Dear all,
I have cross tabulation data with 500 students who belongs to 5 subgroup (5 colleges). If I use this data set to run CFA and SEM, I wonder if the different size of subgroups will make the perceived opinion of students who belongs to big group are dominated the perceived opinion of people who belongs to small group. As a result, the SEM coefficient is unfair for small colleges.
If you know any science document about this problem, pls show me, thanks.
NGOC
If you have 500 students from different groups, don't worry you have enough sample size. You can perform group differences i.e. moderating groups analysis in AMOS which tells you the characteristics of each group.
You can check James Videos of group difference (moderating groups) in Youtube to get idea.
Sincerely,
Thanks for your information.
Actually I do not want to test the different characteristics of each group, because I sure they are different. I want use the SEM coefficients in next analysis about 5 colleges, so I want to sure that they are fair for 5 groups. I intend to collect equall sample size for all groups for my purpose?What do you think?
Do you know about total population from where you can calculate sample size for each group through stratified or cluster method? Or you afraid of small sample size for SEM?
If I understood your question, you need to calculate each sample size for each group based on starta or cluster and apply SMART PLS which does not need a big sample size. If you follow these two sampling method, you shall not need to have equal sample size.
You can search for Hair et al., 2006, 2010, 2017 Book of SMART PLS available free on Google. They have discussed in detail every thing.
Sincerely,
- Badges
- Science topic
- Similar topics
- Biological Science
- Kinesiology
- Running