2
$\begingroup$

For my masters dissertation I am currently preparing a Survey Questionnaire. The dissertation main aim is to find the factors which have an impact on failure of IT related projects in SMEs. I have done some extensive reading in the respective area and shortlisted around 30 factors.

My current plan is to use each of these factors as a Likert item, with 5 options. Then the idea was to find the mean and the standard deviation for each of these item. Then from the survey result shortlist the 10 most selected factors with the highest mean.

But googling around especially on this site I have read that it is advisable not to calculate the mean of the likert item, mostly because they are ordinal.

Is likert approach the right one? if yes, are there methods (other then mean calculation) to find the 10 most selected factors from the survey?

Help on this would really be appreciated.

Thank You

Improve this question
$\endgroup$

1 Answer 1

Reset to default
2
$\begingroup$

I think there are a few issues here:

1) Terminology can get confusing, especially the term "factor" which could be part of a factor analysis or could be used just as a general term. Clearly, you are doing the latter.

2) Testing each "factor" with a single item is almost surely not enough. Single questions get misinterpreted (ask anyone who has done survey research!) and do not cover a "factor" fully. If you have multiple items for each factor, you can then sum them or do a formal factor analysis, and, if the latter, you can take the means of the factor scores.

3) You will surely want to do some pilot testing. I know this isn't a full scale-development exercise, but still....

4) How are you going to test whether the factors that results from the survey are correct? Do you have a measure of "failure of IT projects" or is this just going to be people's opinions of why these projects fail?

5) Finally, if, despite the above, you do wind up using just single Likert scale, you can take the median (although there may be a lot of ties) or you can take the proportion that got a "5" or the proportion that got, e.g. above 3.

Writing a good survey is one of those things that are a lot harder than people think.

Improve this answer
$\endgroup$
2
  • $\begingroup$ Thanks for your input, I fully agree with you that writing a good survey is harder than initiall estimated. $\endgroup$
    – user22117
    Commented Mar 18, 2013 at 8:32
  • $\begingroup$ Thanks for your input, I fully agree with you that writing a good survey is harder than initiall estimated. There are published articles where various factors/reasons for failure of IT project have been evaluated. These were mostly done on large IT project in multinational firms. My focus is on smaller projects in SMEs. Hence the idea to use the available data of the known factors and make the survey participants rate them. Other then the known factors I have also other factors which can be more related to SMEs. $\endgroup$
    – user22117
    Commented Mar 18, 2013 at 8:45

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.