The article does a user testing on the usability of fisheye menus and reveals in terms of text menu they are worse than normal menus; and the best menu is the hierarchical menus.
Fisheye menus are those which when you pass the mouse over an item, this item is focused showing it more bigger than actually is and the other elements are reduced in size. For example [taken from: http://www.cs.umd.edu/class/fall2002/cmsc838s/tichi/fisheye.html]:
Some points are interesting of the research:
However, the level of magnification had significant effect: Participants performed faster at a magnification level of two twice normal size) than at both lower and higher levels of magnification. Also, the three fisheye interfaces were significantly faster than two overview+detail ones.
Scanpath length, … Bonferroni-adjusted post hoc tests indicated that participants’ scanpaths were shorter with the hierarchical
menu than with the multifocus and overview menus, suggesting that participants engaged in less visual search with the former. [...] Bonferroni-adjusted post hoc tests indicated that fixations may be longer with the fisheye menu than with hierarchical and multifocus menus.
Participants’ selection times with fisheye, multifocus, and overview menus show a bell-shaped distribution, meaning that items at the beginning and end of the menus were selected much faster than those in the middle [...]. Linear contrasts indicated that the first 10% of the menu items were selected the fastest, followed by the last 10%, and finally, the selection of middle items . Fast selection of the first menu items is expectable. Fast selection of the last items may be attributable to the fact that these items are in an area that is easy to identify; selecting the menu item “Zen” with the alphabetical dataset is almost certain to involve some of the last menu items, facilitating a decision to make a fast coarse movement toward the end of the menu.
One explanation is that while participants using the hierarchical menu take more time getting close to the target menu item, they take substantially less time selecting it. Getting close to a menu item is slow with the hierarchical menu because of the changes required in the direction of mouse movement, the small delay before menu items at lower levels expand, and the laborious backtracking necessary to correct erroneous selections.
A final explanation for why participants perform well with the hierarchical menu is that it simplifies navigation.With fisheye and overview menus, participants made longer fixations, suggesting increased mental activity [Goldberg and Kotval 1999], compared with the hierarchical menu. Also, participants’ scanpaths were longer with the multifocus and overview menus, indicating
more visual search. Reasons for this could include: (a) the need with nonhierarchical menus to determine or remember which part of the menu structure one is currently in; and (b) difficulties operating the focus-lock mode. While these explanations are somewhat speculative, they are supported by participants’ comments after the experiment and by their usage patterns.
I think fisheye menus should be used only on graphical elements like youtube’s menu at the end of embedded videos.