Securing extended like/dislike feedback from users

Question

On my Android app, I am toying with the idea of replacing simple Like/Dislike feedback with some form of rating system. My concern is that by and large when users are presented with the need to make a complex decision, they elect not to make one at all or make the easiest one.

The original idea for the replacement was along the lines shown in the image below. .

However, this requires the user to spend more than 2s understanding the smileys and then reflecting on which one best corresponds to their feelings. My own thinking is that while there might be some merit in replacing a yay/nay system with something that returns more value it is perhaps best restricted to the middle three smileys above - Yay, Indifferent, Nay.

I'd like to hear views from others here who have already had the experience of experimenting with such issues.

Answer 1

I agree that this is a challenge on mobile (you said Android.)

My $.02 is that certain fickle user tendencies are amplified in the portable format of mobile. Portable = interruptible, so the pain threshold for interaction overhead - which includes making decisions - is probably lower on mobile than on desktop. As such I think it is appropriate to use the same principles as desktop but exaggerate them, like a caricature. Mobile content should be shorter and decisions clearer, and that includes survey content such as multiple choice ratings.

There's also some dissonance between the semantic differential format and the whimsical abstract metaphors used to represent points on your ratings scale. In my experience the 5-point scale with neutral smack in the middle is less common in ratings systems on mobile than 0-to-5 star ratings, which I would wager everyone understands.

On the other hand, I have seen some evidence that "smiley face" are more easily consumed than Likert scales with word labels.

VerifyApp actually went with an aesthetic response format they call mood test which I think splits the difference nicely with just 3 choices - confused, meh or smiley face. I think this makes for a simpler decision for gathering reactions quickly in a forced choice format, but that's only my opinion because I've used it to gauge reactions before (and it works nicely, provided you have enough sample size.)

P.S. - I'd definitely try to put these assumptions to the test and experiment on a research question like:

Will {X} result in higher completion rates and fewer abandoned surveys?

...where {X} is your independent variable. It could be any of the following:

1. Smiley face vs text labels
2. 3 options vs 5 options
3. Required vs optional question format

...or something else entirely, but I'd experiment on them one at a time to improve the signal relative to noise and try to connect cause-and-effect a little better.