With Conjoint.ly, you can upload your own experimental design from the JMP Discrete Choice designer. Conjoint.ly will handle data collection and, if you require, fielding to panel respondents. Once data have been collected, you will be able to download a .csv output for analysis in JMP. However, Conjoint.ly will not automatically produce a report for this type of experiment.
Experimental design files need to comply with the JMP standard. Here are a couple of examples:
- Example 1 (alternatives are not labelled, with multiple blocks/surveys)
- Example 2 (with labelled alternatives, only one block/survey)
Once the file is selected, you will need to choose the type of the "None of the above" option (because it is not specified in the design file):
- Forced choice (i.e., no opt-out option),
- Standard "None of the above" option (part of the choice set), or
- Two-stage response (a confirmation question after the choice set).
Next, the file is uploaded and validated. You will then be able to format the study, tweak the names of attributes and levels, and specify other settings that are available to conjoint experiments on the platform.
Export of collected data is available in the JMP "stacked" format (see example exported file). Please note:
- Each row in these exports corresponds to an alternative. For example, for each respondent that replies to a study with eight questions per block/survey and two alternatives in each question, there will be 16 rows. If a respondent selects a particular alternative, the "Choice Indicator" column for that alternative row will contain "1".
- "None of the above" alternatives are omitted from the file: if a respondent chooses "None of the above", the rows that correspond to the question will contain all zeros in the "Choice Indicator" column.
- Respondent-level information (for example, location, answers to additional questions, and GET variables) are repeated on each row for each respondent.
Partial-profile designs are also supported (see example design file). In order to specify such a design, you need to insert "NULL" in the place of level names for the hidden attribute in specific choice sets.