Page 42 - IJSA, Vol. 4, No 2, 2021

International Journal of Science Annals, Vol. 4, No. 2, 2021
рrint ISSN: 2617-2682; online ISSN: 2707-3637; DOI:10.26697/ijsa

dimension (McElroy et al., 2019; Norr et al, 2015). The respectively, for each construct (Table 2). Finally tests
final list contains 12 items only. of discriminant validity were also acceptable as the
The present study uses the new scale of the Fear of squared correlation between every two constructs in the
COVID-19 (FC-19S) developed recently by Ahorsu et model was less than the AVE of these two constructs
al. (2020). This scale has a stable unidimensional (Fornell & Larcker, 1981).
structure with good psychometric properties. Initial Table 2
results indicated that the FCV-19S had good properties Reliability and Convergent Validity Test
from different types of testing (i.e., Rasch analysis). The Convergent
internal consistency was good as well (α=0.82; Variable Composite validity
composite reliability=0.88). reliability (AVE)
We measured six hedonic shopping motivations: Excessiveness 0.87 0.761
gratification seeking (GRA), idea shopping (IDE),
adventure seeking (ADV), social shopping (SOC), role Compulsiveness 0.93 0.866
play (ROL), and value shopping (VAL), based on Distress 0.89 0.804
Arnold and Reynolds (2003; 2012). This measurement Reassurance 0.91 0.821
was validated in several contexts (Ali et al., 2020; Adventure 0.84 0.717
Horváth & Adıgüzel, 2018). It contains 18 items; three
to each dimension. Gratitude 0.84 0.751
The online shopping enjoyment was measured by a scale Role 0.90 0.766
developed by Babin et al. (1994). This unidimensional Value 0.92 0.813
scale was used in several contexts and had demonstrated Social 0.93 0.878
its validity and its good reliability (α=0.872).
We conducted confirmatory factor analysis (CFAs) and Idea 0.88 0.831
a structural equation model (SEM) was examined. In Fear 0.94 0.689
this model, cyberchondria factors were considered as Shopping
predictors of six hedonic motivations, the fear of enjoyment 0.95 0.816
COVID-19 and the online shopping enjoyment. The
CFAs and SEM were conducted in AMOS 23 with We also tested whether CMB was a potential problem
Maximum of Likelihood (ML) as an estimation method. following a procedure suggested by Podsakoff et al.
We verified first the conditions of the applicability of (2003). A latent construct capturing the common
those methods by verifying the outliers, the method variance was added to the measurement model
multinormality and the multicollinearity. Bootstrapping and allowed to load on all of the indicators. Results
was envisaged as a procedure for dealing with non- indicate that CMB is not a potential threat to the validity
normal data (Byrne, 2001), generated confidence of the findings.
intervals that were used instead of t-values to evaluate We used AMOS 23 to test our research hypotheses. We
the significance of path estimates because the data did tested four alternative models to assess the superiority of
not exhibit multivariate normality. Model fit was then our model as we followed prior procedures in well-
2
assessed using the χ statistic, for which a non- established works (Bajaj et al., 2016; Diallo & Seck,
significant value indicates good model fit. In addition, 2018): the hypothesized structural model, a direct
other fit indices were interpreted to provide an model, and two full mediation models. The
approximate estimate of the model fit. Specifically, we hypothesized structural model (Model 1) fit the data
checked the comparative fit index (CFI), the Tucker- well overall (χ =2929.480; df=706; p=0.000;
2
Lewis Index (TLI), and the root mean square error of NFI=0.824; IFI=0.860; TLI=0.836; and CFI=0.859;
approximation (RMSEA). RMSEA=0.094).
For the direct effect of cyberchondria on online
Results shopping enjoyment (Model 2), results show that three
First, our results indicate that more than 70% of the of the four cyberchondria dimensions were positively
sample reported experiencing high or moderate levels of affecting online shopping enjoyment. The health-related
cyberchondria. Mean item analysis showed that going anxiety that cyberchondriac express influence positively
online to search for symptoms often disrupts their time their online shopping experiences. This could reinforce
spent not only on leisure activities but also on their the hypothesis supposing that during this pandemic,
work. Participants showed difficulty in controlling their online shopping is considered as one of the main
ruminations regarding symptoms that they have solutions for anxious people. Results of testing the
researched online. second hypothesis (H2a-d) are presented in Model 2
The data indicated that the measurement model (Figure 2).
2
exhibited an acceptable fit (χ =2857.269; df=699; For mediating effect of hedonic shopping motivations
p=0.000; CFI=0.863; NFI=0.828; TLI=0.840; (Model 3), we followed current trends in mediation
RMSEA=0.093). Tests of convergent validity were analysis (Preacher et al., 2007; Zhao et al., 2010) and
acceptable as the composite reliability and the average examined the direct and indirect effects of
variance extracted (AVE) both exceeded the cyberchondria through conditional process analysis.
recommended minimum cutoff of 0.7 and 0.5, More precisely, in a third model, we tested a full

40