Likert Scale

Recently I stumbled upon this website regarding the use of Likert Scale, here is another useful website to learn more about this tools.

My reading lead me to this statisticcafeblog which explains furthermore the analysis of likert scale.

I remember my professor said that in order to build an instrument, we need to ensure the validity and reliability of the tools. If I apply that to the likert scale, I would employ some steps:

1. Building the key points that I need to develop by putting up the definition preferably based in available theories or consensus among the experts.
2. Test the content to the experts to provide the validity of the tools and build up temporary classification.
3. Run a factor analysis of the items by sending it out to a number of people
4. Clarify the results and develop the category based on the factors (and subfactors)
5. Run a Cronbach's Alpha Reliability to test the reliability of each category
6. Review and eliminate the items that did not belong to the category.

And the next is to run the actual likert scale to intended responden and analyze the result. Note, if the items are ordinal data avoid to use the mean as this will be useless, use mode instead.

Well, that's the summary of learning today :)