Meta-Analysis of Correlations between Altmetric Attention Score and Citations in Health Sciences

Figure 7

Funnel plots with nonparametric trim-and-fill analysis of publication bias for two groups which involved publication bias comprising all included study groups (a) and the most cited group (b). The linear estimator was used to estimate the number of missing studies. The observed effect size for all included studies was 0.192 (95% confidence interval 0.128 to 0.255), and the observed + imputed effect size was 0.186 (95% confidence interval 0.122 to 0.250) (a). The observed effect size for the most cited group was 0.233 (95% confidence interval 0.127 to 0.340), and the observed + imputed effect size was 0.220 (95% confidence interval 0.112 to 0.327) (b). Please see Figure 6 for more details about publication bias assessments (REML: restricted maximum-likelihood).

(a)

(b)