I still have some questions about both conditional and summed pricing.
I tried to implement summed pricing with different amounts of levels. We calculate that around 800 people will do the CBC. When I tested a design for 11 price levels after having imported the CSV design file, the standard error was still more than 10 percent for each price attribute level. I also varied the number of price levels, but there is almost no difference in the standard errors. Is there any trick for improving the design file, e.g. in the settings for the design I previously exported or what else could I do?
Regarding the analysis: Did I understand correctly that in order to find out the utility of the price attribute in summed pricing, there is no "magic trick" needed? And if so, why does the utility rise with the price? Shouldn't it be the opposite?
The alternative would be conditional pricing. If I am not wrong, in this case testing the design does not provide information about the standard error for the price attribute as only 3 attributes are defined, but those reflect different prices for the different attribute level combinations. How could I find out then how many respondents I need?
Also for conditional pricing, how do I include the real prices in the analysis? I only found the option to include the look-up table in the market simulator. And is there anything else one has to consider when interpreting the utilities?
And more in general: Is there any rule of thumb what the maximum number of price attribute levels will be? Again, we estimate that over 800 people will be doing the CBC.
Another question for both pricing methods: Do I have to indicate interactions with other attributes in attribute coding for HB? And how is this even possible with all five attributes defining the price?
Is there any recommendation for us which pricing method we should use or would you recommend e.g. comparing the RLH in Hierarchical Bayes for both methods in order to decide?
Thank you so much in advance!