Three experiments were conducted to investigate whether graphic organizers (GOs) were more effective than concept maps (CMs) in facilitating text learning. Experiment 1, which involved 56 undergraduates, was a replication of the second experiment of D. H. Robinson and K. A. Kiewra (1995) except that only GO and CM conditions were included. All materials were constructed by outside sources to help avoid researcher bias. Experiment 2, with 46 undergraduates, used a different text and separate GO, CM, and test constructors. Experiment 3 replicated Experiment 2 except that the GO constructor constructed the tests. Results show that GOs only facilitated text learning when they were constructed by the person who also constructed the tests. The implications of this study are that adjunct displays will probably not be useful for classroom learning unless the tests match the types of learning the displays are intended to facilitate. (Contains 15 references.) (Author/SLD)