Have an idea?

Visit Sawtooth Software Feedback to share your ideas on how we can improve our products.

The R-squared in the CVA with OLS utility estimation report is the adjusted R-squared? and for RLH?

I am running a CVA and CBC study and the reports are for utility estimation, in the CVA with OLS and CBC with HB.

Those were the reports I was talking about in the related question. The issue is about the value for the coefficient of determination (R-squared) in the CVA utility estimation and the root likelihood (RLH) in the CBC utility estimation.

For the CVA, the R-squared in Montgomery Design and Analysis of Experiments is calculated as follows: Sum of Squares of the Model (SSModel)/Sum of Squares Total (SSTotal) and as the author also mentions, a potencial problem with this statistic is that it always increases as factors are added to the model, even if these factors are not significant. So an adjusted R-squared is calculated which accounts for the "size", the significant factors, and it uses the next formula: 1-((SSE/dfE)/(SST/dfT)) where SSE is the Sum of Squares of the Error term in the regression and dfE is the degrees of freedom of the Error term.

So, does the Sawtooth Software SSI web estimate an adjusted R-squared? or it does not account for degrees of freedom?
asked Nov 18, 2014 by Alvaro

1 Answer

0 votes
Thanks for clarifying!

We do not compute adjusted R-squared.  

Like you cite, each time a new term is added, the R-squared will increase.
answered Nov 18, 2014 by Bryan Orme Platinum Sawtooth Software, Inc. (175,290 points)