|Some Messy Experimental Designs|
Dallas E. Johnson Professor Emeritus, Kansas State University
Monday, October 17th 12 pm
This talk will provide a short introduction to the process of designing an experiment and then it will talk about some of the more interesting experimental designs that I have encountered. Some are good designs but complicated to analyze, others were really bad designs from which no useful information could be obtained.
Here is one example that will be discussed. It is a good design, but difficult to analyze. An experiment was conducted to determine the effects of protein and energy levels in the diets of lambs. There were two protein levels, high and low, and four energy levels which we will call E1, E2, E3, & E4. E1 and E2 were both lower levels of energy and E3 and E4 were higher levels of energy. Each of the eight diet combinations were randomly assigned to two lambs selected at random from a group of 16 lambs. Data was collected for a week, then each lamb that received E1 during the first week, received E2 during a second week. Those getting E2 the first week, received E1 the second week. Similarly, each lamb that received E3 during the first week, received E4 during the second week. Those getting E4 the first week, received E3 the second week. How should the resulting data be analyzed?