Question

Topic: Research/Metrics

Mean Score Of Nine Close-ended Questions

Posted by Anonymous on 3/30/2008 at 4:02 AM ET 125 Points
Dear Experts,
Kindly, I have the following question:
The mean score of nine close-ended questions is 4.22, 4.03, 3.93, 3.75, 3.65, 3.58, 3.39, 3.18, and 2.17. From five-point scale.
Is there any specific method that can be used to transfer the above data to information?
Many thanks in advance
Omar
To continue reading this question and the solution, sign up ... it's free!
SUBSCRIBE
Or, SIGN IN

RESPONSES

  • Posted by adammjw on 3/30/2008 at 5:39 AM Member
    Omar,

    What was your intention? Why did you carry out the survey?
    Are the questions independent from one another or dependent?
    Are they equally important or not?
    Apart from above the framing of the questions might have skewed the results and biased your respondents, especially when they were close-ended questions.

    Adam
  • Posted on 3/30/2008 at 12:35 PM Member
    Hello Omar,

    In addition to Adam's excellent questions, consider:

    1. What are you measuring? Is it satisfaction, likelihood to purchase or refer? Perceived value?

    2. How many respondents do you have to the survey? Is it a representative sample? If you have too few respondents to run statistical testing, the results should be viewed as directional only, not significant.

    If you have a large number of respondents, you could consider running crosses of the data by demographic or other segments, to try to isolate where there may be more or less acceptance of the idea among the target.

    To answer your question, assuming each question contained the same value scale, you could turn the numbers into information with a simple bar graph of the means, noting top two box score in each bar. If you can stat test it, also note the means that are statistically significant from peers. This will show at a glance the relative differences between what was tested in the questions. Include a bullet point addressing how the ratings relate to any of the learning objectives of the study. For example, if you were testing reaction to product concepts and expected that people would react favorably to some over others, note whether this view played out in the data and what other questions may need to be investigated based on the data.

    Hope this helps.
  • Posted on 4/2/2008 at 10:32 AM Accepted
    Omar - If you are trying to graphically represent these means, I agree with BluGlobe - you can either graph the means or the top-two box. Also, did you aks for importance ratings of these attributes? Respondents may be highly satisfied with certain attributes, but if these attributes are unimportant to them, that will be important information to know. Did you collect other information about respondents? You can calculate means within certain groups (e.g., men versus women) to illustrate differences that way too. You can do stat testing on the means provided that there are at least 30 respondents in each group (although that is still a small base, and groups of 50 each are preferred.)
  • Posted on 4/4/2008 at 7:27 AM Accepted
    That's a good point, jlevin.

    If you are asking satisfaction, Omar, there is a way to infer importance without asking it. This can be more insightful than a straight-out question, in which people may say that everything is important.

    Ask an overall satisfaction question, then satisfaction with each of the provided services. Correlate the individual ratings to the overall. This will yield the derived importance of the individual services. This can be shown in a 2x2 matrix, with satisfaction and importance as the scales. The items in the box corresponding to high importance and low satisfaction should be targeted for remediation.
BACK TO TOP

Post a Comment

Related Discussions

Community Info

Top 25 Experts
(Research/Metrics)