How to treat experiments repeated by year

[mmconeyb]

I am an agricultural researcher and grad student who repeated the same experiment three years in a row. The experiment is a randomized complete block design with field as the block (i.e. it was replicated by location instead of multiple blocks at one location). I have two independent variables (treatment and time) and five dependent variables (height, width, color, yield, and flavor). Each year, data was collected over time, so I am using PROC mixed with repeated measures.

My question is how do I account for the 'year' component in my analysis?

Technically, I conducted three separate experiments (albeit the same experiment), once a year in 2019, 2020, and 2021. However, the point of repeating the experiment over three years is to control for differences by production season and demonstrate that the results are consistent year after year. I am referencing both my statistics and SAS texts, and everything talks about collecting data over time (in the case of repeated measures) and not what to do with your analysis when the whole experiment was repeated multiple times. Any insight is appreciated. Thanks!

[pink_poodle]

I read about a similar example with agricultural fields. It is nice to know someone is applying it to real agricultural fields. This paper is in the right direction:
https://support.sas.com/resources/papers/proceedings/proceedings/sugi29/188-29.pdf

[PaigeMiller]

You can make YEAR a main effect, and test to see if it has any significant effect.

[RosieSAS]

If your experiments was repeated every year but not applied on the same subjects, for instant, the same plants, then YEAR should not be treated as a repeated measure. In this situation, YEAR is just a random or fixed effect in a standard ANOVA. It should be decided that whether year is a random effect or fixed effect when you designed your experiment. If YEAR does represents some special environments, and the purpose to do this multiple year experiment is to obtain more stable results, then YEAR should be treated as a random effect. Please search "RCBD in multiple years in SAS", you will get some sample SAS code.