Negative reactivity and parental warmth in early adolescence and depressive symptoms in emerging adulthood

Abstract

Objective

Cross‐sectional research suggests that relationships between temperamental negative reactivity and adolescent depressive symptoms may be moderated by parental warmth. The primary purpose of this study was to conduct the first prospective analysis of this relationship.

Method

Data from 1,147 families in an Australian population‐based longitudinal study were used to examine: (1) temporal relationships between negative reactivity in early adolescence (13–14 years) and depressive symptoms in emerging adulthood (19–20 years); (2) the moderating role of parent‐reported warmth in early adolescence (13–14 years); and (3) the moderating role of child gender. Hierarchical multiple regression was conducted to test the hypothesis that parental warmth would moderate the relationship between early adolescent negative reactivity and depressive symptoms in emerging adulthood.

Results

After accounting for previous depressive symptoms at age 13–14 years, negative reactivity was positively associated with later depressive symptoms. By contrast, parental warmth at 13–14 years was negatively associated with later depressive symptoms for females but not males. Parental warmth did not moderate the association between early adolescent negative reactivity and subsequent depressive symptoms.

Conclusions

This study was the first to use prospective data to assess the protective effects of early adolescent parental warmth on the association between negative reactive temperaments and early adult depressive symptoms. Findings suggest that parental warmth for negatively reactive children provides only concurrent protection against subsequent depressive risk. This study did not examine parent–child transactional models, which may, in future longitudinal research, improve understanding of how trajectories of parent–child goodness‐of‐fit contribute to depressive symptoms.

Key words:

• adolescence
• depressive symptoms
• goodness‐of‐fit
• negative reactivity
• parental warmth
• temperament

What is already known about this topic?

1. Negative reactive temperament and low‐warmth parenting in adolescence are linear risk factors for depressive symptoms.

2. The interaction, or ‘goodness‐of‐fit’, of child and parent variables can contribute to the heightened risk of psychopathology in addition to linear effects.

3. The relationship between early adolescent negative reactivity and concurrent depressive symptoms is moderated by parental warmth, but this has not been examined longitudinally.

What this paper adds?

1. Young adolescents with negative reactive temperaments have persistent risk for early adulthood depressive symptoms, even after accounting for earlier depressive symptoms.

2. Low parental warmth in adolescence, mostly reported by mothers, has lasting effects on subsequent depressive symptoms in emerging adulthood for girls, but not boys.

3. The interaction of negative reactivity and parental warmth in young adolescence does not directly predict emerging adulthood depressive symptoms.

Reactive temperaments appear early in life and pose a risk for depressive symptoms throughout development for males and female, but particularly females (Kampman & Poutanen, 2011; Katainen, Räikkönen, & Keltikangas‐Järvinen, 1999). In particular, longitudinal research demonstrates a consistent association between depressive symptoms and temperament specifically characterised by intense, frequent or excessive negative reactions to various stimuli and situations (Doucherty, Klein, Durbin, Hayden, & Olino, 2010). This temperament style is variously called negative reactivity (McClowry, Hegvik, & Teglasi, 1993), reactivity (Keogh, Pullis, & Cadwell, 1982) and negative emotionality (Presley & Martin, 1994), amongst other names (McClowry, 1995) and, despite variation in measurement and name, is arguably representative of the same underlying construct (McClowry, 1995).

While numerous theories posit explanations for how temperament—considered innate (Goldsmith et al., 1987)—contributes to depressive symptomology and disorders (Rothbart & Bates, 1998), increasingly, evidence supports models that suggest the environment interacts with temperament to contribute to psychopathology (Zentner & Bates, 2008). Thomas and Chess’ goodness‐of‐fit model is one such example in which risk of maladjustment is said to increase when a mismatch exists between a child's temperament and environmental demands (Thomas, Chess, & Birch, 1968). In relation to parent–child relationships, it is considered goodness‐of‐fit when a child's temperament can thrive in the presence of the parent's demands and behaviour, and as a result, the child's development progresses favourably.

Arguably, the most successful conceptualisation of goodness‐of‐fit in research is behaviour matching, where child temperament and parenting practices are compared for level of fit (Seifer, 2000). Predominantly, literature in this field has used statistical interactions of child temperament dimensions and parenting practices to explain unique variance in child internalising problems in addition to their respective linear effects (Seifer, 2000). This approach allows exploration of how a particular parenting practice may be especially beneficial or detrimental for children of a certain temperament (Seifer, 2000). There is some evidence that the level of fit of temperament traits and parenting practices varies based on the gender of the child (Oldehinkel, Veenstra, Ormel, De Winter, & Verhulst, 2006), but this is not well established.

Parental warmth is widely accepted to contribute to the prevention of depressive symptoms throughout development for males and females (Yap, Pilkington, Ryan, & Jorm, 2014) and has shown some efficacy in moderating risks between temperament and depressive symptoms within the goodness‐of‐fit framework (Gilliom & Shaw, 2004; Kiff, Lengua, & Bush, 2011). For example, a cross‐sectional study of 2230, 10–12‐year‐old children found a range of interaction effects supporting the goodness‐of‐fit model, including one indicating low parental warmth increased the effect of temperament (specifically, the dimension of frustration) on depression (Oldehinkel et al., 2006). Warm parents reflect tendencies to be supportive, affectionate, and sensitive to the child's needs (Zhou et al., 2002), which may soothe a negatively oriented child and prevent negative responses from cascading to ongoing depressive symptoms.

In an earlier exploratory study of participants of the Australian Temperament Project (ATP), from which data are drawn for the current analysis, negative reactivity and low parental warmth were shown to be significant linear predictors of concurrent depressive symptoms in 13–14‐year‐old children (Letcher et al., 2004). The interaction of negative reactivity and warmth also predicted additional variance after accounting for main effects (Letcher et al., 2004). Results showed that while both variables were associated with risk for depressive symptoms, the combined effect of high negative reactivity and low warmth was detrimental, with children in this group experiencing symptoms greater than what would be predicted if only the linear effects were considered. This interaction suggests that an environment with low warmth, characterised by having limited affection or responsiveness, may heighten the risk of concurrent depressive symptoms specifically for negatively reactive children. Given that parental warmth and temperament dimensions have shown longitudinal relationships with depression, the goodness‐of‐fit may also have a longitudinal effect. However, there has yet to be prospective research to examine whether this childhood‐moderating relationship influences depressive symptoms later in development, particularly into emerging adulthood where depressive disorders are more prevalent, and whether this interaction effect is different for males and females (Slade, Johnston, Oakley Browne, Andrews, & Whiteford, 2009).

The current study aimed to investigate the contribution of negative reactivity and low parental warmth in early adolescence to the development of depressive symptoms in emerging adulthood. In light of the absence of longitudinal research on the factors discussed, this study aimed to prospectively investigate for the first time whether negative reactivity and parental warmth interact, specifically whether parental warmth moderates the risk relationship between negative reactivity and subsequent depressive symptoms. Furthermore, on an exploratory basis, this study aimed to identify whether these effects differed for males and females. Our focus was on risk and protective factors in early adolescence at age 13–14 years and depressive symptoms in emerging adulthood at age 19–20 years, a developmental period characterised by increases in depressive symptoms in community samples (Rawana & Morgan, 2014).

It was hypothesised that negative reactivity in early adolescence would be positively associated with depressive symptoms in emerging adulthood and that parental warmth in early adolescence would be negatively associated with emerging adulthood depressive symptoms. Based on the goodness‐of‐fit model, it was hypothesised that parental warmth in early adolescence would moderate the relationship between early adolescent negative reactivity and emerging adulthood depressive symptoms, whereby the association between negative reactivity and emerging adulthood depressive symptoms is reduced in the presence of higher levels of parental warmth. These associations were hypothesised to occur after controlling for early adolescent depressive symptoms, socioeconomic status, and gender.

METHOD

Ethics clearance

Ethics clearance was provided for all waves of data collection by the University of Melbourne Human Research Ethics Committee and at the emerging adulthood wave also by the Human Research Ethics Committee of the Australian Institute of Family Studies.

Participants

Participants were 1,158 emerging adults (male = 511, female = 647) who, in 2002, at age 19–20 years, completed the wave 13 questionnaire of the ATP. The ATP, a representative community birth cohort study (N = 2,443), began in 1983 and has collected psychosocial information over 33 years (16 waves). There has been some selective loss of families with lower socioeconomic status (SES) and those not born in Australia; however, only negligible differences exist between original and retained participants on temperament and behavioural problems during infancy (see Table 1). For more information about recruitment and sampling, see Prior, Sanson, Smart, and Oberklaid (2000).

Table 1. Comparison of retained sample and original cohort on characteristics at recruitment in 1983

Domain	Original cohort	Retained sample 2002
SES Quartile in 1983
Highest	26.7%	32.3%
Medium–high	29.2%	30.4%
Medium–low	24.4%	23.1%
Lowest	19.8%	14.2%
Mother's country of birth
Australian	79.9%	83.6%
UK	6.0%	5.6%
Other	14.1%	10.8%
Father's country of birth
Australian	73.3%	77.0%
UK	7.3%	7.0%
Other	19.4%	16.0%
Infant behaviour problems Mean (SD)	1.73 (.69)	1.72 (.68)
Infant easy–difficult temperament factor Mean (SD)	2.46 (.63)	2.5 (.64)

The sample includes individuals from families of varying SES (see Table 1) and those who at 19–20 years were working (32.2%), studying (30.6%), both studying and working (32.8%), and not currently working or studying (4.4%). Predictor variables were measured at 13–14 years, while early adulthood depressive symptoms were measured at 19–20 years.

Measures

Negative reactivity

Parent‐reported negative reactivity was measured using the School‐Age Temperament Inventory (SATI; McClowry, 1995). The 12‐item scale has shown convergent validity with the same dimension of the Temperament Assessment Battery for Children‐Revised (TABC‐R) (McClowry, 1995). Parents responded to items such as ‘gets upset when he/she can't find something’ by indicating on a scale of 1 ‘never/almost never’ to 5 ‘always/almost always’ how often their child's behaviour is like the behaviour described in each item. A mean score was calculated, ranging from 1 to 5, where higher scores indicate more frequent and intense negative and reactive behaviours (α = .92).

Parental warmth

Self‐reported parental warmth was measured using six items from the ATP Parenting Practices Questionnaire. The warmth scale has shown adequate psychometric properties in various ATP studies, including longitudinal predictive validity for a range of internalising and externalising problems as well as adequate internal consistency (Letcher, Sanson, Smart, & Toumbourou, 2012; Prior et al., 2000; Vassallo, Smart, Sanson, & Dussuyer, 2004). Parents indicated responses on a scale of 1–5 to statements such as ‘I enjoy listening to, and doing things with, my child’. A mean score was derived where higher scores indicate greater warmth (α = .74).

Socioeconomic risk

Socioeconomic risk was derived from a composite of fathers’ and mothers’ occupational and educational levels in 1996. Possible scores ranged from 1 to 7.75, where higher scores indicate lower educational and occupational levels (Prior et al., 2000).

Early adolescent depressive symptoms

Adolescent self‐reported depressive symptoms were assessed using the Short Mood and Feelings Questionnaire (SMFQ) (Angold et al., 1995), a brief, adequately reliable, and valid measure in community samples of young adolescents (Rhew et al., 2010). Participants responded ‘rarely or never’, ‘sometimes’, or ‘very often’ to 12 questions such as ‘I feel miserable or unhappy’. A mean total score was calculated, ranging from 0 to 2, where higher scores indicate more depressive symptoms (α = .80)

Emerging adulthood depressive symptoms

Self‐reported emerging adulthood depressive symptoms were assessed using the Depression Scale of the Depression, Anxiety and Stress Scales (DASS; short form) at 19–20 years (Lovibond & Lovibond, 1995). The Depression scale of the DASS is an adequately reliable and valid measure of depressive symptoms in young adults (Osman et al., 2012). Participants were asked how much each statement applied to them over the past month, such as ‘I couldn't seem to experience any positive feeling’ (seven items). Responses ranged from 0 ‘did not apply’ to 3 ‘applied very much’. A mean score, ranging from 0 to 3, was calculated where higher scores indicate higher symtomatology (α = .89).

Statistical analysis

Eleven participants who did not provide sufficient responses to the outcome measure were excluded from analyses. Missing data on predictor variables (11%) for the remaining sample (male = 505, female = 642) were consistent with being missing completely at random (MCAR; Roderick, 1988) and were accounted for using Expectation Maximisation (EM) in SPSS (V.22). All inferential tests were conducted using bootstrapped standard errors (Guan, Yusoff, Zainal, & Yun, 2012).

Hierarchical multiple regression with bootstrapping was conducted to assess the influence of early adolescent negative reactivity and parental warmth on emerging adulthood depressive symptoms and to examine the interaction between negative reactivity and parental warmth. Continuous variables were centred before creating interaction terms. SES and gender were entered into the model at step 1 to control for basic demographic variance. To control for heterogeneity in early adolescent depressive symptoms, this variable was inserted in the model at step 2. The early adolescent negative reactivity and parental warmth variables, and their interaction, were inserted in step 3. Finally, we explored whether the relationship between the primary variables of interest (i.e., early adolescent depressive symptoms, warmth, negative reactivity, and warmth × negative reactivity interaction) and emerging adulthood depressive symptoms were moderated by gender. Specifically, we systematically examined whether the inclusion of each gender interaction term significantly increased the variance explained by the model.

RESULTS

Mean scores and standard deviations of the variables used in the analyses are presented in Table 2. Emerging adulthood depressive symptoms were comparable to non‐clinical normative data (Henry & Crawford, 2005), and using the defined cut‐off scores, approximately 20% of the sample were found to have at least moderate levels of depression (Lovibond & Lovibond, 1995).

Table 2. Mean scores and standard deviations

	Total		Male		Female
	M	SD	M	SD	M	SD
Emerging adulthood depressive symptoms	0.54	0.58	0.54	0.57	0.55	0.58
Early adolescent depressive symptoms	0.37	0.28	0.34	0.25	0.40	0.30
Negative reactivity	2.85	0.70	2.93	0.70	2.86	0.71
Parental warmth	4.21	0.56	4.16	0.58	4.24	0.54
Socioeconomic risk	3.96	1.34	3.91	1.39	4.00	1.30

Independent samples t‐tests indicated that female participants reported significantly higher scores for early adolescent depressive symptoms (M = 0.40, SD = 0.029) compared to males (M = 0.34, SD = 0.25) (t(1,145) = −3.793, p < .001, d = 1.24) and that females had significantly higher scores for parental warmth (M = 4.24, SD = 0.54) compared to males (M = 4.16, SD = 0.58) (t(1,145) = −2.34, p = .019, d = 1.25). Male and female participants did not significantly differ on mean scores for adulthood depressive symptoms, negative reactivity, or socioeconomic risk. Bivariate correlations for the combined sample and for each gender using Pearson's r are shown in Table 3.

Table 3. Bivariate correlations between variables in the model shown for combined sample and for each gender

	1	2	3	4	5
Combined (male and female)
1 Emerging adulthood depressive symptoms	1	.30^**	.20^**	–.12^**	.02
2 Early adolescent depressive symptoms		1	.30^**	–.24^**	.13^**
3 Negative reactivity			1	–.33^**	.10^**
4 Parental warmth				1	–.02
5 Socioeconomic risk					1
Males
1 Emerging adulthood depressive symptoms	1	.32^**	.15^**	–.01	–.04
2 Early adolescent depressive symptoms		1	.33^*	–.20^**	.11^*
3 Negative reactivity			1	–.30^**	.16^**
4 Parental warmth				1	–.05
5 Socioeconomic risk					1
Females
1 Emerging adulthood depressive symptoms	1	.29^**	.24^**	–.20^**	.06
2 Early adolescent depressive symptoms		1	.29^**	.–29^**	.14^**
3 Negative reactivity			1	–.37^**	.05
4 Parental warmth				1	.01
5 Socioeconomic risk					1

** Correlation is significant at the 0.01 level (two‐tailed),

* Correlation is significant at the 0.05 level (two‐tailed).

Table 4 presents the results of the hierarchical regression. At step 1, the model did not explain a significant amount of variance in emerging adult depressive symptoms. The inclusion of early adolescent depressive symptoms in step 2 explained a significant additional 9.1% of variance in emerging adulthood depressive symptoms. The step 3 model explained a significant additional 1.5% of variance in emerging adulthood depressive symptoms. When examining gender interactions, only the two‐way warmth X gender and the three‐way warmth X negative reactivity X gender interactions were associated with a significant increase in model R ² when entered into the model separately (0.6% and 0.3%, respectively). However, upon further examination (see Step 4(b) → Step 5(a) in Table 4), it was found that the three‐way warmth × negative reactivity × gender interaction did not explain additional significant variance in the model that included the warmth × gender interaction. Consequently, only the warmth × gender interaction term was retained for the final model.

Table 4. Changes in R ² for each step of hierarchical regression analysis

Step	Variables added in step	R	R ^2 (adjusted R ^2)	R ^2 change	F change	p value
Step 1	SES
	Gender	.018	.000 (−.001)	.000	.194	.824
Step 2	T1 Dep	.302	.091 (.089)	.091	113.96	<.001^***
Step 3	Warmth
	NR
	Warmth × NR	.325	.106 (.101)	.015	6.25	<.001^***
Examining gender interactions
Step 4(a)	T1 Dep × gender	.327	.107 (.101)	.001	1.37	.241
Step 4(b)	Warmth × gender	.334	.111 (.106)	.006	7.32	.007^**
Step 4(c)	NR × gender	.328	.107 (.102)	.002	2.18	.140
Step 4(d)	Warmth × NR × gender^3	.330	.109 (.103)	.003	4.04	.045^*
Step 4(e)	Warmth × gender
	Warmth × NR × gender^3	.336	.113 (.106)	.007	4.52	.011^*
Step 4(b) → Step 5(a)	Warmth × NR × gender^3	.336	.113 (.106)	.001	1.71	.191
Step 4(d) → Step 5(b)	Warmth × gender	.336	.113 (.106)	.004	4.98	.026^*

T1 Dep = early adolescent depressive symptoms; NR = negative reactivity; warm = parental warmth; SES = socioeconomic status.

Three‐way interaction.

* Significant F change at the 0.05 level (two‐tailed),

** Significant F change at the 0.01 level (two‐tailed),

*** Significant F change at the 0.001 level (two‐tailed).

Table 5 presents the results of the final model, which included all primary effects of interest and the gender × warmth interaction. This model was found to explain a total 11.1% of variance in emerging adult depressive symptoms. Early adolescent depressive symptoms and negative reactivity were found to be significant predictors of emerging adult depressive symptoms. Additionally, there was a significant parental warmth × gender interaction. A post hoc simple slopes analysis (Figure 1) found that the significant gender × parental warmth interaction was driven by a significant negative relationship between early adulthood depression and parental warmth for females (p = .024). The simple slope for males was not significantly different from zero (p = .174).

Table 5. Final step of hierarchical regression analysis predicting emerging adulthood depressive symptoms

			Bootstrapped (1,000 draws)
	b	SE	p value	CI (95%)	β
Step 4(b)
SES	− .011	.012	.360	− .036, .012	−.026
Gender	−.025	.033	.465	−.089, .040	−.022
T1 Dep	.543	.062	.001^**	.402, .698	.263
NR	.096	.025	.001^**	.048, .145	.117
Warm	.059	.043	.173	−.027, .151	.057
NR × warm	.038	.038	.285	−.036, .108	.030
Warm × gender	−.157	.058	.010^*	−.281, −.035	−.110

Note. T1 Dep = early adolescent depressive symptoms; NR = negative reactivity; warm = parental warmth; SES = socioeconomic status.

* Significant at the 0.05 level (two‐tailed);

** Significant at the 0.01 level (two‐tailed).

short-legendFigure 1.

The supplementary analysis in which the three‐way interaction between negative reactivity, parental warmth, and gender was entered into a fifth step found that the three‐way interaction term was not significant (b = .099, 95% bootstrapped confidence interval (−.23, .05)).

DISCUSSION

This study investigated the combined influence of early adolescent negative reactivity, parental warmth, and their interaction on the development of depressive symptoms in emerging adulthood. Hypothesis 1, that negative reactivity would be positively associated with emerging adulthood depressive symptoms after controlling for early adolescent depressive symptoms, was supported. Hypothesis 2, that parental warmth would be negatively associated with emerging adulthood depressive symptoms, was partially supported via a significant gender interaction showing a significant negative relationship for females but no significant effect for males. Finally, Hypothesis 3, that parental warmth would moderate the relationship between negative reactivity and emerging adulthood depressive symptoms after controlling for early adolescent depressive symptoms, was not supported.

We found that negative reactivity was positively associated with emerging adulthood depressive symptoms after accounting for early adolescent depressive symptoms and the other variables in the model. This effect was not moderated by gender despite a previous study of 302 adolescents finding that negative emotionality at age 15 predicted depressive symptoms at age 20 for girls but not boys (Katainen et al., 1999). In contrast to that study, the current study controlled for early adolescent depressive symptoms.

The hypothesis that parental warmth would be negatively associated with emerging adulthood depressive symptoms in the model after accounting for early adolescent depressive symptoms was supported for females, but there was no evidence of an effect for males. The findings suggest that for females, parental warmth in early adolescence is important for preventing not only concurrent depressive symptoms but also depressive symptoms later in life. These results are consistent with a previous cross‐sectional analysis of this cohort that found negative associations between parental warmth at 13–14 years and concurrent depressive symptoms for girls but not boys (Letcher et al., 2004). ATP surveys are primarily completed by mothers (>85%), and this suggests warmth from the same‐gender parent may be particularly important. Other research has found warmth provided by fathers to be one of the strongest protective factors against depressive symptoms for adolescent boys (Smojveri‐Ažič & Bezinovič, 2011). Given that these developmental periods are marked by transitions in social identity (Tanti, Stukas, Halloran, & Foddy, 2011) and pubertal processes, it is plausible that warmth from a parent who has experienced similar physiological and social changes is reassuring during this developmental stage.

The hypothesis that parental warmth would moderate the relationship between negative reactivity and emerging adulthood depressive symptoms, after accounting for early adolescent depressive symptoms, was not supported. This was despite a previously reported finding from a cross‐sectional analysis in this cohort that parental warmth did moderate the association between reactivity and depressive symptoms at 13–14 years (Letcher et al., 2004). A range of studies have found longitudinal associations between different behaviour matching conceptualisations of goodness‐of‐fit and internalising problems in children and adolescents, but few have examined these associations across a developmental stage while controlling for initial depressive symptoms (Seifer, 2000). Given that depressive symptoms are a risk factor for later depressive symptoms (Wickrama, Conger, Lorenz, & Martin, 2012), the concurrent risk of depressive symptoms associated with goodness‐of‐fit may cascade, and the initial risk of depressive symptoms may present ongoing risk for subsequent depressive symptoms.

However, the findings of the current study suggest that the direct influence of goodness‐of‐fit does not contribute beyond the concurrent risk in relation to the child and parent dimensions examined. These findings suggest that it may be more important to consider the dynamic nature of goodness‐of‐fit (Seifer, 2000). Goodness‐of‐fit is not necessarily constant throughout the relationship, and the ongoing development may be more important than goodness‐of‐fit at a particular moment in time (Seifer, 2000). Parents can change their parenting practices throughout their child's development, often in response to their child's temperament (Lipscomb et al., 2011; Moilanen, Rasmussen, & Padilla‐Walker, 2014), and these changes are often accompanied by changes in offspring depressive symptoms (Watson et al., 2014). Changes in parenting practices or the expression of temperament would cause fundamental changes to goodness‐of‐fit (Seifer, 2000), potentially superseding previous degrees of goodness‐of‐fit. In order to develop greater understanding of features of parent–child relationships and risks for depressive symptoms, there is a need to consider the flexibility of these relationships and explore changes in the relationship over time. Additionally, emerging adulthood is a time characterised by increased independence from parents, and the quality of other social relationships may become of greater importance for preventing depression (Lisznyai, Vida, Németh, & Benczúr, 2014).

Strengths and limitations

In this study, parental warmth and negative reactivity were moderately negatively correlated. Directionality was not possible to ascertain, but transactional models suggest that consideration should be given to a bidirectional nature of the association (Sameroff & Mackenzie, 2003). A child low on negative reactivity, who is easy to soothe and be around, may elicit warm parenting, while the irritable, negative reactive child may elicit parental irritation and withdrawal of parental warmth (Putnam, Sanson, & Rothbart, 2002). Alternately, while temperament is considered somewhat stable over time, there is room for change in its expression (Bould, Joinson, Sterne, & Araya, 2013), which could be influenced by parenting practices. Furthermore, it is important to consider that since parent report was used to measure negative reactivity and warmth, there is potential for non‐independence of measures, whereby characteristics of the parent may have influenced reports of both parental warmth and negative reactivity. Further studies might reduce biases by engaging multiple respondents to tease out the bidirectional nature of temperament and parenting practices (Sameroff & Mackenzie, 2003).

Strengths of the current study include its use of multi‐informant data from a large‐scale community study, which included the assessment of depressive symptoms at two time points, 6 years apart, allowing the prediction of emerging adulthood depressive symptoms after controlling for early adolescent depressive symptoms. This prospective design allows for a basic understanding of risks for depressive symptoms over a developmental time span. Nevertheless, certain aspects of the present study limit the conclusions that may be drawn. The current study only measured parental warmth of one parent and was unable to measure all sources of parental warmth a child receives. In two‐parent families, both parents contribute to protective factors for adolescent depressive symptoms (Rueger, Chen, Jenkins, & Choe, 2014). Furthermore, evidence shows that the adverse effects of negative parenting can be considerably modified by positive parenting from the other parent (Rutter, 1981). This is particularly important to consider when examining difficult temperament, which is associated with tendencies for parents to have low co‐parenting and different beliefs about how best to raise the child (Laxman et al., 2013).

It must be acknowledged that the associations found by the current study were modest; however, they are comparable to previous research on parent–child interactions. While parental warmth and negative reactivity are important, they are but two of a range of early adolescent factors that contribute to vulnerability for adulthood depressive symptoms, including other parenting and temperament characteristics (Saluja et al., 2004). In addition, the current study only examined one conceptualisation of goodness‐of‐fit, and others may prove additionally effective at longitudinally predicting depressive symptoms (Seifer, 2000). Some success has been noted predicting childhood behavioural problems where temperament and expectations of their parents and culture are compared for level of fit instead of being compared with parenting styles (Seifer, 2000). Also of importance, further research that captures how changes in goodness‐of‐fit are associated with changes in depressive symptoms would allow a more complex understanding of the role of goodness‐of‐fit.

Conclusions and clinical implications

The findings reinforce the ongoing risk of early adolescent negative reactive temperament on depressive symptoms in emerging adulthood, suggesting a need for targeted support in early adolescence for those with negative reactive temperament. While temperament‐ and goodness‐of‐fit‐based interventions have been developed, they are limited in number and lack strong empirical support in the form of rigorous randomised controlled trials (Sanson, Hemphill, Yagmurlu, & McClowry, 2010). In addition, practitioners working with children with this temperament style should be mindful of the continuing risk for depressive symptoms, and repeated assessment throughout development may be appropriate.

Also important for practitioners to note are the lasting effects for girls of low parental warmth in adolescence on subsequent depressive symptoms in emerging adulthood. Given the previously identified concurrent risk in early adolescence (Letcher et al., 2004), attention is warranted to the possibility of compounding effects of a low‐warmth parental environment for children with negative reactive temperament. The potential for dynamic change in the goodness‐of‐fit between parenting practices and child temperament requires further investigation with specific consideration for sensitive developmental periods when intervention might affect optimum outcomes. Such research is particularly important given evidence of sustained risk of depressive symptoms associated with negative reactivity.

ACKNOWLEDGEMENTS

The ATP study is located at The Royal Children's Hospital Melbourne and is a collaboration between Deakin University, The University of Melbourne, The Australian Institute of Family Studies, The University of New South Wales, The University of Otago (NZ), and the Royal Children's Hospital; further information is available at http://www.aifs.gov.au/atp. The views expressed in this paper are those of the authors and may not reflect those of their organisational affiliations nor of other collaborating individuals or organisations. We acknowledge all collaborators who have contributed to the Australian Temperament Project, especially Professors Ann Sanson, Margot Prior, Frank Oberklaid, and Ms Diana Smart. We would also like to sincerely thank the participating families for their time and invaluable contribution to the study.