Dear Kezang,

I would question the validity of your findings. Is that what your teacher graduates actually think or are they telling you what they think you want to hear? From a statistical point of view, the main thing you can conclude is that your questions have not been very discriminatory between your respondents.

In future I would recommend throwing in the odd reverse worded question (where the language is easy to follow) and using more points on your Likert questions (we recommend 7).

Kerang,

I agree with Peter that your data have a ceiling effect, and this leads to a loss of information (a bit like giving a third grade math test to fifth graders). 

However, I disagree with the recommendation to use reversed items. I (and many others) have found that doing this typically introduces a second factor, with the reversed and non-reversed items loading on different factors (E.g., Imagine your wife's face if you told her you "absolutely certainly didn't hate her").

Also, using more than four response categories is rarely productive, and having a middle category leads to inconsistent answer patterns ("uncertain" rarely means "uncertain", instead responders often use it to indicate things like "I don't want to tell you", "none of your beeswax", "I'm afraid to tell you", "no answer").

Rense Lange

You obviously have highly skewed data, but that is hardly the same as invalidity. The more practical problem is that you probably do not have much variance in your measure, so it is not likely to produce high correlations with other variables -- unless the distinction between scores of 4 versus 5 turns out to be meaningful.

Thank you Peter, Rense and David for your advice and comments. It was very useful...

There are three things here, Validity, Normality and Sample Size. I agreed with Peter in the use of 7-point likert scale since factor analysis can be used to reduce the items in the instrument. However, it should be noted that some literature have established that items in the questionnaire should either be asked in a positive or negative pattern. The sample size may be too large perhaps approximating to 1,000 or more.

I FOLLOWN THIS QUESTION