Hi all,
Had a quick question regarding the development of some metrics/benchmarks via survey analysis.
Essentially, I'm developing a questionnaire designed to measure program outcomes and evaluating the effectiveness of said program. However, I'm at a bit of a crossroads, because the questionnaire will be sent to two different age groups (7-10 and 14 to 18). One of the items, for instance, would read "do you feel like you learn new skills in this program?")
The 7-10 age group will receive a 3-point Likert scale questionnaire ("Yes", "Sometimes", "Not Really"), where was the older age group will receive that same question, but with more response options (“Always”, “Often”, “Sometimes”, “Rarely”, “Never”).
The reasoning behind the different response options is to minimize cognitive load for the younger age group.
My question is, would to collapsing the response options "Always"/"Often" and "Rarely"/"Never" into one group (such as "Yes" and "Not Really", so it the response options for the older group would be on the same scale as the the one for the younger age group? Or would best practices suggest to analyze the forms separately to keep the data as clean as possible?
Thank you in advance!
I personally suggest conducting a separate analysis since the scales/constructs are very different.
Rodney Duffett Thanks for the feedback!
Both forms (for both the younger age group and older age group) are measuring the same constructs/scales (i.e., Self-Esteem, Courage, Sympathy, Leadership, Ambition).
Unless you're referring to something else that I might be overlooking.
I believe that the scales are not comparable, based on the differences in response options. Your suggestions about collapsing categories are based on your own perceptions, so you cannot guarantee that they match those of your participants.
David L Morgan I certainly see what you mean.
In terms of the constructs, you're suggesting that an item that reads "I am confident" and given the response options
Yes
Sometimes
Not at All
Could be measuring a different construct than that same item, worded the same way, but given the response options:
Strongly Agree
Agree
Disagree
Strongly Disagree
I was not referring to the constructs, but the actual measurement scale items, i.e. "Yes", "Sometimes", "Not Really" versus “Always”, “Often”, “Sometimes”, “Rarely”, “Never” that are very different. I concur with David L Morgan response.
- Badges
- Science topic
- Similar topics
- Mathematical Sciences
- Statistics
- Statistical Software
- SPSS