Hello everyone,
1.I'm trying to test the difference between part-worth utilities in my HB analysis and correct the p-values using the Bonferroni method. Is it necessary to first normalize/standardize the individual-level utilities, and if so, how? Until now the utilities are calculated with zero-centered difference. If I understood correctly, I need to use this formula t=(U1-U2)/(SE1^2+SE2^2)^1/2 for that, where the values are just the calculated average utilities in HB, is that right? In the utility report of the HB analysis, I find only the confidence intervals (95%) and the standard deviations, but no standard errors, unlike in the logit analysis. Can these be calculated, or are they available somewhere for this type of analysis?.If I want to calculate the significance of attribute levels within the HB analysis, I also need the standard error. Can I calculate the standard error using the following formula: Utility estimate + 1.96 * std error = CI (upper limit) --> std error = (CI upper limit - utility estimate) / 1.96? Can I use therefore the calculated average utilities?
2. Is it the same way as in 1. to calculate the significance of the attribute importance?
Thank you in advance.
Best regards, Noemi