# Treatment Randomization without Replacement

3 years 7 months ago #106118 by adduct01
adduct01 created the topic: Treatment Randomization without Replacement
I have a quick question. I’m creating a survey with a number of randomized treatments. For one of the treatments, I would like to present the participants with each level of the treatment but in random order from participant to participant.

For instance, say I’m running a vignette experiment in which each participant assesses 10 separate vignettes. Each vignette consists of three treatments. Treatment A has 5 levels, treatment B has 5 levels, and treatment C has 10 levels. As stated above, I would like each of the 10 vignettes to have a unique random value of treatment C. This would yield a factorial object universe of 250 unique vignettes (5x5x10).

By exploring the forum, I’ve figured out how randomly sample with replacement from this factorial object universe:

Group 1
Question 1a
Question: {TreatmentA1 = rand(1,5)}{TreatmentB1 = rand(1,5)}{TreatmentB1 = rand(1,10)}

Group 2
Question 1b
Question: {TreatmentA2 = rand(1,5)}{TreatmentB2 = rand(1,5)}{TreatmentB2 = rand(1,10)}
.
.
.
Group 10
Question 1j
Question: {TreatmentA10 = rand(1,5)}{TreatmentB10 = rand(1,5)}{TreatmentB10 = rand(1,10)}

Based on the above, is it possible for TreatmentB1,…,B10 to != one another, and to have their order of appearance in each vignette for each participant to be random? In other words, is it possible to randomly sample without replacement across groups for Treatment C but NOT for Treatments A and B? If so, how?

Thanks ahead of time for any assistance.

3 years 7 months ago #106146 by adduct01
adduct01 replied the topic: Treatment Randomization without Replacement
There was an error above. The example was meant to be as follows:

Group 1
Question 1a
Question: {TreatmentA1 = rand(1,5)}{TreatmentB1 = rand(1,5)}{TreatmentC1 = rand(1,10)}

Group 2
Question 1b
Question: {TreatmentA2 = rand(1,5)}{TreatmentB2 = rand(1,5)}{TreatmentC2 = rand(1,10)}
.
.
.
Group 10
Question 1j
Question: {TreatmentA10 = rand(1,5)}{TreatmentB10 = rand(1,5)}{TreatmentC10 = rand(1,10)}