I'm confused on what test to use for an experiment I have conducted over the past months. Let me give you some context:

I ran a germination experiment, sowed 900 seeds total in trays of 40 in a nursery in a way that all can be treated as equal except for the treatment they receive. Treatment groups were divided into 100.

The results are cumulative germination per group, counting all the seeds that germinated. What am I supposed to use here? Can I use an ANOVA with 99 degrees of freedom, should I be using something with 2 degrees of freedom because every treatment was sown into 3 (technically 2.5) trays, or should I be using some sort of binary test with 100 repetitions and 99 degrees of freedom because the only result I am getting is DID GERMINATE (1) or DID NOT GERMINATE (0)?

I'm doing this in R and have a large portion of my script set up already, but I can't determine what test I should use to find out of any treatment groups have significant differences between them.

Your help is much appreciated.


migrated from stackoverflow.com Jan 7 '15 at 21:16

This question came from our site for professional and enthusiast programmers.

  • 2
    $\begingroup$ Choosing a statistical method is off-topic for Stack Overflow. I've noted to migrate to Cross Validated where questions of a statistical nature are more appropriate. $\endgroup$ – MrFlick Jan 7 '15 at 20:55
  • $\begingroup$ We need a little bit more information/clarification please. Are there 9 = 900/100 distinct treatments? How are the treatments divided across trays -- 2 trays of 40 and a tray with only 20 seeds, or are there trays that are split evenly between two different treatments ... ? I think you probably want to use some form of binomial GLM, but it's not clear yet whether binomial, quasi-binomial, beta-binomial or mixed binomial is most appropriate ... $\endgroup$ – Ben Bolker Jan 7 '15 at 21:48
  • $\begingroup$ Sorry about that MrFlick, I didn't even know this existed! Thanks for moving. @BenBolker: There are indeed trays split evenly between two different treatments, and there are 9 distinct treatments indeed, each with seeds. Thanks for looking into it! $\endgroup$ – Yoeri Jan 8 '15 at 13:05

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Browse other questions tagged or ask your own question.