Balance a baseline attitude across conditions
A pre-survey rating puts each participant in a band. Each band has its own line, so the conditions stay equal in each band and in total.
-
Put the rating in bands
A counterbalance variable keeps the number of participants in each condition equal. It does not look at survey answers. Thus one condition can get more participants who agreed with the statement before the study.
To prevent this, put each participant in a band first. Then balance the conditions in each band. This method is stratified randomisation.
Here, the global pre-survey asks for a rating from 1 to 7. The question name is
tip_rating. An Expression variable,tip_band, uses the Conditional shape to put the rating in a band:- 1 to 3 gives
disagree. - 4 gives
neutral. - 5 to 7 gives
agree.
Use two or three bands. With more bands, each band gets too few participants for good balance.
The rating question in the global pre-survey, and tip_band, which reads it. - 1 to 3 gives
-
A counterbalance variable for the condition
In Variables, add a counterbalance variable,
condition, with the valuessupportandagainst. Keep the default strategy, Randomized blocks.The band, not the condition, selects the chamberline. The condition only changes what the participant sees in the line.
condition, with one value for each message. -
One seat accepts both values, and each message shows for one
The line starts with a Welcome chamber. Then it has one chamber for each message.
Open the seat in the Welcome chamber, and add the condition
conditioninsupport, against. All participants meet this condition. But it tells Carrier to drawconditionafter it assigns the line. Then Carrier counts the draws in that line only.Add this condition to one seat in each line only. Each seat with the condition adds weight to each value. With two seats, each block holds two of each value, and the balance is less exact.
Then open each message chamber. In Visibility Conditions, add
condition=supportto Support message, andcondition=againstto Against message.The Welcome seat accepts both values. Each message chamber shows for one value. -
One line for each band
Complete the first line before you copy it. Duplicate it two times, and give the copies the names Neutral and Agree. The copies keep the seat condition and the visibility conditions.
In the experiment settings, open Chamberline Assignment:
- Set Assignment Method to Variables-based.
- Select
tip_band. - Add a rule for each band, with the name of its line.
Deploy to save.
Three lines, and a rule for each band: disagree,neutralandagree. -
Each band gets both conditions
Ana and Cleo give ratings of 2 and 3, so they go to the Disagree line. Ben and Dev give ratings of 6 and 7, so they go to the Agree line. In each line, one participant reads the support message and one reads the against message.
In each band, the two conditions differ by one participant maximum. Thus the totals of the two conditions also stay almost equal.
In the export, the Participants sheet has the columns
var.tip_bandandvar.condition. TheassignedChamberLinecolumn shows the line of each band.Two participants who disagree and two who agree. Each pair gets one message of each type.