Advertisement

Journal of Youth and Adolescence

, Volume 47, Issue 6, pp 1252–1266 | Cite as

The Rise and Fall of Depressive Symptoms and Academic Stress in Two Samples of University Students

  • Erin T. Barker
  • Andrea L. Howard
  • Rosanne Villemaire-Krajden
  • Nancy L. Galambos
Empirical Research

Abstract

Self-reported depressive experiences are common among university students. However, most studies assessing depression in university students are cross-sectional, limiting our understanding of when in the academic year risk for depression is greatest and when interventions may be most needed. We examined within-person change in depressive symptoms from September to April. Study 1 (N = 198; 57% female; 72% white; Mage = 18.4): Depressive symptoms rose from September, peaked in December, and fell across the second semester. The rise in depressive symptoms was associated with higher perceived stress in December. Study 2 (N = 267; 78.7% female; 67.87% white; Mage = 21.25): Depressive symptoms peaked in December and covaried within persons with perceived stress and academic demands. The results have implications for understanding when and for whom there is increased risk for depressive experiences among university students.

Keywords

Depressive symptoms Academic stress University students Longitudinal 

Introduction

On average, psychological well-being improves across the transition to adulthood (Schulenberg and Zarrett 2006) and depressive symptoms in particular decline from age 18 to 25 (Galambos et al. 2006). At the same time, prevalence rates for major depression peak during this transition (Pearson et al. 2013; Rohde et al. 2012) and many university students in particular screen above clinical cut-off scores for major depression (Eisenberg et al. 2007). A recent review of prevalence rates of depression among university students from around the world showed that, on average, 30% of undergraduate students experience depression (Ibrahim et al. 2013). Moreover, a recent birth cohort analysis in the United States showed that rates of major depression among college students rose substantially from the 1930s through 2010 (Twenge et al. 2010).

With greater proportions of the young adult population pursuing post-secondary education (Bureau of Labor Statistics U.S. Department of Labor 2014; Galarneau et al. 2013), high rates of depression in this population likely reflect, at least in part, the fact that more people vulnerable to depression are attending university (e.g., Rohde et al. 2012). That said, experiences specific to the university context may also contribute to depressive experiences. Given that depressive symptoms vary across the academic year (e.g., Cooke Bewick et al. 2006) and that both objective and subjective experiences of academic stress are associated with depressive symptoms (e.g., Chambel and Curral 2005), it is likely that stress inherent in the university context contributes to variation in university students’ depressive experiences.

Although many studies have measured depressive experiences in university students, most have done so cross-sectionally. Results from cross-sectional studies tell us only who is at greater risk—that some students are more or less at risk than others—but do not speak to when students are at greatest risk. In the current study, we examined within-person patterns of change in depressive symptoms in two samples of Canadian university students, each of which was followed across one academic year. We aimed to determine when risk for depression may be highest by assessing whether increases in depressive symptoms coincided with expected periods of high academic stress (e.g., the end of semester when final papers and exams are due) and whether within-person patterns of change in depressive symptoms were related to academic demands (i.e., workload) and perceived stress.

Depression in University Students

University students around the world experience elevated depressive symptoms and many are at risk for clinical depression (for a review, see Ibrahim et al. 2013). In a study of U.S. students, 53% indicated that they had experienced what they considered depression since starting college (Furr et al. 2001), with similar percentages found in countries as diverse as France (Bouteyre et al. 2006), Malaysia (Shamsuddin et al. 2013), and Kenya (Othieno et al. 2014). Of the few longitudinal studies that assessed change in depressive experiences in university samples, symptom levels tended to increase over time. For example, depressive symptoms rose between orientation and seven months later in first-year students in the eastern U. S. (Alfred-Liro and Sigelman 1998). Similar increases were found in three longitudinal studies conducted in the United Kingdom. In the first, students assessed before entering and six weeks after starting university showed significant increases in depressive symptoms (Fisher and Hood 1987). In the second study, depressive symptoms were higher during students’ second year of university compared to the month before starting university (Andrews and Wilding 2004). In the third study, incoming students were surveyed before starting university and three times across the first year; depressive symptoms were highest at the end of first semester (Cooke et al. 2006). Thus, it appears that the university context increases risk for depression, and that symptoms may fluctuate over time.

Stress in University

University students are confronted with multiple stressors, including academic demands (Montgomery and Côté 2003), and show high levels of subjective distress associated with these demands. Recent results from the American College Health Association National College Health Assessment (American College Health Association 2016) survey showed that 42.3% of Canadian university student participants reported overwhelming levels of anxiety during the previous year. More (58.1%) found their academic work very difficult to handle. In a large sample of first-year university students in Germany, Poland, and Bulgaria, course work was rated as the greatest burden compared to relationships or concerns about the future (Mikolajczyk et al. 2008).

Depressive symptoms and perceived stress have been shown to correlate with academic demands in several university samples from different parts of the world (e.g., in Europe, Haldorsen et al. 2014; Mikolajczyk et al. 2008; and in China, Sha and Xia 2004). In a recent cross-sectional study of 900 Canadian university students (Newcomb-Anjo et al. 2017), perceived stress associated with academic demands was related to elevated depressive symptoms after controlling for the effects of 13 established risk factors for depression (e.g., demographic characteristics, history of abuse, cognitive style and personality, recent stressful life events, social support). Others have shown that students who report many demands and who also appraise them as stressful report the most depressive symptoms (Chambel and Curral 2005; Haldorsen et al. 2014).

These associations could be explained by models of stress and coping that emphasize the role of appraisal in the stress response (Lazarus and Folkman 1984), including Pancer and colleagues’ university adjustment model (Pancer et al. 2004). Appraisal involves evaluating demands in the environment against the availability of resources, such as effective coping styles and social support. According to these models, if a demand is deemed threatening and resources considered insufficient, perceived stress and associated negative affect will be elevated, possibly leading to adjustment difficulties such as mental health problems. Drawing on general appraisal models, Pancer et al. (2004) developed their model to explain positive and negative adjustment outcomes for students coping with university challenges. According to Pancer et al., students appraise the stressors they face during their transition to university, and deem them manageable if resources are adequate. Conversely, if a mismatch between stressors and resources is perceived, risk of experiencing negative outcomes such as depression increases. Stressors and resources in the academic domain pertain to workloads, aptitude, and study and time management skills.

Current Study

Studies examining the association between depressive symptoms and academic stress in university students at a single point during the academic year do not capture within-person change in symptoms and stress. Cross-sectional studies might provide information about which students are more at risk, but cannot impart knowledge about when students are at greater risk. Furthermore, results from cross-sectional studies cannot determine whether risk is occurring between individuals or within individuals as a function of the oscillating demands inherent in the university context. Separating between and within-person effects is important for understanding whether risk is a function of the individual or the context, or both (Curran et al. 2012; Hoffman and Stawski 2009). If depressive symptoms rise and fall in conjunction with fluctuating demands, the academic context is likely partly accountable for changes in students’ mood. Determining whether depressive symptoms ebb and flow with academic demands will clarify when and for whom risk for depression is highest.

In the current study, depressive symptoms were assessed multiple times across one academic year in two separate samples of Canadian university students. In both samples, waves of measurement coincided with periods of relatively low (start-of-semester) and relatively high (end-of-semester) academic workloads. In Study 1, global academic workload and global perceived stress were assessed at the end of the first semester. In Study 2, global academic workload and perceived stress specific to academic demands were measured multiple times across the academic year during both low and high stress periods.

Given that depressive symptoms in university students may represent, at least in part, a response to stress associated with academic demands, we hypothesized that depressive symptoms would rise toward the end of the semester, and that their peak would be associated with periods of increased academic workload as measured by the number of recent academic demands. Given the role of appraisal in the stress process, we further hypothesized that depressive symptoms would be higher for students who appraise their academic demands as unmanageable.

Study 1: Method

Participants

Participants were 198 full-time first-year students at a large Canadian university taking part in Making the Transition II, a web-based study of health-related behaviors, well-being, and academic performance. Sixty percent of students identified as female (n = 113), and students’ ages ranged from 17.5 to 19.8 years (M = 18.4, SD = .44). Self-reports showed that the racial/ethnic distribution was 72% white, 16% Asian or South Asian, 5% mixed ethnicity, and 5% another visible minority (two students declined to report). About half of students lived at home with parents (53%). Most students’ mothers (73%) and fathers (75%) had completed two-year college or four-year university degrees. Given these characteristics, this sample is similar to student populations at other large 4-year Canadian universities.

Procedures

In Fall 2005, participants were recruited from compulsory first-year classes across campus. Research assistants described the study to students who were present on the day of their recruitment visit, and students interested in participating provided contact information. Students were then contacted by email and invited to complete an initial paper-and-pencil questionnaire in groups at the beginning of the semester. A total of 198 students attended initial sessions in September or October (baseline), where they completed demographic questions and measures of depressive symptoms. Participants were then invited to complete web-based questionnaires each month across their first year (through April). Depressive symptoms measures were assessed at the baseline, December and April waves of measurement. Measures of academic demands and perceived stress were included in the November and December assessments. At all waves of measurement, informed consent was obtained from participants who completed that wave. The study was approved by the university research ethics review committee in accordance with the Government of Canada’s Tri-Council Policy Statement: Ethical Conduct for Research Involving Humans.

All 198 participants had complete data on the outcome measure (depressive symptoms) at baseline; four participants were missing demographic variables used as covariates in the analysis, reducing the sample to 194. Of these participants, 171 (88%) completed end-of-semester assessments in November or December and 152 (78%) completed the end-of-semester assessment in April. Participants who completed all waves (n = 144, 74%) were compared on all study variables to participants who only completed baseline. The groups differed on one variable: participants who only completed baseline had higher baseline depressive symptoms scores than participants who completed all three waves (t = 2.40, p < .05).

Measures

Depressive symptoms

The 10-item version (CESD-10; Andresen et al. 1994) of the Center for Epidemiologic Studies Depression Scale (CES-D; Radloff 1991) measured depressive symptoms. Participants completed the scale in September or early October (baseline/start-of-semester 1), in December (end-of-semester 1), and in April (end-of-semester 2) of their first year of university. Participants were asked how often in the past week they had experienced each of 10 depressive symptoms (e.g., felt depressed, felt fearful, felt lonely). Responses ranged from 0 [rarely or none of the time (less than 1 day)] to 3 [most or all of the time (5-7 days)]. The CESD-10 was developed and found to be reliable and valid in samples of older adults, and it compared well with the full 20-item version (Andresen et al. 1994). Total scores were calculated for descriptive purposes to assess clinically significant levels of symptomatology (scores of 10 or greater; Andresen et al. 1994). Mean scores were used in our main analyses, with higher scores indicating more symptoms in the past week. Cronbach’s alphas for the three waves were .80, .82, and .83, respectively.

End-of-semester academic workload

In November and December students were asked to indicate the number of grades that had been returned to them (e.g., on tests, assignments) in the past 14 days. Some students were missing values for either November (n = 23) or December (n = 30). To maximize the available n, reports for these two months were averaged and used as an indicator of end-of-semester workload or academic demands.

End-of-semester perceived stress

The 4-item version of the Perceived Stress Scale (PSS; Cohen et al. 1983) was administered in November and December. Participants were asked to indicate how often over the past 2 weeks they felt 1) unable to control the important things in their lives; 2) confident about their ability to handle personal problems (reverse scored); 3) that things were going their way (reverse scored); and 4) that difficulties were piling up so high that they could not overcome them. Items were rated on a 5-point scale ranging from 0 (never) to 4 (very often). Reports for both months were averaged and used as a measure of perceived stress at the end of the semester. Higher scores indicated greater perceived stress. Cronbach’s alpha for the combined November and December scale was .85.

Control variables

Main analyses controlled for demographic variables that could be related to depressive symptoms. These included mother’s education (a proxy for family socioeconomic status/SES), race/ethnicity, and sex. Living situation (whether the student lived with parents or not) was also controlled.

Data Analysis Plan

Multilevel linear models were estimated using HLM version 7.01 (Raudenbush et al. 2011), using restricted maximum likelihood estimation to account for missing data. This procedure computes an individual likelihood function for each participant based on available data, provided that complete predictor data are present for each wave of available outcome data. Importantly, cases contributing partial outcome data are retained and leveraged to improve the accuracy of the model estimates. The Level 1 within-person model included a random intercept to estimate baseline levels of depressive symptoms, and linear and quadratic effects of time to assess the pattern of change in depressive symptoms across the academic year. Eq. 1 shows the Level 1 model predicting depressive symptoms (DEPRESS) at each wave t for each person i, from the linear combination of a random intercept (β0i), effects of linear time (β1iLINEAR) and quadratic time trends (β2iQUAD), and residual deviations of depressive symptom scores at each wave t for each person i from the average trajectory (r ti ):
$$DEPRESS_{ti}{\mathrm{ = }}\beta _{0i}{\mathrm{ + }}\beta _{1i}LINEAR{\mathrm{ + }}\beta _{2i}QUAD + r_{ti}$$
(1)
The Level 2 between-persons model included the effects of demographic covariates, perceived stress and academic workload on the intercept and the effects of perceived stress, academic workload, and their interaction on the linear and quadratic effects of time. Eq. 2 shows a simplified Level 2 model containing stress and academic workload covariates:
$$\begin{array}{*{20}{l}} \beta _{0i} &= \gamma _{00} + \gamma _{01} STRESS_i + \gamma _{02} WORKLOAD_i + u_{0i} \\ \beta _{1i} &= \gamma _{10}+ \gamma _{11} STRESS_i + \gamma _{12} WORKLOAD_i \\ \beta _{2i} &= \gamma _{20} + \gamma _{21}STRESS_i + \gamma _{22} WORKLOAD_i\end{array}$$
(2)

Estimates include average levels of depressive symptoms at baseline (γ00), the instantaneous rate of change in symptoms (γ10; “linear” growth), average acceleration/deceleration in the rate of change in symptoms (γ20; “quadratic” growth), associations between perceived stress and academic workload at the end of semester 1 and baseline depressive symptoms (γ01 and γ02, respectively), and estimates of stress and workload differences in the rates of change in depressive symptoms (γ11 and γ21 for perceived stress; γ12 and γ22 for academic workload). Deviations of individual mean scores from the conditional mean level of depressive symptoms are captured in the error term u0i. The linear and quadratic time trends were treated as fixed (nonrandomly varying), because preliminary model testing showed that these effects did not vary significantly across participants. Thus, any deviations of individual trajectories from the mean rates of change, conditional on stress and workload, are captured by residual error at Level 1. Additional effects not shown in Eq. 2 (intercept covariates; interaction terms for stress and workload) are entered in Eq. 2, as we have shown for stress and workload.

Models were tested in two steps, beginning with unconditional growth models to establish an optimal functional form of growth in depressive symptoms. Perceived stress, academic workload, their interaction, and all covariates were added simultaneously in a second step.

Study 1: Results

Descriptive statistics and intercorrelations are presented in Table 1. A score of 10 or above on the 10-item version of the CES-D represents clinically significant symptom levels and is comparable to a score of 16 or above on the 20-item version of the CES-D (Andresen et al. 1994). Total scores for depressive symptoms at all three waves were around the clinically significant level, on average. Furthermore, 38.8%, 50.3%, and 41.8% of students had depressive symptoms scores equal to or greater than 10 at baseline, December, and April, respectively. The average perceived stress score was low, falling between the Likert-scale anchors of “almost never” and “sometimes.” Students reported having had on average between 4 and 5 academic demands in the past 14 days in November/December. Scores on the CES-D were highly intercorrelated across waves and with end-of-semester perceived stress. End-of-semester 1 academic demands was not correlated with CES-D scores from any wave nor with end-of-semester 1 perceived stress.
Table 1

Sample 1 Descriptive Statistics and Intercorrelations among Study Variables

 

Depressive Symptoms Start of Semester 1

Depressive Symptoms End of Semester 1

Depressive Symptoms End of Semester 2

Perceived Stress End of Semester 1

Academic Workload End of Semester 1

Depressive Symptoms Start of Semester 1

 

.60***

.50***

.59***

.05

Depressive Symptoms End of Semester 1

  

.56***

.76***

.00

Depressive Symptoms End of Semester 2

   

.57***

−.03

Perceived Stress End of Semester 1

    

.09

Mean (SD)

9.88 (5.29)

11.01 (5.74)

10.37 (5.84)

1.77 (.72)

4.41 (2.40)

*p < .05. **p < .01. ***p < .001

Multilevel models of within-person change in CES-D scores (Table 2) showed significant linear and quadratic effects of time for depressive symptoms (controlling for the effects of covariates on initial status), with symptoms peaking in December (Fig. 1). In these models, linear time was coded 0, 1, 2, corresponding to each wave of measurement. Symptoms initially increased at a steeper rate for students with higher end-of-semester levels of perceived stress, but switched direction and were declining by the end of the academic year. At the December peak, estimated depressive symptoms for students with higher stress were 82% of a standard deviation above average, falling slightly to 72% of a standard deviation above average by the end of April. End-of-semester 1 academic demands and the interaction between perceived stress and academic demands were not related to depressive symptom trajectories. None of the control variables were significant predictors of depressive symptom levels.
Table 2

Sample 1 Multilevel Model of Between-Person Effects of End-of-Semester 1 Perceived Stress and Academic Workload on Level (intercept) and Change in Depressive Symptoms Across One Year

Fixed Effects

Coefficient

SE

Intercept (random), π 0

 Intercept (initial status September)

1.081***

.105

 Mother’s Education

−.042

.030

 Race/Ethnicity (white = 0; another = 1)

−.073

.056

 Sex (male = 0; female = 1)

.026

.044

 Living Situation (with parents = 1; other = 0)

.076

.046

 Perceived Stress (end-of-semester 1 average)

.435***

.044

 Academic Workload (end-of-semester 1 average)

−.003

.012

Linear time slope (fixed), π 1

 Intercept

.253***

.066

 Perceived Stress (end-of-semester 1 average)

.339***

.088

 Academic Workload (end-of-semester 1 average)

−.019

.022

 Perceived Stress X Academic Workload

−.026

.032

Quadratic time slope (fixed), π 2

 Intercept

−.106**

.033

 Perceived Stress (end-of-semester 1 average)

−.158***

.043

 Academic Workload (end-of-semester 1 average)

.003

.012

 Perceived Stress X Academic Workload

.012

.019

 Random Effects

Estimate

SE

 Intercept

.047***

.216

*p < .05. **p < .01. ***p < = .001

Fig. 1

Study 1: Trajectories of depressive symptoms across the academic year for students with high perceived stress levels at the end of semester 1 (+1 SD above the mean) and low perceived stress levels at the end of semester 1 (−1 SD below the mean). Dashed lines represent 95% confidence intervals

Study 2: Method

Study 2 replicates and extends Study 1 with a different sample of Canadian university students. Study 2 collected four waves of data from students enrolled in any year of study (across the 2013–2014 academic year). Depressive symptoms, perceived stress, and academic workload were measured at all waves. This afforded the opportunity to 1) test the hypothesis that academic workload and stress associated with those demands would rise toward the end of the semester; 2) examine within-person associations of perceived stress and academic workload with depressive symptoms; and 3) to test whether the pattern of change in depressive symptoms and associations with perceived stress and academic workload were the same across years of study.

Participants

Participants in Study 2 were 267 (78.7% identified as female) full-time undergraduate students between the ages of 18 and 25 years (M = 21.25, SD = 1.67) enrolled in their first university degree at a large urban university located in a different Canadian province than that of the Study 1 sample. Participants in Study 2 were relatively evenly spread over the first (26.0%), second (32.0%), third (29.0%), and fourth (13.0%) years of university study or beyond (maximum 6th year). Over two-thirds of participants (67.8%) self-identified as white and not belonging to a visible racial or ethnic minority group, 7.8% identified as Black, 10.5% as Asian, 3.3% as Native North American, 6.7% as Hispanic, and 4.5% as Arab. A further 13.9% identified as both white and belonging to a visible racial or ethnic minority group. Similar to the Study 1 sample, about half (51.3%) of Study 2 participants reported living at home with their parents and 75.1% percent of participants’ mothers and 76.1% of participants’ fathers had completed two-year (college), four-year (university), or higher education degrees.

Procedures

Potential participants were recruited at student events held at the start of an academic year and by flyers posted on campus. Those who agreed to participate were sent an email that described the study and provided a link to an online survey. Similar to Study 1, the survey assessed many constructs relevant to the well-being of university students. These procedures likely contributed to the greater proportion of female participants in Study 2. It has been shown that more women than men participate in this type of student survey research (Gosling et al. 2004).

Data collection for the study proceeded over four waves spanning one academic year: Wave 1 = start-of-semester 1 (September/early October); Wave 2 = end-of-semester 1 (late November/early December); Wave 3 = start-of-semester 2 (late January); Wave 4 = end-of-semester 2 (late March/early April). At each wave, participants who had at least partially completed one or more previous online survey(s) were contacted via email and invited to complete a new survey. For each survey, participants were required to give their free and informed consent, and indicate that they were between the ages of 18 and 25 years and enrolled in their first university degree program. Participants who completed surveys at each wave were sent an electronic gift card of their choice. The study was approved by the university research ethics review committee in accordance with the Government of Canada’s Tri-Council Policy Statement: Ethical Conduct for Research Involving Humans.

Of the 267 full-time students who completed the Wave 1 survey, 256 (96%) had complete data on the outcome measure (depressive symptoms) at Wave 1. Across the study variables, 23 participants were missing information on demographic variables used as covariates or predictor variables in the main analysis. Of the 244 participants with complete Wave 1 data, 198 (74.2%) completed Wave 2 assessments, 165 (61.8%) completed Wave 3 assessments, and 146 (54.7%) completed Wave 4 assessments. Participants who completed all waves (n = 127; 47.6%) were compared on all study variables to participants who completed only Wave 1. The groups differed on only two variables. First, men were more likely to drop out of the study. Of participants who did not complete all waves, 30% were men, and only 13.4% of students who completed all waves were men (χ2 = 9.92, p < .05). Second, students who completed all waves reported having more academic demands (6.75 tests or assignments) at Wave 1 compared to students who did not complete all waves (6.0 tests or assignments; t = 2.31, p < .05). The same patterns emerged when an ANOVA was conducted to compare participants based on the total number of waves completed. In total 57 participants (21%) completed only Wave 1, 39 (15%) completed Wave 1 plus one other wave, and 44 (16%) completed Wave 1 plus two other waves.

Measures

Depressive symptoms

In Study 2, the complete 20-item Center for Epidemiologic Studies Depression Scale (CES-D) was administered at all four waves (Radloff 1977). Total scores were calculated for descriptive purposes to assess clinically significant levels of symptomatology (scores of 16 or greater). Mean scores were calculated for main analyses. Higher scores reflected elevated frequency of depressive symptoms experienced in the previous week. Cronbach’s alphas ranged from .88 to .92 across waves.

Current academic workload

At all four waves, participants reported the number of quizzes, exams, papers, presentations, labs, readings, problem sets, and other course work they completed in the past two weeks and how many of these assessments they would have to complete in the upcoming two weeks. Reports for both two-week periods were totaled to create an index of academic workload. Frequencies for each type of academic demand at each wave of measurement showed that more students had labs, readings, problem sets, and quizzes earlier in the semester and more exams, papers, and presentations later in the semester. For example, at the start of semester 1, 42% of students had a quiz in the past two weeks and 61% expected another quiz in the next 2 weeks. At the end of semester 1, 63% had a quiz in the past 2 weeks and 24% expected a quiz in the next two weeks. In contrast, at the start of semester 1, 16% of students had an exam in the previous 2 weeks and 56% reported an upcoming exam in the next two weeks. At the end of semester 1, 66% of students had an exam in the previous 2 weeks and 89% expected an exam within two weeks.

Perceived stress associated with current academic workload

As in Study 1, the 4-item version of the Perceived Stress Scale (PSS; Cohen et al. 1983) measured perceived stress. The items were the same as the PSS 4-item scale described in Study 1, but were posed after participants completed their reports of recent academic workload and were framed around these demands rather than life in general (e.g., “How often have you felt that difficulties in your academic work were piling up so high that you could not overcome them”; “How often did you feel confident about your ability to handle your academic work”). Mean scores were calculated with higher scores indicating higher perceived academic stress. Cronbach’s alphas ranged from .76 to .81 across waves.

Control variables

Main analyses controlled for demographic variables: mother’s education (to assess SES), race/ethnicity, and sex. Living situation (whether the student lived with parents or not) was also controlled, as was year of university study.

Data Analysis Plan

Multilevel linear models were estimated using Mplus software (Version 7.0). Multiple imputation for clustered data was used to retain all 267 participants in our analyses (joint modeling using an unstructured within-cluster covariance matrix; Asparouhov and Muthén 2010; Enders et al. 2016). Results pool over estimates from 50 imputed datasets. Within-person (time-varying) estimates of academic demands and perceived stress were respectively calculated by subtracting, for each person, the mean of his/her own set of repeated measures from their score at each wave of assessment (person mean centering, e.g., Curran and Bauer 2011; Howard 2015). Between-person (average) estimates were the means for each person across their own set of repeated measures of academic demands and perceived stress, respectively. These averages were included in the model to examine individual differences in depressive symptoms explained by students carrying higher average academic loads and students reporting higher levels of stress on average. Separate estimates of the time-varying and average components allow us to test unique hypotheses about effects of within-person variation and between-person differences in academic demands and stress on depression over time. Tests of within-person effects are orthogonal to their between-person counterparts, allowing us to triangulate on specific sources of influence on students’ depressive symptoms.

As in Study 1, our Level 1 model included linear and quadratic time trends, with the addition of terms capturing time-varying effects of perceived stress and academic workload, as follows:
$$\begin{array}{l}DEPRESS_{ti} = \beta _{0i} + \beta _{1i}LINEAR + \beta _{2i}QUAD \\ + \beta _{3i}\mathrm {STRESS}_{ti} + \beta _{4i}WORKLOAD_{ti} + r_{ti}\end{array}$$
(3)
Our Level 2 equations were similar to those shown in Study 1, except that we had sufficient variability to estimate random effects for each of the time trends (u1i and u2i added to Eq. 2 for the linear and quadratic time trends, respectively), capturing systematic individual differences in rates of change in depressive symptoms. We treated our time-varying estimates of stress and academic workload as fixed effects at Level 2, as follows:
$$\begin{array}{l}\beta _{3i} = \gamma _{30}\\ \beta _{4i} = \gamma _{40}\end{array}$$
(4)
Model testing proceeded as in Study 1, beginning with unconditional growth models to establish an optimal functional form of growth in depressive symptoms. Perceived stress, academic workload, and all covariates were included simultaneously. We also included several exploratory interactions between stress (time-varying and person mean), workload (time-varying and person mean), and linear time. None were found to be statistically significant and we trimmed these terms from our model one at a time, from largest to smallest p-values (Aiken and West 1991).

Study 2: Results

Descriptive statistics for depressive symptoms, perceived stress, and academic workload are reported in Table 3. At each wave of measurement, average total scores for depressive symptoms were around the clinically significant score of 16 (Radloff 1977). Across the academic year, at Waves 1 through 4 respectively, 38.2%, 46.0%, 39.3%, and 41.2% of students scored equal to or greater than 16. Means for perceived stress were low, falling between the Likert scale anchors of “almost never” to “sometimes” across the year. Students reported having between six and seven academic demands, on average, at each wave.
Table 3

Sample 2 Descriptive Statistics, by Wave of Measurement

 

Start of Semester 1

End of Semester 1

Start of Semester 2

End of Semester 2

Variable

M (SD)

M (SD)

M (SD)

M (SD)

Depressive Symptoms (Total Scores)

14.90 (9.39)

17.72 (11.64)

15.51 (10.48)

15.78 (9.55)

Perceived Stress

1.62 (1.68)

1.81 (.78)

1.51 (.76)

1.65 (.77)

Academic Workload

6.40 (2.49)

6.67 (2.49)

6.03 (2.63)

6.33 (2.84)

Intercorrelations between depressive symptoms with perceived stress and counts of academic workload for each wave of measurement are presented in Table 4. Within and across waves, depressive symptoms and perceived stress were consistently moderately correlated. Depressive symptoms and counts of academic workload were uncorrelated at most waves. Additionally, three correlations between perceived stress scores and academic workload were significant: Wave 2 workload (end-of-semester 1) with both Wave 1 perceived stress (r = .16, p < .05) and Wave 2 perceived stress (r = .20, p < .05); Wave 3 perceived stress with Wave 3 workload (r = .17, p < .05). Autocorrelations for CES-D scores across waves were all significant (p < .05) and ranged from r = .43 to r = .70, as were autocorrelations for perceived stress (r = .48 to r = .71) and academic workload (r = .34 to r = .52).
Table 4

Sample 2 Correlations Between Depressive Symptoms with Perceived Stress and Academic Workload by Wave of Measurement

Depressive Symptoms

 

Start of Semester 1

End of Semester 1

Start of Semester 2

End of Semester 2

Perceived Stress

 Start of Semester 1

.52**

.46**

.42**

.33**

 End of Semester 1

.46**

.60**

.55**

.42**

 Start of Semester 2

.42**

.45**

.46**

.40**

 End of Semester 2

.49**

.48**

.54**

.51**

Academic Workload

 Start of Semester 1

.08

.12

.01

−.08

 End of Semester 1

.07

.16*

.08

.10

 Start of Semester 2

.21**

.12

.14

.09

 End of Semester 2

−.01

−.02

.09

.09

*p < .05. **p < .01

To test the hypothesis that academic workload and perceived stress associated with those demands rise toward the end of the semester, we ran unconditional growth models for both academic workload and perceived stress. Linear time was coded 0, 1, 2, 3, corresponding to each wave of measurement. Results showed non-significant linear change (coefficient = .086, SE = .056, p = .12), but a significant quadratic effect for perceived stress (coefficient = -.036, SE = .018, p = .04), with stress levels rising from September and peaking in December. For academic workload, the linear (coefficient = -.090, SE = .188, p = .63) and quadratic rates of change (coefficient = .005, SE = .062, p = .94) were not significantly different from zero.

Next, our multilevel model tested the associations of academic workload and perceived stress with depressive symptoms. Similar to results for Study 1, the multilevel model for Study 2 (Table 5) showed significant linear and quadratic rates of change in depressive symptoms (controlling for the effects of covariates on initial status), with symptoms peaking in December. An effect of average perceived stress on the intercept showed that students with higher average levels of perceived stress started semester 1 with higher depressive symptoms, a difference in level that was maintained until the December peak. After that, there was a deceleration in depressive symptoms among students with higher average levels of perceived stress (shown by the significant negative effect for average perceived stress on the quadratic time slope). As shown in Fig. 2, at the end-of-semester 1 peak, predicted depressive symptoms for students with higher average perceived stress were nearly two thirds of a standard deviation (62%) above average, falling to just under half a standard deviation (47%) above average by the end of semester 2. Average levels of current academic workload were not related to level or rate of change in depressive symptoms across the academic year. Within individuals, depressive symptoms were highest in months when students reported greater within-person perceived stress and more academic demands than usual (time-varying effects in Table 5). However, these effects were small: when students felt higher-than-usual stress (+1 SD above their own mean stress level), depressive symptoms were 11% of a standard deviation higher than usual. At times when students reported higher-than-usual academic workload (+1 SD above their own mean number of demands), depressive symptoms were 4.5% of a standard deviation higher than usual. None of the control variables was a significant predictor of the intercept. The interaction between time-varying perceived stress and academic workload was not significant and thus was not presented in the final model.
Table 5

Sample 2 Multilevel Model of Between-Person Effects of Average Perceived Stress and Academic Workload on Level (intercept) and Change in Depressive Symptoms and Within-Person Effects of Perceived Stress and Academic Workload on Change in Depressive Symptoms Across One Year

  

Coefficient

SE

Fixed Effects

 Intercept (initial status September)

 

0.719***

.068

 Mother’s Education

 

.026

.017

 Race/Ethnicity (white = 0; visible minority = 1)

 

−.008

.043

 Sex (male = 0; female = 1)

 

.019

.048

 Living Situation (with parents = 1; other = 0)

 

−.041

.041

 Year of Study (1, 2, 3, 4)

 

−.009

.021

 Average Perceived Stress (mean across waves)

 

.450***

.042

 Average Academic Workload (mean across waves)

 

.011

.017

 Linear time slope

 Rate of change in depressive symptoms

 

.124***

.036

 Average Perceived Stress (mean across waves)

 

.102

.057

 Average Academic Workload (mean across waves)

 

.001

.023

 Quadratic time slope

 Deceleration in rate of change in depressive symptoms

 

−.039***

.012

 Average Perceived Stress (mean across waves)

 

−.041*

.019

 Average Academic Workload (mean across waves)

 

−.001

.007

 Time-varying effect of perceived stress (person-centered)

 

.151***

.031

 Time-varying effect of academic workload (person-centered)

 

.019*

.007

 Random effects

 Residual (σ2)

 

.097***

.012

 Covariance matrix of L2 random effects

u 0

u 1

u 2

 Random intercept (u 0 )

.057** (.018)

.

.

 Linear time slope (u 1 )

.017 (.020)

.021 (.035)

.

 Quadratic time slope (u 2 )

−.007 (.006)

−.006 (.011)

.002 (.003)

*p < .05. **p < .01. ***p < .001

Fig. 2

Study 2: Trajectories of depressive symptoms across the academic year for students with high average perceived stress levels (+1 SD above the mean) and low average perceived stress levels (−1 SD below the mean). Dashed lines represent 95% confidence intervals

Discussion

Given concerns about the prevalence of poor mental health on university campuses (e.g., Canadian Association of College and University Student Services and the Canadian Mental Health Association 2013), it is critical to examine patterns of change in depressive symptoms across the academic year to identify when, in the typical academic cycle, risk for depression may be highest. Knowledge about fluctuations in depressive symptoms and how they relate to students’ challenges is essential for designing strategies aimed at promoting mental health and academic success. Surprisingly, few studies of university students have monitored change in depressive symptoms. To address this gap, we examined within-person patterns of change in depressive symptoms across one academic year in two samples of Canadian university students. We hypothesized that depressive symptoms would rise toward the end of the semester, peaking when academic workloads and perceived stress are likely highest (i.e., final exam period and major project deadlines). Multilevel models for both samples showed significant increases in depressive symptoms over time, with symptoms peaking in December, at the end of semester 1.

After determining peak timing of students’ depressive symptoms in the academic cycle, we examined whether that pattern was related to students’ objective and subjective experiences of academic stress. We examined both between-person and within-person associations of academic workload (i.e., counts of academic demands) and ratings of perceived stress with depressive symptoms. In Study 1, we found that students with higher levels of global perceived stress at the end of semester 1, when depressive symptoms peaked, experienced more depressive symptoms at that time compared to students whose perceived stress was lower. Number of end-of-semester academic demands, however, was not related to depressive symptoms in Study 1.

In Study 2, we improved upon Study 1 by assessing academic workload, level of perceived stress associated with these academic demands, and depressive symptoms at four times across the academic year. Study 2 assessed both between- and within-person associations of depressive symptoms with counts of academic demands and ratings of perceived stress. Between-persons results showed that students whose levels of perceived stress were higher on average across the academic year started off with higher depressive symptoms, which increased at a steeper rate toward peak symptoms around the end of semester 1, but moved slightly toward average symptom levels by the end of semester 2. Within persons, depressive symptoms rose to greater levels on occasions when perceived stress was higher than usual for a given person. Importantly, this association is over and above the effect of average perceived stress shown in Fig. 2. Even among students reporting high average perceived stress, occasions marked by unusually high levels of perceived stress were associated with additional increases in depressive symptoms.

In addition to assessing the number of academic demands along with depressive symptoms at each wave of the study, Study 2 also improved the measurement of academic workload relative to Study 1. In Study 2 participants were asked to indicate the number of assessments they had to complete in the previous and upcoming weeks, rather than the number of grades returned to them. Results showed that average academic workload levels (the mean number of academic demands across the year) was not related to depressive symptoms, but depressive symptoms were higher on occasions when counts of academic demands were also slightly higher than one’s typical levels (the time-varying effect). That is, when students reported having more academic work to complete than usual, they reported more depressive symptoms. The within-person effect of academic workload in the absence of a between-person effect rules out the possibility that students who were enrolled in more demanding programs (e.g., disciplines with more frequent assignments and supplemental lab courses) experienced more depressive symptoms in general. Instead, it only appears to matter whether academic workload at a particular time exceeded what is typical for a given person.

Both Study 1 and Study 2 results replicate the robust finding that students who experience greater perceived stress associated with academic demands also report more depressive symptoms (e.g., Newcomb-Anjo et al. 2017). Additionally, the current study contributes two novel findings to the literature on depressive experiences in university students, both of which arise from the repeated measurement of depressive symptoms across the academic year. First, the results of Studies 1 and 2 revealed when in the academic year risk for depression may be highest. Depressive symptoms fluctuated within-individuals across the academic year as expected, rising from the beginning of the first semester and peaking at the end of the first semester, when final assessments typically occur. These results have implications for the accurate assessment of depression in university students. It has been suggested that two of the most commonly used screening measures for depression, the Beck Depression Inventory (BDI-II; Beck et al. 1996) and the Center for Epidemiologic Studies Depression Scale (CES-D; Radloff 1991) overestimate rates of depression in university samples (Santor et al. 1995). Our results showed that average levels of depressive symptoms exceeded what is considered clinically significant symptomatology in December, at the end of semester 1, and that more students fell into the “at-risk” group at that time relative to other points in the academic year. Thus, at least a portion of the discrepancy between rates of clinically significant levels of depressive symptoms obtained through self-report screening measures in university students (around 50%) and population-level prevalence rates obtained via diagnostic interviews (10% to 25%; Pearson et al. 2013; Rohde et al. 2012) may be accounted for by the timing of assessment. Given that most research on depression in university samples to date has been cross-sectional, levels of clinically significant symptoms may be somewhat inflated or deflated if time of year is not taken into account.

The second contribution of this research comes from Study 2, wherein we showed that on occasions when students experienced greater academic workloads and more perceived stress than their own average levels, they also reported increases in depressive symptoms. Although the sizes of these effects were small, these findings are important: they suggest that risk for depression increases when students are challenged to manage more academic demands than is typical for them. These results correspond with models of stress and coping suggesting that perceived stress will increase when demands in the environment are perceived as exceeding one’s individual resources (Lazarus and Folkman 1984; Pancer et al. 2004). Importantly, we were able to rule out the possibility that students with heavier academic demands in general were also the most depressed. Average workload was unrelated to depressive symptoms. In general models of depression, stress is a central contributing component to the onset of depression (e.g., diathesis-stress models, Hammen 2005; allostatic load models, McEwen 2003; developmental models, Compas 2004). And, in models of student burnout (Salmela-Aro et al. 2008), burnout has been observed in students who were initially immersed in their studies, but later came to develop a chronic stress response after exposure to repeated achievement pressures (Maroco and Campos 2012). Thus, if depressive symptoms arise from students appraising their academic demands as exceeding what they typically can manage, risk for major depression and burnout increase.

Limitations and Future Research

Several limitations of the current studies point to further directions for research on the role of academic demands in university students’ depressive experiences. First, future studies should explore the temporal ordering of demands, perceived stress, and depressive symptoms to better understand the progression from stress response to depression in university students. In the university context where demands, perceived stress, and depressive symptoms rise and fall together, pathways to depression may not be linear. Thus, it would be important to conduct path models to map out the temporal ordering of risk for depression in this context.

Future research should also identify the factors that distinguish students for whom elevated depressive symptoms represent compromised mental health from those for whom short-lived increases in depressive symptoms associated with academic stress serve an adaptive motivational function. Although negative affect reflected in clinical depression and student burnout is considered maladaptive, negative affect in other contexts can serve adaptive self-regulatory functions by motivating cognitive and behavioral responses to challenges, including academic ones (Harmon-Jones et al. 2013; for an example with adolescents see Wrosch and Miller 2009). For example, daily diary results showed that university students who experienced occasional periods of negative affect across one semester achieved the greatest academic success (Oishi et al. 2007). In another study spanning four years of university, students who were generally happy across all years achieved the greatest levels of academic success in semesters during which they experienced increases in negative affect. In fact, these students benefited only when within-person negative affect was high (Barker et al. 2016). Thus, for some students, increases in depressive symptoms may reflect productive investment in academic pursuits motivated by negative affect. In the current studies, emotion and self-regulatory strategies students may employ to adaptively manage demands and perceived stress and direct them toward academic success were not assessed, however.

A third limitation of the current research was the assessment of academic workload. In Studies 1 and 2, counts were not weighted by the effort students invested in the workload, or by the type of work or assignment they completed and which might have differed in its impact on final grades (e.g., quiz vs. assignment vs. final exam). Moreover, perceived stress associated with demands was not assessed at the exact time they were experienced. Daily diary, experience sampling, and stress interviews are methods that would more precisely identify which academic demands are particularly stressful for students and better capture the temporal relationship between type of demand and associated perceived stress.

Finally, future research should examine factors that may exacerbate or ameliorate perceived stress associated with academic demands and its association with depressive experiences. Although academic stress has been shown to be independently associated with depressive symptoms, over and above the effects of established risk and protective factors (Newcomb-Anjo et al. 2017), moderators of the effects of perceived stress on trajectories of depressive symptoms should be examined. In the Pancer et al. model (2004), stressors and resources are not only described in the academic domain, but in the personal and social domains as well. These additional domains include, for example, finances, individual personality characteristics, expectations for university life, and social environments and levels of social support. Certainly, the university experience can be characterized as a context in which socialization occurs broadly and wherein many developmental challenges are encountered en route to adulthood (Montgomery and Côté 2003). In the current study, the fact that depressive symptoms peaked in December and did not rise to the same levels in April, when, theoretically, demands should be similarly high, suggests that additional stressors and resources should be assessed in future research. On the one hand, it is possible that December is particularly challenging due to other factors associated with time of year, such as seasonal effects, financial burdens associated with holiday travel, and a briefer break before resuming studies. On the other hand, resources such as coping behaviors, personality traits and social support may, for some students, buffer the effects of stress on depressive experiences across the academic year. More generally, future research aimed at replicating and extending these findings may benefit from the use of recruitment strategies other than convenience sampling. This could improve representativeness in terms of demographic variables, possible range in depressive experiences, and factors other than academics that may contribute to or protect against these experiences for university students.

Conclusion

In recent years, media outlets have drawn attention to a possible mental health crisis on Canadian university campuses (e.g., Goffin 2017; Lunau 2012) and campus communities have responded by paying closer attention to the emotional needs of students. For example, in 2013, the Canadian Association of College and University Student Services and the Canadian Mental Health Association jointly published Post-Secondary Student Mental Health: Guide to a Systemic Approach in which they outline different ways in which universities can organize to support student well-being. Our results showed that symptom levels peak (i.e., that risk is highest) at the end of the first semester, constituting a first important step toward clarifying when the risk for depression may be highest in university samples. The second important finding from the current research is that symptoms were higher when students appraised their academic demands as exceeding what they typically could manage. This implies that enhancing individual psychological resources that involve stress appraisal may be a prime target for programs aimed at supporting student well-being and mental health. Together, the key findings from the current research have implications for the assessment, prevention, and treatment of depression in the university student population. Effective intervention at this stage of development is important, given that experiencing one major depressive episode increases the likelihood of experiencing another episode in the future (Kendler et al. 2001) and of experiencing poor educational and future economic outcomes (Fergusson et al. 2007). More generally, well-being and success across the transition to adulthood accumulate and predict future well-being in adult occupational and social roles (e.g., Howard et al. 2010). Considering that promoting university students’ self-management and coping skills is one avenue of support identified by the Canadian Association of College and University Student Services and the Canadian Mental Health Association, evidence from the current study may be used in efforts to reduce the incidence of depression and to enhance the concurrent and future well-being of university students.

Notes

Authors’ Contributions

All authors contributed to the development of the study concept. Study 1 design and data collection were the responsibility of N.L.G., A.L.H., and E.T.B. Study 2 design and data collection were the responsibility of E.T.B., A.L.H., and R.V.K. E.T.B. and A.L.H. performed the data analyses and drafted the Methods and Results sections. E.T.B. wrote the Introduction and Discussion sections. All co-authors provided critical revisions and approved the final version of the manuscript.

Funding

Study 1 was supported by a Social Sciences and Humanities Research Council of Canada operating grant awarded to N.L.G. and J.L.M. Study 2 was supported by a Social Sciences and Humanities Research Council of Canada Insight Development Grant awarded to E.T.B. and A.L.H.

Data Sharing Declaration

The datasets generated and/or analyzed during the current study are not publicly available but are available from the corresponding author on reasonable request.

Compliance with Ethical Standards

These studies were conducted in compliance with ethical standards outlined by the Government of Canada’s Tri-Council Policy Statement: Ethical Conduct for Research Involving Humans.

Conflict of interest

The authors declare that they have no conflict of interest.

Ethical Approval

Both studies were approved by their respective university research ethics review committees in accordance with the Government of Canada’s Tri-Council Policy Statement: Ethical Conduct for Research Involving Humans.

Informed Consent

Informed consent was received from all participants who participated at each wave of measurement as per the Government of Canada’s Tri-Council Policy Statement: Ethical Conduct for Research Involving Humans.

References

  1. Aiken, L. S., & West, S. G. (1991). Multiple regression: Testing and interpreting interactions. Thousand Oaks, CA: Sage.Google Scholar
  2. Alfred-Liro, C., & Sigelman, C. K. (1998). Sex differences in self-concept and symptoms of depression during the transition to college. Journal of Youth and Adolescence, 27, 219–244.  https://doi.org/10.1023/A:1021667813858.CrossRefGoogle Scholar
  3. American College Health Association. (2016). American College Health Association-National College Health Assessment II: Canadian Reference Group Executive Summary. Hanover, MD: American College Health Association.Google Scholar
  4. Andresen, E. M., Malmgren, J. A., Carter, W. B., & Patrick, D. L. (1994). Screening for depression in well older adults: Evaluation of a short form of the CES-D. American Journal of Preventive Medicine, 10, 77–84.CrossRefPubMedGoogle Scholar
  5. Andrews, B., & Wilding, J. M. (2004). The relation of depression and anxiety to life-stress and achievement in students. British Journal of Psychology, 95, 509–521.  https://doi.org/10.1348/0007126042369802.CrossRefPubMedGoogle Scholar
  6. Asparouhov, T., & Muthén, B. (2010). Weighted least squares estimation with missing data. Mplus Technical Appendix, 2010, 1–10.Google Scholar
  7. Barker, E. T., Howard, A. L., Galambos, N. L., & Wrosch, C. (2016). Tracking affect and academic success across university: Happy students benefit from bouts of negative mood. Developmental Psychology, 52, 2022–2030.  https://doi.org/10.1037/dev0000231.CrossRefPubMedGoogle Scholar
  8. Beck, A. T., Steer, R. A., & Brown, G. K. (1996). Manual for the Beck Depression Inventory-II. San Antonio, TX: Psychological Corporation.Google Scholar
  9. Bouteyre, E., Maurel, M., & Bernaud, J.-L. (2006). Daily hassles and depressive symptoms among first year psychology students in France: The role of coping and social support. Stress and Health, 23, 93–99.  https://doi.org/10.1002/smi.1125.CrossRefGoogle Scholar
  10. Bureau of Labor Statistics U.S. Department of Labor (2014). College Enrollment and Work Activity of2013 High School Graduates. Economic News Release. http://www.bls.gov/news.release/hsgec.nr0.htm.
  11. Canadian Association of College & University Student Services and the Canadian Mental Health Association. (2013). Post-Secondary Student Mental Health: Guide to a Systemic Approach. Vancouver, BC: Author.Google Scholar
  12. Chambel, M. J., & Curral, L. (2005). Stress in academic life: Work characteristics as predictors of student well-being and performance. Applied Psychology: An International Review, 54, 135–147.  https://doi.org/10.1111/j.1464-0597.2005.00200.x.CrossRefGoogle Scholar
  13. Cohen, S., Kamark, T., & Mermelstein, R. (1983). A global measure of perceived stress. Journal of Health and Social Behavior, 24, 385–396.  https://doi.org/10.2307/2136404.CrossRefPubMedGoogle Scholar
  14. Compas, B. E. (2004). Processes of risk and resilience during adolescence: Linking context and individuals. In R. M. Lerner, & L. Steinberg (Eds.), Handbook of adolescence (2nd ed., pp. 263–296). Hoboken, NJ: Wiley.Google Scholar
  15. Cooke, R., Bewick, B. M., Barkham, M., Bradley, M., & Audin, K. (2006). Measuring, monitoring and managing the psychological well-being of first year university students. British Journal of Guidance and Counselling, 34, 505–517.  https://doi.org/10.1080/03069880600942624.CrossRefGoogle Scholar
  16. Curran, P. J., & Bauer, D. J. (2011). The disaggregation of within-person and between-person effects in longitudinal models of change. Annual Review of Psychology, 62, 583–619.CrossRefPubMedPubMedCentralGoogle Scholar
  17. Curran, P. J., Lee, T., Howard, A. L., Lane, S., & MacCallum, R. (2012). Disaggregating within-person and between-person effects in multilevel and structural equation growth models. In J. R. Harring, & G. R. Hancock (Eds.), CILVR series on latent variable methodology. Advances in longitudinal methods in the social and behavioral sciences (pp. 217–253). Charlotte, NC: IAP Information Age Publishing.Google Scholar
  18. Eisenberg, D., Gollust, S. E., Golberstein, E., & Hefner, J. L. (2007). Prevalence and correlates of depression, anxiety, and suicidality among university students. American Journal of Orthopsychiatry, 77, 534–542.  https://doi.org/10.1037/0002-9432.77.4.534.CrossRefPubMedGoogle Scholar
  19. Enders, C. K., Mistler, S. A., & Keller, B. T. (2016). Multilevel multiple imputation: A review and evaluation of joint modeling and chained equations imputation. Psychological Methods, 21, 222–240.  https://doi.org/10.1037/met0000063.CrossRefPubMedGoogle Scholar
  20. Fisher, S., & Hood, B. (1987). The stress of the transition to university: A longitudinal study of psychological disturbance and vulnerability to homesickness. British Journal of Psychology, 78, 425–441.  https://doi.org/10.1111/j.2044-8295.1987.tb02260.x.CrossRefPubMedGoogle Scholar
  21. Fergusson, D. M., Boden, J. M., & Horwood, L. J. (2007). Recurrence of major depression in adolescence and early adulthood, and later mental health, educational and economic outcomes. British Journal of Psychiatry, 191, 335–342.  https://doi.org/10.1192/bjp.bp.107.036079.CrossRefPubMedGoogle Scholar
  22. Furr, S. R., Westefeld, J. S., McConnell, G. N., & Jenkins, J. M. (2001). Suicide and depression among college students: A decade later. Professional Psychology: Research and Practice, 32, 97–100.  https://doi.org/10.1037/0735-7028.32.1.97.CrossRefGoogle Scholar
  23. Galambos, N. L., Barker, E. T., & Krahn, H. J. (2006). Depression, self-esteem, and anger in emerging adulthood: Seven-year trajectories. Developmental Psychology, 42, 350–365.  https://doi.org/10.1037/0012-1649.42.2.350.CrossRefPubMedGoogle Scholar
  24. Galarneau, D., Marissette, R., & Usalcas, J. (2013). What has changed for young people in Canada? In Statistics Canada. http://www.statcan.gc.ca/pub/75-006-x/2013001/article/11847-eng.htm.
  25. Goffin, P. (2017). ‘We’re not a treatment facility’: The struggle for campuses to provide students mental health care. The Toronto Star. https://www.thestar.com/news/gta/2017/08/13/were-not-a-treatment-facility-the-struggle-for-campuses-to-provide-students-mental-health-care.html.
  26. Gosling, S. D., Vazire, S., Srivastava, S., & John, O. P. (2004). Should we trust web-based studies? A comparative analysis of six pre-conceptions about internet questionnaires. American Psychologist, 59, 93–104.  https://doi.org/10.1037/0003-066X.59.2.93.CrossRefPubMedGoogle Scholar
  27. Haldorsen, H., Bak, N. H., Dissing, A., & Petersson, B. (2014). Stress and symptoms of depression among medical students at the University of Copenhagen. Scandinavian Journal of Public Health, 42, 89–95.  https://doi.org/10.1177/1403494813503055.CrossRefPubMedGoogle Scholar
  28. Hammen, C. (2005). Stress and depression. Annual Review of Clinical Psychology, 1, 293–319.CrossRefPubMedGoogle Scholar
  29. Harmon-Jones, E., Gable, P. A., & Price, T. F. (2013). Does negative affect always narrow and positive affect always broaden the mind? Considering the influence of motivational intensity on cognitive scope. Current Directions in Psychological Science, 22, 301–307.  https://doi.org/10.1177/0963721413481353.CrossRefGoogle Scholar
  30. Hoffman, L., & Stawski, R. S. (2009). Persons as contexts: Evaluating between-person and within-person effects in longitudinal analysis. Research in Human Development, 6, 97–120.  https://doi.org/10.1080/15427600902911189.CrossRefGoogle Scholar
  31. Howard, A. L. (2015). Leveraging time-varying covariates to test within- and between-person effects and interactions in the multilevel linear model. Emerging Adulthood, 3, 400–412.  https://doi.org/10.1177/2167696815592726.CrossRefGoogle Scholar
  32. Howard, A. L., Galambos, N. L., & Krahn, H. J. (2010). Paths to success in young adulthood from mental health and life transitions in emerging adulthood. International Journal of Behavioral Development, 34, 538–546.  https://doi.org/10.1177/0165025410365803.CrossRefGoogle Scholar
  33. Ibrahim, A. K., Kelly, S. J., Adams, C. E., & Glazebrook, C. (2013). A systematic review of studies of depression prevalence in university students. Journal of Psychiatric Research, 47, 391–400.  https://doi.org/10.1037/t06165-000.CrossRefPubMedGoogle Scholar
  34. Kendler, K. S., Thornton, L. M., & Gardner, C. O. (2001). Genetic risk, number of previous depressive episodes, and stressful life events in predicting onset of major depression. American Journal of Psychiatry, 158, 582–586.  https://doi.org/10.1176/appi.ajp.158.4.582.CrossRefPubMedGoogle Scholar
  35. Lazarus, R. S., & Folkman, S. (1984). Stress, appraisal, and coping. New York, NY: Springer.Google Scholar
  36. Lunau, K. (2012). Campus crisis: the broken generation. Why so many of our best and brightest students report feeling hopeless, depressed, even suicidal. McLean’s Magazine. http://www.macleans.ca/news/canada/the-broken-generation/.
  37. Maroco, J., & Campos, J. A. D. B. (2012). Defining the student burnout construct: A structural analysis from three burnout inventories. Psychological Reports, 111, 814–830.  https://doi.org/10.2466/14.10.20.PR0.111.6.814-830.CrossRefPubMedGoogle Scholar
  38. McEwen, B. S. (2003). Mood disorders and allostatic load. Biological Psychiatry, 54, 200–207.CrossRefPubMedGoogle Scholar
  39. Mikolajczyk, R.T., Maxwell, A.E., Naydenova, V., Meier, S., & El Ansari, W. (2008). Depressive symptoms and perceived burdens related to being a student: Survey in three European countries. Clinical Practice and Epidemiology in Mental Health, 4.  https://doi.org/10.1186/1745-0179-4-19.
  40. Montgomery, M. J., & Côté, J. E. (2003). College as a transition to adulthood. In G. R. Adams & M. D. Berzonsky (Eds.), Blackwell handbook of adolescence (pp. 149–172). Oxford, UK: Blackwell.Google Scholar
  41. Newcomb-Anjo, S., Villemaire-Krajden, R., Takefman, K., & Barker, E. T. (2017). The unique associations of university experiences with depressive symptoms in emerging adulthood. Emerging Adulthood, 5, 75–80.  https://doi.org/10.1177/2167696816657233.CrossRefGoogle Scholar
  42. Oishi, S., Diener, E., & Lucas, R. E. (2007). The optimum level of well-being: Can people be too happy. Perspectives on Psychological Science, 2, 346–360.  https://doi.org/10.1111/j.1745-6916.2007.00048.x.CrossRefPubMedGoogle Scholar
  43. Othieno, C. J., Okoth, R. O., Peltzer, K., Pengpid, S., & Malla, L. O. (2014). Depression among university students in Kenya: Prevalence and sociodemographic correlates. Journal of Affective Disorders, 165, 120–125.  https://doi.org/10.1016/j.jad.2014.04.070.CrossRefPubMedGoogle Scholar
  44. Pancer, S. M., Pratt, M., Hunsberger, B., & Alisat, S. (2004). Bridging troubled waters: Helping students make the transition from high school to university. Guidance and Counselling, 9, 184–190.Google Scholar
  45. Pearson, C., Janz, T., & Ali, J. (2013). Mental and substance use disorders in Canada. In Health at a Glance, Statistics Canada. http://www.statcan.gc.ca/pub/82-624-x/2013001/article/11855-eng.htm.
  46. Radloff, L. S. (1977). The CES-D scale: A self-report depression scale or research in the general population. Applied Psychological Measurement, 1, 385–401.CrossRefGoogle Scholar
  47. Radloff, L. S. (1991). The use of the Center for Epidemiologic Studies Depression Scale in adolescents and young adults. Journal of Youth and Adolescence, 20, 149–166.  https://doi.org/10.1007/BF01537606.CrossRefPubMedGoogle Scholar
  48. Raudenbush, S. W., Bryk, A. S., Cheong, Y. F., Congdon, R. T., & Du Toit, M. (2011). HLM 7. Lincolnwood, IL: Scientific Software International Inc.Google Scholar
  49. Rohde, P., Lewinsohn, P. M., Klein, D. N., Seeley, J. R., & Gau, J. M. (2012). Key characteristics of major depressive disorder occurring in childhood, adolescence, emerging adulthood, and adulthood. Clinical Psychological Science, 1, 41–53.  https://doi.org/10.1177/2167702612457599.CrossRefGoogle Scholar
  50. Salmela-Aro, K., Kiuru, N., Pietikäinen, M., & Jokela, J. (2008). Does school matter? The role of school context in adolescents’ school-related burnout. European Psychologist, 13, 12–23.  https://doi.org/10.1027/1016-9040.13.1.12.CrossRefGoogle Scholar
  51. Santor, D. A., Zuroff, D. C., Ramsay, J. O., Cervantes, P., & Palacios, J. (1995). Examining scale discriminability in the BDI and CES-D as a function of depressive severity. Psychological Assessment, 7, 131–139.  https://doi.org/10.1037/1040-3590.7.2.131.CrossRefGoogle Scholar
  52. Sha, T., & Xia, L. (2004). Mediation effect of cognitive appraisal orientation on the relationship between stress and negative feelings in university students. Chinese Mental Health Journal, 18, 107–110.Google Scholar
  53. Shamsuddin, K., Fadzil, F., Wan Ismail, W. S., Shah, S. A., Omar, K., Muhammad, N. A., Jaffar, A., Ismail, A., & Mahadevan, R. (2013). Correlates of depression, anxiety and stress among Malaysian university students. Asian Journal of Psychiatry, 6, 318–323.  https://doi.org/10.1016/j.ajp.2013.01.014.CrossRefPubMedGoogle Scholar
  54. Schulenberg, J. E., & Zarrett, N. R. (2006). Mental health during emerging adulthood: Continuity and discontinuity in courses, causes, and functions. In J. J. Arnett & J. Tanner (Eds.), Emerging Adults in America: Coming of Age in the 21st Century (pp. 135–72). Washington, DC: American Psychological Association.CrossRefGoogle Scholar
  55. Twenge, J. M., Gentile, B., DeWall, C. N., Ma, D., Lacefield, K., & Schurtz, D. R. (2010). Birth cohort increases in psychopathology among young Americans, 1938–2007: A cross-temporal meta-analysis of the MMPI. Clinical Psychology Review, 30, 145–154.  https://doi.org/10.1016/j.cpr.2009.10.005.CrossRefPubMedGoogle Scholar
  56. Wrosch, C., & Miller, G. E. (2009). Depressive symptoms can be useful: Self-regulatory and emotional benefits of dysphoric mood in adolescence. Journal of Personality and Social Psychology, 96, 1181–1190.  https://doi.org/10.1037/a0015172.CrossRefPubMedGoogle Scholar

Copyright information

© Springer Science+Business Media, LLC, part of Springer Nature 2018

Authors and Affiliations

  • Erin T. Barker
    • 1
  • Andrea L. Howard
    • 2
  • Rosanne Villemaire-Krajden
    • 1
  • Nancy L. Galambos
    • 3
  1. 1.Department of PsychologyConcordia UniversityMontrealCanada
  2. 2.Department of PsychologyCarleton UniversityOttawaCanada
  3. 3.Department of PsychologyUniversity of AlbertaEdmontonCanada

Personalised recommendations