Andrea Onofri
@onofriandreapg
Followers
109
Following
28
Statuses
62
Data analyses at the University of Perugia. Teaching Experimental methods in agriculture. #rstats, #Rmarkdown, #DataScience
Perugia, Umbria
Joined January 2016
The Pearson correlation coefficient is often used to describe the joint variability of two response variables from the same plots in designed field experiments. Inferences therof may be totally wrong: check out my results with #rstat and ‘sommer’ pkg at:
0
0
3
Just wondering about how to go back and forth from a variance-covariance matrix to a correlation matrix with #rstats . For a few methods, goto: . If you have other methods to suggest, pls. drop me a line…
0
0
3
Sometimes we want to derive information from a model fit, such as the half-life or the LD50. Here is how we can derive standard errors as well. Part 2: go to
0
0
2
Sometimes we want to derive information from a model fit, such as the half-life or the LD50. Here is how we can derive standard errors as well. Part 1, go to:
0
0
2
@jonathanbinder Hello Jonathan, these models are nice, but, in practice, the fit is often bad with many datasets! I’d be happy to discuss this with you: drop me a note at my email address!
0
0
1
In agriculture and pesticide research it is fundamental to ask the data the correct questions. In a new post, I am showing an example relating to the check for basic assumptions with #rstats. Check it out at:
0
0
2
In agriculture and pesticide research, we might like to compare different degradation or dose-response curves in a pairwise fashion (is the whole response curve for A different from the whole response curve for B?). A 'how-to' post in my blog, at:
1
3
9
@RHenry_UPL If you take the dose as a factor, with a common control, the design is not fully factorial (see this post: . Furthermore, pairwise comparisons with quantitative variables may be either illogical or inefficient (e.g., see
0
0
0
In pesticide research or, in general, agriculture research, we very commonly encouter experiments with two/three crossed factors and some other treatment (usually a check) that is not included in the factorial structure. Tips for data analysis are here:
0
0
3
Subsampling and repeated measures can occur together. What model do we fit, in that situation? Some hints in my new post, at , #rstats
2
2
5
Multi-year data with perennial crops are very different from multi-year data with annual crops and they should be analysed by using different methods. I make this point in a new post in my blog. Go to: #rstats
0
2
5
As an editor of International Journals, I am surprised to see so many submissions where true-replicates and pseudo-replicates are wrongly put on equal footing. I added a post to my blog to discuss such an issue. Go to:
1
1
13
A new post is out in my blog. This time, it is a dive in the coefficient of determination, that is often misused in the agricultural literature. Check it out here:
0
1
2
Split-plot experiments are rather common in agriculture, but, to my experience, the resulting datasets are not always analysed properly. Some hints here:
0
0
2
As an editor, I am surprised of how many manuscripts use GLMs… and how often the authors forget to check for the basic assumptions! Counts are not necessarily Poisson distributed and proportions are not necessarily binomial! We need to check that!
1
1
3
Yes, I know, #RStats reports the Pr(>F) column in ANOVA tables. But, please, do not write ‘Prob > F’ in your manuscript tables. It makes no sense apart from denoting lack of accuracy…
0
2
3