This is something I've been confused ever since I started my undergraduate research.

My research related to User Interfaces. Based on some of the previous years undergraduate research or from some research in the same year as me, as far as I see, have these steps:

  1. Suppose we wants to design new interface for a system-X
  2. We do the survey about the interface of system-like-X (but not X) to the users
  3. From the survey, we get the data: what the users don't like from the present system interface, which part should be improved, etc
  4. Then, we propose the new design based on that data
  5. We then, again, test that new design with users

And now, my steps, according to my advisor

  1. You design the new interface
  2. Test it

I'm so confused, because, based on what would I design the interface? What standard or principle I should used? Also, in my case, I have to implement or develop the testing tools myself, which means that I got more work to do compared to my friend's workload. Could that be why step 1-3 skipped? But this is academic, I can't say the reason I choose that design is because "well, because that feels good/looks nice!". But my advisor never told me by what should I based my design on, she just "Yes, you design it first". But that's not my question here.

My questions are: In this case, where if I do not base my research on data, by what should I based it on? Should I use a principle to back it on or what? Is there a research method like this? What is the name of this research method?

In case I ask in wrong place or there's lack of information, please tell me.

  • 2
    I think you will find more experts in UX.SE. Anyway, my guess is that you have limited time, and the full process for a more complex UI would run out of time.
    – Davidmh
    Jun 3, 2014 at 19:20
  • @Davidmh Thank you very much for your suggestion Sir, I'll try that for the UX-related question. I put all the detail because honestly I had a hard time on conveying my question, so I thought better to tell what I've experienced. Also, your guess is true, but don't worry about that.
    – Konayuki
    Jun 3, 2014 at 19:47
  • 1
    How did your advisor answer this question?
    – JeffE
    Jun 8, 2014 at 11:40
  • Goodness, I forgot to ask in our last meeting. I've touched about this before, but that time the answer wasn't clear. But maybe also because I have difficulty in talking, especially in telling a story and give long explanation (always messed up), so maybe she didn't really get what I meant, and I'm afraid talk about it longer will make she's tired of me or even it becomes more abstract.
    – Konayuki
    Jun 8, 2014 at 17:23
  • If nothing else works, you can use a modified version of steps 1 to 3 in which, instead of doing a genuine survey, you sample the opinions of one person, namely yourself. In other words, think about your own experience with systems-like-X, consider what was good and what was bad, and use those opinions as the basis for your design. (For a better survey, average your opinions with those of 2 or 3 friends.) Aug 14, 2015 at 18:40

2 Answers 2


You can base it on the data collected by other people already. There are many facts about user behaviors available in various research papers. This list has a seemingly good compilation of conferences.

So you base your design on facts and observations and user preferences that someone has published. Then your contribution can be using those principles to create a user interface for a domain, device, or a situation that no one else has explored yet.

I am not a usability person, but since the question was on research method, I thought I would provide my opinion.


In addition to bkd.online's answer, from your limited description I would call the extra steps in the first list pilot testing the user-design. Pilot tests are typically constructed to have a low time and cost data collection to get immediate feedback to improve the design.

In terms of research, the pilot study is typically not considered research in and of itself, but basically the same strategy can be used later on down the line to conduct actual research. For instance, you may have a question about the location of a widget in your design, and whether the location impacts users behavior. You can then set up a proper experiment where you randomly expose different uses to the different widget locations and then measure users behavior under these different conditions.

You must log in to answer this question.

Not the answer you're looking for? Browse other questions tagged .