4
$\begingroup$

I've just started a field experiment which my supervisors have already designed and planted for me (It's a maize trial). It is a Randomised Complete Block Design with four replicates, however, the first replicate is not randomised. It was laid it out this way for a field day so that visitors could see the treatments in logical order. As far as I know, including data that I measure within this block could lead to biases in my results, especially since there may be varying conditions (fertility, moisture etc.) throughout the field. My question is whether there is still any way of including the data I measure for plots within this block or if it is useless?

The image below shows the design, the first row is the block which isn't random. Colored boxes indicate the plots that I will be working on for my project, whilst grey boxes are treatments related to my experiment but being worked on by other people. Any help would be appreciated!

Semi! Randomised Complete Block Design

$\endgroup$
1
  • 2
    $\begingroup$ Seems like you can still use the data, but that the main treatment effects maybe confounded by hidden variables. $\endgroup$
    – Curt F.
    Jun 25, 2018 at 21:09

1 Answer 1

2
$\begingroup$

with only 4 replicates for 6 conditions, even randomization wouldn't smear the confounding factors well enough. you can definitely still see some pattern in the lower 3 rows. Adding first row may not be a bad thing. I suggest you keep the first row to have a stronger statistical power but do some additional test to make sure the main conclusion you want to get is not because of the other factors.

$\endgroup$
2
  • $\begingroup$ Thanks! Could you suggest what type of test I would do to test for these confounding factors? $\endgroup$ Aug 11, 2018 at 1:21
  • 1
    $\begingroup$ I don't know if there is dedicated test for that. What I thought of was like constructing a new variable e.g. the position along X axis, and see if it correlates with your yields $\endgroup$ Aug 16, 2018 at 13:09

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge that you have read and understand our privacy policy and code of conduct.

Not the answer you're looking for? Browse other questions tagged or ask your own question.