How to Convert Alchemer MaxDiff Data for Analysis in Q

In order to analyze the results of Alchemer MaxDiff data, you will need to first convert the data file.   

This article describes how to convert a MaxDiff data file provided by Alchemer (formerly known as Survey Gizmo) and add the responses to the main survey data set in a format suitable for analysis in Q.

Requirements

• An Excel Alchemer MaxDiff data file containing variables with the following Labels: Response ID, Combination, your Attribute variables with the shown attribute, and the Best and Worst selection variables.  
  
• A survey data set with the Label Response ID.
  • If you don't have a survey data set, but still want to do the MaxDiff analysis: you can create a dummy dataset with one variable named Response ID for the unique response id numbers from the MaxDiff data. This is most easily done in Excel:
    

NOTE: If your data sets do not have variables with the exact labels above, this tool will not run. You can change the labels of the variables as needed by selecting the variable you wish to rename on the Variables and Questions tab and changing the Question field. 

Method

1. Import the survey data set into Q.
2. Import the Alchemer MaxDiff Data into Q.
3. From the toolbar menu, select Automate > Browse Online Library > Marketing > MaxDiff > Convert Alchemer (Survey Gizmo) MaxDiff Data for Analysis.
4. At the first prompt, select the Alchemer MaxDiff data set and click OK. 
5. At the next prompt, select the respondent data set and click OK. 
   
   New variables containing best/worst choices along will be added to the respondent data set. A MaxDiff experimental design output will be added to the report tree as well.
   
   

Use the outputs in step 5 as the inputs in the steps in either How to Use Hierarchical Bayes for MaxDiff or How to Do MaxDiff Latent Class Analysis.

• • EXPERIMENTAL DESIGN > Design source > Use an existing table and select the experimental design table output created above.
  • RESPONDENT DATA > Version > Version from the respondent data set.
  • RESPONDENT DATA > Best/Worst from the best/worst variables added to the respondent data set. 

Next