Sample characteristics
A recruitment response target of participants of 2000 was the aim for the baseline survey, with the total response rate at 2501. After several exclusions were applied concerning data quality control (please see inclusion criteria), a final effective sample of 1989 participants from Great Britain and Northern Ireland aged between 18 and 87 (M = 37.11, SD = 12.86) were eligible to be included in baseline data. Overall, the sample was mostly female (70.4%) and white (92.7%). Overall, 30.3% of the sample met the threshold criteria for GAD, mean score on GAD-7 scale, 7.27 (SD = 6.02), while 34% met the criteria for MDD, mean score on PHQ-9 scale, 7.84 (SD = 6.48). The remaining sample characteristics are presented in Table 1.
Table 1 Social demographic, mental health variables, media consumption by GAD-7 and PHQ-9 scores Factors for increasing MDD and GAD scores
Univariate analysis
The results demonstrated that as media usage increased, so too did scores on the GAD-7 and MDD (Table 1). Both traditional and social media consumption variables were statistically significant and positively correlated with GAD-7 and PHQ-9 scales. GAD-7 and PHQ-9 scales were highly and positively correlated with each other (Table 2).
Table 2 Correlations for study variables Results from an independent t-test demonstrated there was a significant difference in GAD scores for low social media consumption (M = 6.15, SD = 5.56) and high social media consumption (M = 9.98, SD = 6.57; t(284.18) = −8.12, p < 0.001). The magnitude of differences in the means (mean difference = −3.83, 95% CI = − 4.76 to −2.90) was large (Cohen’s d = 0.63). Similarly, there was a significant difference in GAD scores for low traditional media consumption (M = 6.91, SD = 5.90) and high traditional media consumption (M = 10.08, SD = 7.08); t(82.58) = −3.88, p = 0.00). The magnitude of differences in the means (mean difference = −3.17, 95% CI = − 4.79 to −1.54) was medium (Cohen’s d = 0.49). There was no significant difference in GAD scores (p > 0.05) for keyworker status (p = 0.73) or living status (p = 0.42).
Statistically significant differences were shown in MDD scores for low traditional media consumption (M = 7.45, SD = 6.32) and high traditional media consumption (M = 10.86, SD = 7.37; t(82.94) = −4.02, p < 0.001). The magnitude of differences in the means (mean difference = −3.41, 95% CI = − 5.10 to −1.72) was medium (Cohen’s d = 0.49). There was a significant difference in MDD scores for low social media consumption (M = 6.78, SD = 6.10) and high social media consumption (M = 10.56, SD = 6.97; t(285.64) = −7.49, p < 0.001). The magnitude of differences in the means (mean difference = −3.78, 95% CI = − 4.77 to −2.79) was large (Cohen’s d = 0.58). There was no significant difference in MDD scores (p > 0.05) for keyworker status (p = 0.30) or living status (p = 0.07).
Multivariate analysis
A statistically significant difference between all social media consumption categories (low, moderate and high) was determined by a one-way ANOVA (F(2, 1967) = 55.98, p < 0.001). A tukey post hoc test revealed that total GAD scores were statistically significantly higher with high social media consumption (M = 10.08, SD = 7.08) when compared to low (M = 6.91, SD = 5.90) or moderate (M = 8.14, SD = 6.04) social media consumption. Traditional media consumption categories also revealed a statistically significant difference with total GAD scores between which was determined by a one-way ANOVA (F(2, 1967) = 15.37, p < 0.001). A tukey post hoc test revealed that total GAD scores were statistically significantly higher with higher traditional media consumption (M = 10.08, SD = 7.08) when compared to low (M = 6.91, SD = 5.90) or moderate (8.14, 6.04) traditional media consumption.
MDD scores also demonstrated a statistically significant difference between all traditional media consumption categories (low, moderate and high), determined by a one-way ANOVA (F(2, 1962) = 43.45, p < 0.001). A tukey post hoc test revealed that total MDD scores were significantly higher with high traditional media consumption (M = 10.86, SD = 7.37) when compared to low (M = 7.45, SD = 6.32) or moderate (M = 8.80, SD = 6.66) traditional media consumption. Similarly, a statistically significant difference between all social media consumption groups (low, moderate and high) was determined by a one-way ANOVA (F(2, 1962) = 43.45, p < 0.001). A tukey post hoc test revealed that total MDD scores were significantly higher with high social media consumption (M = 10.56, SD = 6.97) when compared to low (M = 6.78, SD = 6.10) or moderate (M = 8.84, SD = 6.56) social media consumption.
A total of 4 of the included variables that may be influencing factors were significant (≤0.05) in the univariate analysis and were entered into a standard logistic regression. The living alone and keyworker status variables were excluded from the analysis as no significant differences were reported, p > 0.05. The final model for the various predictor variables is presented in Table 3.
Table 3 Logistic regression analysis of factors predicting meeting GAD or MDD clinical threshold criteria The logistic model was statistically significant, x2 (4) = 128.31, p < 0.001. The model explained 9.0% of the variance in meeting the clinical threshold for GAD and correctly classified 70.6% of cases. Females were 42.9 times more likely to meet the clinical threshold criteria for GAD than males. Increasing age was associated with an increased likelihood of meeting the clinical threshold criteria for GAD. Both high traditional and social media consumption variables were linked to increased GAD scores and an increased likelihood of meeting the clinical threshold criteria for GAD.
Similarly, the logistic model for MDD was statistically significant x2 (4) = 132.92, p < 0.001. The model explained 9.1% of the variance in meeting the clinical threshold for MDD and correctly classified 67.8% of cases. Females were 12.03 times more likely to meet the clinical threshold criteria for MDD than males. Lower age was associated with a lower likelihood of meeting the clinical threshold criteria for MDD. Both high traditional and social media consumption variables were linked to increased MDD scores and an increased likelihood of meeting the clinical threshold criteria for MDD.