# Thread: One-Way ANOVA and tongue testing

1. ## One-Way ANOVA and tongue testing

I have a pretty basic question about using a one-way ANOVA. I am performing an oral tribology experiment on pig tongues which basically means I am testing the friction of various mouthwash products after applying it to the tongue. There are approximately 10 of these products that I need to test.

Looking through the literature on ANOVA I can perform a within-subjects or between-subjects ANOVA. If my understanding is correct the within-subjects would mean that I need a set number of tongues but I would need to test each of the mouthwash products on each one. This is not plausible because of the limited surface area of the tongue and it is for this reason that I thought between-subjects would work better because I could have 10 groups and test each tongue once. I conducted a power analysis in R and it showed that I may need around 250-300 tongues. So now I am looking for a way to use these tongues more than once.

I can probably test each tongue at least twice if I cut the them into two separate pieces. The question is if I cut each tongue into two different pieces can I count them as two different "subjects" or assuming the tongues are not radically different, can I just say I'm testing one material 10 different times? It seems like I could, but I'm not 100% sure.

Thanks.

2. ## Re: One-Way ANOVA and tongue testing

The question is if I cut each tongue into two different pieces can I count them as two different "subjects"
No.
or assuming the tongues are not radically different, can I just say I'm testing one material 10 different times?
Sorry, I don't understand this.

Why don't you just limit yourself to a smaller set of
products, if you can neither apply all of them in
repeated measures, nor can afford enough subjects
for between-measures.

3. ## Re: One-Way ANOVA and tongue testing

Originally Posted by Karabiner
No.

Sorry, I don't understand this.

Why don't you just limit yourself to a smaller set of
products, if you can neither apply all of them in
repeated measures, nor can afford enough subjects
for between-measures.
Let me explain with the following analogy. If I produce a sheet of steel and want to determine the effects of 10 different treatments on the steel. I would cut smaller pieces of the steel out, put them into ten different groups, and test each sample once after it had been treated.

If I assume that variations in the pig tongues are negligible, can I effectively assume that all these tongues from the same material? If I did this it seems I could cut the tongues in half and double my samples.

The people I am doing this experiment for want about 10 different mouthwash samples tested, so I can’t change that. When I told them approximately how many tongues we may need they asked if I could just cut the tongues in half. I told them that we either needed to have one group where each sample was exposed to each mouthwash or that we needed 10 groups where each member was tested only once but I told them I would look into it anyway. If I can't cut the tongues into two pieces, I don’t think it will matter, but I would like to give the client and my boss a reason as to why it wouldn’t be a good idea. Unfortunately, my knowledge of statistics is limited.

 Tweet