Introduction

The World Health Organization (2001) has identified suicide as one of the three most important causes of death during the development of adolescence, but the extent of this phenomenon can be considered more pervasive if we take into account both suicidal ideation and failed suicide attempts. Lifetime prevalence estimates of suicidal ideation among non clinical populations of adolescents generally range from 60% (Strang and Orlofsky 1990) to 75% (Bonner and Rich 1987), and in many cases its severity increases the risk of completed suicide (Lewinsohn et al. 1996). For this reason, the individuation of the risk factors for suicidal ideation may be very important in suicide prevention.

Depression has been commonly identified as the most significant clinical risk for suicidal ideation and behaviors (Chabrol et al. 2007; De Man 1999). Since depression is a prototypic multifactor syndrome, many processes, mechanisms, and risk factors need to be evaluated simultaneously to provide a complete understanding of the etiology of depression starting in adolescence. For example, as Lewinsohn has affirmed in his behavioral theory (Libet and Lewinsohn 1973), poor social skills may be crucial in influencing the experience of depression. People who lack adequate social skills are, by definition, those persons who will not be able to produce many positive and rewarding outcomes and to avoid negative and punishing outcomes as a result of their behaviors. From a socio-emotional perspective, depression during development may be associated to a failure of the attachment system in regulating affect (Sroufe 1989). When the caregiver reacts to the child’s distress in a responsive way, the child develops trust in the parental capability to regulate his or her emotional state. Conversely, if the emotional signals are ignored, the caregiver acts as a poor regulator of negative and harmful feelings (Liotti 2000; Sloman et al. 2003). In this sense, as Bowlby (1980) has argued, the unavailability of parents as attachment figures may be experienced as a loss that may result in depression during adolescence (Pace and Zappulla in press).

The importance of depression in suicidal ideation has been largely stressed, especially in clinical population (Rutter 2007). Most adolescent suicides (attempted or completed) are the result of suffering from psychiatric disorders, among which depression represents the most important (Hallfors et al. 2004). Both retrospective and prospective studies have revealed a strong connection between major depression and suicide (Blair-West et al. 1997; Isacsson 2000).

However, most people who suffer from depression do not exhibit severe suicidal ideation and do not make attempts (Levy and Deykin 1989), and not all the individuals who are likely to consider suicide are clinically depressed. Therefore it is imperative to focus on those factors that may get involved in the relation between depression and suicidal ideation. Several authors have underlined the relationship between suicidal ideation and family variables, such as low cohesion, conflicts with parents, and a lack of connectedness and subsequent independence from them (Marcenko et al. 1999). In this perspective, the adolescent psychological move away from parents, or emotional autonomy, may assume the feature of separation throughout psychological individuation or, conversely, may take the shape of the detachment and may be characterized by alienation and distrust toward the parents’ role (Frank et al. 1988). In any case, both as separation and detachment, emotional autonomy may be singly linked to suicidal thoughts, but it may also intervene in the relation between depression and suicidal ideation.

Traditionally, emotional autonomy has been described as the process by means of which individuals relinquish their dependency on parents and begin to develop a more mature, realistic and balanced perception of their own parents and of their roles (Beyers and Goossens 2003; Ryan and Lynch 1989; Steinberg 2005). The relation between emotional autonomy and adjustment has been controversial (Pace and Zappulla 2009). Some authors have linked it to adaptive outcomes; others have related it to indexes of maladjustment. The first position has been claimed by Steinberg and Silverberg (1986) who have attributed an adaptive meaning to emotional autonomy. Researchers, on the basis of the psychoanalytic perspective (Blos 1979) conceptualizing emotional autonomy as a second process of separation-individuation, have positively defined it as a physical and emotional separation from family. According to the authors, for an adaptive outcome, adolescents should develop a sense of non-dependency from their parental figures and reach a personal individuation. The second position has been claimed by Ryan and Lynch (1989) who have affirmed that emotional autonomy is linked to a negative family functioning. They have shown that the more adolescents express emotional autonomy (or as the authors have renamed it, emotional detachment), the less they feel connected or secure within the family, the less they experience their parents as conveying love and understanding, and the less they report willingness to draw upon parental resources. In the same direction, it has been confirmed the role that emotional autonomy as detachment plays in predicting psychosocial maladjustment, such as depression and other internalizing problem behaviors (Lo Coco et al. 2001, 2000; Ryan and Lynch 1989). Recently, some authors have tried to address the controversy (Beyers and Goossens 2003; Beyers et al. 2005, 2003) by distinguishing two different types of emotional autonomy, a healthy separation from parents and a more hostile detachment from parents. Separation has a positive meaning and “involves a move away from childhood representations of parents and is not accompanied by negative feelings towards them” (Beyers et al. 2005, p. 154). Detachment has a negative significance because it is seen as disengagement from earlier infantile representations of parents due to feelings of mistrust and perceived alienation towards them. According to this distinction, separation would be associated with adolescents’ well-being, whereas detachment would be related to adolescents’ maladjustment (Beyers et al. 2005).

Although no specific literature exits linking the emotional autonomy process to suicidal ideation, a great amount of research has pointed out the influences that family variables, as the process towards emotional autonomy has been largely considered (Steinberg and Silverberg 1986), may have on suicidal ideation. If on the one hand a positive familial climate may be a buffer from negative outcomes, on the other hand a negative one may exacerbate any individual problem (Marcenko et al. 1999; Sun and Hui 2007). Most of the negative feelings that detached adolescents have about family functioning, such as the absence of parental support, the lack of closeness to parents, the low level of security and family cohesion, and the high level of parental rejection (Lo Coco et al. 2000, 2001; Ryan and Lynch 1989), have been found to contribute to suicidal ideation (Lee et al. 2006; Kerr et al. 2006; Sun and Hui 2007). Families characterized by excessive detachment often produce an atmosphere of isolation and intolerance which increase suicide risk (Ledgerwood 1999). From the attachment theory, feelings of insecurity and rejection, typical of detached adolescents, may explain self-destructive behaviors. For example, Adam (1994) has proposed a relation between the negative quality of attachment and suicidal behavior. He has suggested a model similar to that proposed by Sroufe (1989) that involved adolescents’ experiences of distress, their inability to acknowledge the difficulties deriving from it and to establish trusted relationships through which regulating their mood. Indeed, a history of suicidal ideation or suicide attempts has been associated with parental unavailability and failure to achieve secure attachment (West et al. 1999).

Several studies have shown the existence of relevant contradictory results on gender differences in each of the dimensions here considered. With regards to emotional autonomy from parents, in a number of studies, girls have shown higher levels of emotional autonomy than boys (Lo Coco et al. 2001; Steinberg and Silverberg 1986); in others, no gender differences have stood out (Huey and Henry 2005). With regards to detachment and separation, Beyers et al. (2005) have underlined that the new factor structure of emotional autonomy proved invariant and equal between males and females, allowing the comparison of average scores across gender.

Some study has also shown that girls in comparison to boys report more depression starting in early adolescence and exhibit higher levels of depressed mood. Pubertal development and timing have been related to sex differences in adolescent depression. Those girls who started puberty earlier than their peers were more likely to become depressed (Twenge and Nolen-Hoeksema 2002; Wichstrom 1999). With regards to suicidal ideation, Lewinsohn et al. (2001) have found that ideation is prevalent among females than among males in non-clinical samples.

In summary, both depression and negative feelings about family functioning may be particularly critical for suicidal ideation. From a multidimensional model we might take into account the relation between suicidal ideation, depression and familial quality-related variables. In depressed adolescents detachment from parents may exacerbate their vulnerability and their tendency to think about suicide (Kerr et al. 2006). From this perspective, on the contrary, a healthy separation from parents might play a moderator role.

In our study we examined the relations between depression, emotional autonomy quality-related constructs of separation and detachment, and suicidal ideation, focusing on the unique and common contribution that depression, separation and detachment made to suicidal ideation. We hypothesized that depression and detachment from parents would have significant effects on suicidal ideation, both singly and jointly. We expected that high levels of depression and high levels of detachment from parents would be related to suicidal ideation. Moreover, we expected that at high levels of depression the rate of suicidal ideation would be more elevated at high levels of detachment from parents and lower at high levels of separation from parents. According to literature reporting differences between boys and girls in all the variables considered, we examined gender differences. We also took into account the age in the preliminary analyses in order to verify any possible difference between the two degrees.

Method

Participants

Participants in the study were 403 adolescents (196 boys and 207 girls), attending all the third (n = 233; age: 15–16 years, M = 16.05, SD = .48) and fifth (n = 170; age: 17–19 years, M = 17.90, SD = .51) classes of two public high schools located in a big city in Southern Italy. We selected these schools as representative of the large part of Italian adolescents’ middle-class population. We recruited participants by sending their parents an informed consent letter. All the adolescents who were approached participated in the study and completed all the measures. We obtained demographic information by the adolescents by means of a brief list of questions included in the self-report measures. All the participants were Caucasian. Schools served a middle-class community of skilled workers and professionals. The majority (88%) of the participants’ parents had a college degree or higher. Most of the participants (93%) came from two-parent intact families.

Procedure

Participants completed self-report measures on depression, emotional autonomy, and suicidal ideation. Two researchers collected data during class hours in the absence of classroom teachers. This research respected the ethical norms of the research and was approved by the Italian Psychology Association.

Measures

Depression

We administered the Italian validation (Camuffo et al. 1988) of the Children’s Depression Inventory (CDI; Kovacs 1980) to assess adolescents’ feelings of depression. The measure consists of 27 items (e.g., “I feel so sad that I can not tolerate it”) to which subjects could answer on a 3-point Likert-type scale ranging from 1 (not at all true) to 3 (always true). The internal consistency of the Italian validation was α = .76. A weighted sum (weighted by the number of items) of depression was computed for each participant, with higher scores indicating greater feelings of depression.

Emotional Autonomy from Parents

We administered the Italian validation (Meleddu and Scalas 2002) of the Emotional Autonomy Scale (EAS, Steinberg and Silverberg 1986) to assess the level of adolescents’ emotional autonomy from parents. The measure consists of 20 items to which adolescents could answer on a 5-point Likert-type scale ranging from 1 (strongly disagree) to 5 (strongly agree). The original version of the scale allowed obtaining four scores about emotional autonomy: perception of parents as people, parental deidealization, non-dependency on parents, and individuation. In this study, we considered the alternative factor structure proposed by Beyers et al. (2005), in which two higher-order factors, called separation and detachment, substituted the original four factor structure. Separation was obtained by summing items of the original scales “deidealization” and “non-dependency” and two items of “individuation”. Detachment was obtained by summing items of the original scale perception of “parents as people” and two items of “individuation”. A first confirmatory factor analysis conducted on our data confirmed the alternative structure with two factors, separation (α = .76) and detachment (α = .60). Because the α obtained for the detachment scale was not very high, to improve the internal consistency of the scale, we removed the item 19 on the basis of the Italian validation of the EAS (Meleddu and Scalas 2002). The coefficient obtained after removing the item 19 (α = .64) was quite similar to that obtained by Beyers et al. (2005) and then considered acceptable. A weighted sum (weighted by the number of items in each scale) of separation and detachment was computed for each participant, with higher scores indicating higher levels of separation and detachment.

Suicidal Ideation

We assessed suicidal ideation by using a single item of the CDI: (a) I do not think to kill myself (score = 1); (b) Sometime I think to kill myself but I would not do it (score = 2); (c) I want to kill myself (score = 3). This item was not computed in the depression score. Scores ranging from 2 to 3 were considered indicative of suicidal ideation. The use of a single item to assess suicidal ideation is supported by a great amount of research examining this construct in childhood and adolescence by means of single items of depression questionnaires (Miranda and Nolen-Hoeksema 2007; Overholsen et al. 1995; Sun and Hui 2007). Each time a student expressed high suicidal ideation, we planned a meeting with correspondent teachers to increase their availability to listen at the general class’ problems: in that way we kept adolescents’ anonymity and alerted teachers about the need to inform the public health system (the only one authorized in Italy to treat scholars’ health problems).

Analysis Plan

The first step in the analyses comprises preliminary analyses, including descriptive statistics on the independent and dependent variables and intercorrelations between all the variables.

The second step in the analyses explores direct prediction from depression and from separation and detachment to suicidal ideation, first for boys and then for girls. We conducted simple regression analyses to test direct relations.

The third step in the analyses examines the additive and interactive contributions of the variables in predicting suicidal ideation. We conducted two hierarchical multiple regression analyses, one for boys and one for girls, by adding depression, separation and detachment in each of three steps, followed by the interaction terms in a fourth step. Before creating the interaction terms (depression × detachment, depression × separation), we centered all the predictors on their means. We entered the centered predictors in the first step and the interactions terms on the second step.

The final step probes the significant interactions, by using the procedure recommended by Aiken and West (1991) and Holmbeck (2002). We restructured the significant regression equations to express the regression of suicidal ideation on depression at levels of the moderator variables.

Results

Preliminary Analyses: Descriptive Statistics and Intercorrelations

To explore gender and age differences in the levels of depression, separation and detachment, and suicidal ideation, we conducted one-way ANOVAs. Data (Table 1) showed significant gender differences for depression (F(1,402) = 10.47, p < .001) and suicidal ideation scores (F(1,402) = 4.59, p < .03). In comparison to boys, girls reported higher levels of depression and suicidal ideation. None of the variables differed significantly across ages. Consequently, we did not consider age in the further analyses.

Table 1 Descriptive statistics of all of the study variables
Full size table

We used correlational analyses to investigate the interrelationships between depression, emotional autonomy quality-related constructs of separation and detachment, and suicidal ideation. We conducted separate analyses for boys and girls. Data for boys showed that depression was positively correlated to detachment (r = .20, p < .01) and to suicidal ideation (r = .27, p < .01). A positive correlation also emerged between detachment and suicidal ideation (r = .21, p < .01). Separation and detachment were negatively intercorrelated (r = −.22, p < .01). Data for girls showed that depression was positively correlated to detachment (r = .31, p < .01) and to suicidal ideation (r = .31, p < .01). We did not find significant correlations between emotional autonomy variables and suicidal ideation for girls.

Direct Relations

Depression and Suicidal Ideation

The simple regression analyses conducted to test boys and girls’ depression as predictors of suicidal ideation revealed a significant relation between boys’ depression and suicidal ideation, F(1,195) = 15.09, p = .000, R 2 = .07. Boys with higher levels of depression showed more suicidal ideation (Table 2). Girls’ depression did predict suicidal ideation, F(1,206) = 21.93, p = .000, R 2 = .10. Girls with higher levels of depression showed more suicidal ideation (Table 3).

Table 2 Regression analyses predicting boys’ suicidal ideation from depression and emotional autonomy quality-related constructs of separation and detachment
Full size table
Table 3 Regression analyses predicting girls’ suicidal ideation from depression and emotional autonomy quality-related constructs of separation and detachment
Full size table

Separation and Suicidal Ideation

The simple regression analyses conducted to test boys and girls’ separation as predictors of suicidal ideation did not reveal significant effects, both for boys (R 2 = .01) and girls (R 2 = .01) (Tables 2 and 3).

Detachment and Suicidal Ideation

The simple regression analyses conducted to test boys and girls’ detachment as predictors of suicidal ideation revealed a significant relation between boys’ detachment and suicidal ideation, F(1,195) = 9.08, p = .003, R 2 = .05. Boys with higher levels of detachment showed more suicidal ideation (Table 2). Girls’ detachment did not predict suicidal ideation, R 2 = .01 (Table 3).

Additive and Interactive Model of Depression and Emotional Autonomy Quality-Related Constructs of Separation and Detachment Predicting Boys and Girls’ Suicidal Ideation

The hierarchical multiple regression analysis conducted for boys showed that depression and detachment added significantly to the prediction of suicidal ideation, β = .24, p = .01 and β = .16, p = .03 respectively, with higher depression and higher detachment each uniquely predicted suicidal ideation (Table 2). The overall regression was significant, F(3,195) = 6.95, p = .000, R 2 = .31. This supports an additive model of boys’ depression and detachment predicting suicidal ideation. The regression showed an interactive contribution of depression and detachment (Table 2) to the prediction of boy’s suicidal ideation, F(5,193) = 6.07, p = .000, R 2 = .37. The observed effect size was small, η2 = .04. The interaction between depression and separation did not add significantly to the prediction of suicidal ideation.

The hierarchical multiple regression analysis conducted for girls did not show an additive contribution of the variables, with only depression entering significantly to the prediction, β = .24, p = .000, with higher depression predicting more suicidal ideation (Table 3). The regression showed that the interaction between depression and detachment (Table 3) added significantly to the prediction of girls’ suicidal ideation, F(5,206) = 4.96, p = .000, R 2 = .33. The observed effect size was small, η2 = .03. The interaction between depression and separation did not add significantly to the prediction of suicidal ideation.

We probed the above interactions following the procedure recommended by Aiken and West (1991) and Holmbeck (2002). We restructured the regression equations to express the regression of suicidal ideation on depression at levels of detachment. The values of detachment chosen corresponded to 1 SD above the mean (high) and 1 SD below the mean (low). These equations are plotted in Fig. 1 (boys) and in Fig. 2 (girls) to display the interactions.

Fig. 1
figure 1

Regression lines for relations between boys’ depression and suicidal ideation as moderated by detachment (a 2-way interaction)

Full size image
Fig. 2
figure 2

Regression lines for relations between girls’ depression and suicidal ideation as moderated by detachment (a 2-way interaction)

Full size image

With regards to boys, the simple slope for the high level of detachment was negatively and significantly different from zero (β = −.18, p = .01), whereas the simple slope for the low level of detachment was positively and significantly different from zero (β = .16, p = .02). As indicated, the simple slopes of suicidal ideation tended to differ from one another as a function of the value of detachment. Thus, at higher levels of depression, suicidal ideation tended to be higher when detachment was high and tended to be lower when detachment was low. On the contrary, at lower levels of depression, suicidal ideation tended to be higher when detachment was low and tended to be lower when detachment was high.

Also for girls, the simple slope for the high level of detachment was negatively and significantly different from zero (β = −.20, p = .004), whereas the simple slope for the low level of detachment was positively and significantly different from zero (β = .23, p = .001). As indicated, the simple slopes of suicidal ideation tended to differ from one another as a function of the value of detachment. Thus, at higher levels of depression, suicidal ideation tended to be higher when detachment was high and tended to be lower when detachment was low. On the contrary, at lower levels of depression, suicidal ideation tended to be higher when detachment was low and tended to be lower when detachment was high.

Discussion

In our study we investigated the relations between depression, emotional autonomy quality-related constructs of separation and detachment, and suicidal ideation, focusing on the unique and common contribution that depression, separation as positive result of the individuation process, and detachment as negative result, made to suicidal ideation during adolescence. We also examined differences between boys and girls.

With regards to gender differences, data confirmed previous findings: girls showed higher score than boys only on depression and suicidal ideation. Differently, regarding the relation between variables, depression predicted suicidal ideation for both boys and girls. Detachment predicted suicidal ideation only for boys. Separation did not predict suicidal ideation for both boys and girls. The data for boys supported an additive model of prediction to suicidal ideation, such that depression and detachment each contributed unique variance to the prediction of suicidal ideation. Furthermore, the data for boys supported an interactive model of prediction to suicidal ideation, such that detachment contributed to exacerbate the risk of suicidal ideation when adolescents were already at risk because of depression. The data for girls supported an interactive, but not additive, model of prediction to suicidal ideation, such that depression and detachment did not contribute independently to suicidal ideation but in a joint way.

Depression predicted suicidal ideation for both boys and girls. These results confirm how feelings of depression should be considered as an important risk factor during adolescence (Dori and Overholser, 1999) regardless of gender. Detachment was related to suicidal ideation for boys but not for girls. Data seem to suggest that the move away from parents in a way similar to an emotional detachment may be particularly critical for males during adolescence. We might read it in the light of literature that underlines that girls report higher levels of empathic concern and emotional support from friendships than boys (Clark and Ayers 1993): we may hypothesize that emotional detachment from parents during adolescence represents a condition that drives to self-destructive thoughts for boys but not for girls, who seem more able to rely on peer support (Moller et al. 1992).

The results supported an additive model of depression and detachment, but not of separation, in the prediction of suicidal ideation only for boys, such that depression and detachment contributed independently to suicidal ideation. On the contrary, data did not support an additive model of girls’ depression and detachment predicting suicidal ideation. These findings underline the importance of focusing on the two variables independently from each other. It seems that only for boys detachment adds up to depressive feelings increasing the emergence of suicidal thoughts. Conversely, for girls these factors do not sum up enough evidence in the prediction of self-destructive thoughts.

For both boys and girls our data supported an interactive model of prediction to suicidal ideation, such that detachment contributed to exacerbate the risk of suicidal ideation when adolescents were already at risk because of depression. These results are in line with recent literature’s findings that illustrate the role that the negative feelings about their own individuation process have on adolescents’ general maladjusted conditions of growth (Meeus et al. 2005).

We might hypothesize that adolescents’ detachment represents an important factor through which evaluating the goodness with which members of the family-system face an important passage. The role that family may play in determining the valence of emotional autonomy takes relevance: those adolescents who are highly detached from parents would probably feel an excessive amount of disaffection toward them (Huey and Henry 2005). In this sense, emotional detachment would increase the possibility that from transitive and non-clinical depression adolescents fall into suicidal ideation. As a matter of fact, we found that suicidal ideation, among depressed adolescents, tended to be higher when they showed higher levels of detachment. Findings seem to suggest the idea that detachment may be more similar to an emotional distance between parents and adolescents than to the achievement of a mature autonomy from parents. For this reason, it would have a negative influence on the individual capability to face the numerous developmental tasks characterizing this period of life (Lo Coco et al. 2000).

Nevertheless, since the role played by detachment as variable that exacerbates the effects of adolescents’ depressive feelings on suicidal ideation seems quite understandable, what remains unclear is the role that detachment plays among non depressed adolescents. Data showed that at low levels of depression, suicidal ideation tended to be higher when detachment was lower and tended to be lower when detachment was higher, especially for boys. Thus, contrary to the data from the simple regression analyses, highlighting that boys with higher levels of detachment showed more suicidal ideation, data from the interactive model drew a picture in which non depressed adolescents thought about suicide if they were low detached from parents. It is quite complex to explain these unexpected results. It appears that when we considered the unique contribution of detachment in the prediction of suicidal ideation the path of prediction was clear. Conversely, when we considered the common contribution of detachment and depression, the path of prediction became more complex. The interplay between the different levels of depression and detachment seemed to change the nature of the outcome. In this sense, if the exacerbating role that detachment from parents played toward suicidal ideation among depressed adolescent may be considered in line with literature, the link between low level of depression and low level of detachment in the prediction of suicidal ideation is unexpected and needs some more consideration.

According to some authors, the process toward emotional autonomy, of which detachment is part of, is one of the most complex and controversial developmental tasks in adolescence (Beyers et al. 2003). Particularly, the relation between autonomy and well-being is controversial: not only autonomy sometime predicts well-being (Steinberg and Silverberg 1986) and sometime maladjustment (Ryan and Lynch 1989), but it may represent on the one hand a buffer against malaise in a stressful familial environment (Fuhrman and Holmbeck 1995), or surprisingly, on the other hand, it is related to positive outcomes only when the familial climate is supportive (Lamborn and Steinberg 1993). Equally surprisingly, from our data it seems that low level of detachment intervenes among non depressed adolescents promoting self-destructive thoughts.

Certainly, these data need to be better investigated by more complex models of the relations studied here, for example by focusing on additional adjustment outcomes or on a possible deny of parents-adolescents conflicts during the development of emotional autonomy. Anyway, they add a piece to the “vicissitudes” of this controversial topic.

The results did not support a significant role of separation. On the basis of literature (Beyers and Goossens 2003; Beyers et al. 2005), we expected that separation from parents would have played a moderator role in the relation between depression and suicidal ideation, such that suicidal ideation would have been lower among depressed adolescents if they could perceive to begin a soft move away from parents in a climate of closeness and connection. Data did not support our hypothesis. Separation seems to be independent of depression and suicidal ideation. It is possible that separation could have moderator effects on adjustment only when adolescents get older; for example as they become involved with the transition to university (Beyers and Goossens 2003).

Our study has some limitation, all of which suggesting fruitful avenues for further research. First of all, it included only Italian participants: it could be interesting to reproduce the same research in other countries, where separation and detachment might assume a different connotation and have a different influence on adolescent adjustment. Some limitations regarded the measures. First, besides the measures obtained directly by adolescents, it could have been useful to collect information by means of other informants, such as peers and adults. Moreover, we only used one item for measuring suicidal ideation, which yields a less stable estimate of the construct. Also, the final coefficient obtained for the EAS’s detachment scale (α = .64), even if considered acceptable, remained low. Further, we collected data for this study at a single point in time in order to focus specifically on the issues of co-occurrence of depression, separation, detachment, and suicidal ideation during a phase of adolescence. Longitudinal data would allow the tracking of depression and emotional autonomy constructs to see how they are interrelated temporally and to delve the peculiar relation between low detachment and low levels of depression. Moreover, longitudinal research would permit to focus on the emotional autonomy as a process and to study the role of separation more profoundly. Finally, considering epidemiological data about suicidal behaviors, longitudinal data would permit to observe to what extent suicidal ideation predicts suicidal attempts and which role the interrelations between depression and detachment from parents play in these paths. Such a design would allow examining the possibility that detachment might be a symptom of depression.

In spite of these limitations, the study represents an important contribution to the study of suicidal ideation during adolescent development because it emphasizes the role that feelings about the quality of path toward emotional autonomy from parents could play in the relation between depression and self-destructive thoughts. Moreover our findings could be considered as a contribution to the literature underlining how poor family functioning, such as lack of cohesion (Friedrich et al. 1982) or support (Dubow et al. 1989), plays a crucial role in the relationship between personal and negative ways to react (for example, by developing a general negative mood) to changes in adolescence and the propensity that potential risk factors could become maladaptive symptoms, such as suicidal ideation.