I'm looking particularly in the context of data annotation, but any scientific sources would be helpful really.
The following options I can think of:
- The potential options (annotation labels) in the dropdown menu are fixed, i.e. are always at the same order. New elements might be inserted at the bottom or throughout. The assumed benefit would be that the user has an easier time memorising the layout for lengthy tasks. Might result in a lot of scrolling when many options are available.
- The options are ordered by usage, i.e. the most used option moves to the top. Quicker access for often needed items.
- A recommendation system orders the options based on the predicted likelihood of each option in the current instance. Might speed things up, but could also be frustrating for the user.
Is there any research into the qualitative and/or quantitative impact comparing these or other forms of ordering?