Inferences from Mouse Movement
September 29th, 2006
Over on the SIGIA-L list (a nifty discussion list on all things information architecture), there’s been a discussion about the use of rollovers. Not to miss an opportunity to remind people about something out of our research, I shared the following thoughts:
If you watch users long enough, you’ll notice that they rarely move their mouse until *after* they’ve decided what they want to click on. So, if the rollover contains any information they would need to make the right choice, it’s likely they’ll choose wrong.
Test two designs: one with rollovers and one without (with the same content always visible on the page). Almost always, the users will perform and prefer the latter.
More info: Users Decide First; Move Second
Andrew Boyd from Humane Information Architecture responded:
I’ve spent a lot of time watching people use web interfaces, and some of them definitely do use the mouse pointer as a guide when they are browsing around the page – they do look for rollovers and tooltips prior to making a click decision. Not all of them, but enough so that meaningful tooltips are worthwhile. On three separate occasions in the last week-and-a-half on the road watching other people test a design I have contributed to, we’ve had comments to the effect that the tooltips were not meaningful enough. This is all anecdotal, of course, but I’ve seen people move the mouse prior to making click decisions (rollovers or no).
I agree you’ll see some people move their mouse. However, It’s important we don’t confuse observation with inference.
Andrew’s observation (based on what he wrote) is he’s seen users move the mouse over items to explore them. He’s also observed users commenting that they felt tooltips weren’t meaningful enough.
Andrew’s inference (also based on what he wrote) seems to be he believes the problem is the tooltips need improvement and this would benefit those users who explore with their mouse.
I don’t disagree with Andrew’s observations. I wasn’t there and have no way of knowing what users did or didn’t do. Plus, I too have seen some users move their mouse around the screen, waiting for the browser to give them the “finger” and looking for additional clues.
However, I do think his inference (as I interpreted it) could be faulty. From what he wrote, I don’t think he asked enough questions:
- Why were the users looking for rollovers and tooltips in the first place?
- Do those same users always look for rollovers and tooltips regardless of the page they’re on?
- Or is it the function of the page’s design that suggests they should do that?
- If you increased the effectiveness of the on-screen design, would it eliminate those users’ desire to move their mouse?
- What happens for those users who need to scroll and the mouse is needed to control the scroll bar?
- Do they have more problems making the right choices than users who end up moving their mouse?
It could be that his users are more likely to move their mouse (thereby suggesting he invest more in improving tooltips and rollovers).
It could also be that all users are more likely to move their mouse on his page, because his design demands it to succeed (thereby suggesting he invest more in the on-screen page design and reduce the need for tooltips and rollovers).
What observations could Andrew have collected to figure out which it is? How could he test to see which is more likely?
It’s important we ask the right questions before we draw our inferences from the observations. Different inferences will lead us to different design solutions. We need to ensure our inferences are leading us in the right direction.
