Dear colleagues,
Recently, I submitted some data analysis results to a significant organization. All work submitted undergoes blind peer-review. For the data analysis, I always use R.
Yesterday, I received the comments of a reviewer who says that I must redo all the analysis using Mplus or SAS because the R package in general is not ready to be used for serious governmental contracts. Obviously, the reviewer is seriously misled and biased but my opinion is not enough.
My question is: How can I respond to that reviewer? How can I reassure the organization that R is a reliable package for data analysis?
While the body of R users increases, this evolves to a significant question.
Please help with ideas!!!!
Dear Ιάσωνα,
You can start by reading this thread:
https://www.researchgate.net/post/R_or_SAS_which_one_is_the_best_statistical_software_used_in_medical_field
My contribution (which was downvoted) is here:
https://www.researchgate.net/post/R_or_SAS_which_one_is_the_best_statistical_software_used_in_medical_field/4
where I compared many programs for the same task.
I think Jochen will help you at the theoretical level, my current advice is that R has a better 'engine', is more accurate than SAS.
Dear Lasonas,
I suggest you to browse the Revolution Analytics web site :
http://www.revolutionanalytics.com/why-revolution-analytics
You could notably find a list of companies that use R :
http://www.revolutionanalytics.com/companies-using-r
This list prove that using R is appropriate even for serious governmental contracts.
For perfect reproducibility, you just have to mention in your manuscript the number version of R and packages you used.
HTH
I think having a list of "references" showing where R was used in "serious governmental contracts" should be the most convincing argument.
May question to the reviewers was how they judge if a software is "ready to be used for serious governmental contracts". What exactly does a software have to provide, what are the criteria?
I generally do not understand why companies or governments often trust proprietary software more than open source software. Sometimes companies want to use somehow "officially certified" software only (e.g. NASA). ISO/IEC 90003 is a common cetrification for software. Not sure if there exists special certificates for statistical software.
Here is "A Guidance Document for the Use of R in Regulated Clinical Trial Environments"
http://www.r-project.org/doc/R-FDA.pdf
this is one example which demostrates that a package of R software for conducting generalized estimating equations in a clinical trial has a better performance
http://www.sciencedirect.com/science/article/pii/S0167947314000863
Also, most of the packages are developed by people working in scientific research, so using those packages is the best choice.
I have never found a package with misleading results, however, as the R is an open sofware it can always be an important question...
- Badges
- Science topic