My instrument has 50 item from 6 IV and 1 DV, which is measured using likert scale. Prior to EFA, i need to remove outliers. what method should use? Univariate method or multivariate method?
While the univariate approach to removing outliers is aimed at removing the extreme values that can affect the mean score, the multivariate approach is aimed at detecting unusual combinations of variables. Therefore it depends on what your goal is. Usually, the outliers detected through the univariate method also turn out to be outliers in the multivariate method.
s. Rama Gokula Krishnan my goal is simply to remove extreme values. Then use the data to finalise my instrument
First ask yourself why you would remove outliers? You should only do that if they are really mistakes, absolutely not because "they don't seem to fit in the model". It would be helpful for us here if you could share some data.
Then you could opt for the univariate approach Girija Ramdas I have a Youtube video on how to remove outliers and secure normality, that might be helpful.
https://www.youtube.com/watch?v=nm4a9bMOoEE&t=443s
- Badges
- Science topic
- Similar topics
- Psychology