9

I was asked to give feedback after chatting with Apple support.

form with options from Very Satisfied (5) on the left to Very Dissatisfied (1) on the right

It seems odd that Apple would ask for feedback in this manner: High to Low (5 - 1). I would expect it to ask for feedback this way: Low to High (1 - 5).

How does the ordering of these options impact responses?

4
  • 1
    Welcome to the site, @Joe. I've reworded your question a bit so that it asks about the principle underlying the ordering of the options rather than asking for speculation about the motives of Apple designers. If you think the change is too large, you can revert the edit. Aug 25 '14 at 15:36
  • Combined with the yes no I think it makes sense. The positive response is on the left. I think yes before no makes sense. I like the design and don't feel like it was selected to impact responses.
    – paparazzo
    Aug 25 '14 at 15:38
  • 1
    @3nafish Thanks for the rewording. I never knew what a "Likert Scale" was until today – good looking out!
    – Joe Clay
    Aug 25 '14 at 15:46
  • As was pointed out in a comment on an answer of mine recently, there is a strong bias towards options on the left of the scale. Presumably Apple want people to rate things as Positive, so have skewed the scale accordingly.
    – JonW
    Aug 26 '14 at 8:00
7

Data and research make me believe that:

  1. Apple's choice to order values from 5 to 1 is a) intentional and b) evidence-based (references below);
  2. Apple is adopting this structure to increase the value of the average rating they get from users ;
  3. and - at the same time - the (un)conscious user's perception of the quality of their service.

Here's why.

In this article you can find the results of a couple of studies. The results of the studies basically demonstrate that ordering values from 5 to 1 (instead of from 1 to 5) leads users to give an higher average rating.

In the first study quoted in the article you can read:

This indicates the existence of a bias towards the response category that is listed first (i.e. on the left side) on the horizontally presented agreement scale, consistent with Holmes (1974).

Does it always happen? No:

In addition, this effect manifested itself in the current study when subjects were presented with a favorable statement. This effect did not appear for any of the negative statements. One possible explanation for this may be that, in this particular study, students clearly held positive attitudes towards their college.

So, in my opinion:

  1. Apple knows that most of its users' hold a positive attitude towards its services and products;
  2. Tries therefore to maximise the results of the feedback survey presenting higher ratings first.

My third thesis is that Apples use this structure to increase the (un)conscious user's perception of the quality of their service. This effect can be explained with many psychological models, for instance the primacy effect, which is

a cognitive bias that results in a subject recalling primary information presented better than information presented later on.

Basically in this case the user's mind will connect Apple's services with the words "Very satisfied" longer and more easily than with the words "Very dissatisfied".

Definitely not bad.

2
  • 1
    Excellent answer and sources to back it up like a dump truck - bravo.
    – Joe Clay
    Aug 26 '14 at 16:58
  • 1
    Interesting study and viewpoint
    – sysscore
    Aug 27 '14 at 12:33
7

I think this would then go against Jakob's law of Internet User Experience. Which would indeed make this negative UX. To expand on this further, I have stated that consumers are accustomed to evaluations being in ascending order. Which would be supported by this 2004 study of questionnaires which clearly shows all the values being in ascending order.

If we take it a step farther and look at the Kano Model we see that user's believe that numbers increasing show more value, so 1 being the least valuable and 5 being the most. Now the study doesn't say what would happen if the numbers were reversed, just that increasing numbers provide increased value in the mind set of customer supports. So this would only imply negative UX.

I did a little more digging into Likert Scale and according to this article, a Likert scale is supposed to be in order from least to most. This has been in effect since 1932 and if we hold Jakob's law true, then the user would be accustomed to seeing this in order from least to most.

Then again, if it does successfully influence consumers to give a more positive answer. Would that not be considered success for the internal users (call center employees).

To further support this theory, I present the Nealson study, Horizontal Attention Leans to the Left, which clearly shows users tend to pay more attention to things on the left and lose interest in things on the right.

Maybe we'll get lucky and Apple will conduct an A/B test and publish the results. (no, I'm not holding my breath.)

4
  • -1 This answer appears to be random speculation. Do you have any evidence? Aug 25 '14 at 15:40
  • 2
    Sure, I apologize, I though this was an obvious answer and did not perform my due diligence. Thank you for knocking me down a peg.
    – Johnny UX
    Aug 25 '14 at 15:51
  • Solid answer. Just to add to this, personally, I think a reversed Likert scale can also break the flow. This is probably mainly because we are accustomed to the 1 to 5 scale and when it's reversed, you have to at least take a few seconds to decipher it properly. To push it even further, I think reversed Likert scales have a negative impact on the overall usability of surveys because we now need to pay attention to the scale (whether it's 1-5 or 5-1) just to make sure.
    – Vince C
    Aug 26 '14 at 7:18
  • Notably, Nielsen Norman Group gives the same label ordering for the Likert scales (but not the sentiment scale) in this article: nngroup.com/articles/rating-scales
    – Merchako
    Feb 5 at 15:29
2

From the perspective of psychology, as well as getting accurate statistics, its encouraged to not always make the "positive" answer on one side and the "negative" answer on another. This can lead to people not thinking too much about their answers... Just checking mostly one side or the other. http://m.epm.sagepub.com/content/51/3/531.short Also, another interesting link about using likert scale: http://www.mededuc.com/articles/toptenarticles/1LikertScales.pdf

1

As a non-designer, I 100% of the time expect to read anything from left to right. And generally, in a scale of how well I enjoyed something my mind functions in dislike to gradual build to like.

I clicked on "dislike" immediately anticipating it to be the location of "like".

If I hadn't taken a second look, which let's be honest, I don't want to spend anymore time on this survey to begin with as a consumer, I would have given the customer support a very bad review unintentionally.

2
  • I'm in your boat too! If I wasn't interested in UX, my helpful support rep would of gotten a poor rating as well.
    – Joe Clay
    Aug 25 '14 at 17:13
  • 1
    Welcome to the site, @Lici. Do you have any evidence to suggest that these ideas apply more broadly (to people other than yourself)? Aug 25 '14 at 17:27

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.