Hi, all.

New to Qualtrics with a conditional block-randomisation procedure to implement for a within-subjects survey experiment. 

My stimuli are true and false statements made by UK politicians from the Labour and Conservative parties. I have a total of 16 stimuli (statements) comprised of four of each of the following categories:

4x Labour-TRUE (LT),

4x Labour-FALSE (LF),

4x Conservative-TRUE (CT),

4x Conservative-FALSE (CF).

SURVEY FLOW:

Control condition task: participants presented with a series of 8 of the 16 stimuli, comprised of exactly 2 of each of the four previously defined categories: 2x LT, 2x LF, 2x CT, 2x CF. The 2 stimuli selected for each of the 4 categories in this condition are randomly selected from the total of 4 stimuli in their category. All 8 stimuli presented in this condition are presented in random order.

Treatment intervention: after responding to the 8 control condition stimuli, all participants receive a treatment intervention.

Treatment condition task: after receiving the treatment, participants are presented with the remaining 8 stimuli (all stimuli that were not presented in the control condition). Necessarily, this would mean that the series will be comprised of exactly 2 stimuli for each of the four types, again in random order.  

QUESTIONS:

Any advice on how to implement this randomisation procedure would be much appreciated. 

WHAT I HAVE TRIED:

I have created blocks for each of the four categories, in my survey flow I have randomised these blocks and I have randomised questions within the blocks to randomly show only two from each block. The obvious problem is that the questions from each category are shown consecutively e.g. cCF1, CF4, TL3, TL2, …]. The less obvious problem is that despite adding page breaks between questions within a block, the two randomly selected questions from each category are shown simultaneously on the screen (also undesirable). 

WHAT I HAVE DISCOVERED:

I gather that I need to use embedded data to store what questions happened to be presented to participants in the control condition so that I can make the randomisation of stimuli in the treatment condition select from the remaining stimuli only.

I believe that JavaScript will be involved at some point but I don’t know at what stage, would it be easier to write the whole procedure in JS or to let natural Qualtrics methods handle some of the procedure and use JS only where it is required.

I think it would be best to handle the stimulus categories in some data structure readable in JS (not blocks).

Any advice much appreciated, many thanks!!