Theorem 1. If something can go wrong in conducting an experiment, it will.

Theorem 2. The probability of successfully completing an experiment is inversely proportional to the number of runs.

Theorem 3. Never let one person design and conduct an experiment alone, particularly if that person is a subject-matter expert in the field of study.

Theorem 4. All experiments are designed experiments; some of them are designed well, and some of them are designed really badly. The badly designed ones often tell you nothing.

Theorem 5. About 80 percent of your success in conducting a designed experiment results directly from how well you do the pre-experimental planning.

Theorems by Dr. Douglas Montgomery

Basic principles of experimental design

Statistical design of the experiment

Process of planning experiment so that the appropriate data can be analyzed by statistical methods….experiment answers the question it was proposed to answer

Statistics Consulting Lab is located in BR 211

METHOD OF ANALYSIS DEPENDS DIRECTLY ON THE DESIGN EMPLOYED!!!

Statistics Consulting Lab is located in BR 211

Principles of experimental design

1. Randomization

Randomization is one of the most important ideas in designing an experiment; without it, there can be no conclusive evidence!!

Statistics Consulting Lab is located in BR 211

2. Replication

Important question…how many replicates do I need? (replicate versus repeated)

Without replication, there can be no test for interaction

Statistics Consulting Lab is located in BR 211

3. Blocking

Block or control variables that will influence the outcome, but which you are not interested in. But be careful as to what your blocking will do.

Statistics Consulting Lab is located in BR 211

The Statistics faculty

Dr. James Blum
BR 211B / Varying Coefficient Models, Non-parametric Regression, Categorical Data Analysis, Statistics Education
Dr. Edward Boone
BR 225 / Bayesian Modeling and Inference, Missing Data, Spatial Statistics, Time Series, Model Selection, Statistical Simulation and Computation, Biological, Ecological and Environmental Applications
Dr. Dargan Frierson
BR 211A / Mathematical Statistics (Sequential Analysis, Nonparametric Statistics), Data Analysis, Statistics Education
Dr. Subramanyam Kasala
BR 251 / Multivariate Analysis, Time Series, Inference
Dr. Shiva K. Saksena
BR 244 / Bayesian Methods, Inference, Simulation, Biostatistics
Dr. Susan Simmons
BR 230 / Hierarchical Models, Bayesian Inference, Decision Theory, Statistical Computing, Statistics Education

Preparing for the First Meeting

The primary purpose of the first meeting is to familiarize the statistical consultant with your project and objectives so that appropriate advice may be given. Depending on the current state of your project, this may involve many different issues. You should be prepared to discuss any of the following as appropriate to your progress to date:

  • Objectives of your research and/or hypotheses you wish to test.
  • Methods of data collection and recording.
  • Experimental and/or survey designs.
  • Analysis methods--including those that are considered "traditional" in your field of study.
  • Statistical software packages.
  • Your timetable for completion of the project and its various stages.

With the above in mind, we offer answers to some general questions you may have:

When should the first consulting meeting take place?

The easiest answer is: As soon as possible. Ideally, data collection and management methods, experimental design(s), summary procedures and analysis techniques would all be in place from the outset. Early assistance on design and data collection issues can often result in less work for the experimenter, a smoother transition from the experimental phase to the analysis phase and, possibly, better results. Of course, this does not mean that someone who has already conducted and experiment and gather data would not be able to receive help. We can offer assistance at any stage of your project, but our motto is: "The earlier, the better."

What materials should I plan to bring to the first meeting?

Any formal or informal documents related to your project (plans for experiment/research, prospectus, articles/references to similar work in the field) will be helpful of course. Any data you have gathered and analyses/summaries you have done--tables, graphs and the like. Also, any collaborators who are available to aid in the description of efforts undertaken so far and questions of interest. In fact, we require that the advisor or another committee member be present at the first meeting for any projects being undertaken by a graduate student.

What happens after the first meeting?

For people looking for confirmation of or suggestions on a completed analysis, the first meeting may well be the last. However, for most people, further consulting services will be necessary. It is one of the goals of the first meeting to determine what level of involvement is required on the part of the consultant and to outline a plan of future collaboration.