I run two-layer measurement model, a sub-construct has 6 measured items.
one item has standardized estimate of 0.754 amd R-squared value is 0.568. based on standardized estimate the weigtage of particular item is substantial, but R-squared value indiates that the item is poorly represented by the construct.
why such contradicting results.
thanks in advance
thank you prof. Solomon Kurz.
Really, I do not know the relationship between Standardized Estimate and R-squared.
But, if squared value of standardized estimate is equal to R-squared, does standardized estimate indicate correlation between DV and IV.
i could not understand it.
Srikanth,
Yes, in your model, the standardized estimate may be interpreted as the correlation of a manifest (observed or measured) and latent variable. So, Solomon's arithmetic simply shows that, if there is only one variable hypothesized to influence a given variable, and you square the observed correlation (standardized coefficient) between the two, then you should have the amount of variance in one "explained" by the other.
As an aside, an R-squared of .568 may be superb explanatory power; it depends on the variables and context. Of course, if you're trying to explain variance in a person's height (in cm) by using their height in furlongs as the predictor, you'd expect far higher explanatory power (with any reduction from 1.0 being due solely to imprecision in the measurement--those furlong sticks are quite unwieldy!).
Good luck in your work.
- Badges
- Science method
- Similar topics
- Mathematics
- Statistics
- Statistical Techniques
- Structural Equation Modeling