Sections Deleted From the CTSA

To help with the study design for Vanderbilt internal investigator initiated VICTR studies, before an investigator submits a VICTR study protocol, he/she will be required to consult a VICTR study design biostatistician from the core unless he/she has formal biostatistical support such as that through the biostatistical collaboration plan established between the Department of Biostatistics and various departments in the medical school. The biostatistician will function as a collaborator to help the investigator with study design issues and help with putting the biostatistical component into the study protocol.

Specifically, the design component of the core will assist in refining study objectives, response variables, and measurement techniques. It will assist in developing research protocols and will provide experimental design, study design, and an initial statistical analysis plan. It will ensure that rigorous standards of scientific inquiry and inference are applied to all projects consulting the core. The core will maximize the yield of human and animal studies, making sure that sample sizes are appropriate, so that investment in time and effort and exposure to risks by subjects is worthwhile. Factorial experimental designs and flexible sequential Bayesian and frequentist designs will be incorporated when feasible to reduce the total sample size for a series of related studies and to, in some cases, shorten the time needed to obtain the needed statistical evidence.

As stated by Hahn, p. 349 1, "proper experimental design is more important than sophisticated statistical analysis ... The main reason for designing an experiment statistically is to obtain unambiguous results at minimum cost." According to Cochran 2 (Vol. 1, p. 286),

The core will ensure that studies and experiments done in the projects meet the standards of good statistical design. Before selecting a final design, alternative designs will be evaluated with respect to feasibility, sensitivity, and validity. Investigators will be trained concerning the execution of the final chosen design as discussed in the Training section. Nowhere are the needs of sound design more pressing than in the context of small patient studies. Employing optimal designs will help small studies to be more informative. Sequential designs will also play key roles, for example, by allowing small equivocal studies to be extended.

These principles include the following:

  • Use randomization to control for unknown factors.
  • Use blocking and factorial experiments when more than one variable has an effect.
  • When possible, validate the results of one technology using a different, complementary technology.
  • Perform the sequence of experiments to provide valid estimates of the sources and magnitude of variability.

Statisticians are full of stories of experimenters who came to the statisticians only after the wrong data were collected and experimenters are likely full of stories that surely document missteps that they would like to make sure that their colleagues avoid.

[The following was delete because it was redundant.]

Communication for VICTR Biostatistics Program

We will make the Department of Biostatistics' twiki web server available to all other CTSA Biostatistics Cores and will develop template pages for sharing of information, teaching materials, and ideas for fostering biomedical research and making it more efficient. A staff biostatistician will coordinate the use of this site by all interested CTSA grantees.
Topic revision: r1 - 10 Mar 2006, FrankHarrell
 

This site is powered by FoswikiCopyright © 2013-2017 by the contributing authors. All material on this collaboration platform is the property of the contributing authors.
Ideas, requests, problems regarding Vanderbilt Biostatistics Wiki? Send feedback