Educational Measurement: Issues and PracticeVolume 36, Issue 3 p. 4-4 Data VisualizationFree Access On the Cover First published: 18 September 2017 https://doi.org/10.1111/emip.12167AboutSectionsPDF ToolsRequest permissionExport citationAdd to favoritesTrack citation ShareShare Give accessShare full text accessShare full-text accessPlease review our Terms and Conditions of Use and check box below to share full-text version of article.I have read and accept the Wiley Online Library Terms and Conditions of UseShareable LinkUse the link below to share a full-text version of this article with your friends and colleagues. Learn more.Copy URL Share a linkShare onFacebookTwitterLinkedInRedditWechat A Visualization of Test Takers’ Reading Skills by Item-Level Clusters, designed by Yuyu Fan of Fordham University and Joseph H. Grochowalski of the College Board, graces this issue's cover. This richly detailed visual depicts examinees’ reading abilities across the multiple dimensions of a large-scale measure of reading ability. These creative researchers described their work this way. This visual answers the following question: “Based on examinees’ performance on 40+ reading items, is it possible to visually identify clusters of examinees that have different reading skills?” The plot answers the question by showing three panels: the outer two visualize different views of a cluster analysis, and the middle panel visualizes differences in skills for the identified clusters.The plots show test takers’ strengths and weaknesses in their reading abilities for a large-scale assessment. The outer panels show different views of low-dimensional clustering of n = 4,930 test takers’ performance on 40+ reading items, and the middle panel represents reading skill profiles of eight clusters of test takers.The outer panels are visual representations of n = 4,930 test takers’ performances on reading items of a large-scale assessment. With more than 40 items, it is typically difficult to visualize test takers’ individual performances, especially when attempting to identify differences in subgroups of test takers by skill. Using an innovative approach, the researchers first reduced the data into three dimensions using T-distributed stochastic neighbor embedding (T-SNE), a relatively new machine learning technique that facilitates visualization of high-dimensional data in low dimensions. Then they applied model-based cluster analysis based on Gaussian mixture modeling, and classified all test takers into eight clusters based on item-level performance. Finally, they generated an interactive 3D plot using the R plotly package (R Core Team, 2016). The outer panels contain four different angles of the 3D plot.The middle panel shows the skill profiles of the eight clusters. Content experts theorized three primary skills for reading, and then identified which skill is primarily measured by each item. We calculated each cluster's mean proportion of correct responses for each skill and then plotted the cluster profiles. The eight profiles have various patterns, but the major difference lies in the elevation/level, which indicates that the three skills are highly correlated and compensatory.The visualization confirmed that the total reading score was a sufficient metric for understanding examinee achievement, and that clusters of similar-performing students were mainly differentiated by their “level” of performance and not by individual strengths or weaknesses in particular skills. If you are interested in learning more about how these informative visualizations were created, contact the authors: Yuyu Fan (yfan2@fordham.edu) or Joseph H. Grochowalski (jgrochowalski@collegeboard.org). Reference R Core Team. (2016). R: A language and environment for statistical computing. Vienna, Austria: R Foundation for Statistical Computing. https://www.R-project.org/package=plotly Volume36, Issue3Fall 2017Pages 4-4 ReferencesRelatedInformation