We are planning some Multi-location trials (MLTs). In which, in one location has limited field area, so we reduced the number plants per plot as compared with other locations. Is it possible to compare all the locations with respect to these MLTs. Actually, how many plants are required as minimum as to carry out the experiment. Please guide me in this regard. Thanks.
You may want to read: http://www.banglajol.info/index.php/BJAR/article/download/542/558
There is no simple answer to this question as it depends on the the number of replicates in the split-plot design, the effect you are interested in, the sizes of the vaiance components, and the size of the diferences you want to be able to detect. So you need first to find some estimates of the expected variance components and concretisize the differences you want to detect. Then use these informations i the power calculations (e.g. see the references already given) in order to ensure a reasonable power (e.g. 80%).
Thanks.. if bigger plots are choosen (say 1000 plants/plot), how much percentage of plants are included in the data collection (as a optimum case). Please share some information on data collection from the plots in a split plot experiment.
Derek R Blanchette provided you a very good tool to address you question. In order to have an "optimum case" - as you call it - you need an objective function; what are you trying to minimize/maximize? power of the test? (moves you towards large sample sizes) cost of the experiment? (moves you to small sample sizes). How many levels do your have? What is the model that you will fit your data? ... Please, check the link provided by Derek to have an appreciation of all issues involved in computing a desired sample size.
- Badges
- Science topic
- Similar topics
- Biological Science
- Biota
- Eukaryota
- Viridiplantae
- Streptophyta
- Plants