Assessment of Biostatistics Expertise Available to Biomedical Research Groups
The following scale is suggested for assessing the degree of biostatistical expertise that a lab or research group has available to them. If the group uses a lab member for their statistical design and analysis they can use this scale to roughly rate that person's level of expertise. If the group uses a statistician outside their group, than can "rate" that person. The highest applicable score should be chosen. In what follows, "full course" refers to a full semester or quarter course. "Statistics" refers to biostatistics or statistics.
Score |
Definition |
0 |
No formal statistics courses or no statistician in group |
1 |
1 or 2 full course in statistics |
2 |
3+ full courses in statistics |
3 |
5+ full courses in statistics |
4 |
minor in statistics with B.S. or B.A. in quantitative field |
5 |
4+ full courses in statistics with 5+ years of experience in statistical analysis |
6 |
B.S. or B.A. in statistics |
7 |
M.S or M.A. in economics, psychology, math, computer science, or related fields with 3+ years of experience in statistical analysis |
8 |
4+ graduate courses in statistics |
9 |
4+ doctoral level courses in statistics |
10 |
PhD in quantitative psychology or econometrics with 5+ years of experience analyzing biomedical data |
11 |
Master's degree in statistics |
12 |
PhD in statistics |
Basic Assessment of Reproducible Statistical Computing Practices
Select the highest applicable score. In what follows, "statistical package" refers to a software system or platform dedicated to statistical analysis or having high-level statistical functions, for example R, Stata, SAS, SPSS, Python. What doesn't qualify as "statistical package" is anything related to Excel or that requires the analyst to write low-level functions even for regression analysis (e.g., C, Fortran, Matlab). "Interactive" refers to point and click software, but not exclusively. "Script" refers to high-level statistical language commands, or to statistical programming code. For example, you can write scripts in R, Stata, SAS, SPSS, Python, and SQL.
Score |
Definition |
5 |
Always write your own analysis scripts |
6 |
Generally embed analysis scripts inside reports and using literate programming techniques for maximum reproducibility |
7 |
Same as 5 but in addition you script all data manipulation/management and do not let the data pass through Excel at any stage |
4 |
Use an interactive statistical package and always saving the script the package developed |
3 |
Use an interactive statistical package having the capability of saving scripts but not always turning on this option |
2 |
Use an interactive statistical package that does not have the capability of creating/saving scripts defining what you did |
0 |
Use Excel |
1 |
Use Excel for inputting or manipulating some of the data but not for any statistical analysis or graphics |
If you selected category 5 or 6 above, this means that you did not copy and paste results into a report document.