Cross-hairs: a scatterplot for meta-analysis in R Michael T. Brannick a * and Mehmet Gültaş b We describe a meta-analytic scatterplot that indicates precision of points for two variables paired within studies; this is equivalent in form to a ‘cross-hairs’ plot used to portray specificity and sensitivity in diagnostic testing. At the user’s discretion, the plot also displays boxplots for each of the X and Y variable distributions, means for each of the variables, and the correlation between the two. The cross-hairs may be suppressed for dense point clouds. The program is written in R, so it can be modified by the user and can serve as a companion to existing meta-analysis programs. Some of the program’s novel uses are described and illustrated with (1) independent effect sizes, (2) dependent effect sizes, and (3) shrunken estimates. Copyright © 2016 John Wiley & Sons, Ltd. Keywords: graph; plot; shrunken estimates; visualization 1. Graphs in meta-analysis Graphs are an excellent way to represent data in a meta-analysis (Anzures-Cabrera and Higgins, 2010; Borenstein et al., 2009). Despite the value of graphs for providing insight, several kinds of graphs are rarely used, and graphs are more likely to be used in medical journals than in social science journals, at least for psychology and business (Schild and Voracek, 2013). In their primer, Anzures-Cabrera and Higgins (2010) noted four main types of graphs used in meta-analysis: forest plots, funnel plots, Galbraith plots, and L’Abbé plots. Forest plots appear to be the most commonly published (Schild and Voracek, 2013). Effect sizes in meta-analysis usually have different standard errors (within-studies variance) and often display heterogeneity (between-studies variance). The forest plot uses ‘whiskers’ to communicate information about the standard error of each effect size (typically whiskers extend 1.96 standard errors to either side of a central point). The forest plot uses box sizes to provide information about the study weights in calculating a weighted average or summary effect size. The forest plot may include a prediction interval (Anzures-Cabrera and Higgins, 2010; Borenstein et al., 2009) to indicate a likely range of infinite sample effect sizes and thus illustrates the implications of heterogeneity. However, even excellent graphs such as the forest plot can be improved to help users interpret their data. For example, the rainforest plot, a modification of the forest plot that illustrates the assumed probability density of each effect size, helps the user to better interpret the meta-analytic data (Schild and Voracek, 2014). At least two additional types of visual representations have appeared in the meta-analytic literature, ‘cross- hairs’ plots (Gatsonis and Paliwal, 2006; Phillips et al., 2010) and effect direction plots (Thompson and Thomas, 2012). The effect direction plot is intended for situations in which several of the studies in the systematic review lack confidence intervals. In such situations, a forest plot will exclude much of the information contained in the review. The cross-hairs plot was introduced to present information about the sensitivity and specificity of a test in a single graph. For each study, the estimate of specificity will be plotted on one axis, and the estimate of sensitivity will be plotted on the other. The error of each estimate will be portrayed by ‘whiskers’ so that the uncertainty for each estimate is included. There is often a trade-off between sensitivity and specificity, and such a trade-off is easier to see in a graph than in a table. a Psychology Department, University of South Florida, Tampa, FL 33620, USA b Department of Psychology, Middle East Technical University, 06800, Ankara, Turkey *Correspondence to: Michael T. Brannick, Psychology Department, University of South Florida, Tampa, FL 33620-7200, USA. E-mail: mbrannick@usf.edu Copyright © 2016 John Wiley & Sons, Ltd. Res. Syn. Meth. 2016 1–11 Method Note Received 30 July 2015, Revised 03 July 2016, Accepted 11 July 2016 Published online in Wiley Online Library (wileyonlinelibrary.com) DOI: 10.1002/jrsm.1220