Have an idea?

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

Merging ACBC data from different markets with different prices for HB analysis

Hello everybody,

I am writing a survey for a study including an ACBC which will be conducted in different markets. Due to differences in language and price levels I need separated survey files for each market.

Price levels for the tested products vary between markets. Therefore in every market different price levels need to be used in the ACBC. However in the end, I would like to make general recommendations regarding prices and utilities across all markets.

I am using summed pricing. Even with a variation of prices of +/- 30% during the ACBC, there will be markets for which price ranges do not overlap at all.
E.g.: A Czech participant sees summed prices between 60€-90€ during the ACBC. A Finnish participant sees summed prices between 80€-120€.

Does anybody has any experience with merging data that differs in this regard before doing the HB analysis?
1. Is this even technically feasible?
2. Is it a problem regarding quality of results, that there are prices which a Czech participant sees and a Finnish participant never sees (e.g. 60€) and vice versa?
3. Can this problem maybe be solved by using country as a covariate?

Best regards and thank you very much in advance,

Jonathan
asked Jul 23, 2018 by JoGu (250 points)

1 Answer

0 votes
I'm assuming your attribute levels (other than price and the translation into languages) is the same for all countries.  Prices differ across countries.

Here's a procedure that can work:

1.  Each country gets its own ACBC exercise, translated into its own language and using the currency and prices that are realistic to each country.

2.  Choose the reference currency (say Euros) that will be used to unify the market simulator onto the same price scale during the final analysis.  This is the currency that will be used in the market simulator, which all other currencies used will be converted to during analysis.  

3.  Next, create a common currency mapping to specific price amounts. This must cover the entire range of prices (in real terms) shown across all countries.  
 
 For example, let's imagine that the "common currency map" (in Euros) for constructing the final market simulator will have the following prices:

100, 110, 120, 130, 140, 150, 160, 170, 180

...for the 9 price levels on the map (which cover the full price, expressed in Euros, ever shown across all countries).

After collecting the data for each country separately (where different countries used different currencies and may have gotten very different absolute price ranges shown), for each country separately estimate HB utilities using the piecewise summed price function with the 9 "common currency map" cutpoints and endpoints that convert to the common Euro price points.  

Imagine that we are estimating HB utilities for USA respondents who saw prices in US dollars (1.2 dollars to the Euro).  In the HB estimation dialog for USA respondents, we select piecewise coding for the summed price variable, and we type the following prices in the resulting dialog:

120, 132, 144, 156, 168, 180, 192, 204, 216

...and the utility estimation for the USA respondents will compute the discrete utility associated with each of those dollar prices, which map back to the common currency map in Euros.

4.  To bring all the utilities from the different countries into a common Sawtooth Software simulator, from the Analysis manager within each country's ACBC project you will Export the utilities (select raw utilities!) to a .HBU file (Save as type "Legacy conjoint file (*.hbu)").  

You will now have a separate .hbu file (a text file that you can open with Notepad or Wordpad) for each country.  Each .HBU file has a set of global header rows which give labels and utility layout for the file.  Then, below the header you'll see the respondent records (one row per respondent) with utility values.

Using a text editor, stack all the respondent data from the different countries (if the respondent numbers aren't already unique across countries, you'll want to modify them so they are unique).  Select the header that is the language you want in the simulator to be the global header for your stacked .HBU file.  Save the new .HBU file that has the (one, single) global header (block of header rows) followed by all respondent data.

5.  Using Sawtooth Software's standalone choice simulator (looks and acts essentially the same as the simulator included within Lighthouse Studio), create a new simulator project by importing the .HBU file (that contains the global block of header rows followed by all respondent utility data, where the respondents have unique ID numbers).  Using the market simulator interface, click that the Price attribute is "continuous" as "is price" and edit the Values associated with the 9 levels of price to be the common Euro mapping of the prices.  Merge a new demographic variable (country) into the simulator project that assigns each respondent to his/her country.  

6. To create a unified market simulator that simulates for all respondents simultaneously and allows you to specify different prices (in terms of the Euro scale) for different countries, use (on the My Scenario Settings tab) the "Apply Product Availability (Multi-Store)" option.  Each country becomes its own "store/region" where you can assign respondents to visit only their own "store".  You can thus specify the same product multiple times with different prices, where each product is the product applying to each different country market.  Only the respondents belonging to the country are allowed to "shop" in that country's "store."  You can use the "Netted Shares" option in the simulator to sum and report as a net the shares of preference for each product across the different countries.
answered Jul 23, 2018 by Bryan Orme Platinum Sawtooth Software, Inc. (154,505 points)
edited Jul 23, 2018 by Bryan Orme
Hi Bryan,

Thank you very much for your detailed answer. I tried to follow your instructions but got stuck after merging the data like you described in 4. Unfortunately I do not have access to the Sawtooth Choice simulator.

I work with SSI Web 8.4.8. Is there any possibility to import the named .hbu file with merged data from different markets to SSI Web and conduct  an HB analysis?

As I extended my survey a related issue occured: I basically need to do the same steps (merge data and conduct an HB analysis) with MaxDiff data.
I asked a single question concerning this yesterday, but probably solutions are somehow related:
https://sawtoothsoftware.com/forum/18688/how-can-merge-raw-maxdiff-data-across-markets-before-analysis

Any further hints would be highly appreciated!


Best regards

Jonathan
If you have a license to SSI Web 8.4.8, then you have access to the SMRT Market Simulator (you can download from our website).  SMRT Market Simulator can import .HBU utilities.  You just create a new CBC or ACBC-like project when you click File > New.  Then, you go to Analysis > Run Manager > Import, then you browse to the .HBU file.  It reads it in and automatically sets up all your attribute levels.

If you don't want to use our SMRT Market simulator and if you want to write your own in Excel, then the part-worth utilities are all in the .HBU file for you to copy into Excel and use in your own simulator.  And (if you followed the previous steps I outlined), you will have put all the price part-worths on a common currency scale for your easy use and easy combination of all countries within the same simulator.
...