Order of 'genotypes' in TD and rest -- can you also split up "genotype" into two factors?

[envanloo]

1. Why is the "genotype" factor changed to levels indicated by numbers? Why not keep the levels of the original factor?
2. Why do you sort the new genotype factor in the TD and the outcomes of statgenSTA lexicographically and not numerically as it is presented as numbers?
3. I sometimes have a set of factors (e.g. genotypes tested at two irrigation levels but not in separate trials but e.g. completely randomized in rows and columns). I now make a single factor representing all combinations of genotype x Factor2, but would like to get an ANOVA like output showing main and interactions effects. I now take the BLUEs and the residuals and use that to do a two factor ANOVA (with slightly to high number of df for residuals). Is it possible to implement a factorial design in statgenSTA, of course only for cases where it fits the assumptions on row column effects, so the spatial design of the experiment has to be fitting.

[BartJanvanRossum]

1. This sounds like a bug to me. "genotype" is converted to a factor internally since some of the packages used for modeling require it to be a factor. However this conversion is only done if it wasn't a factor in the input. Therefore the original levels of the column representing the genotype should always be kept. I tried to reproduce the problem, but I couldn't, for me it is working fine. Are you able to share an example where this goes wrong? If there are issues here with confidentiality, I think we are working in the same building, so I could also pass by your office if you want to just have a look.

2. I assume this is related to the first issue. If R converts a numerical column to a factor, it orders the factor levels lexicographically instead of numerically. statgenSTA then presents the levels in this order. However if 'correct' levels were already present, they shouldn't be converted at all and the original order should be kept.

3. I think this could be a nice extension to the package. I will discuss internally if we want to add this.