3.6. Data Analyses
The data analyses included four stages. In the first stage, descriptive statistics were employed to calculate the means and standard deviations of the continuous variables and the percentage and frequency of the categorical variables. In the second stage, in order to test the first hypothesis regarding the difference in subjective age before and during the COVID-19 pandemic, a paired t-test was used. In the third stage, bivariate analyses were performed to examine the association between subjective age and the independent variable, mediator variables, and socioeconomic variables using an independent t-test, one-way ANOVA, and Pearson or Spearman correlation tests.
In the fourth stage, mediation analyses were then computed, and the selected mediators (depressive symptoms and malnutrition) were entered to test the components of the mediation model (Model 4). The bootstrapping method was used to assess the indirect effects of the mediation model [62,63]. Thus, the mediation model was examined by directly testing the significance of the indirect effect of the independent variable (IV; feelings of loneliness) on the dependent variable (DV; subjective age) through the mediators (MeV; depressive symptoms and malnutrition), while controlling for background variables that had been identified in the bivariate analyses as significant.
This method is based on regression analysis and calculates the direct effect (weight C’, with a mediator), total effect (C, without mediator) and indirect effects (a × b weights) of an independent variable on a dependent variable. The total and specific indirect effects were calculated through bootstrapping, set at 5000 samples. Confidence intervals were calculated using this method by sorting the lowest to highest of these samples, yielding a 95-percentile confidence interval (if the number 0 falls within the confidence intervals, the tested effect is nonsignificant).
All analyses were run using SPSS 25.0 with the PROCESS statistical program [62]. All estimated effects reported by PROCESS are unstandardized regression coefficients.