Statistics: controlled experiments vs observational studies
Applications of Statistics
- Medical,
- Salk Vaccine: the biggest public health experiment ever.
There were two studies done, one was a randomized controll double blind and
the other was not randomized or double blind (NFIP). (p.6)
randomized | NFIP |
| Rate | | Rate |
treatment | 28 | Grade 2 (vaccine) | 25 |
control | 71 | Grades 1 & 3 (control) | 54 |
no consent | 46 | Grade 2 (no consent) | 44 |
- safety of anasthetics
- setting drug dosages
- smoking
- epidemics
- literature: authorship of the Federalist Papers
- Government: do speed limits reduce accidents? Discrimination cases.
- market research
- election polls
- quality control for manufacturing
- accounting
Controlled Experiments
When comparing numbers for 2 (or more) groups, it is important that the groups be
similar enough, otherwise the comparison is not valid. For example, in the NFIP Salk
vaccine study, the groups were not similar with respect to some important factors or
variables. When this happens, the results of the two variables are confounded.
Descriptive Statistics
- histograms
- mean (average)
- standard deviation