Suppose that in a questionnaire five factors is anticipated, that are formed from different questions(questions measured by Likert-Scale), for example score of factor-one is calculated from questions one and two ,score of factor-two is calculated from questions three to 13 and so on because of different number of questions in each factor, we get unequal score range between factors .
The question is:
How can we compare mentioned factors in this questionnaire?
Why don't you just take the average score per scale?
e.g. F2 = sum(Q3+Q4+...+Q13)/11
Just in case your Likert scales have different ranges (e.g., 1-5 and 1-7) you first have to standardize a question to a range [-1,1] centered at 0. For a 7 levels Liker scale this is:
score_norm = (score - 4)/3
Why don't you just take the average score per scale?
e.g. F2 = sum(Q3+Q4+...+Q13)/11
Just in case your Likert scales have different ranges (e.g., 1-5 and 1-7) you first have to standardize a question to a range [-1,1] centered at 0. For a 7 levels Liker scale this is:
score_norm = (score - 4)/3
Frequently, Timo, the scale points are equidistant in this type of situation. Martin's suggestion is a very reasonable place to start.
Masoumeh, are the questions part of a previously validated, published instrument of some sort, such as the SF36? If they are, I recommend that you find the scoring methodology that was used, and start from there. If not, try Martin's idea.
- Badges
- Science topic
- Similar topics
- Mathematics
- Statistics
- Applied Statistics
- Survey Design
- Questionnaire