Introduction
Loneliness is associated with numerous indices of maladaptive adjustment and there is an attendant need to identify the outcomes of changes in loneliness during adolescence. Empirical study has begun to investigate these relationships but has neither considered positive mental wellbeing as an outcome nor addressed the multidimensional nature of loneliness. This study identified trajectories of change in relation to two dimensions of loneliness (Isolation loneliness and Friendship loneliness) and considered the effects of these on both positive mental wellbeing and symptoms of depression.
Loneliness is the sense that one’s social relationships are not commensurate with one’s desired social relationships (Cacioppo et al.,
2015a). Considering loneliness across the life-span, it is evident that adolescence and early adulthood are times when loneliness peaks (Heinrich & Gullone,
2006; Xerxa et al.,
2023). According to the Office for National Statistics (
2018) nearly half of 10–12-year-olds report feeling lonely at least some of the time, with this rising to almost 60% in 16–24-year-olds. Brain development during adolescence in regions involved in social processing, a desire for increased peer interaction and friendships (Orben et al.,
2020), but increased sensitivity to social rejection (Blakemore & Mills,
2014), and developmental shifts and transitions in social networks (e.g., moving from primary to secondary school, or leaving home) are all known risk factors for increased loneliness (Siva,
2020; Sundqvist & Hemberg,
2021).
Adolescence is a particularly vulnerable period for the development of mental health problems, with half of all mental health problems emerging before the age of 14 (Kessler et al.,
2007) and the peak age of incidence coinciding with the transition from “childhood/adolescence” to “adult” life (Thapar & Riglin,
2020). Loneliness in adolescence is a known risk factor for anxiety (Maes et al.,
2019), depression (Fontaine et al.,
2009; Lasgaard et al.,
2011), suicidal ideation (Gallagher et al.,
2014) and diminished positive mental health (Lyyra et al.,
2021). Moreover, loneliness is associated with later reports of mental health problems, physical health risk behaviours, poorer employment prospects (Leigh-Hunt et al.,
2017; Matthews et al.,
2018), poorer sleep quality (Matthews et al.,
2017), withdrawal from social activities and relationships (Böger & Huxhold,
2018), suicidal behaviours (Heinrich & Gullone,
2006), cardiovascular disease (Leigh-Hunt et al.,
2017) and both morbidity and mortality (see Hawkley & Cacioppo,
2010). Given that loneliness experienced in adolescence can have both immediate and lasting implications for outcomes in later years, studying loneliness in young people is important.
The degree to which different trajectories of loneliness are associated with psycho-social adjustment has often been the focus of published work (see Table
1). Confidently naming trajectories relating to loneliness is problematic since there is no accepted definition of low, medium, or high loneliness. However, across studies, youth in stable-low trajectory groups tend to report better psychological and social adjustment than those in other groups, including lower depressive symptomatology (Jobe-Shields et al.,
2011; Ladd & Ettekal,
2013; Qualter et al.,
2013; Schinka et al.,
2013; Vanhalst et al.,
2013), higher social skills (Schinka et al.,
2013), better self-esteem (Vanhalst et al.,
2013), and higher academic outcomes (Benner,
2011). Young people with chronically high levels of loneliness tend to be most at risk for depressive symptomatology (Ladd & Ettekal,
2013; Qualter et al.,
2013; Schinka et al.,
2013; Vanhalst et al.,
2013), alcohol misuse (Qualter et al.,
2013), suicidal ideation (Schinka et al.,
2013), and poorer general health (Harris et al.,
2013; Qualter et al.,
2013). Among those with trajectories of loneliness that display change over time, there is evidence that increasing levels of loneliness are a marker for later maladjustment (Benner,
2011; Jobe-Shields et al.,
2011; Qualter et al.,
2013) and there is some support for a scar hypothesis (Rohde et al.,
1990) whereby higher starting levels that subsequently reduce are still associated with poor outcomes (Harris et al.,
2013).
Table 1
Studies estimating trajectory classes of loneliness among children and young people
| N = 209. Aged 8 years at T1. | Data collected three times, at 18-month intervals. | Two-class solution: Relatively High, Reducing (48%), and Low Stable (52%). | At 11 years, the Relatively High, Reducing group had higher levels of depressive symptoms, poorer general health, and poorer sleep. |
| N = 640. Grade 9 at T1. | Data collected twice in Grades 9 and 10. | Three-class solution: Consistently Low (78%), High (11%), and Low but Increasing (11%). | High and Low Increasing classes had poorer academic outcomes than the Low class. |
Jobe-Shields et al. ( 2011) | N = 170. Aged 9 years at T1. | Data collected annually at ages 9, 10 and 11 years. | Three-class solution: Stable Low (65%), Increasers (23%), and Decreasers (12%). | Stable Low had generally positive peer functioning; Increasers at risk group for developing later internalising symptomatology; and Decreasers had mixed pattern of peer functioning at age 9 years, but were indistinguishable from the Stable Low group subsequently. |
| N-130. 9 years. | Data collected at ages 9, 13, 16, and 21. | Three-class solution: Stable-Low (51–61%), Low-Increasing (21–32%), and High-Declining (7–22%). NB. Ranges are reported here as three separate clustering techniques were employed. | |
| N = 586. Age 7 years at T1. | Data collected biannually for ten years. | Four-class solution: Low Stable (37%), Moderate Decliners (23%), Moderate Increasers (18%), Relatively High Stable (22%). | Both the High Stable and Moderate Increasing trajectories were associated with depressive symptoms at age 17 years. |
| N = 478. Age 12 years at T1. | Data collected annually for six years. | Five-class solution: Stable Non-Lonely (19%), Stable Low (20%), Stable High (Chronic) (14%), Moderate Decliners (42%), and Steep Decliners (6%). | Stable High had the highest level of symptoms of depression. |
| N = 832. Age 9 years at T1. | Loneliness assessed at ages 9, 11, and 15 years. | Five-class solution: Stable Low (49%), Moderate Increasing (32%), High Increasing (5%), Decreasing (11%), and Chronic (4%). | Depression assessed at 7 years old predicted greater likelihood of being in: High Increasing/Chronic groups than Stable Low. All groups reported higher levels of depression at age 15 when compared to the Stable. |
| N = 389. Mage at T1 = 15.22 years. | Cohort-sequential (two-groups, ages 15 and 16 at T1). Data collected annually for 5 years. | Five-class solution: Chronically High (3%), High Decreasing (6%), Moderate Decreasing (8%), Low Increasing (17%), and Stable Low (65%). | Those who never experienced loneliness were best adjusted (lowest stress, least depressive symptoms, highest self-esteem), whereas the Chronically High group had the most problems (highest stress, most depressive symptoms, lowest self-esteem). |
| N = 730. Mage at T1 = 15.43 years. | Data collected annually for 4 years. | Five-class solution: Chronic (3%), Low Stable (47%), Moderate-Stable (27%), Moderate-Increasing (14%), High Decreasing (9%). | Cognitive and emotional responses of the Chronic group seem to perpetuate, rather than reduce, loneliness (e.g., hypersensitivity to social exclusion and hyposensitivity to social inclusion). |
As is clear from Table
1, symptoms of depression have frequently been the focus of research seeking to document the outcomes of different loneliness trajectories. However, notably absent from this literature examining trajectories of change in loneliness is information concerning associations with positive mental wellbeing. Positive mental wellbeing is neither the polar opposite, nor absence, of psychological maladjustment (World Health Organisation WHO (
2004)) and includes both hedonic (i.e., happiness, subjective well-being) and eudemonic (i.e., positive functioning) aspects of wellbeing (Clarke et al.,
2011). Promoting positive mental wellbeing among adolescents is a national priority in many countries and has been placed at the centre of Government policy involving children and young people in the UK (Garrat et al.,
2022; Scottish Government,
2020). Cross-sectional work with adolescents has supported that a negative association exists between positive mental wellbeing and loneliness (Houghton et al.,
2016; Lyyra et al.,
2021). Cross-lagged panel analyses using longitudinal data collected across the COVID-19 pandemic, in contrast, indicates that levels of loneliness may not be associated with positive mental wellbeing (Houghton et al.,
2022).
While positive mental wellbeing is negatively correlated with indices of psychological maladjustment such as depressive symptomatology (Clarke et al.,
2011) it is also the case that depressive symptomatology is more strongly associated with subsequent positive mental wellbeing than vice-versa among adolescents (Zadow et al.,
2017). Indeed, in adolescence, low levels of positive mental wellbeing can occur in the presence of low levels of depressive symptomatology and high levels of positive mental wellbeing can occur even in the presence of mental illness (Patalay & Fitzsimons,
2018), and there is significant change between groups over time (Petersen et al.,
2022). Thus, negative and positive mental wellbeing are distinctive concepts and the degree to which tackling loneliness can contribute to improving either health outcome is not clear.
As well as focussing only on negative indices of adjustment as outcomes of loneliness, the current literature examining trajectories of loneliness has not addressed the multi-dimensional nature of the construct. Contemporary theory and measurement of loneliness considers it to have between two and four dimensions (Goossens et al.,
2009; Houghton et al.,
2014; Majorano et al.,
2015). Parent (or family)-related and peer-related loneliness sub-scales have been reported in two separate studies (Goossens et al.,
2009; Marjorano et al.,
2015), and to these have been added two attitudinal factors reflecting positive and negative attitudes toward solitude (Goossens et al.,
2009). While there exists support for the existence of positive and negative attitudinal factors, they may reflect attitudes toward “aloneness” (e.g., “I have discovered the benefits of being alone”) rather than loneliness per se (Houghton et al.,
2014). Furthermore, parent (or family)-related and peer-related factors risk confusing the construct with the situations in which it may be expressed and thus it has been proposed that factors relating to friendship loneliness (e.g., “I feel part of a group of friends”, where higher scores reflect a positive outcome) and isolation loneliness (e.g., “I have nobody to talk to”, where higher scores reflect a negative outcome) are more theoretically distinct constructs (Houghton et al.,
2014). There is clear evidence that feelings of isolation and quality of friendships are highly, inversely correlated among adolescents (Houghton et al.,
2020).
Studying both isolation and friendships during adolescence is important because it is a period where belonging to a peer group is a major concern among young people (Rubin et al.,
2008) and peer interactions and relationships become increasingly more important (Qualter et al.,
2015; Rubin et al.,
2009). For some young people, insufficient connections to others can lead to profound and long-standing negative consequences, while having quality friendships can provide numerous social and emotional benefits (Houghton et al.,
2014). The turbulence of significant life transitions during adolescence means some young people drift in or out of loneliness while others experience loneliness persistently (Matthews et al.,
2023). Cross-sectional research with adolescents shows that positive mental wellbeing is positively associated with “friendship” loneliness (where higher scores represent lower loneliness) and is negatively associated with “isolation” loneliness (where higher scores represent higher loneliness) (Houghton et al.,
2016). This speaks to the importance of considering loneliness as a multi-dimensional construct, which no published work on loneliness trajectories and adjustment has yet considered.
Discussion
There is a need to identify the outcomes of changes in loneliness during adolescence, and to consider this within a multidimensional framework of loneliness. When experienced in adolescence, loneliness can impose long-term distress and significant adverse outcomes in later years, regardless of whether it recurs or persists over time (Matthews et al.,
2023). Therefore, the importance of interventions to break the cycle of loneliness during the early years is critical. To achieve this, the present study identified trajectories of change in relation to two dimensions of loneliness (Isolation loneliness and Friendship loneliness) and considered the effects of these on both positive mental wellbeing and symptoms of depression. Expectations relating to the number of trajectories (four or five), and the form that trajectories would take, were largely confirmed. Identification of different numbers of trajectory groups for each of the two different forms of loneliness is consistent with a multidimensional conceptualisation of loneliness. Additionally, youth in different trajectory groups evidenced clear differences in their mental health outcomes, supporting the argument that it is important to consider change and development in loneliness.
It was hypothesised that there would exist four or five trajectories of loneliness. Four trajectories were found for Isolation loneliness and five trajectories for Friendship related loneliness. The largest trajectory group in both cases, and as expected, is a consistently unproblematic one: 75% of youth were in the Low Stable Isolation Loneliness trajectory groups and 55% were in the High Stable Friendship Loneliness trajectory (with a further 26% reporting Average Stable Friendship Loneliness). However, contrary to expectation (Ladd & Ettekal,
2013; Qualter et al.,
2013; Schinka et al.,
2013; Vanhalst et al.,
2013,
2015,
2018), there was not a chronically high trajectory for Isolation loneliness, though in line with expectations there was a group of adolescents who reported consistently low levels of Friendship related loneliness. Given the relationships between trajectory membership and adjustment, discussed below, these prevalence rates are very encouraging because they indicate that most youth were not experiencing problematic levels of loneliness and were not therefore at risk for the problematic outcomes associated with such membership.
Since this is the first report of trajectories for two separate loneliness factors, the ways in which these may co-occur was documented. Four combinations were most prevalent: Low Stable Isolation Loneliness with either High Stable Friendship Loneliness (48%), High Decreasing Friendship Loneliness (9%), or Average Stable Friendship Loneliness (15%), and Elevated Stable Isolation Loneliness with Average Stable Friendship Loneliness (10%). All other trajectory group combinations occur at a prevalence of under 3%, though there were at least some young people in all 20 possible combination of trajectories. These combinations speak to the need for theory to consider the ways in which dimensions of loneliness may interact to produce outcomes. Of particular interest here is the possible disparity between desired and experienced levels of loneliness (Cacioppo et al.,
2015b; Qualter et al.,
2013; Rook,
1984) as the existence of these combinations suggests that youth may hold different beliefs about such a mismatch across different factors of loneliness. Alternatively, it may be that one or other dimension is associated with shorter, or fewer, fluctuations over time, and experience sampling assessments (e.g., van Roekel et al.,
2014) offer appropriate ways to assess such a proposition in future. However, it is encouraging to see that the most prevalent group is that which combines both of the least problematic trajectories of the two forms of loneliness.
These results speak to the multidimensionality of loneliness (Goossens et al.,
2009; Houghton et al.,
2014; Majorano et al.,
2015). Different numbers of trajectory classes for each type of loneliness is evidence in support of a multidimensional conceptualisation. In addition, youth did not always report “equivalent” trajectories on each measure of loneliness as might be expected if the two scales were simply mirror-images of one another. For example, only 23% of young people in the Low Increasing Isolation Loneliness trajectory were also in the High Decreasing Friendship Loneliness trajectory. Similarly, only 13% of young people in the High Decreasing Isolation Loneliness trajectory were also in the Low-Increasing Friendship Loneliness trajectory. These results strengthen the case for considering Friendship loneliness and Isolation loneliness as distinctive dimensions, reflecting the importance of considering both the quality of friendships and the degree to which youth feel isolated from others.
The benefits of having stable, low trajectories of loneliness with respect to symptoms of depression have been reported in previous research (e.g., Harris et al.,
2013; Qualter et al.,
2013; Schinka et al.,
2013; Vanhalst et al.,
2013). This was echoed on the current findings, with the Low Stable Isolation Loneliness trajectory group reporting fewer symptoms of depression than all other youth in this study. The difference was particularly notable concerning other youth in the Low Increasing Isolation loneliness trajectory group, whose levels of depressive symptomatology increased by more than a standard deviation from the start to the end of the study. The reasons underpinning why young people experiencing
increasing levels of loneliness across adolescence are at particular risk of experiencing symptoms of depression is unknown, but the disconnection between desired and perceived relationships which is the core of loneliness (e.g. Cacioppo et al.,
2015b) may offer some insight. Specifically, an increasing awareness of how far one’s desires are from what one actually has, especially in the context of having had it previously, may differentiate this group from those who experience decreasing, or consistently high, levels of Isolation Loneliness. Such a pattern of results could provide support for a scar hypothesis (Rohde et al.,
1990) pertaining to loneliness. Future research should seek to interrogate this possibility in more detail.
Novel findings reported here also indicated that specific trajectories of loneliness during adolescence can negatively impact positive mental wellbeing even after controlling for earlier levels of that construct. Outcomes for positive mental wellbeing were very similar to those reported above for symptoms of depression, although the risk associated with the Low Increasing Isolation loneliness trajectory group for symptoms of depression was not apparent for positive mental wellbeing. This is the first research to establish a link between trajectories of loneliness during adolescence and low levels of positive mental health and extends results reported in correlational studies (Houghton et al.,
2016; Lyyra et al.,
2021). While the present results contrast with Houghton et al.’s (
2022), the unique context of the COVID-19 pandemic that took place during that study may have had an impact on their results. Thus, interventions that can successfully tackle loneliness (Eccles & Qualter (
2021)) may usefully be used alongside, or integrated into, existing interventions to enhance positive mental wellbeing for young people (see Cilar et al.,
2020).
The results reported here may not generalise to adolescents outside of Western Australia. In addition, a reliance on self-reports may mean the results are subject to limitations relating to mono-method approaches such as shared-method variance. However, accurate and reliable reports of subjective dispositions such as those which were the subject of the current study can be difficult to obtain from third parties such as teachers or parents (see Frick et al.,
2009). The number of trajectories for each of the two forms of loneliness that were investigated here may be sample-specific, though one of the strengths of the study was the recruitment of a large demographically representative sample, a longitudinal design with low levels of attrition, and the use of state-of-the-art statistical techniques. A final limitation relates to the collection of both loneliness and outcome data at the same final time point, which may restrict causal interpretations of these data.
Publisher’s note Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.