VCU STAT 210 - Lecture5 (44 pages)

Previewing pages 1, 2, 3, 21, 22, 23, 42, 43, 44 of 44 page document View the full content.
View Full Document

Lecture5



Previewing pages 1, 2, 3, 21, 22, 23, 42, 43, 44 of actual document.

View the full content.
View Full Document
View Full Document

Lecture5

49 views


Pages:
44
School:
Virginia Commonwealth University
Course:
Stat 210 - Basic Practice of Statistics
Basic Practice of Statistics Documents
Unformatted text preview:

STAT 210 Lecture 5 Design of Experiments September 6 2017 Example to think about Suppose the population is all students enrolled in STAT 210 this semester There are ten sections and to select a sample of 40 students the following was done 4 students were randomly selected from each of the 10 sections What type of sampling procedure is this Test 1 Friday September 8 Questions from 1 00 to 1 10 Test from 1 10 to 1 50 papers due at 10 50 Covers chapters 1 and 2 pages 1 42 Combination of multiple choice questions and a few written questions Practice tests posted on Blackboard Practice Problems Pages 38 42 Relevant problems II 7 II 8 II 9 II 10 Recommended problem II 10 Additional Reading and Examples Pages 36 37 Top Hat B Design of Experiments Design and carry out an experiment that will allow us to evaluate the effect of a treatment or treatments Example Suppose we are interested in the effect of a new drug in reducing blood pressure Design of Experiments 1 Experimental Units the subjects or individuals or units on which the measurements are made Example People medically identified as suffering from high blood pressure diastolic pressure measurement of 90 or higher Design of Experiments Experimental Units the subjects or individuals or units on which the measurements are made Treatment an experimental condition applied to the experimental units xample A new blood pressure medicine being evaluated and the three leading blood pressure medicines that have been previously approved Design of Experiments Response the characteristic that is measured on each experimental unit xample The diastolic blood pressure measurement of the patients in the study or possibly the reduction in the diastolic blood pressure measurement before and after treatment Design of Experiments Response the characteristic that is measured on each experimental unit Comparison we want to compare the results for one group or treatment with the results for another group or treatment or groups xample Compare the results for the new blood pressure drug to those of the three leading blood pressure medicines and no medication Design of Experiments Comparison we want to compare the results for one group or treatment with the results for another group or treatment or groups Treatment Group s the group or groups of experimental units who receive the treatment s xample The patients who get the new blood pressure drug are one treatment group those who get each of the other three leading blood pressure drugs would be other treatment groups for a total of 4 Design of Experiments 5 Treatment Group the group s of experimental units who receive the treatment s 6 Control Group a group of experimental units who do not receive the treatment but instead receive either a fake treatment or nothing at all Example A group of patients who receive no active medication is the control group A control group is not required if there are multiple treatment groups Design of Experiments 5 Treatment Group the group of experimental units who receive the treatment 6 Control Group a group of experimental units who do not receive the treatment but instead receive either an alternative treatment or nothing at all 7 Placebo a fake treatment that is given to the experimental units in the control group to prevent them from knowing to which group they belong This is important if there is not another treatment to compare the new treatment to Design of Experiments 7 Placebo a fake treatment that is given to the experimental units in the control group to prevent them from knowing to which group they belong This is important if there is not another treatment to compare the new treatment to Example If there is a group of people not receiving any medication a control group then they should receive a fake drug that looks similar to actual medication treatments Design of Experiments 8 Replication Each group should contain more than one experimental unit which should reduce chance variation in the results and reduce the effects of fluke occurrences Example Select 500 high blood pressure patients to participate Randomly assign 100 of these patients to each of the five groups 100 to the new blood pressure medicine group 100 each to each of the three leading blood pressure drugs already approved and 100 to the control group to receive the placebo but no active medicine Design of Experiments 9 Randomization The experimental units used in the experiment should be randomly chosen and they should be randomly assigned to the different groups This can involve using the table of random digits Example Have each of the 500 patients draw a marble Red they receive the new blood pressure medicine Blue they receive leading medicine A Yellow they receive leading medicine B White they receive leading medicine C Green they receive the placebo Design of Experiments 9 Randomization The experimental units used in the experiment should be randomly chosen and they should be randomly assigned to the different groups This can involve using the table of random digits Example The randomization can also be done by using the Table of Random Digits Design of Experiments 10 Blinding The experimental units should not know to which group they have been assigned This will hopefully reduce the occurrence of psychological effects and is the reason for using placebos Example If the new and old blood pressure medications and the placebo all look the same and if the patients are randomly assigned to the 5 groups then the experiment would be blind Design of Experiments 11 Double blinding To reduce the possibility of those running the experiment influencing the results in any way those who are in contact with the experimental units also should not know to which group they belong Example Have one person recording to which group each patient belongs and recording the results but those who are administering the drugs and measuring the results of the medication should not know which group the patient is in Design of Experiments 12 Bias Anything that systematically favors some outcomes over others and which would cause us to question the validity of our results Example If the treatment group consists only of people in a high risk blood pressure group while the control group consists of those in the marginal group then the results may be biased to indicate that the new drug is better or worse than it actually is To avoid bias the treatment group s and control group should be as similar as possible except for the


View Full Document

Access the best Study Guides, Lecture Notes and Practice Exams

Loading Unlocking...
Login

Join to view Lecture5 and access 3M+ class-specific study document.

or
We will never post anything without your permission.
Don't have an account?
Sign Up

Join to view Lecture5 and access 3M+ class-specific study document.

or

By creating an account you agree to our Privacy Policy and Terms Of Use

Already a member?