Timeline for Is lmer or glmer appropriate for nested continuous proportion data
Current License: CC BY-SA 4.0
9 events
| when toggle format | what | by | license | comment | |
|---|---|---|---|---|---|
| Jan 21, 2019 at 9:05 | comment | added | amoeba |
Based on your explanations, (1|Genotype) can indeed make sense. Also, you need (1|Pair) to address reviewer's concern. However, given how you coded Pair, I don't think you need to specify any nesting, and also it does not make sense to have (Range|Genotype) random slope because each genotype has only 1 value of range. Apart from that, I am not sure if it is valid to use both rows that add up to 1 -- but maybe it does (@DimitrisRizopoulos?). Regarding the model, beta regression with glmmTMB would be a reasonable choice, but it's a good idea to start with lmer.
|
|
| Jan 21, 2019 at 0:34 | comment | added | CJ123 | @DimitrisRizopoulos I put it as a random effect because I don't actually care about the specific genotypes. But I think that they are probably correlated. | |
| Jan 21, 2019 at 0:30 | comment | added | CJ123 | @amoeba. each genotype in the pair adds to 1. Here is the reviewer comment: a t-test is not an appropriate method to analyse these data. The t-test does not account for pseudo-replication within genotypes (each pair was replicated in 5 petri dishes). If you did not use genotype means but data on the petri dish level, you should rather fit a Mixed Effects Model with the respective random effect Have you used both plants from each of the petri dishes or only one plant per petri dish to build the data set? If you used both, you need an additional random factor for petri dish nested in genotype | |
| Jan 20, 2019 at 20:41 | comment | added | Dimitris Rizopoulos |
It is not clear which random effects you want to use. Note that random effects are used to model correlation in the outcome data. That is, if you put a random effect for Genotype, you assume that measurements for your outcome variable Con_Ratio from the same genotype are correlated. Is this a reasonable assumption?
|
|
| Jan 20, 2019 at 0:00 | comment | added | amoeba | Do I understand correctly that for each pair you have two ratios in your data table that add up to 1? | |
| Jan 19, 2019 at 23:57 | comment | added | amoeba |
Regarding your specific dataset, I am not quite sure what Pair means. Why is it nested in Genotype (as Genotype/Pair suggests)?
|
|
| Jan 19, 2019 at 23:55 | comment | added | amoeba | You don't seem to have count data, so binomial (logistic) model is not appropriate. See stats.stackexchange.com/questions/233366. | |
| Jan 19, 2019 at 21:10 | review | First posts | |||
| Jan 19, 2019 at 21:37 | |||||
| Jan 19, 2019 at 21:10 | history | asked | CJ123 | CC BY-SA 4.0 |