When using chimpsky, it is useful to be aware of its limitations.
Chimpsky imposes some limits on usage to protect its server from overload. In particular, each user is limited to 5 tasks and 500 files overall. No file can have more than 5000 words; files longer than that will be listed as 'failed uploads'.
Chimpsky is inherently dumb and can only detect matches of text that are strictly identical for a certain minimum number of words. If such a match is found, it is advisable to also consider the similarity of the adjacent paragraphs. It will often become apparent a student spent his time just slightly rephrasing somebody else's work rather than putting in the time and effort to do it independently. It is of course up to you to judge the significance of such similarity.
The tree view of comparison results, may be entertaining (or so I find) and useful for a quick overview, but its accuracy is decidedly limited. The underlying algorithm will shoehorn any set of results, however heterogeneous, into a single tree, and it therefore sometimes suggests similarity where none exists. If there is a discrepancy between the tree and the tables, disregard the tree.
The Google search function only covers the web, not books. While it would technically be easy to also include it, the Google book search does not return any sample text with its results, so that it is not easily possible to evaluate the significance of these results.