Hi everybody,
I have a questionnaire with 4 point Likert scale questions on participant experience. I want to score the results into low/moderate/high experience categories. How do I do this on SPSS?
I have assigned the following values on SPSS.
1-never
2-sometimes
3-frequently
4-always
Appreciate any help. Thank you
Why do you want to create a high, medium, low categorization? In general, scale variables are created by summing the individual items, so that you have a continuous variables which is characterized by a mean and standardization. Reducing that continuous variable to a three-category variable would be a serious loss of information.
If you have single item measuring experience, then never, sometimes, frequently and always are already four categories. Collapsing these four categories to three categories is losing information, as David said.
If you have 2+ items measuring experience. Check for the reliability (Cronbach's alpha), if it is 0.6 or higher, compute a mean score, and treat it as a continuous variable by using mean and SD.
If you have many 6+ items measuring experience, and the items are not from the an established scale, run an exploratory factor analysis, and perhaps you may get multi-dimensional experience (2 or more sub-scales) which in my opinion is even better.
Interval data is better for most analyses, but I'll assume you have a good reason for needed 3 categories.
If you have only 1 item you can't split it into tertiles (3 equal categories)
If you have several items then you can create the split for tertiles. Unlike a prior comment, I do not think Cronbach's alpha is necessary in order to create the split, UNLESS the task is psychometric evaluation of a scale (which would be a different question).
First, in SPSS use the pathway to create new variable of the mean score: transform>compute variable> then use the function group in the right panel to locate the "statistical" function group, then pull "mean" up into the top window and proceed to put the items in and make the new variable.
Once you have the new variable for the mean, use this pathway to get the tertile cut-points: analyze>descriptives>frequencies, and then open the statistics button, click "cut points for ___ equal groups" and type in the number 3
the output will have a little box at top that tells you two different percentiles...those are the precise percentiles that form the tertile split.
Use the information from the previous step to create a new variable using the percentiles and assign a number of 1,2, or3 for this new group.
If you have a small number of items forming the mean scale then you might have unequal groups (because many people have identical means).
- Badges
- Science topic