The three-year trajectory of students’ school adaptation in secondary school and its longitudinal associations with trust, prospects, and positivity towards stress

ABSTRACT

This three-year longitudinal study attempted to clarify the overall trajectory of secondary school students’ school adaptation, its individual patterns, and longitudinal relations with three internal protective factors: interpersonal trust, prospects, and positivity towards stress. An anonymous survey on school adaptation, recent stressor severity, and the three internal factors was conducted every semester for three years. Data from 147 students (78 boys, 68 girls, and 1 unknown) were analysed. The results showed a tendency for school adaptation to decline around the second year, which was the same as in previous research; however, there were eight patterns, including stable or increased trajectories. Moreover, a positive longitudinal relationship was confirmed between school adaptation and interpersonal trust. The results highlighted the importance of timely preventive interventions creating a trustful environment. Also, based on the development of interpersonal trust, a possible mechanism of the decline in school adaptation and reinterpretation of the phenomenon was suggested.

KEYWORDS:

• Longitudinal research
• school adaptation
• secondary school
• trust
• Latent profile analysis (LPA)
• Autoregressive latent trajectory models (ALTMs)

Introduction

In September 2015, the United Nations agreed on 17 Sustainable Development Goals, and promotion of mental health and well-being was included in the third goal ‘Good health and wellbeing’ (Votruba & Thornicroft, 2016). The mental health of adolescents has been of particular concern because half of all mental illnesses’ onset occurs during adolescence (Kessler et al., 2005). Despite this, there are still one in seven teenagers worldwide experiencing mental illness (World Health Organization [WHO], 2021). School is often the first place where the symptoms of adolescents’ mental health issues are recognized, in addition to family (Froeschle & Moyer, 2004; Ryan et al., 2015). Furthermore, schools can provide mental health services in less stigmatized environments (Anglin et al., 1996). Considering these advantages, sustaining a specific level of school adaptation plays a significant role in protecting adolescents’ mental health.

School adaptation and associated factors

School adaptation refers to the degree to which students feel comfortable, committed to, and accepted by their school environment (Tomás et al., 2020). It plays a vital role in the lives of students because it affects their well-being, not only during school age (Tomás et al., 2020) but also in adulthood (McCarty et al., 2008; Russell et al., 2014). Support for maintaining school adaptation may be important for adolescents, especially for those in secondary school. It has been found that individuals aged 14 and 15 years (i.e. around 8^th grade) tend to show a decline in various mental health measures, including self-esteem (Kato et al., 2018), academic motivation (B. Kim et al., 2014), and well-being (Andreou et al., 2020). There are several reports that school adaptation also declines during this period and increases afterwards (Bae, 2022; Yamamoto & Wapner, 1991).

Researchers have identified various factors that affect students’ school adaptation. It is a robust finding that healthy social relationships, especially with peers (Hatzichristou & Hopf, 1996; Zhang et al., 2021), teachers (Roeser et al., 1996), and family members (Shek et al., 2001) induce better school adaptation (Wentzel, 1998). However, because social relationships depend on the environment and secondary school students have a limited range of environmental choices and changes, these students cannot control the factors through their own efforts in many cases (Dzurilla et al., 1998). Thus, instead of environmental factors, this study focused on students’ internal factors that were positively correlated with school adaptation. As several internal factors, such as academic motivation (Mizuno, 2016; Xie et al., 2022) and social skills (Mizuno, 2016; Punia & Sangwan, 2011), are already well-considered, this study examines three recently discovered internal factors: interpersonal trust, prospects, and positivity towards stress (Amai, 2020).

Three Internal Protective Factors for School Adaptation

Interpersonal trust has been defined in various ways depending on the research (Mayer et al., 1995). However, a common characteristic among the various definitions of trust is the risk assumption (Johnson-George & Swap, 1982). For example, in one study involving interviews with students who conceal their problems, the interviewees said that trusting someone and disclosing their personal problems included the risks of leaking information and eliciting unpleasant reactions (Amai & Emi, 2022). The level of trust may be related to the level of invasiveness assumed by a person. Thus, this study regards interpersonal trust as the belief that others are sincere and non-invasive to those who trust. Although basic trust advocated by Erikson (1959) is a developmental task that one needs to acquire during infancy, interpersonal trust is developed through life-long experiences (Tani, 1998). This study focuses not on unchangeable traits acquired in infancy but on changeable traits. Therefore, in this research, interpersonal trust was defined as a belief in others’ sincerity and non-invasiveness, which is developmentally acquired through experiences in real human relationships.

As for prospects, this study follows Amai’s (2020) definition of prospects as the level of expectation for an individual’s potential in the future. According to the Socioemotional Selectivity Theory (Carstensen et al., 1999), the perception of the future enhances people’s activeness in learning and social goals. In the Time Perspective Theory (Zimbardo & Boyd, 2010), a higher future time perspective reflects a person’s tendency to behave in a manner that is oriented towards future goals. Both theories insist that human behaviour varies depending on expectations for the future. This implies that students who expect a better future may be more patient in difficult situations and more active in learning than students who have low expectations for their future; this inference matches the present study’s definition.

Positivity towards stress is defined as the degree to which an individual views a stressful situation positively. Although general positivity is the tendency to view life and experiences with a positive outlook like optimism (Caprara et al., 2012), the exceeded emphasis on positivity has been critiqued, and a second wave of positive psychology has emerged (Lomas & Ivtzan, 2016). It involves a dynamic interplay of the negative and positive aspects of life, such as distress and experiences to overcome the challenges. Multiple contextual factors cause stress in adolescents, including family, school, and peers. However, most environmental factors are not controllable for young students (Parikh et al., 2019). Since the recent wave of positive psychology and the inevitability of stress for teenagers, this study particularly focuses on positive thinking in stressful situations.

These internal factors are not only associated with school adaptation but also with well-being (Catalino et al., 2014; Lauriola & Iani, 2016; Zimbardo & Boyd, 2010). Research on the association of interpersonal trust, prospects, and positivity towards stress with mental health has become increasingly active in recent years, and the importance of these factors has been recognized regardless of country or generation (Al-Ajlani et al., 2019; Lukács, 2021; Nizeyumukiza et al., 2020). However, there are still few studies targeting adolescents, and much of the existing research is cross-sectional (Clarke et al., 2020). Thus, the longitudinal relationship between these three factors and secondary school students’ mental health has not yet been examined. Moreover, longitudinal studies on school adaptation are scarce (Onda et al., 2017), and neither the tendencies nor individual patterns of change over three years in secondary school are well known.

The current study

This was a three-year longitudinal study from the first semester of secondary school to graduation. It aimed to clarify the longitudinal association between secondary school students’ school adaptation and the three internal factors described above: trust, prospects, and positivity towards stress. Before exploring these factors, for an integrated understanding of school adaptation, this study investigated the overall three-year trajectory of school adaptation and its patterns of change. Although this research is exploratory owing to a lack of sufficient previous research, it was conducted based on three hypotheses.

Hypothesis 1.

School adaptation declines during the second year of secondary school.

Hypothesis 2.

Several patterns of school adaptation changes, such as rising, declining, and stable trajectories, are observed.

Hypothesis 3.

Interpersonal trust, prospects, and positivity towards stress demonstrate positive longitudinal associations with school adaptation.

Methods

Design and participants

This prospective, longitudinal study was conducted from the beginning of the 2018 academic year to the end of the 2020 academic year. Data collection began when 147 students (aged 12–13 years; 78 boys, 68 girls, and 1 unknown) entered a public secondary school in northern Japan¹; it ended when they graduated. An anonymous questionnaire² was administered every semester, thrice per year at four-month intervals, except for the semester of early 2020 when the school was closed due to the COVID-19 pandemic³. Participants reported their school adaptation, trust, prospects, positivity towards stress, and stressor events they had experienced in the past month. Only those who reported having experienced stressful events in the past month reported the severity of the stressor events. The response rate ranged from 92.5% to 95.9% at each time point (Table 1). The final number of samples was 145 due to two students’ school transfers.

Table 1. The numbers of participants, mean scores, and standard deviations (SD) at each time point.

	T1	T2	T3	T4	T5	T6	T7	T8
n =	141	136	138	139	138	141	138	138
School adaptation	3.42	3.44	3.37	3.4	3.32	3.28	3.24	3.18
	(.53)	(.55)	(.61)	(.71)	(.69)	(.70)	(.73)	(.82)
Trust	3.07	3.09	3.08	3.16	3.15	3.14	3.17	3.15
	(.61)	(.71)	(.72)	(.70)	(.70)	(.70)	(.72)	(.72)
Prospects	3.16	3.20	3.25	3.31	3.30	3.35	3.39	3.42
	(.81)	(.85)	(.79)	(.81)	(.78)	(.78)	(.83)	(.74)
Positivity	2.98	2.96	3.16	3.06	3.02	3.05	3.01	3.07
	(.72)	(.74)	(.69)	(.85)	(.77)	(.79)	(.81)	(.80)

All students and their parents were informed in writing about the purpose, content, and privacy policy of the survey beforehand, and participation was voluntary. Written informed consent was obtained from school principals, and all participants were included in the study. No incentives were provided to participants. The university research ethics committee approved all methods and materials used in this study.

Measures

The questionnaire used in this study consisted of the following measuring scales described below, with verified reliability and validity. The reported Cronbach’s alphas belonged to the current sample. The principal and teachers at the participating schools previewed all items beforehand and confirmed the validity for secondary school students.

School adaptation

We extracted five items of the School Adaptation Scale (Furuichi & Tamaki, 1994; α = .86–.94), a widely used 10-item scale in Japan, for measuring enjoyment of school life. The five items used in the current research were as follows: (a) I look forward to going to school, (b) every day passes quickly because school is fun, (c) I want to go to school even if I feel a little bad, (d) there are many things to enjoy at school, and (e) I like this school. These items were selected because they showed high factor loading in the preliminary surveys. Participants used a 4-point scale ranging from 1 (never true for me) to 4 (always true for me). The three internal variables – trust, prospects, and positivity towards stress – were also measured using the same 4-point Likert scale.

Trust

Interpersonal trust was measured by three items of the Sense of Basic Trust Scale (Tani, 1998; α = .65–.80) that was developed based on Rasmussen’s (1964) Ego Identity Scale. The items were ‘I can expect help from people around me when I am in trouble,’ ‘commonly, people relate to one another sincerely,’ and ‘I think people are credible in general.’

Prospects

Three items were derived from the Japanese version of the Future Time Perspective Scale (Ikeuchi & Osada, 2013; α = .65–.71), which focuses on opportunities expected in the future. The items were ‘many opportunities await me in the future,’ ‘in the future, I will be able to do whatever I want,’ and ‘my future is filled with possibilities.’

Positivity towards stress

The three-item stress-is-enhancing subscale of the Stress Mindset Measure (youth version; Park et al., 2018; α = .60–.78) was used for measuring positivity towards stress. The items were ‘experiencing stress facilitates my learning and growth,’ ‘I try to find the positive side of things even when things get tough,’ and ‘the effect of stress is positive and should be utilized.’

Severity of stressor events

This study used three items of the Cognitive Appraisal Questionnaire (Okayasu, 1992; e.g. ‘it threatens my daily life’; α = .69–.84), a widely used five-item measure in Japan, to measure the severity of stressor events that the students experienced in the past month. Participants evaluated their stressor events at each time point by rating them on a 4-point scale ranging from 1 (never felt so) to 4 (always felt so).

Data analysis

The analysis consisted of three phases. First, latent growth curve models (LGCMs) were used to assess the change trajectory of the three-year school adaptation (Analysis 1). Second, Latent Profile Analysis (LPA) was employed to obtain typologies of school adaptation changes (Analysis 2). Third, Autoregressive Latent Trajectory Models (ALTMs) using the three internal protective factors, respectively, were estimated and compared to determine the model that provided the most precise and parsimonious representation of longitudinal relationships with school adaptation (Analysis 3). In LGCMs and ALTMs, the model selection was mainly based on the Chi-square test. Other fit measures, including the root mean square error of approximation (RMSEA), were also referred to because the determination of practical significance depends on the context (Preacher et al., 2008). In LPA, the bootstrap likelihood ratio test was mainly used for the determination of the number of classes: The Bayesian information criterion (BIC) and sample size-adjusted BIC (adBIC) were supplementarily used. Other fit measures were referred to, but the tests and fit measures mentioned above were considered more reliable in latent mixture modelling (Jenn-Yun et al., 2013; Nylund et al., 2007). The, LGCMs and ALTMs were performed using the RAMpath and Lavaan packages of R. The LPA was performed using Mplus software (version 8.8). The analysis codes are available for download from the Open Science Framework Directory (https://osf.io/se8pn/?view_only=0bb3358aa13e440d8ca9e7320b577fed).

Results

Of the entire sample, the data from 111 students (55 boys and 56 girls) who provided complete sets of data from T1 to T8 were used for the LGCMs and ALTMs. The LPA used the data of 145 students (77 boys, 67 girls, and 1 unknown) because the analysis method could estimate and complement missing data with the maximum likelihood method using all available data. Table 1 presents descriptive statistics for each time point. Figure 1 illustrates the change in the mean scores for school adaptation and the measured protective factors. All repeated measures were positively and significantly correlated across time (school adaptation, r = .63–.82; trust, r = .48–.79; prospects, r = .66–.78; positivity, r = .61–.68; severity of stressor events, r = .37–.65).

Figure 1. Mean scores regarding (a) school adaptation, (b) trust, (c) prospects, and (d) positivity towards stress at each time point.

The range of possible values is from 1.0 to 4.0.

Analysis 1: the three-year trajectory of school adaptation

Unconditional LGCMs compared four types of change in school adaptation: (a) no-growth model, (b) linear change model, (c) quadratic change model, and (d) latent basis model; the quadratic model had the best fitting (Table 2;χ² = 51.657, df = 34, p = .027, CFI = .975, RMSEA = .068, SRMR = .057). In this quadratic model, the factor loadings were set to 0, 1, 4, and 9 (0², 1², 2², and 3²) on a quadratic slope factor by default. Therefore, the model must be improved by aligning the data. In detail, the model with its division point of the slope set at T4 showed the best fitting (χ2 = 45.938, df = 34, p = .083, CFI = .983, RMSEA = .056, SRMR = .046). In this model, the factor loading of the slope in the first half was fixed at the same value after T4, and the factor loading of the slope in the second half was fixed at zero until T4.

Table 2. Model fitting scores of LGCMs by each model.

	No	Linear	Quadratic Default	Latent	Quadratic Divide at T3	Quadratic Divide at T4		Quadratic Divide at T5
P-value (Chi-square)	.000	.000	.027	.000	.003	.083	.000
CFI	.756	.933	.975	.907	.961	.983	.983
AIC	1189.448	1067.390	1041.208	1161.450	1051.245	1035.489	1035.784
BIC	1197.576	1083.647	1068.304	1191.255	1078.341	1062.584	1062.879
adBIC	1188.096	1064.686	1036.702	1156.493	1046.739	1030.982	1031.278
RMSEA	.195	.106	.068	.193	.086	.056	.057
SRMR	.223	.109	.057	.227	.066	.046	.056

CFI = comparative fit index; AIC = Akaike information criterion; BIC = the Bayesian information criterion; adBIC = sample-size adjusted BIC; RMSEA = the root mean square error of approximation; SRMR = standardized root mean square residual; No = No linear model, Linear = Linear model, Quadratic = Quadratic model, Latent = Latent model; The accepted model in Analysis 1 is in bold.

The intercept was 3.43 (SE = .044, p = .000). It showed statistically significant covariances with the slope (S = .017, p = .035) but not with the quadratic slope (S = −.012, p = .071). The estimated average latent changed in the first half (i.e. slope) was −.018 (SE = .016, p = .287), indicating that there was no statistically significant change during the first year in secondary school; and the estimated average latent changed in the last half (i.e. quadratic slope) was −.038 (SE = .014, p = .008), indicating a slight decline of school adaptation after the first semester of the second year. All variances of the intercept (s² = .143, p = .000), slope (s² = .014, p = .001) and quadratic slope (s² = .014, p = .000) were statistically significant. Altogether, the mean score of secondary school students’ school adaptation tended not to change significantly during 7^th grade and gradually declined after they reached 8^th grade. The trajectory of school adaptation has individual patterns, although these are not explained by the LGCM.

Analysis 2: the patterns of school-adaptation change

When LPA was performed with models assuming two to nine latent classes, the analyses converged in all cases⁴. Although the sample size of this study was not enriched, factors such as a higher quality of indicators and a larger class separation might have led to model convergence and stability (Tein et al., 2013; Wurpts & Geiser, 2014). The results suggested an eight-class model illustrating individual differences in the school adaptation trajectory (Table 3). The comparison of residual deviance differences between k-class and (k + 1)-class models using the likelihood ratio test (i.e. bootstrap test) showed p > 0.05 for the nine-class model. The BIC (1538.26, 1527.51, 1534.06, and 1549.08 from the six- to nine-class models, respectively) and adBIC (1345.21, 1306.00, 1284.07, and 1270.62) suggested seven or more classes. Considering these test results and fit measures, an eight-class model was adopted.

Table 3. Model fitting scores by each latent class model.

	2	3	4	5	6	7	8	9
AIC	1674.91	1507.52	1423.47	1392.96	1356.66	1319.14	1298.89	1287.13
BIC	1749.33	1608.73	1551.47	1547.75	1538.26	1527.51	1534.06	1549.08
adBIC	1670.22	1501.14	1415.40	1383.20	1345.21	1306.00	1284.07	1270.62
entropy	0.96	0.97	0.92	0.94	0.94	0.94	0.94	0.95
BLRT	0.00	0.00	0.00	0.00	0.00	0.00	0.00	0.06

AIC = Akaike information criterion; BIC = the Bayesian information criterion; adBIC = sample-size adjusted BIC, BLRT = bootstrap likelihood ratio test; The accepted model in Analysis 2 is in bold.

The number of participants in each latent class and mean scores for school adaptation are shown in Table 4 and Figure 2. Classes 1–4 started with relatively high school adaptation, Classes 5 and 6 started with middle adaptation scores, and Classes 7 and 8 started with relatively low school adaptation. Class 1 (labelled ‘High’) had participants who showed stable high school-adaptation from the beginning to the end of their secondary school life. The other three latent classes that started from relatively high school-adaptation were Class 2 (labelled ‘High to middle’; H to M), Class 3 (labelled ‘High to low’; H to L), and Class 4 (labelled ‘Plummets’), which showed a rapid decline of school adaptation and the lowest scores in the end. Other latent classes included Class 5 (labelled ‘Middle’) that stayed in the middle through the time points, Class 6 (labelled ‘Middle to High’; M to H), Class 7 (labelled ‘Low to middle’; L to M), and Class 8 (labelled ‘Low’).

Figure 2. Mean scores of school adaptation by latent class at each time point.

The vertical axis represents the mean scores, and the horizontal axis represents the time points. Class 1 = High, class 2 = High to Middle (H to M), class 3 = High to Low (H to L), class 4 = Plummets, class 5 = Middle, class 6 = Middle to High (M to H), class 7 = Low to Middle (L to M), and class 8 = Low; the range of possible values is from 1.0 to 4.0.

Table 4. The numbers of participants and mean scores of school adaptation in each latent class.

	class 1	class 2	class 3	class 4	class 5	class 6	class 7	class 8
	High	H->M	H->L	Plummets	Middle	M->H	L->M	Low
n =	47	15	14	5	31	17	9	7
T1	3.76	3.74	3.61	3.60	3.22	3.18	2.56	2.43
T2	3.86	3.77	3.70	3.80	3.07	3.22	2.67	2.50
T3	3.85	3.80	3.57	3.26	2.90	3.23	2.51	1.92
T4	3.94	3.95	3.55	2.35	2.95	3.68	2.07	2.03
T5	3.91	3.74	3.29	2.60	2.75	3.38	2.43	1.90
T6	3.92	3.55	3.16	2.07	2.81	3.45	2.78	1.57
T7	3.95	3.10	3.03	1.53	2.87	3.69	2.68	1.57
T8	3.95	3.34	2.35	1.10	2.79	3.62	2.81	1.63

Class 1 = High, class 2 = High to Middle (H to M), class 3 = High to Low (H to L), class 4 = Plummets, class 5 = Middle, class 6 = Middle to High (M to H), class 7 = Low to Middle (L to M), and class 8 = Low; The range of possible values is from 1.0 to 4.0.

The low-starting class (Class 7, ‘L to M’; & Class 8, ‘Low’) showed significantly lower levels of school adaptation compared to the other six classes at T1; however, at the end, Class 4 (‘Plummets’) and Class 8 (‘Low’) demonstrated lower levels than the other classes did. One of the high-starting classes, Class 3 (‘H to L’), showed the third-lowest school adaptation at the end, significantly lower than Class 1, ‘High’; Class 2 (‘H to M’); Class 5 (‘Middle’); and Class 6 (‘M to H’). Half of the sample belonged to Class 1, ‘High,’ or Class 5, ‘Middle,’ and illustrated their stable school adaptation. Of the other half, the participants who showed high school-adaptation at T1 mostly demonstrated a decline in school adaptation. In contrast, Class 6 (‘M to H’) and Class 7 (‘L to M’) raised their school adaptation, especially during the second year. Through Analyses 1 and 2, the trajectory and pattern of the three-year school adaptation in a secondary school were revealed. In the following analysis, we examined the association between changes in school adaptation and the internal protective factors.

Analysis 3: the association between school adaptation and internal protective factors

Regarding the association between school adaptation and trust, prospects, or positivity towards stress, nine possible models were considered in all the ALTM analyses. Based on the results of Analysis 1, all models adopted a quadratic curve model for school adaptation. Instead, the internal factors, autoregression, cross-lagged effects, and equality constraints varied depending on the model. In Model 1, the internal variables were linear with no regression and Model 2 was a quadratic curve with no regression. In Model 3 and later models, the internal variables were quadratic curve models, with autoregression in school adaptation (Model 3), internal variables (Model 4), and both (Model 5). Model 6 added cross-lag effects to Model 5. Later models also added equality constraints to the autoregressive coefficients (Model 7), cross-lag effects (Model 8), and both variables (Model 9). The fit measures for each model are listed in Table 5. When trust (χ² = 82.479, df = 73, p = .210, CFI = .994, RMSEA = .035, SRMR = .053) or prospects (χ² = 95.065, df = 73, p = .042, CFI = .986, RMSEA = .053, SRMR = .055) were examined, Model 6 (i.e. a model that includes autoregression and cross-lag effect with no equality constraint) showed the best fit. When positivity towards stress was examined, Model 5 (i.e. a model that includes autoregression but no cross-lag effect) showed the best fit (χ² = 110.153, df = 87, p = .048, CFI = .983, RMSEA = .049, SRMR = .050). To avoid excessive complexity, Table 6 explains the results of the growth-curve model portion and Figure 3 shows the cross-lagged model portion of the ALTM.

Figure 3. The part of cross-lagged model of ALTM with school adaptation and (a) trust, (b) prospects, (c) positivity through three years in secondary school.

*p < .05, **p < .01, ***p < .001; Illustrated only significant paths.

Table 5. The fit measures of each considered ALTM model.

	Model 1	Model 2	Model 3	Model 4	Model 5	Model 6	Model 7	Model 8	Model 9
Trust
P-value (Chi-square)	.000	.011	.026	.008	.030	.210	.013	.022	.012
CFI	.958	.976	.981	.975	.982	.994	.979	.981	.977
AIC	2151.920	2131.993	2132.322	2140.511	2137.441	2134.393	2144.412	2141.296	2135.094
BIC	2270.339	2269.252	2288.420	2296.609	2312.379	2347.009	2324.732	2321.616	2283.118
adBIC	2131.305	2108.098	2105.148	2113.336	2106.987	2097.380	2113.021	2109.905	2109.325
RMSEA	.073	.056	.053	.060	.053	.035	.058	.055	.057
SRMR	.078	.056	.055	.055	.054	.053	.051	.054	.056
Prospects
P-value (Chi-square)	.000	.003	.011	.002	.010	.042	.009	.008	.003
CFI	.945	.972	.978	.971	.978	.986	.978	.977	.972
AIC	2335.450	2301.397	2299.183	2309.361	2305.688	2308.041	2308.116	2308.520	2305.327
BIC	2453.869	2438.655	2455.282	2465.459	2480.625	2520.658	2488.437	2488.840	2453.351
adBIC	2314.835	2277.502	2272.009	2282.186	2275.234	2271.028	2276.725	2277.128	2279.558
RMSEA	.085	.063	.058	.066	.060	.053	.061	.061	.064
SRMR	.062	.050	.050	.050	.049	.055	.053	.049	.051
Positivity towards stress
P-value (Chi-square)	.000	.001	.016	.005	.048	.030	.007	.040	.000
CFI	.933	.962	.977	.972	.983	.982	.974	.982	.960
AIC	2443.466	2410.834	2397.375	2404.462	2395.940	2411.214	2409.917	2398.998	2417.478
BIC	2561.885	2548.093	2553.473	2560.561	2570.877	2623.831	2590.237	2579.318	2565.502
adBIC	2422.851	2386.940	2370.100	2377.288	2365.486	2374.201	2378.526	2367.607	2391.709
RMSEA	.088	.069	.056	.061	.049	.055	.062	.051	.072
SRMR	.061	.051	.053	.051	.050	.046	.050	.048	.051

adBIC = sample-size adjusted BIC; The accepted model in Analysis 2 is in bold.

Model 1 = Internal variable linear model with no regression.

Model 2 = Internal variable quadratic curve model with no regression.

Model 3 = Internal variable quadratic curve model with autoregression in school adaptation.

Model 4 = Internal variable quadratic curve model with autoregression in internal variables.

Model 5 = Internal variable quadratic curve model with autoregression in school adaptation and internal variables.

Model 6 = Internal variable quadratic curve model with autoregression and cross-lag effect in school adaptation and internal variables.

Model 7 = Internal variable quadratic curve model with autoregression and cross-lag effect in school adaptation and internal variables, added equality constraint on autoregressive coefficients.

Model 8 = Internal variable quadratic curve model with autoregression and cross-lag effect in school adaptation and internal variables, added equality constraint on cross-lag effects.

Model 9 = Internal variable quadratic curve model with autoregression and cross-lag effect in school adaptation and internal variables, added equality constraint on autoregression and cross-lag effect.

Table 6. The results of the growth-curve-model portion of ALTM.

	Trust	Prospects	Positivity
Intercepts
School adaptation
intercept (iX)	3.430***	3.430***	3.431***
slope (sX)	0.096	−0.211	−0.164
quadratic (qX)	−1.326***	−0.386	−0.321
Internal factor
intercept (iY)	3.072***	3.143***	3.015***
slope (sY)	0.092	−0.068	−0.057
quadratic (qY)	−1.170***	−0.454*	−0.035
Covariances
iX ~~ sX	0.018	0.003	0.006
iX ~~ qX	−0.070***	−0.027	−0.026*
iX ~~ iY	0.103**	0.153***	0.155***
iX ~~ sY	0.018	−0.002	−0.007
iX ~~ qY	−0.065***	−0.035**	−0.005
sX ~~qX	−0.021	−0.001	−0.002
sX ~~ iY	0.022	0.001	−0.006
sX ~~ sY	0.010	0.006	0.013**
sX ~~ qY	−0.018*	−0.005	−0.006
qX ~~ iY	−0.074***	−0.037*	−0.014
qX ~~ sY	−0.021*	−0.002	−0.011**
qX ~~ qY	0.039*	0.008	0.010**
iY ~~ sY	0.038*	−0.002	−0.052*
iY ~~ qY	−0.074*	−0.044*	−0.003
sY ~~ qY	−0.01	−0.008	−0.011*

*p < .05, **p < .01, ***p < .001.

School adaptation and trust

First, the suggested model contained the autoregressive effects of trust from T5 to T7 (i.e. from the second semester of 8th grade and a year later). During this term, increased (or decreased) trust over a given time point was significantly positively correlated with increased (or decreased) trust in the following semester (β = .34 at T5 and T6, p < .01; β = .72 at T6 and T7, p < .01). This implied that trust remained stable over time in the middle of secondary-school life. Furthermore, cross-lagged effects were found; for instance, decline of trust over a given semester was significantly and positively correlated with decline of school adaptation both over the same semester (time-concurrent effects: β = .07 at T5, β = .08 at T6, β = .13 at T7, β = .34 at T8,, p < .001 at all the time points) and in the following semester (cross-lagged effects from trust to school adaptation: β = .20 at T5 and T6, p < .05; β = .60 at T6 and T7, p < .001; β = .57 at T7 and T8, p < .001). Thus, school adaptation and trust did not influence each other in the first half of secondary school, but gradually influenced each other in the last half.

A notable feature of the growth-curve-model portion was the negative association between the intercept of trust and the quadratic slope of school adaptation, and the negative association between the intercept of school adaptation and the quadratic slope of trust; both covariances were approximately .07 and were significant at the 1% level. These associations indicated that higher trust (or school adaptation) at T1 might have slightly suppressed the decline in school adaptation (or trust) after T4.

School adaptation and prospects

The model revealed fragmented time-concurrent and cross-lagged effects between school adaptation and prospects. These two variables correlated at several time points (time-concurrent effects: β = .04 at T2, p < .05; β = .07 at T3, p < .01; β = .04 at T5, p < .05; β = .08 at T7, p < .01; β = .08 at T8, p < .01), but the correlations were weak. The model illustrated that there were cross-lagged effects between school adaptation and prospects only in the last half of secondary-school life, such as increased school adaptation at T5, T6, and T7 being significantly positively correlated with increased prospects in the following semester (cross-lagged effects from school adaptation to prospects: β = .32 at T5 and T6, p < .01; β = .40 at T6 and T7, p < .01; β = .36 at T7 and T8, p < .05). The cross-lagged effect from prospects to school adaptation (β = .23, p < .05) and autoregressive effect of prospects (β = .33, p < .01) were observed only between T7 and T8. Thus, school adaptation influenced prospects after the second semester in 8th grade, but changes in prospects rarely affected school adaptation.

For the growth-curve-model portion, we found significant covariances between the intercept of school adaptations and the quadratic slope of prospects, whereas the covariance in the intercept of prospects and the quadratic slope of school adaptation were not significant. This implies a weaker longitudinal effect of prospects on school adaptation than the effect of early trust on later school adaptation.

School adaptation and positivity towards stress

Contrary to the two models explained above, this model did not confirm any cross-lagged effects between school adaptation and positivity towards stress. We only observed time-concurrent effects at T3 (β = .05, p < .01) and T7 (β = .05, p < .05). In addition, positivity towards stress did not show an autoregressive effect over time. Therefore, an increase in positivity towards stress did not increase school adaptation – neither at the same time nor in the following semester – and vice versa. The growth curve model supports this hypothesis. No significant covariance was found between the intercept of positivity and any of the slopes of school adaptation. Although significant covariances were found between the slopes and quadratic slopes of the two variables, their coefficients were .01; thus, the effects were small.

Discussion

The main objective of this study was to clarify the longitudinal association between secondary-school students’ school adaptation and their trust, prospects, and positivity towards stress. First, the three-year trajectory of school adaptation was investigated, and several patterns of school adaptation change were detected. Furthermore, the longitudinal relations between school adaptation and trust, prospects, and positivity towards stress, respectively, were revealed.

The three-year trajectory of school adaptation in secondary school

The LGCM confirmed no significant change in school adaptation during the first year after entering secondary school, and school adaptation tended to decline gradually after the first semester of the second year, thereby supporting Hypothesis 1. The results aligned with previous research illustrating that individuals around 14 years of age were more likely to show a decline in mental health (e.g. Andreou et al., 2020; B. Kim et al., 2014). This can be attributed to two factors: developmental and school systems. Adolescence is a period in which many students experience an imbalance between strong bottom-up emotionally charged impulses and the still-developing brain function for regulating urges (Casey et al., 2011). This imbalance might be considerable around the age of 14. Additionally, 8th grade falls between the transition term to a secondary school (first year) and the high school exam period (third year). About 90% of secondary school students in Japan take part in club activities, and after the 9th graders retire in the first semester, the 8th graders are required to take a leadership role (Agency for Cultural Affairs, 2018; Japan Sports Agency, Government of Japan, 2017); thus, they experience more trouble in peer relationships. Accordingly, this study further strengthens the finding that individuals are more likely to experience mental health crises at the age of 14.

The patterns of school-adaptation changes

LPA identified eight patterns of change in school adaptation, thus supporting Hypothesis 2. The pattern of change was broadly divided into stable, rising, and declining. Whether it was rising or declining, significant changes were seen from the end of the first year to the second year of secondary school. Half of the students who showed higher school-adaptation scores initially showed a gradual decline or a sharp drop in adaptation thereafter. This finding indicates that students’ school adaptation is not always stable. Even if students look well adapted to school at the time of entering school, parents, teachers, and other supporters should not be complacent, but pay close attention to changes in their students. While several existing studies have targeted abuse victims (Paik et al., 2019) and rural minority youth (Estell et al., 2007), no study has so far longitudinally examined individual differences in school adaptation among secondary school students in a nonclinical sample. Further studies are required to confirm the generalizability of this study’s results.

The longitudinal association between school adaptation and internal factors

Hypothesis 3 was partially supported. Of the three internal protective factors examined in this study, interpersonal trust showed the clearest longitudinal relationship with school adaptation. The cross-lag effect between school adaptation and interpersonal trust stabilized in the latter half of secondary school. Considering that a higher level of interpersonal trust mitigates school adaptation even when the students’ personal worries are serious (Amai, 2020), stressors are less likely to have serious effects on those who have someone to trust. Given that a longitudinal study of immigrants also reported an association between trust and adaptation (Nickerson et al., 2019), the present study’s result confirms a positive longitudinal relationship between trust and adaptation.

However, prospects and positivity towards stress did not show the same level of longitudinal effect on school adaptation as interpersonal trust. One possible reason for this is the difference in the tendencies of the variables. Prospects are the idea that ‘the future will get better from now on’ and positivity towards stress is the idea that ‘the present hardship will make me grow up.’ These two variables share the common belief that they will see an upturn or growth in the future, even if the current situation is difficult. Meanwhile, the scale of school adaptation this study used examined ‘whether a student enjoys the school now.’ The difference in the tenses with which the variables deal may have made it difficult to observe a longitudinal relationship. Another possible factor is the stage of mental development. A concept similar to prospects – hope—has been reported to be longitudinally associated with well-being in college students (Rand et al., 2020). There is also a longitudinal study showing that people with high positivity towards stress are less likely to leave their jobs soon (J. Kim et al., 2020). Compared to college students and adults who have already overcome some major life events such as high school entrance exams and university entrance exams, secondary school students’ time perspectives are still developing (Loose & Vásquez-Echeverría, 2022). Such developmental differences may have emerged as variations in the long-term relationship between prospects and adaptation.

Nevertheless, both prospects and positivity towards stress showed significant covariance with school adaptation at multiple time points. A cross-sectional survey showed that higher prospects and positivity towards stress, as well as trust, were associated with higher school adaptation (Amai, 2020). This indicated a cross-cutting relationship between them. To deepen our understanding of the cross-sectional and longitudinal effects of interpersonal trust, prospects, and positivity towards stress, future research should clarify the factors of the frail longitudinal relationship between school adaptation and prospects or positivity and examine the relationship between the three internal variables and other psychological variables such as well-being.

Implications

The results of this study have three practical implications. The first concerns the timing of interventions to protect the mental health of children and youth. School adaptation tended to change from the end of the first year to the second year of secondary school and stabilized thereafter. Because the effectiveness of interventions, such as educational programs and counselling, declines when mental health deteriorates significantly (Ye et al., 2014), it is desirable to implement preventive interventions in advance rather than just dealing with the crisis. Therefore, preventive intervention until early in the second grade of secondary school is recommended. Considering that half of all mental illnesses emerge by the age of 14 (WHO, 2021), this timing is reasonable.

Second, trust in others may contribute to school adaptation. This study clarifies the longitudinal and cyclical relationships between interpersonal trust and school adaptation. Especially after the second year of secondary school, when interpersonal trust increases (or decreases), school adaptation tends to increase (or decrease) simultaneously; this tendency can also be seen in the next semester. Based on this, we put forward three practical proposals at the individual level, research level, and organizational level such as schools and governments. First, at the individual level, people who support teenagers should be aware of the role that their trustful behaviour can play in protecting students’ mental health. Knowing that trust in others is enhanced by experiences of acceptance and approval (Amagai, 2001), as well as a sense of being trusted (Sakai, 2005), will be helpful for them. However, more detailed knowledge at the verbal and behavioural levels may be necessary to facilitate the effective behaviour of non-mental health professionals, such as parents and teachers. Therefore, as a proposal for the research level, a detailed examination of the process by which adolescents acquire a sense of trust in others is expected in the future. Finally, at the organizational level, faculty mental health support and the development of a supportive environment may bring diverse benefits. Teachers’ well-being (Maricuțoiu et al., 2023) and daily emotions (Koenen et al., 2017) are related to the quality of the teacher-student relationship. Governments should encourage school-wide interventions such as positive behaviour interventions and support (PBIS) that not only sustain students’ motivation (Petrasek et al., 2021), but also enhance trust towards staff (Houchens et al., 2017).

The third implication is concerned with understanding the possible mechanism of the decline in school adaptation and reinterpreting the phenomenon. A group of students showed high school adaptation at the time of entering secondary school; however, it dropped sharply around the second year and to significantly lower levels towards the end. This may be partly due to the developmental stage of trust in relationships with others. According to Amagai (2001), from secondary school to high school, there are three developmental phases of interpersonal trust: a state of not being conscious of trust, a temporary decrease in trust through conflict experiences with others, and the recovery of trust by overcoming interpersonal conflicts. The tendency to experience more social problems in secondary school than in high school (Baidoo-Anu & Acquah, 2021) also suggests that interpersonal trust may change during secondary school life. Some of those who showed a high level of school adaptation at the time of entering secondary school may have had little experience with interpersonal conflicts before and would have confronted difficulties for the first time in secondary school. In other words, for some students, a sharp drop in school adaptation might be a sign of transition from the stage of not being conscious of trust to a temporary decline due to interpersonal conflict. Even if a decline in school adaptation is observed, it may be important to regard it as a developmental stage and to intervene appropriately regarding the factors that promote recovery.

Limitations

This study has several limitations. First, the sample was collected from a single school and was not large enough. Further research in other schools and countries is needed to generalize the findings of this study. In addition, the sample did not include long-term absentee students. Future continuous surveys with a large sample will not only confirm the results of this study, but may also detect signs of school refusal, such as a sudden drop in interpersonal trust and school adaptation. Finally, this study did not examine the detailed processes and triggers of changes in school adaptation. More research is needed to utilize the results of this study in actual schools, such as a study on behaviours that help build trust or a study regarding the qualitative differences of respondents who belonged to each class in Analysis 2.

Conclusion

The results of this study provide insight into adolescent school adaptation and the possible support to students. The second year of secondary school is a critical term in which school adaptation drastically changes, and experiences of mental distress might be likely. Although the effects of prospects and positivity towards stress require further investigation, the results suggest that interpersonal trust may protect students’ mental health. To overcome the interpersonal conflicts that students experience during this period and strengthen their sense of trust, the environment created by adults around them is important. Supporters, including parents and teachers, should be sensitive to the positive and negative effects of their daily behaviours. Even if they do not provide direct approaches, their sincere behaviour may indirectly affect the students’ mental health.

Ethical approval

All procedures performed in this study involving human participants were in accordance with the ethical standards of Life Science Research Ethics and Safety in the University of Tokyo under the approval number 18–93, and with the 1964 Helsinki declaration and its later amendments or comparable ethical standards.

Informed consent

Informed consent was obtained from the school principals and all individual participants included in the study. Parents were informed in writing of the purpose, contents, and privacy policy of the survey beforehand; and the school principal submitted a consent form on behalf of the students’ parents.

Disclosure statement

No potential conflict of interest was reported by the author.

Data availability statement

The data that support the findings of this study are available from the corresponding author, Kyoko Amai, upon reasonable request. The analysis codes are available for download from the open science framework directory (https://osf.io/se8pn/?view_only = 0bb3358aa13e440d8ca9e7320b577fed).

Additional information

Funding

This work was supported by the Japan Society for the Promotion of Science (JSPS) under Grant number 18J20397 and 21J40150. The content is solely the responsibility of the author and does not reflect the views of the funder, JSPS.

Notes on contributors

Kyoko Amai

Kyoko Amai is a project associate professor at Chiba University specializing in school psychology and developmental psychology. She conducts research regarding mental health in adolescence. Especially, the mentality of adolescents who do not seek nor accept support from others and novel support systems for them are her recent research focuses. For more information, please refer to https://researchmap.jp/KyokoAmai?lang=en.

Notes

1. The research site and school were selected for two reasons. First, the city was a typical provincial city without extreme gaps in wealth and education that are observed in large metropolitan areas. The school comprised an ordinary public school that was not designated as a special research school in an area with an average economic situation. Second, the first author had conducted another research beforehand at the school.

2. Students wrote their grade, class, gender, and birthday on the face sheet of the questionnaire. The information was used for connecting the data from different time points.

3. Time 7 (T7) and Time 8 (T8) were observed during the COVID-19 pandemic. There were several months of school closure and online classes between Time 6 (T6) and T7, but none of the variables at T7 showed a statistically significant decline compared to the pre-pandemic scores.