Standardised coefficients in path analysis are derived from differences in SD hence two different metrics could be brought to the same level.
So as far as I understand standardised effect of .2 indicates difference of .2 of one SD.
Does it happen that this difference would be above one SD, hence the standardised effect would be above 1?
Thank you for clarifying.
It is indeed possible. Standardized path coefficients are standardized regression coefficients ("beta weights"), and in general may exceed 1 in absolute value. Standardized coefficients are only correlation coefficients in special cases--either (1) all correlations among predictor variables are zero, or (2) there is only one predictor variable--and naturally, correlations cannot be greater than 1.
It is not only possible but very likely if there is a high degree of multicollinearity (and very possible with relatively modest collinearity).
- Badges
- Science topic
- Similar topics
- Phytochemistry
- Extracts