Sample Testing in an Incomplete Block Design
May 20, 2022 10:15 AM Subscribe
I cannot even begin how to work this conundrum, please help me with a testing plan for sample distribution with a minimum of participants.
I am working on a plan for product testing. Everything was all fine and good, we had 6 products and 60 participants, and an incomplete block design where each participant tested 3 products and each product had a total of 30 participants testing them. We even had a plan drawn up grouping the products into bundles of 3 products so that the products were evenly distributed (i.e., no two products were grouped together more often than others).
Well, there was a client meeting this morning and now we have 7 products to test. Preferably the same number of participants, as that was what we'd budgeted and been approved for. Here is my question: what is the minimum number of participants I can have testing these 7 products and still have 30 participants test each one in an incomplete block design? Preferably we would still have all participants testing only 3 products, not some testing 4 and some testing 3.
I was not the one who came up with the original design and that person has left for the weekend already. I need an answer so I can send in a recruitment request by the end of today. Help!
I am working on a plan for product testing. Everything was all fine and good, we had 6 products and 60 participants, and an incomplete block design where each participant tested 3 products and each product had a total of 30 participants testing them. We even had a plan drawn up grouping the products into bundles of 3 products so that the products were evenly distributed (i.e., no two products were grouped together more often than others).
Well, there was a client meeting this morning and now we have 7 products to test. Preferably the same number of participants, as that was what we'd budgeted and been approved for. Here is my question: what is the minimum number of participants I can have testing these 7 products and still have 30 participants test each one in an incomplete block design? Preferably we would still have all participants testing only 3 products, not some testing 4 and some testing 3.
I was not the one who came up with the original design and that person has left for the weekend already. I need an answer so I can send in a recruitment request by the end of today. Help!
Response by poster: Oh my god, thank you so much! I might have been able to figure this out on my own eventually but...it's been a hell of a week. I knew Metafilter would come through.
posted by Fuego at 10:39 AM on May 20, 2022
posted by Fuego at 10:39 AM on May 20, 2022
« Older Where in the world is this scene in film/tv? | Microphone suited for deep voice, spoken word... Newer »
This thread is closed to new comments.
Now you want 7 products tested 30 times each, 7*3=210 total tests; if everybody still tests 3 products, that's 70 people -- 70*3=210.
Conveniently, there are 35 unique combinations (if the order doesn't matter) when you choose 3 alternatives out of 7 products (7c3=35) and 70 is evenly divisible by 35, so with 70 people you can have a balanced incomplete block design where every 3-way combination of products is tested exactly twice, if you don't mind the more complex coordination.
posted by Superilla at 10:33 AM on May 20, 2022 [2 favorites]