I can, perhaps, make what I mean by the term “structure” more apparent by focusing on a specific concrete example. More than two years ago, David Reinking published a hypertext in Reading Online (Reinking, 1997, online document). His article consisted of 29 nodes (i.e., pages) linked together into a larger structure. After carrying out an analysis of the connectedness of the pages in his hypertext I used a graph-drawing program (Ellson, Koutsofios, & North, 1998; downloadable open source software) to create the image on the right that essentially “maps” the macrostructure of the article as reflected in its links.

The map has a number of interesting aspects. Node 20, for instance, emerges as a clear central node that serves to organize several other node clusters. Note also that the two clusters at the bottom left and right are completely unconnected except for the bridge provided at Node 20 and other indirect routes. Structural maps like this one also turn out to be fairly useful as navigational devices, so they are often included in hypertext documents as navigational aids. As I recall, only a portion of David's “hypertext map” could actually be viewed.

Research I have carried out recently shows that user movement in hypertext can be displayed in a similar fashion, and that the structure of movement in hypertext is a better predictor of effective hypertext use than number of pages viewed, print reading ability, and even prior reported experience using hypertext reading environments (McEneaney, 1998 [online document], 1999 [online document]).