Hello! Can any one help?
I am collecting data through an online survey. My survey includes ranking questions. For analysis, the answers to the ranking questions will be replaced by the numbers 1 to 5. In other words, my data type is ordinal.
So far, 6 participants answered each question. At this stage, I must simulate how a larger number of participants (totally 20 or 30 participants) answer the ranking questions.
1) Is resampling from the data collected from the 6 participants an appropriate method to simulate how totally 30 participants answer the ranking questions?
2) If yes, how I can resample from the current data using JMP?
Thank you in advance.
Hello Fatemeh,
For your second question, try this link: https://community.jmp.com/t5/JMP-Academic-Knowledge-Base/Sample-With-Replacement-Add-In/ta-p/22406
For your first question, with 6 cases, you may not have sufficient variation to genuinely "represent" a larger sample, though you may certainly generate a meta-sample that's 5X larger than your current data set.
As well, remember that ranking is an ipsative process, so that you'd have to sample the full set of rankings (1-5), not just a single ranked element for each case.
Good luck with your work.
Dr Sharma, thank you for your response. Can cross-validation be used for ordinal data? I could find that it can be used for nominal and continuous data but I did not find that it can also be used for ordinal data.
- Badges
- Science topic
- Similar topics
- Statistics
- Survey