We don't automatically output a report, but it isn't that hard to do the analysis using your own tools (Excel, SPSS, R, etc.) using our output files. A big question is whether you want to do it in the Bayesian way or the frequentist (traditional) way.

If each of your attributes has two levels, then typically the question is whether the enhanced or \"on\" level is preferred to the reduced or \"off\" level.

Frequentist Approach (matched t-test):

1. From our software (let's assume Lighthouse CBC), export the utilities to a normalized format (Zero-Centered Diffs). Or, you may see that the utilities are in a multi-tabbed report where one of the tabs contains the normalized (zero-centered diffs) utilities.

2. Read into Excel (or similar). Create a new variable which represents the \"on\" minus the \"off\" level utility.

3. Compute the mean of that new variable. Also, compute the standard deviation of that new variable. Then, divide that standard deviation by the square root of the sample size. This is the standard error for that new variable (representing the difference between the \"on\" and \"off\" levels).

4. Compute the t-ratio for each of those new variables for each attribute, by taking the mean difference divided by the standard error. If it has a magnitude of 1.96 or greater, you are 95% confident that the sample thinks that one level is preferred to the other.

Bayesian Test:

1. Open the \"Alpha Draws\" file in a program like Excel (or similar) that contains the estimates of the population preference at each iteration of the HB algorithm. (If you are using Lighthouse Studio CBC, you will need to request the advanced Alpha file prior to running estimation, by going into the settings area for HB estimation, clicking the \"Advanced Output Options\" link, and then requesting the Successive Estimates of Alpha). Only pay attention the last (say) 10000 iterations, which we'll assume are the iterations after convergence.

2. Count for what percent of the last 10000 draws the \"on\" level is preferred to the \"off\" level. This is a direct representation of your confidence that the one level is preferred to the other for the population.

Significance in HB Estimation