Deck 16: Design of Experiments

Question 1

What is meant by a trial and error experiment?

Answer 1

A trial and error experiment refers to experimenters who just try anything they can think of without any plan or rhyme or reason. They use guesswork and hope that a shot in the dark will provide a direct hit on the best solution.

Question 2

A study was performed on gas mileage at a local automotive testing ground. The purpose of the study was to try and determine whether tire pressure, timing setting, type of oil and/or type of gas had significant effects on gas mileage. Timing settings were low and high. Types of oil were 10W30 and 10W40. Two types of gasoline were used: mid-grade and hi-test. Tire pressures were tested at levels of 30 and 34 psi.

Answer 2

a. What is the response variable?
Gas mileage
b. What are the factors?
Timing, Oil, Gas, Pressure
c. What are the levels of the factors? Put these into a table. d. Describe a treatment condition.
A treatment condition is a single experiment for instance: L, 10w30, Mid-grade, 30 psi.
e. How many different experiments would need to be run to cover all of the permutations? How did you arrive at that figure?
2x2x2x2 = 16

Question 3

A paint manufacturer is studying two different paint formulas to determine which of two different types of paints would provide the most protection from the elements. This is judged by how quickly the paint peels under trial. The formulas are similar, but they use different levels of additives. Additive A can be 4 or 6 ounces/gallon. Additive B can be 1 or 2 ounces/gallon. Additive C can be ½ or 1 ounce/gallon. Answer the following questions about the experiment.

Answer 3

a. What is the response variable?
Time for paint to begin peeling.
b. What are the factors?
Additive A, Additive B, Additive C
c. What are the levels of the factors? Put these into a table. d. Describe a treatment condition.
A treatment condition is a single experiment for instance: Additive A 4 ounces, Additive B 2 ounces, Additive C 1 ounce.
e. How many different experiments would need to be run to cover all of the permutations? How did you arrive at that figure?
2x2x2 = 8

Question 4

What is the response variable?

Question 5

What is a factor?

Question 6

What are two aspects of an experiment design that will enable you to tell whether or not the experiment is well designed?

Question 7

A company is trying to increase production rates at a particular machine. They are considering making two changes, but they aren't sure which changes will be effective. They would like to run an experiment to determine the best combination of these changes. The changes are:

Question 8

What is an interaction?

Question 9

When experimenters talk about the effects of an experiment, what are they referring to?

Question 10

What is a level?