I am using structural equation modeling and need to transform the data before using maximum likelihood estimation. I am using the book Principles and Practice of Structural Equation Modeling by Kline 2016 for reference.
The book says "before applying normalizing transformations, add a constant to the scores so that the lowest value is 1.0." I follow it to add 0.99 to the value so that the lowest value is 1.
However, after adding the value, it seems impossible to normalize the data by transformation. If not add a constant, the data can be successfully transformed using log-transformation. My data value is very small, range from 0.01 to 0.726. So adding 0.99 has strong effect on the value.
I did search on google but didn't find any other reference saying "add a constant to the scores so that the lowest value is 1.0." The only thing I can find is that "add a constant to make sure the lowest value become positive so that log-transformation can be used."
Is it really necesssary to add a constant so that the lowest value is 1.0 before normalizing transformation?
- Badges
- Science method
- Similar topics
- Mathematics
- Statistics
- Statistical Techniques
- Meta-Analysis