dying by stating that finiteness itself should give meaning to human existence--not deprive it of meaning.

Religious belief systems seem to play an important role in providing a framework to cope with existential topics. Moreover, religion is often seen as a way of coping with the unpredictable. Because death is pre-eminently an unpredictable phenomenon, it is understandable that the great religions make it a central part of their belief system (Spilka, Hood, Hunsberger, & Gorsuch, 2003). Several theorists indeed link the phenomena of religion and death, and some scholars even state that if there were no death, there would be no religion (Becker, 1973; Weisman, 1972).

Religion and Death Attitudes: Multidimensional Concepts

A substantial body of research has addressed the relation between religiosity and death attitudes. However, this research seems to have a number of limitations (Neimeyer, Wittkowski, & Moser, 2004). First, the dominant focus of many studies has been on the measurement of death anxiety, without considering other (potentially more positive) death attitudes. Second, many studies have adopted a rather robust, unidimensional conceptualisation of religiosity without tapping into people's underlying attitudes towards religiosity. Perhaps as a consequence of these limitations, evidence for a link between religiosity and death anxiety is ambivalent. Consequently, one way to obtain a more precise picture of the religiosity-death attitudes link is to expand on the construct of death attitudes (Neimeyer et al., 2004; Spilka et al., 2003). Recent research already pointed to the importance of distinguishing between different death attitudes and, more specifically, to the usefulness of focusing on both negative (i.e., anxiety and avoidance) and relatively more positive attitudes toward death (i.e., a neutral or even accepting orientation towards death). For example, Christopher, Drummond, Jones, Marek, and Therriault (2006) found that religiosity was positively related to positive death attitudes (e.g., death as a natural end of life) and negatively to negative death attitudes (e.g., death as a failure).

As mentioned before, the operationalisation of religion can also be seen as a limitation in the field. In the past, researchers often operationalised religion in terms of frequency of church attendance or prayer behavior (e.g., Duff & Hong, 1995). However, more recently, scholars point to the importance of focusing on religious attitudes instead of religious behavior (Dezutter, Soenens, & Hutsebaut, 2006; Francis, Robbins, Lewis, Quigley, & Wheeler, 2004). Religious attitudes are deeper-rooted aspects of religion and will therefore be more salient in the psychological functioning of individuals. Consequently, focusing on religious attitudes as a measurement of religion might broaden our insight in the link between religion and death attitudes.

When focusing on religious attitudes, the dominant model is the model of Allport and Ross (1967). This model pertains to people's underlying goals and motives for being religious, and distinguishes between two religious orientations. In the extrinsic orientation, religion is used for non-religious ends (e.g., emotional or social support, or feelings of security). In the intrinsic orientation, religion is experienced as a master motive in life. Earlier research already indicated that these two orientations hold distinct relations with death attitudes. For example, Cohen et al. (2005) showed that the intrinsic religious orientation was negatively related and the extrinsic religious orientation was positively related to death anxiety. In a sample of elderly persons, Ardelt and Koenig (2006) found that, whereas intrinsic religiosity was positively related to an accepting attitude towards death, extrinsic religiosity was positively related to death anxiety.

A disadvantage of this model is that the distinction between intrinsic and extrinsic religiosity pertains to religious people only. Given the steadily waning impact of institutionalized religion in many European countries (Kerkhofs, Peetermans, & Geeraerts, 1997), the utility of this model and accompanied instrument (Religious Orientation Scale, ROS) can be questioned. In Belgium--as in other Western European countries--most people are not associated to any religious denomination. Only 45% finds religion important and only 11% attends church weekly (Dobbelaere & Voye, 2000). Consequently, a major part of the population is unable to answer the ROS because its questions are too remote from their frame of reference. Indeed, recent studies have already stressed that the model of Allport and Ross is not as useful in secularized European countries as it is in the USA due to the differences in religious landscape (Cohen et al., 2005; Dezutter, Soenens, & Hutsebaut, 2006; Flannelly, Koenig, Ellison, Galek, & Krause, 2006).

The Need for Multidimensional Instruments

In a recent community study, some of the limitations in the field concerning the conceptualisation and operationalisation of religion and death attitudes were tackled (Dezutter et al., 2009). To this end, attention was paid to the multidimensional assessment of both religion and death attitudes. Concerning death attitudes, this study relied on the Death Attitude Profile- Revised (DAP-R; Gesser, Wong, & Reker, 1987), an instrument that comprises scales for both death anxiety and death acceptance and thus investigates both positive and negative aspects of death attitudes. Wong, Reker, and Gesser (1994) made a number of additional distinctions within the concepts of death anxiety and death acceptance, resulting in five different death attitudes: (a) Neutral Acceptance, involving the view that death is an integral part of life, (b) Approach Acceptance, a positive outlook on death rooted in the belief in a happy afterlife, (c) Escape Acceptance, in which death is a welcome alternative for a life full of pain and misery, (d) Fear of Death, involving feelings of fear evoked by confrontations with death, and (e) Death Avoidance, involving avoidance of thinking or talking about death in order to reduce death anxiety.

In addition, the study by Dezutter et al. (2009) made use of a more refined and theory-driven conceptualisation of religion fitting with the secularized, postmodern, and Catholic history of the sample. To this end, the authors relied on the Post-Critical Belief Scale (Hutsebaut, 1996), an instrument based on Wulff's theory (1991). Wulff argued that all possible approaches to religion can be located in a two-dimensional space. The Exclusion versus Inclusion of Transcendence dimension specifies whether a person participates in a transcendent reality and thus pertains to religiosity as such. The Literal versus Symbolic dimension indicates whether religious contents are interpreted literally or symbolically and thus pertains to the way individuals process religious contents. These dimensions define four religious attitudes: Literal Inclusion, Literal Exclusion, Symbolic Inclusion, and Symbolic Exclusion. Literal Inclusion is typical of individuals defining themselves as religious and interpreting religious contents in a rather rigid, closed-minded, and dogmatic fashion. These individuals uncritically and strictly adopt religious contents the way they are taught within a particular religious tradition. Individuals with an attitude of Symbolic Inclusion also define themselves as religious persons yet process religious contents in a more symbolic and open fashion. They assume that several interpretations of religious contents are possible and try to find an interpretation of religious contents that they find personally meaningful. Individuals with an attitude of Literal Exclusion reject the possibility of a religious reality on the basis of strict and literal arguments (e.g., the lack of scientific evidence for the miraculous stories in the Bible). Finally, people holding an attitude of Symbolic Exclusion also define themselves as non-religious, yet are less judgmental and closed-minded towards religiosity compared to people with an attitude of Literal Exclusion. These individuals consider religiosity as one way of giving meaning to life among many others and respect other people's choice to be religious but do not need religiosity themselves to find personal meaning in life.

It is important to note that the model of Wulff (1991) and Hutsebaut (1996) is distinct from the model of Allport and Ross (1967). These two models differ in at least two ways. First, whereas Allport and Ross's distinction between intrinsic and extrinsic religiosity pertains to the motivational sources of people's religious attitudes (see also Neyrinck, Lens, & Vansteenkiste, 2005), the four attitudes of Wulff and Hutsebaut represent four different social-cognitive approaches or attitudes to religiosity. Second, whereas the model of Allport and Ross does not take into account the attitudes of non-religious people, the model of Wulff and Hutsebaut allows one to chart individuals' attitudes irrespective of whether they are religious or not. Perhaps as a consequence of their greater applicability and relevance in modern-day secularized societies, Wulff's religious attitudes were found to be strongly predictive of a broad range of variables (for an overview, see Duriez & Hutsebaut, in press).

Purpose of this Study

Recent research studying religion and death attitudes seems to focus mainly on adulthood and the last stages of life and not on adolescent or young adult groups (e.g., Duff & Hong, 1995; Falkenhain & Handal, 2003; Former & Neimeyer, 1999). The few studies focusing on adolescents are (mostly) situated in less-secularized countries such as islamitic countries (e.g., Al-Sabwah & Abdel-Khalek, 2006; Lester & Abdel-Khalek, 2008) or in countries with a different religious landscape than in West-European countries (e.g., Chinese adolescents; Hui & Fung, 2009). Studies exploring the association between religious attitudes and death attitudes in a group of adolescents in a secularized or postmodern society are rather scarce. Since previous research (Dezutter et al., 2009) indicated that also in secularized societies meaningful associations between religiosity and death attitudes emerge in adulthood, the present study investigated whether these associations are expressions of deep-rooted basic human concerns and, therefore, emerge in other life stages as well.

According to Piaget (Inhelder & Piaget, 1958), adolescents obtain the cognitive ability to think abstractly on topics such as life and death. Existential questions concerning existence and non-existence and possible frameworks to cope with these questions become important in this life stage. Distinct theoretical frameworks such as the psychosocial theory of Erikson (1968) and existential theorizing (Frankl, 1958) also emphasize the need to investigate this topic in the adolescent years. Concerning Erikson's psychosocial theory, the main developmental task during adolescence involves forming a stable sense of identity. As a part of this process, adolescents need to establish an integrated personal stance on existential issues, including the role of religion in their life (Markstrom, 1999). From the existentialist perspective, we can assume that adolescence, with the task of identity formation, is a stage where the tension between existence and nonexistence comes into prominence. The emphasis placed on being someone also implies the possibility of not being.

Empirical studies indeed testify to the importance of assessing death attitudes in adolescence and early adulthood. For instance, Lavoie and de Vries (2003) showed that meaningful relations exist between certain identity statuses and specific death orientations. Sterling and Van Horn (1989) indicated that in a group of male college students significant associations were found between identity statuses and death anxiety. In sum, both on theoretical and empirical grounds, we hypothesized that death and dying--and how it is related to religious issues--could be an important issue in the psychological functioning of adolescents and young adults.

In order to investigate the link between religion and death attitudes, we examined associations between Wulff's (1991) four religious attitudes (i.e., Literal Inclusion, Literal Exclusion, Symbolic Inclusion, and Symbolic Exclusion) and four death attitudes (i.e., Fear of Death, Death Avoidance, Neutral Acceptance, and Approach Acceptance). (1) Based on a recent study in an adult sample (Dezutter et al., 2009), the following three hypotheses were forwarded. First, we hypothesized that religious individuals (Literal and Symbolic Inclusion) adopt an Approach Acceptance attitude towards death taking into account that religious participants adhere to the picture of the afterlife provided by their religion. Second, we hypothesized that individuals with a literal approach towards religion (Literal Inclusion and Literal Exclusion) show more Fear of Death and Death Avoidance. Previous research (Duriez, Van Hiel, & Kossowska, 2005) already indicated that these literal thinkers show higher levels of need for closure and function in a more defensive way thereby avoiding ambiguity and feelings of insecurity. Because death is pre-eminently an unpredictable phenomenon, awareness of death may arouse anxiety resulting in the avoidance of death-related thoughts. Third, we hypothesized that in the prediction of a Neutral Acceptance attitude, where death is seen as a natural end point of life, both dimensions play a role. Based on previous findings (Dezutter et al., 2009), we expected Literal Inclusion and Symbolic Exclusion, which are opposite religious attitudes, to be respectively negatively and positively associated with the Neutral Acceptance attitude.

This variable-centered approach--with a focus on separate death attitudes and how they relate to religious attitudes--is also the main perspective in most previous studies. Consequently, little is known about how these death attitudes integrate to form different profiles in coping with the idea of death and dying. A person-centered perspective allows for such an approach through its focus on identifying meaningful profiles of death attitudes which can enhance our understanding of the complex co-occurrence of death attitudes (Magnusson, 1998). Wong, Reker, and Gesser (1994) already stated that "it is the pattern of different death attitudes rather than the magnitude of a single death attitude that best captures individual differences" (p. 141). We indeed assume that individuals do not adhere to one specific attitude when coping with a complex phenomenon as their own death but we hypothesize that they integrate different attitudes in a meaningful profile. Within the framework of a person-centered approach, cluster analysis can be used to classify individuals in such way that individuals within one cluster have more in common than they do with individuals assigned to other clusters (Gore, 2000).

However, because no earlier studies used this person-centered approach in investigating death attitudes, these analyses are exploratory in nature. As such, we had no specific theory-based hypotheses concerning the profiles we would expect. Nevertheless, we hypothesized that at least two clusters would emerge: individuals who fear death will probably also try to avoid feelings or thoughts concerning death and dying whereas a positive acceptance attitude will probably not be present. These individuals may show a profile with an emphasis on fear, avoidance, and anxiety. Individuals who approach death as an integral part of life, on the other hand, will probably show lower levels of both fear and anxiety whereas higher levels of both neutral and approach acceptance. Their profile would be characterized by elevated scores on a resigning and accepting attitude.

METHOD

Participants and Procedure

The sample consisted of 213 participants from the Faculty of Psychology and Educational Sciences from a large university in the Dutch-speaking part (Flanders) of Belgium. All participants were freshmen and a breakdown by gender yielded 169 women (79.3%) and 44 men (20.7%). This unbalanced distribution of gender mirrored the freshmen population who studies Psychology and Educational Sciences at this university. All participants were Caucasian students with a middle-class background. Students participated during group sessions, supervised by the first author, and received course credit for their participation. Students were told that anonymity at the individual level was guaranteed and signed a standard consent form before participating. Mean age for our participating sample was 18 years and 2 months (SD = 12 months) with a range from 17 to 26 years.

With respect to religious profile, 59.2% of the participants never went to church, 34.3% went only on important religious days (e.g., Christmas, Easter), 3.8% went once a month, and 2.8% went once a week. Furthermore, 37.1% of the participants rated religion as not important, 46.5% only of minor importance, 12.2% as important, and for 4.2% religion was very important.

Measures

All measures were in Dutch, the native language of the participants.

Socio-demographic variables. Gender and age were measured at the beginning of the questionnaire.

Post-Critical Belief Scale (PCBS). To measure the different religious attitudes identified by Wulff (1991), the 33-item Post-Critical Belief scale (Duriez, Fontaine, & Hutsebaut, 2000) was administered. All items were scored on a 7-point Likert scale ranging from 1 (strongly disagree) to 7 (strongly agree). Estimates of internal consistency (Cronbach's alpha) were .75 for Symbolic Inclusion, .80 for Literal Inclusion, .83 for Literal Exclusion, and .64 for Symbolic Exclusion. Examples of items are "I think that Bible stories should be taken literally, as they are written" (Literal Inclusion), "Faith is an expression of a weak personality" (Literal Exclusion), "I am well aware my ideology is only one possibility among so many others" (Symbolic Exclusion), and "The Bible holds a deeper truth which can only be revealed by personal reflection" (Symbolic Inclusion).

Death Attitude Profile-Revised (DAP-R). The DAP-R is a revision of the DAP, a multidimensional measure of attitudes toward death developed by Gesser, Wong, and Reker (1987). The scale consists of 32 items scored on a 7-point Likert scale ranging from 1 (strongly disagree) to 7 (strongly agree). Estimates of internal consistency (Cronbach's alpha) were .85 for Fear of death, .92 for Death avoidance, .64 for Neutral Acceptance, and .90 for Approach Acceptance.

RESULTS

Preliminary Analyses

All analyses reported were conducted in SPSS version 15. All correlations among the religious and death attitudes are presented in Table 1 As expected, Literal Inclusion and Symbolic Exclusion were negatively related, and so were Symbolic Inclusion and Literal Exclusion.

TABLE 1

Correlations between religious attitudes and death attitudes

Variable             Literal    Literal    Symbolic   Symbolic
                     Inclusion  Exclusion  Inclusion  Exclusion

Literal Inclusion
Literal Exclusion    -.07
Symbolic Inclusion    .28 ***    -.61 ***
Symbolic Exclusion   -.25 ***    -.03       .25 ***
Fear of death         .20 **      .16 *     .04        -.03
Death avoidance       .22 **      .19 **   -.05        -.11
Neutral acceptance   -.21 **      .03       .02         .26 ***
Approach acceptance   .37 ***    -.21 **    .40 ***    -.06

Variable             Fear of   Death Avoidance   Neutral     Approach
                     Death                      Acceptance  Acceptance

Literal Inclusion
Literal Exclusion
Symbolic Inclusion
Symbolic Exclusion
Fear of death
Death avoidance       .68 ***
Neutral acceptance   -.47 ***      -.37 **
Approach acceptance   .23 **        .13           -.05

* p <.05,** p <.01, *** p<.001

The religious attitudes showed a specific correlational profile with the death attitudes. As expected, Literal Inclusion correlated positively with Fear of Death, Death Avoidance, and Approach Acceptance and negatively with Neutral Acceptance. Literal Exclusion showed similar results for Fear of Death and Death Avoidance. However, Literal Exclusion correlated negatively with Approach Acceptance. Symbolic Inclusion correlated positively with Approach Acceptance while Symbolic Exclusion correlated positively with Neutral Acceptance.

Regression Analyses

To assess the contribution of religious attitudes in the prediction of death attitudes, a series of hierarchical multiple regression analyses were performed. Separate regression analyses were performed for each of the death attitudes. Because our sample was predominantly female, gender was entered in Step 1 as a background variable. The religious attitudes were entered In Step 2. Results can be found in Table 2.

TABLE 2

Hierarchical regressions predicting death attitudes

                       Approach          Neutral       Fear of Death
                      Acceptance        Acceptance

                    Step 1   Step 2   Step 1  Step 2   Step 1  Step 2
                    [beta]   [beta]   [beta]  [beta]   [beta]  [beta]

Gender               .12     .16 *     -.12   -.09       .09   .12
Literal Inclusion            .25 ***          -.19 *           .20 **
Literal Exclusion            .02              -.02             .32 ***
Symbolic Inclusion           .39 ***           .03             .19
Symbolic Exclusion          -.08               .19 *           .02
[DELTA][R.sup.2]     .01     .27 ***    .01    .09 **    .01   .12 ***

                    Death Avoidance

                    Step 1   Step 2

                    [beta]   [beta]

Gender               .05     .04
Literal Inclusion            .24 **
Literal Exclusion            .26 **
Symbolic Inclusion           .04
Symbolic Exclusion          -.06
[DELTA][R.sup.2]     .00     .12 ***

* P<.05 ** p<.01 *** p<.001

Gender did not predict any of the death attitudes in Step 1. In Step 2, the religious attitudes added to the prediction of each of the death attitudes above and beyond the contribution of gender and significantly predicted Fear of Death (F (5,204) = 5.68, p < .01, [DELTA][R.sup.2] = .12), Death Avoidance (F (5, 204) = 5.67, p < .01, [DELTA][R.sup.2] = .12), Neutral Acceptance (F (5, 204) = 4.57, p <.01, [DELTA][R.sup.2], and Approach Acceptance (F (5, 204) = 15.56, p < [DELTA][R.sup.2] = .27). Both aspects of death anxiety, that is, Fear of Death and Death Avoidance, were positively predicted by Literal Inclusion and Literal Exclusion. Approach Acceptance was positively predicted by both Literal Inclusion and Symbolic Inclusion. Finally, Neutral Acceptance was negatively predicted by Literal Inclusion and positively by Symbolic Exclusion.

Cluster Analysis

Internal validity. After removal of 3 univariate outliers (i.e., values more than 3 SD below or above the mean) and 5 multivariate outliers (i.e. individuals with high values for the Mahalanobis distance statistic), cluster analysis on the remaining subjects (n = 205) was conducted using a two-step procedure (Gore, 2000). Removal of the outliers is conducted both on statistical grounds, that is, the disturbance of cluster analysis by multivariate outliers as well on theoretical grounds, that is, the possibility that measurement errors create a small number of 'unique' individuals who should not be forced in a cluster (Bergman & Magnusson, 1997).

In the first step, a hierarchical cluster analysis was carried out using Ward's method on squared Euclidian distances. In the second step, these initial cluster centers were used as starting partitions in the k-means clustering. This iterative procedure remedies one of the major shortcomings of the hierarchical method, namely that once an object is clustered using the latter method, it cannot be reassigned to another cluster at a subsequent stage. Iterative clustering, however, minimizes within-cluster and maximizes between-cluster variability, allowing reassignments to "better fitting" clusters and thus optimizing cluster membership of the different objects (Gore, 2000).

We considered 2 to 4 cluster solutions and inspected the percentage of variance explained in the four death attitudes in each of these solutions. The 3-cluster solution explained more variance than the 2-cluster solution and therefore, the 2-cluster solution was not considered for further analyses. The 3-cluster solution explained 50% of the variance in each of the defining variables, except for Neutral Acceptance. The 4-cluster solution explained 50% of the variance in each of the death attitudes but was less parsimonious. Furthermore, the 3-cluster solution was theoretically more meaningful than the 4-cluster solution.

A double split cross-validation procedure was used to compare the 3- and 4-cluster solution (Breck-enridge, 2000; Tinsley & Brown, 2000). The sample was randomly split into halves. The full two-step procedure (Ward, followed by k-means) is then applied to each half and the two solutions are compared for agreement as follows. The participants of each half of the sample are assigned to new clusters on the basis of their Euclidean distances to the cluster centers of the other half of the sample (SPSS procedure QUICK CLUSTER, option CLASSIFY). These new clusters are then compared for agreement with the original cluster by means of Cohen's kappa (k). The two resulting kappa's are averaged. An agreement of at least .60 is considered acceptable (Asendorpf, Borkenau, Ostendorf, & van Aken, 2001). The cluster solution with the highest kappa is preferred because this solution is more stable and replicable. The 3-cluster solution was found to be more stable (k = .92) than the 4-cluster solution (k = .64). In sum, the 3-cluster solution was chosen on theoretical and statistical grounds.

Figure 1 presents the final cluster solution. The Y-axis represents the z-scores. The distances between the cluster means and the total sample standardized mean, in standard deviation units, were interpreted as effect sizes (Scholte, van Lieshout, de Wit, & van Aken, 2005). Analogous to Cohen's d, 0.2 SD is a small effect, 0.5 SD is a moderate effect, and 0.8 SD is a large effect. As can be noted, all three groups were characterized by z-scores that reflected moderate to strong deviations from the sample mean, suggesting that the three groups differed considerably in terms of their death attitudes profile. More specifically, we found an Acceptance cluster (n = 89), which consisted of individuals high on Approach Acceptance and modest on Neutral Acceptance and relatively low on Fear of Death and Death Avoidance, an Anxiety cluster (n = 64) which consisted of individuals high on Fear of Death and Death Avoidance while relatively low on Neutral Acceptance, and, finally, a Natural Process cluster (n = 52) which consisted of individuals high on Neutral Acceptance while low on Approach Acceptance, Fear of Death and Death Avoidance.

[FIGURE 1 OMITTED]

External validity. A one-way MANOVA was conducted and revealed significant cluster differences for the religious variables, except for Symbolic Exclusion. The univariate F-values and multiple pairwise combinations with the Tukey HSD Test are shown in Table 3. Individuals in the Anxiety and Acceptance cluster had significant higher levels of Literal Inclusion in comparison with individuals in the Natural Process cluster. Further, individuals in the Acceptance cluster had also significant lower levels of Literal Exclusion and higher levels of Symbolic Inclusion in comparison with both the Natural Process cluster and the Anxiety cluster.

TABLE 3

Univariate ANOVA's and Post-hoc Cluster Comparisons Based Upon Tukey
HSD Tests for the Four Religious Attitudes

                                 Clusters

           Acceptance   Anxiety   Natural    F(2,200)  [Eta.sup.2]
                                  process

Literal     2.11 (a)    2.18 (a)   1.55 (b)  11.95 **      .11
Inclusion
           (0.81)      (0.82)     (0.49)

Literal     3.4 (a)     4.02 (b)   3.86 (b)   7.11 **      .07
Exclusion
           (1.00)      (0.92)     (1.20)

Symbolic    4.43 (a)    3.85 (b)   3.46 (b)  14.63 **      .13
Inclusion
           (1.01)      (0.96)     (1.24)

Symbolic    4.91        4.80       4.89       0.43         .00
Exclusion
           (0.69)      (0.70)     (0.83)

Note. A cluster mean is significantly different from another mean if
they have different superscripts. A mean without a superscript is not
significantly different from any other mean. Standard deviations are in
parentheses.

* p <.01 ** p <.001.

DISCUSSION

The merits of the present study were twofold. First, previous findings concerning the influence of religion on death attitudes in adulthood could be generalized to earlier stages of the life span, that is, adolescence. Second, the use of a data-driven person-oriented approach made it possible to distinguish between three specific profiles of death attitudes, each characterized by a different combination of death attitudes levels and by different scores on the religious attitudes.

In this study, we found similar associations between religious attitudes and death attitudes as in our previous study on adults (Dezutter et al., 2009). In line with our hypotheses, Literal Inclusion predicted higher levels of Approach Acceptance, Fear of Death, and Death Avoidance, and lower levels of Neutral Acceptance, whereas Symbolic Inclusion predicted higher levels of Approach Acceptance. Furthermore, Literal Exclusion predicted higher levels of Fear of Death and Death Avoidance as was the case in the adult sample but, contrary to the results in the adult sample, Literal Exclusion was not negatively related to Approach Acceptance. Finally, Symbolic Exclusion predicted higher levels of Neutral Acceptance similar to the adults but did not show an association with Death Avoidance which we did find in the adult sample. In sum, similar patterns of associations appeared in the adolescent sample indicative of processes which are active in younger as well as in older individuals.

Findings indicated that irrespective of the way religious contents are interpreted (literal or symbolical), religious people are more likely to believe in an afterlife compared to non-religious people. Since most, if not all, religions provide a framework with elaborated ideas of an afterlife, it is most likely that religious individuals rely on these ideas and agree with the Approach Acceptance attitude. Furthermore, the combination of the Literal versus Symbolic dimension and the Inclusion versus Exclusion dimension seems necessary in the prediction of Neutral Acceptance. Religious individuals who believe in a rather literal fashion (i.e., Literal Inclusion) strongly reject the idea that death is no more than a natural part of life itself which is indeed contrary to the detailed picture of the afterlife that their religion provides. Nonbelievers with a symbolic approach (i.e., Symbolic Exclusion), on the other hand, try to find answers on the existential issues of death and dying without relying on a transcendent reality. They strongly agree with the idea that death is a natural part of life itself. Finally, a literal approach to religion, irrespective of whether people believe (i.e., Literal Inclusion) or not (i.e., Literal Exclusion), predicted both Fear of Death and Death Avoidance. This finding strengthens our assumption that individuals with a literal approach to religion have more difficulties coping with the uncertainty of death and dying (Dezutter et al., 2009).

The persistence of these associations in both age groups also qualifies some conclusions in the field. Spilka and colleagues (2003), for example, stated after reviewing the literature that extant research points in the direction of a reductive effect of religious commitment and religiosity on death anxiety. Based on our results, we would like to qualify this conclusion somewhat. Especially in a secularized Western European society, it seems necessary to take into account the approach towards religion besides religiosity. Religiosity does not, per se, reduce death anxiety. In contrary, religiosity combined with a literal approach is associated with higher levels of death anxiety.

A second aim of this study was to heed some recent calls in the field (Wong, Reker, & Gesser, 1994) concerning the use of a person-centered approach when investigating death attitudes. This approach enables us to search for specific profiles integrating different death attitudes and, hence, provides a more factual picture of how individuals approach a complex phenomenon as death and dying. In accordance with Wong and colleagues, we expected that individuals rely on several attitudes instead of limiting themselves to one specific approach. Cluster analysis indeed retained three distinct and theoretically meaningful clusters. The clusters were quite heterogeneous in terms of their attitudinal profiles. The Approach attitude was high in the Acceptance cluster, close to the mean in the Anxiety cluster, and very low in the Natural Process cluster. The Fear and Avoidance attitudes showed a similar pattern with close to the mean scores in the Acceptance cluster, high scores in the Anxiety cluster, and low scores in the Natural Process cluster. Finally, the Neutral attitude showed close to the mean scores in the Acceptance cluster, a low level in the Anxiety cluster and a high level in the Natural Process cluster.

This 3-cluster solution was found to be highly replicable and externally valid, as the three clusters differed in theoretically predictable ways in terms of religious attitudes. Moreover, individuals both in the Anxiety and Acceptance cluster showed significant higher levels of Literal Inclusion in comparison with individuals in the Natural Process cluster. Further, individuals in the Acceptance cluster had also significant lower levels of Literal Exclusion and higher levels of Symbolic Inclusion in comparison with both the Natural Process cluster and the Anxiety cluster.

Limitations

One must keep in mind the limitations of the present study that provide avenues for further research. First, the rather homogeneous sample consisted of Flemish students with a majority of females. Therefore, we cannot state with certainty that our findings can be generalized to all adolescent populations. However, since we controlled in the analyses for gender and we were able to replicate the findings from a previous study, we assume that similar associations would emerge in adolescent populations with a similar religious landscape. Second, all data in this study were collected by use of self-report questionnaires. Although questionnaires are appropriate to gather information about subjective and internal concepts such as religious and death attitudes, the sole reliance on self-report measures may have led to an overestimation of some of the correlations due to shared method variance. Future research would do well to use more diverse methods in the assessment of religious and death attitudes. Furthermore, the cross-sectional character of the study limits conclusions considering the nature of the relationship in terms of directionality. With respect to causality, future research should focus on longitudinal studies or experimental designs in order to sort out whether religious attitudes indeed influence death attitudes or whether it is the other way around. Finally, although a strength of this study was the replication of previous findings in another stage of the lifespan, it might be interesting to investigate these specific associations in a sample where the topic of death and dying is more salient. Therefore, future research should focus on specific subgroups of the population confronted with these existential themes.

Conclusions

Despite these limitations, the overall conclusions of this study are important. The study indicated that the associations between religion and death attitudes are probably expressions of deep-rooted basic human concerns which are present during diverse stages of the life span. The study also sheds light on the naturally occurring patterns of death attitudes and emphasizes the need for further research on this topic.

REFERENCES

Allport, G. W.,&Ross, J. M. (1967). Personal religious orientation and prejudice. Journal of Personality and Social Psychology, 5, 432-443.

Al-Sabwah, M., & Abdel-Khalek, A. (2006). Religiosity and death distress in Arabic college students. Death Studies, 30, 365-375.

Ardelt, M., & Koenig, C. (2006). The role of religion for hospice patients and relatively healthy older adults. Research on Aging, 28, 184-215.

Asendorpf, J. B., Borkenau, P., Ostendorf, F., & van Aken, M. A. G. (2001). Carving personality at its joints: Confirmation of three replicable personality prototypes for both children and adults. European Journal of Personality, 15, 169-198.

Becker, E. (1973). The denial of death. New York: the Free Press.

Bergman, L., & Magnusson, D. (1997). A person-oriented approach in research on developmental psychopathology. Development and Psychopathology, 9,291-319.

Breckenridge, J. N. (2000). Validating cluster analysis: Consistent replication and symmetry. Multivariate Behavioral Research, 35, 261-285.

Christopher, A., Drummond, K., Jones, J., Marek, P., & Therriault, K. (2006). Beliefs about one's own death, personal insecurity, and materialism. Personality and Individual Differences, 40, 441-451.

Cohen, A., Pierce, J., Chambers, J., Meade, R., Gorvine, B., & Koenig, H. (2005). Intrinsic and extrinsic religiosity, belief in the afterlife, death anxiety, and life satisfaction in young Catholics and Protestants. Journal of Research in Personality, 39, 307-324.

Dezutter, J., Soenens, B., & Hutsebaut, D. (2006). Religiosity and mental health: A further exploration of the relative importance of religious behaviors versus religious attitudes. Personality and Individual Differences, 40, 807-818.

Dezutter, J., Soenens, B., Luyckx, K., Bruyneel, S., Vansteenkiste, M., Duriez, B., & Hutsebaut, D. (2009). The role of religion in death attitudes: Distinguishing between religious belief and style of processing religious contents. Death Studies, 33, 73-92.

Dobbelaere, K., & Voye, L. (2000). Religie en kerkbetrokkenheid: Ambivalentie en vervreemding [Religion and church involvement: Ambivalence and alienation]. In K. Dobbelaere, M. Elchardus, J. Kerkhofs, L. Voye, and B. Bawin-Legros (Eds.), Verloren zekerheid: De Belgen en hun waarden, overtuigingen en houndingen (117-152). Tielt, Belgium: Lannoo.

Duff, R. W., & Hong, L. K. (1995). Age density, religiosity and death anxiety in retirment communities. Review of Religious Research, 37, 19-32.

Duriez, B., Fontaine, J. R., & Hutsebaut, D. (2000). A further elaboration of the post-critical belief scale: Evidence for the existence of four different approaches to religion in Flanders-Belgium. Psychologica Belgica, 40, 153-181.

Duriez, B., & Hutsebaut, D. (in press). A slow and easy introduction to the Post-Critical Belief Scale. Internal structure and external relationships. In D.M. Wulff (Ed.), Handbook of the Psychology of Religion. Oxford: University Press.

Duriez, B., Van Hiel, A, & Kossowska, M. (2005). Authoritarianism and social dominance in Western and Eastern Europe: The importance of the socio-political context and of political interest and involvement. Political Psychology, 26, 299-320.

Erikson, E. H. (1968). Identity: Youth and crisis. New York: Norton.

Falkenhain, M., & Handal, P. (2003). Religion, death attitudes, and belief in afterlife in the elderly: Untangling the relationships. Journal of Religion and Health, 42, 67-76.

Flannelly, K. J., Koenig, H. G., Ellison, C. G., Galek, K. & Krause, N. (2006). Belief in life after death and mental health. The Journal of Nervous and Mental Disease, 194(7), 524-529.

Fortner, B. V., & Neimeyer, R. A. (1999). Death anxiety in older adults: A quantitative review. Death Studies, 23, 387-411.

Francis, L. J., Robbins, M., Lewis, C. A., Quigley, C. F., & Wheeler, C. (2004). Religiosity and general health among undergraduate students: A response to O'Connor, Cobb, and O'Connor (2003). Personality and Individual Differences, 37, 485-494.

Frankl, V. E. (1958). The will to meaning. Journal of Pastoral Care, 12,82-88.

Gesser, G., Wong, P., & Reker, G. (1987). Death attitudes across the life-span: The development and validation of the Death Attitude Profile. Omega, 18, 113-128.

Gore, P. A. Jr. (2000). Cluster analysis. In H. E. A. Tinsley & S. D. Brown (Eds.), Handbook of applied multivariate statistics and mathematical modeling (pp. 297-321). San Diego, CA: Academic Press.

Hui, V., & Fung, H. (2009). Mortality anxiety as a function of intrinsic religiosity and perceived purpose in life. Death Studies, 33, 30-50.

Hutsebaut, D. (1996). Post-Critical Belief: A new approach of the religious attitude problem. Journal of Empirical Theology, 9(2), 48-66.

Inhelder, B. & Piaget, J. (1958). The growth of logical thinking from childhood to adolescence. New York: Basic Books.

Kerkhofs, J., Peetermans, D., & Geeraerts, D. (1997). De Europeanen en hun waarden: Wat wij denken en voelen [The Europeans and their valueas: What we think and feel]. Leuven, Belgium: Davidsfonds.

Lavoie, J., & de Vries, B. (2003). Identity and death: An empirical investigation. Omega-Journal of Death and Dying, 48, 223-243.

Lester, D. & Abdel-Khalek, A. (2008). Religiosity and death anxiety using non-western scales. Psychological Reports, 103, 652-652.

Magnusson, D. (1998). The logic and implications of a personcentered approach. In: R. Cairns R., L. Bergman, & J. Kagan (Eds.), Methods and models for studying the individual (pp. 3-36). Thousand Oaks: Sage.

Markstrom, C. (1999). Religious involvement and adolescent psychosocial development. Journal of Adolescence, 22, 205-221.

Neimeyer, R., Wittkowski, J., & Moser, R. P. (2004). Psychological research on death attitudes: An overview and evaluation. Death Studies, 28,309-340.

Neyrinck, B., Lens, W., & Vansteenkiste, M. (2005). Goals and regulations of religiosity: A motivational analysis. In M. L. Maehr & S. Karabenick (Eds.). Advances in motivation and achievement, Vol. 14 (pp. 77-106). Greenwich, Conn.: Jai Press Inc.

Scholte, R. H. J., van Lieshout, C. F. M., de Wit, C. A. M., & van Aken, M. A. G. (2005). Adolescent personality types and subtypes and their psychosocial development. Merrill-Palmer Quarterly, 51, 258-286.

Sterling, C., & Van Horn, K. (1989). Identity and death anxiety. Adolescence, 24, 321-326.

Spilka, B., Hood, R., Hunsberger, B., & Gorsuch, R. (2003). The psychology of religion: An empirical approach. New York: Guilford Press.

Tillich, P. (1952). The courage to be. London: Nisbet.

Tinsley, H. E. A., & Brown, S. D. (2000). Multivariate statistics and mathematical modeling. In H. E. A., Tinsley & S. D. Brown (Eds.), Handbook of applied multivariate statistics and mathematical modeling (pp. 3-36). San Diego, CA: Academic Press.

Weisman, A. (1972). On dying and denying: A psychiatric study of terminality. New York: Behavioral Publications.

Wong, P., Reker, G., & Gesser, G. (1994). Death Attitude Profile-Revised: A multidimensional measure of attitudes toward death. In R. A. Neimeyer (Ed.), Death anxiety handbook: research, instrumentation, and application (pp. 121-148). London: Taylor and Francis.

Wulff, D. M. (1991). Psychology of religion: classic and contemporary views. New York: Wiley.

Yalom, I. (1980). Existential psychotherapy. New York: Basic Books.

AUTHORS

DEZUTTER, JESSIE. Address: Center for the Psychology of Religion, Tiensestraat 102 bus 3715, Leuven. Title: Doctoral Researcher, Center for the Psychology of Religion, Catholic University of Leuven, Belgium. Degree: Master in Psychology Specializations: Psychology of Religion.

LUYCKX, KOEN. Address: Center for Developmental Psychology, Tiensestraat 102 bus 3715, Leuven Title: Post-Doctoral researcher at the Fund for Scientific Research Flanders (FWO) Center for Developmental Psychology Catholic University of Leu ven, Belgium Degrees: PhD. Specializations: Identity, Developmental Psychology.

HUTSEBAUT, DIRK. Address: Center for the Psychology of Religion, Tiensestraat 102 bus 3715, Leuven Title: Professor Emeritus, Center for the Psychology of Religion, Catholic University of Leuven, Belgium. Degree: PhD. Specializations: Psychology of Religion.

The contribution of the second author was supported by the Fund for Scientific Research Flanders (FWO). Correspondence concerning this article should be sent to Jessie Dezutter, Catholic University of Leuven, Department of Psychology, Tiensestraat 102 bus 3715, 3000 Leuven, Belgium. Email: [e-mail].

(1) We did not include the "Escape Acceptance" scale from the DAP-R in this study (a) because we did not have a priori hypotheses on the link between religiosity and Escape Acceptance and (b) because this scale has a specific content (i.e., death as deliverance from pain and suffering) that may be of little relevance in our community adolescent sample.