1
$\begingroup$

Recently, I came across survey responses which I was to analyze. There were simple questions with four response levels for each item/question.

  • Excellent (10)
  • Good (7)
  • Satisfactory (4)
  • Poor (0)

This is a modification of a Likert scale. The client suggested that, for analyses, we add the values of all the responses and dividing by the no. of respondents to get a percentage.

For Example

  • 7 + 7 = 14 == 70% of 20
  • 7 + 10 = 17 == 85% of 20

Is this right? Usual Likert values require that there should be an equal gap between the values assigned to each response.

$\endgroup$

1 Answer 1

2
$\begingroup$

It totally depends on what you mean by "analysis". For example, do you just want to describe the responses to that question, or do you want to relate that question to something else, for example analyse whether there is a difference in response between males and females, or if the response was related to the length of time the respondent had been using the company's products.

In general, there is no "right" or "wrong" way to deal with this kind of variable. There are multiple methods, each with their pros and cons. If you want to treat the question as if it was interval scaled (as your question implies), then it's best to use at least 5 points, and to label the extremes but not the intermediate points. If you must label the intermediate points, use balanced labels, e.g. 2="Agree" and 4="Disagree".

But you do not have a Likert scale. A Likert scale is either a single variable that measures agreement, or a set of variables that measure roughly the same thing, which are then summed to create a total score. You have an unbalanced (Osgood) semantic differential scale that would be difficult to interpret numerically. For this reason, I think the best thing is to do as your client suggests, i.e. report the percentages.

For future work, I suggest you work with an expert in this area. Satisfaction research is easy to conceptualise but VERY hard to do in a meaningful sense. I suggest starting with the "bible" of satisfaction measurement:

Satisfaction: A Behavioral Perspective on the Consumer (2nd Edition) by Richard L. Oliver. ISBN-13: 978-0765617705; ISBN-10: 0765617706.

Good luck,

John

$\endgroup$
4
  • $\begingroup$ But I am saving all the responses in 1 variable. We have multiple-choice but the respondent can select only 1 variable out of the available four choices. $\endgroup$ Jul 14, 2015 at 1:23
  • $\begingroup$ Sorry, I don't understand. Do you intend to ask a question? Or are you just making a comment which needs no reply? $\endgroup$ Jul 15, 2015 at 2:15
  • $\begingroup$ I was asking a question. $\endgroup$ Jul 16, 2015 at 4:20
  • $\begingroup$ What is your question? I don't want to offend you, but I can tell by what you right that you are out of your depth, i.e. you don't really understand what you're doing. I strongly counsel you to see expert help in real life, not online. $\endgroup$ Jul 17, 2015 at 21:13

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

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