3. Linear regression analysis
The correlation coefficient (R) and the determination coefficient (R2)
are equal to 0.645 and 0.414, respectively. Thus, a positive correlation
exists, and the statistical model explains more than 40% of the
variation in the number of 2018 WoS citations obtained for TE articles
from 2015 Altmetrics scores. Furthermore, the result of the analysis of
variance (ANOVA) test is significative for P < .001.
The number of Mendeley readers constitutes the best citation predictor
as it holds a higher correlation coefficient with R = 0.599. The rest of
the altmetrics scores have a positive correlation coefficient; however,
the association is weaker.
The prediction equation for 2018 TE WoS citations counts from 2015
altmetrics scores can be expressed as follows:
Ln (1 + WoS) = -27.25 + 5.37 x Ln (1 + Blog) + 0.82 x Ln(1 + News) +
12.78 x Ln(1 + Mendeley readers) + 5.83 x Ln(1 + Patent) + 0.75 x Ln(1 +
Twitter)