Carrier Log in
← All tutorials

Group matchingSurveyVariablesChat

Control group composition with a survey answer

Two conditions of discussions with six persons. One chamberline makes groups of 4:2 and the other makes groups of 3:3, from the answer to "Nuclear power is safe."

6 steps 12 min read

  1. Ask the question before the discussion

    Each chamberline starts with an Intake chamber for one person. This chamber has the survey question "Nuclear power is safe." with the answers agree and disagree. An alias variable, nuclear_safe, reads the answer. To make the question and the variable, see Match participants on a survey answer.

    Matching needs the answer, so the question must come before the discussion chamber.

    The intake question, and the nuclear_safe variable that reads it.

  2. Give the discussion its group rules

    Open the discussion chamber with six seats. Under Group Composition (Matching), add one rule for each side:

    • At least 4 have nuclear_safe = agree.
    • At least 2 have nuclear_safe = disagree.

    The rules apply to the full group, so the seats have no conditions. Matching looks at all participants who wait in the chamberline. It finds six participants who agree with all the rules, and puts them together. With six seats, these two rules allow only a 4:2 split.

    Other rules can set a maximum count (at most N), the same answer for all, all different answers, or a majority.

    The discussion in line A: at least 4 agree and at least 2 disagree.

  3. Duplicate the line for the second condition

    Duplicate the chamberline. Give the copy the name of its condition. Change its rules to at least 3 agree and at least 3 disagree. All other settings stay the same, so the split is the only difference between the conditions.

    Balanced 3:3, a copy of the first line with 3 and 3.

  4. Read both lines' answers, and assign lines

    The copy has its own intake survey, so the variable needs a second source. Do these steps:

    • In Variables, click + Add fallback source.
    • Select the question of the copy.

    Each participant answers only one of the two questions. The variable reads the question that the participant answered.

    Set Chamberline Assignment to Counterbalance. The two chamberlines then fill equally, and the assignment does not use the answers. Then deploy the study.

    nuclear_safe reads both intake questions, and the chamberlines are counterbalanced.

  5. Line A: a 4:2 group

    Twelve participants arrive, six for each chamberline. Each participant answers and waits for more participants. Carrier puts a group together when six participants agree with the rules. Here, four answered agree and two answered disagree.

    Carrier does not make a group that does not agree with the rules. If three agree and three disagree wait in line A, no participant gets a seat until a fourth "agree" arrives.

    The clip shows the answer of each window. Participants do not see the answers of other participants.

    Six windows in line A: four answered agree, two answered disagree, and they are in one discussion.

  6. Line B: a 3:3 group, and what to plan for

    The balanced chamberline makes its group in the same way.

    Carrier assigns the chamberline before the participant answers. Thus each chamberline needs sufficient participants of each answer in its own queue. If most of your sample agrees, participants who agree wait longer for a 3:3 group.

    Set the matching timeout of the chamber and the action after the timeout. Recruit participants with the necessary split. The chatrooms on the study dashboard show which participants were in each group.

    Six windows in line B: three agree and three disagree.