XTREE: A Multivariate Graphical Icon Applicable in the Evaluation of Statistical Estimators

Both computer simulations and analytical research are often used to compare statistical estimators.A common structure (e.g., Andrews, Bickel, Hampel, Huber, Rogers, and Tukey 1972) is to compare a number of estimators for a number of distributions under a number of side conditions (i.e., various sample sizes).The investigator is faced with a substantial problem in trying to interpret voluminous results.The problem is exacerbated if there are multiple dependent variables (e.g., variance and bias).In this article we propose a graphical icon, the XTREE, which aids perception and interpretation of such results.The method is illustrated with data from three recently published studies.