Graphical display of data

“A picture is worth thousand words”- this analogy aptly describes the scenario in statistical analysis of research data. Graphical display of key statistical findings certainly optimizes the interpretative ease of result outputs. Commonly cited graphs include bar diagram, histogram and pie chart. Box plot provides five-point summary of data (median, 1st and 3rd quartile values, minimum and maximum values) and error bars add interval estimates to point estimates of mean or median. Specialized graphs include receiver operating characteristic (ROC) curve, Kaplan Meier survival or cumulative incidence curves and marginal probability plots in regression analysis. In meta-analysis, key graphs are funnel plot (for publication bias), forest plots (for effect size measures) and Galbraith plot (for assessing heterogeneity in meta-analysis).