Fig 1.
Timing of social network and mental health data collection, and empirical comparisons.
Students were enrolled in one of two study majors and one of two year groups.
Table 1.
Demographics of each sample.
Fig 2.
Numbers of reported ties and isolation.
Students reported fewer pleasant interactions and fewer studying partners during lockdown than prior (within-subjects comparisons, panel a and b). Comparing the subset from one major to the previous year’s cohort of that major at the same time one year earlier (between-cohort comparison, panel c and d), outdegrees in the same two networks were significantly lower. Lines indicate one standard error above and below the mean estimate. *** p < .001; * p < .05.
Fig 3.
Social interaction networks between students during the COVID-19 crisis (April 2020).
(a) Few face-to-face interactions were reported (pink ties). (b) Interaction networks through digital communication technologies were dense (social media: yellow ties, messaging apps: red, video and voice calls: brown, other pleasant interaction: blue). The Venn diagram indicates relative density and overlap between networks. Data from the current-year cohort, Major II (N = 294. This includes, in addition to all respondents to the April 2020 survey, all students enrolled in Major II who were nominated as connections by those who filled out the survey).
Fig 4.
Proportion of stable (a) interaction and (b) co-study ties since September 2019 by the number of existing other network dimensions in September 2019.
Dashed horizontal lines represents the network’s proportion of stable ties. Error bars indicate 95% bootstrapped Confidence Intervals.
Fig 5.
Specific COVID-19 stressor items and their distributions.
N = 277, all current-year, April 2020 participants. Frequency distributions and t-test results (H0: M = 0) of each COVID-19-related stressor item. Rated on a 7-point scale (green = much less than before (-3), yellow = neutral (0), red = much more than before (+3)).
Fig 6.
Mean scores and standard errors of the (a) depression, (b) anxiety, (c) stress, and (d) loneliness scale by measurement time-point (x-axis).
The scales theoretically ranged between 0–60 (a), 0–21 (b), 0–40 (c), and 1–4 (d). The significance levels of paired t-tests of previous time-points with the COVID-19 wave (April 2020) are indicated by asterisks, *** p < .001; ** p < .01; * p < .05; n.s. = not significant.
Table 2.
Results of backwards-selected regression models predicting changes in depression, anxiety, stress, and loneliness between September 2019 and April 2020.