What do all the acronyms mean in the Excel export?


After downloading the Excel export of an experiment, you may start wondering what do the acronyms represent and how to interpret the corresponding data. You can find the meaning of the acronyms below:

Tab: Experimental design

The experimental design is an important core component of all statistical experiments. It defines the choice tasks that respondents will see. You can think of these as the independent (or x) variables in the model, and respondents’ selections will be the dependent (or y) variables.

  • BLOCK: Design block, or version of the design. The experimental design contains several different versions, each respondent is randomly assigned to one of these design blocks.
  • QES: Question, choice task number. Each block contains several tasks, usually around 12, each of which contain information on what respondents will see on a page.
  • ALT: Alternative, within a choice task (or question) there will be several choices shown to respondents, usually around 5. This variable indicates the number of the choices in a question.
  • A1: Attribute 1, contains the level of attribute 1 that will be displayed for a particular alternative. There will be as many such variables, A2, A3… as there are attributes in the experiment.

Tab: Design matrix

  • BLOCK: Design block, or version of the design. The experimental design contains several different versions, each respondent is randomly assigned to one of these design blocks.
  • QES: Question, choice task number. Each block contains several tasks, usually around 12, each of which contain information on what respondents will see on a page.
  • ALT: Alternative, within a choice task (or question) there will be several choices shown to respondents, usually around 5. This variable indicates the number of the choices in a question.
  • ASC: Alternative specific constant, a value between 1 and 0, where 1 indicates the row is an alternative (a choice in the task) and 0 if the row corresponds to the “None” option.
  • A1L2: Attribute 1 Level 2, if value is 1 it means the second level of attribute 1 was shown. If it is 0 it means it wasn’t shown. If none of the levels of an attribute has a value of 1 it means that the level 1 of the attribute was shown (this would be A1L1). There will be as many such variables, A1L2, A1L3, A2L2… as attributes multiplied by levels minus the number of attributes (corresponding at the first level not shown).
Note: A1L1, A2L1, A3L1, and all first levels of each attribute aren't added as columns, because it would cause multicollinearity with the combination of the other levels.

Tab: Raw responses

  • ID: participant_id who answered that specific row.
  • BLOCK: Design block, or version of the design. The experimental design contains several different versions, each respondent is randomly assigned to one of these design blocks.
  • q1: Question, choice task number 1. This corresponds to the value 1 in the column QES of the Design matrix. There will be as many such variables q2, q3, … as there are choice tasks in the experiment.

Tab: Model matrix

  • ID: participant_id who answered that specific row.
  • BLOCK: Design block, or version of the design. The experimental design contains several different versions, each respondent is randomly assigned one of these design blocks.
  • QES: Question, choice task number. Each block contains several tasks, usually around 12, each of which contain information on what respondents will see on a page.
  • ALT: Alternative, within a choice task (or question) there will be several choices shown to respondents, usually around 5. This variable indicates the number of the choices in a question.
  • RES: Response, this is the respondent’s selection, in this particular question, 0 indicates that the alternative was NOT selected, and 1 indicates that it was selected.
  • ASC: Alternative Specific Constant, a value between 1 and 0, where 1 indicates the row is an alternative (a choice in the task) and 0 if the row corresponds to the “None” option.
  • A1L2: Attribute 1 Level 2, if value is 1 it means the second level of attribute 1 was shown. If it is 0 it means it wasn’t shown. If none of the levels of an attribute has a value of 1 it means that the level 1 of the attribute was shown (this would be A1L1). There will be as many such variables, A1L2, A1L3, A2L2… as attributes multiplied by levels minus the number of attributes (corresponding at the first level not shown).
  • STR: Unique index for the task.
Note: A1L1, A2L1, A3L1, and all first levels of each attribute aren't added as columns, because it would cause multicollinearity with the combination of the other levels.