Comparing treatments over time - which test?

[Elly]

Hi,

I have been trying to find the effect of different treatments on fruit decay.

The experiment: Fruit was monitored for degradation over a ten day period. I scored each of 6 fruits in each punnet every day and there are six punnets per treatment. Each punnet is being treated as a replicate here.
I now have a mean degradation % for each of six punnets every day for four treatments.

I now have a graph of this which shows a general pattern of the four groups decaying at different rates over the ten days .
I want to know if there is a significant difference between the treatments over time.

I have tried a repeated measures ANOVA, but not sure if this is correct as I used the same punnets everyday.

I've been reading about different tests I think might work, but I'm getting very lost in it all.


Thank you for reading, any help would be much appreciated

 

[victorxstc]

Hi and welcome

Yes repeated-measures ANOVA seems one of the correct choices. However, you should test whether its requirements are met or not. If not, you have to run its non-parametric alternative, Friedman.

 

[hlsmith]

If these options appear to meet your needs, you may be set. However, if you were interested, linear or generalized linear regression modeling may be of interest as well. I did not quite understand the groupings of fruits (multiple types of fruit or just one). With this ambiguity, I did not know if incorporating hierarchical levels may also help explain some of the inter-punnet conditions you were inquiring about.

 

[Elly]

Hi, thank you for the responses.

Yes, the fruit are all the same (strawberries) and were all picked at the same time.

I'm still having trouble with trying to use repeated measures ANOVA. I have looked at lots of examples (using Minitab) but they don't appear to correspond well to what I'm trying to do.
I think I should possibly be doing it as a nested repeated measures as each punnet is within a group which is within a day?

What do you think, I am completely wrong?

 

[victorxstc]

I couldn't quite understand your design. Can you embed the variables in a "time * punnets" table?

If so, I think your case might be a two-way repeated-measures ANOVA. I don't know if Minitab has it or not, but SPSS can do it.

Could you summarize your design in a table? If so, how would that table look like?