Page 1 of 1


Posted: Thu Sep 07, 2017 3:52 pm
by sellitto
Dear friends:
In calculating the predictive relevance of an endogenous reflective construct, we must calculate Q2. In the book, Hair et al. (2017, p. 207) state that zero values and below have no predictive relevance. I ask: If Q2 is a square, can Q2 reach a value below zero? Is there something wrong or Q2 doesn't mean a square?
Best Regards

Re: Q2

Posted: Thu Sep 07, 2017 6:53 pm
by jmbecker
The name Q² might be a little bit confusing as it is not a real square. In contrast to the R² it is also not bound between 0 and 1. Yet, it has some similarities to R² and that is probably the origin of its name. It is calculated as 1-(SSE/SST), where SSE is the sum of squared errors and SST is the total sum of squares. In contrast to the R² the SSE is based on the prediction error and this prediction error can be smaller or larger than the SST (which would be a prediction using the mean). Hence, you can have negative and positive values for Q².