This preview shows pages 1–3. Sign up to view the full content.
This preview has intentionally blurred sections. Sign up to view the full version.
View Full Document
Unformatted text preview: STAT 218 Week #6, 7(Monday) Chapter 9 / Chapter 8 Questions from Previous Lecture / HW (10 minutes) ANNOUNCEMENT: Test next Week Chapters 69, one page of notes, tables in book, CHAPTER 9  PAIRED DATA Paired Design / Paired Data Often possible to study your population in pairs The Concept the subject (or experimental unit) becomes his/her own control Used with two treatments The subject receives BOTH treatments Examples Test a sunscreen use left and right side of body for each treatment Sunglass protection put a different lens in the two eye lens Hand Lotion Different lotion on left and right hands Evaluate 6 minute walking distance before and after a treatment Evaluate growth factors in twins Important Subjects are selected at random from your population, as they are in all experiments The application of the treatment to the left or right side is done via a randomization process (and/or the data is representative of the population) The observations from subject to subject are independent But the observation of the two treatments on the same subject are NOT independent Illustration Compare A natural herbal oil to a chemically developed oil for sunscreen protection 10 subjects Measure redness of sunburn on both sides of a back after 2 hours in the sun 10 backs Left/Right randomization (subject, application person unblinded, observer) Measurement of data and differences Analysis is done as a one sample problem 1 Statistical View The Concept the subject becomes his/her own control The data is looked at as differences between the two Again, the key point: The data we analyze is not the original data but the difference in response of the two treatments Hypothesis tests and confidence intervals are done based upon comparing two treatments But the analysis is identical to the one sample methods of Chapter 6. o CI = Estimate +/ (t/Z) * SE o Hypothesis test: Ho: u 1u 2 = 0 or d = 0 vs Ha: u 1u 2 0 or d 0 o The sample mean and SE are calculated the same as Chapter 6 Example Data Set: Chronic Lung Hypertension Patients with inability to breath normally are given a new product and a placebo, on two different days. The order they receive these products is randomized via a coin flip. The endpoint is the distance walked (meters) in 6 minutes. The research is interested is evaluating whether his new product is effective Assume the subjects were obtained randomly and the data (the difference) follows a normal distribution. Assume this is the data Subject NEW DRUG Walk Distance (feet) PLACEBO Walk Distance (ft) DIFFERENCE NewPlac (feet) Sam 400 375 25 Fred 250 220 30 Joe 404 41915 Mary 500 450 50 Clyde 350 330 20 Bob 430 4355 Sally 460 420 40 Dude 190 175 15 ESTIMATION Confidence Interval What is the 95% CI for the difference in walking distance?...
View
Full
Document
This note was uploaded on 05/07/2008 for the course STAT 218 taught by Professor Staff during the Spring '08 term at Cal Poly.
 Spring '08
 staff

Click to edit the document details