Dear RG Members,
I am handling missing value in confirmatory factor analysis and has some questions.
I want to do multiple Imputation in confirmatory factor analysis and use factor scores to do some analysis. In detail, I want to use the 1SD of the factor scores as a cutoff to determine two groups.
And I found an R package "semTools" and several functions to complete the imputation step automatically. Especially, the "plausible values" function is to draw plausible values of factor scores for each imputation dataset.
For example, if I impute 20 datasets, there are 20 groups of factor scores.
But it confused me.
- Should I use factor scores of each dataset to determine two groups and take the most results? But I cannot find the correctness of this method.
- Are there some functions to pool factor scores as one result in the packages "semTools" like the function "pool" in the "mice" packages, but I ignore it?
So, I require your help.
What is appropriate for me to analyse these groups of factor scores as an integral?
Thank you for your advice!
I would appreciate your help.
Yuyang.
Hi Liu Yy ;
I think this will hep you ==> https://www.statisticssolutions.com/multiple-imputation-for-missing-data/
But in summary, you can use average of factor scores to obtain cut point.
Hope it helps.
Bests.
AFK
Thanks for suggestion, AFK! Abdullah Faruk Kılıç
My reaserch limits the cutoff as 1SD of the factor scores refer to the previous researches. It seems to be difficult to change the cut point.
Do you mean that I should take the average derive from all dataset (for example, 50 groups as a integral dataset) as a cut point?
Sincerely,
Liu Yy
Hi Liu Yy Liu Yy ;
Yes i mean, you can use average derive from all dataset as a cut point.
Hope helps,
Bests
AFK
Check https://www.researchgate.net/publication/301305175_Multiple_Imputation_of_missing_values_in_exploratory_factor_analysis_of_multidimensional_scales_estimating_latent_trait_scores
- Badges
- Science topic
- Similar topics
- Counseling Psychology
- Confusion