0
$\begingroup$

It will be difficult for me to describe it in one sentence, I will use an example. I would like to know if I can arrange the cards in the block combining scenarios other than randomly.

In Aizaki, H., & Nishimura, K. (2008). Design and analysis of choice experiments using R: a brief introduction. Agricultural Information Research, 17(2), 86-94. https://www.jstage.jst.go.jp/article/air/17/2/17_2_86/_pdf one can find exaple of creating choice sets using random selection. One can do a fairly similar example, just for a bigger plan:

library(AlgDesign)
library(choiceDes)
# I am interested in a fairly large example, where subject will 
# have 6 dichotomous features to compare.
ffd <- gen.factorial(2,6)
# Then, for example, I want 16 cards that I want 
# to present in four blocks of 4 cards.
des <- optFederov(~.,ffd,16,crit = "D",evaluateI=TRUE,nRepeats = 100)
blocks_list <- optBlock(~.,des$design,c(4,4,4,4),nRepeats=100)

and for a single block I get, for example:

> blocks_list$Blocks$B1
   X1 X2 X3 X4 X5 X6
10  1 -1 -1  1 -1 -1
32  1  1  1  1  1 -1
33 -1 -1 -1 -1 -1  1
55 -1  1  1 -1  1  1

then, i understand that for the correct making of the conditions that will be compared by the subject using different techniques, I should randomly sort the block card and compile it to scenarios, for example like this:

   option A               optionB
   X1 X2 X3 X4 X5 X6      X1 X2 X3 X4 X5 X6
SA  1 -1 -1  1 -1 -1   SD -1  1  1 -1  1  1
SB  1  1  1  1  1 -1   SC -1 -1 -1 -1 -1  1
SC -1 -1 -1 -1 -1  1   SB  1  1  1  1  1 -1
SD -1  1  1 -1  1  1   SA  1 -1 -1  1 -1 -1

The subject in one card compares two scenarios with each other, for example in the first card SA and SD scenario and so on. And what interests me is the problem of difficulties from the perspective of the respondent/subject. Some scenarios are very different, some almost similar. The comparison of SB and SC scenarios

SB  1  1  1  1  1 -1
SC -1 -1 -1 -1 -1  1

assumes that the subject compares all X1...X6 features that are different, which may be cognitively more difficult for subject/him/her than the comparison of features between SB and SD scenarios

SB  1  1  1  1  1 -1
SD -1  1  1 -1  1  1 

where only three features are different. If the subject has more such cards to compare, he probably has even greater difficulty in doing so. So my question is, instead of completely randomizing the scenarios for fixing them together, can I write a randomizer that takes into account the maximum similarity for all blocks and all cards?

Of course, I assume that I will randomize them to some extent, for example in 100 trials I will select versions that have the highest degree of similarity of the scenarios in the cards compared, taking into account the blocks generated by the example above. If I keep the blocks, but within the blocks I will combine scenarios in cards so that the similarity between them is maximum (and thus the subject will have a lower cognitive load), do I violate some fundamental assumptions of randomization? Could this prejudice the results of research? How? How one can check how it affects the results?

$\endgroup$

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Browse other questions tagged or ask your own question.