This action might not be possible to undo. Are you sure you want to continue?

4 Sign test

**Chapter 8 Paired observations
**

Timothy Hanson

Department of Statistics, University of South Carolina Stat 205: Elementary Statistics for the Biological and Life Sciences

1 / 19

2 Paired observations 8.4 Sign test Book review of two-sample t-test ingredients 2 / 19 .Section 8.

(b) HA : µD < 0. one of (a) HA : µD = 0. 3 / 19 .4 Sign test Paired designs Paired data arise when two of the same measurements are taken from the same subject.2 Paired observations 8. Let µD be the mean diﬀerence for the entire population. but under diﬀerent experimental conditions. The analysis focuses on the diﬀerence in response from treatment to control.Section 8. We want a conﬁdence interval for µD and will want to test H0 : µD = 0 vs. or (c) HA : µD > 0. Subjects often receive both a treatment Y1 and a control Y2 . Pairing observations reduces the subject-to-subject variability in the response.

1 Coﬀee and blood ﬂow Doctors studying healthy subjects measured myocardial blood ﬂow (MBF) (ml/min/g) during bicycle exercise before and after giving the subjects the equivalent of two cups of coﬀee (200 mg of caﬀeine). 4 / 19 .Section 8. Others have low blood ﬂow before and after. How does this analyis diﬀer from those in Chapters 6 and 7? Observations are collected on the same person. By focusing on the diﬀerences from the same individual before and after.1.4 Sign test Example 8.2 Paired observations 8. Some people have high blood ﬂow both before and after caﬀeine. we adjust for individuals tendancy to be high or low regardless.

1 blood ﬂow data 5 / 19 .Section 8.1.4 Sign test Example 8.2 Paired observations 8.

What does caﬀeine do to bloodﬂow? 6 / 19 .1 blood ﬂow data Each subject has a connected line (control and treatment).Section 8.4 Sign test Example 8.2 Paired observations 8.1.

alternative=”less”) gives P-value for HA : µD < 0.paired=TRUE.sample2.4 Sign test Paired analysis in R Null is H0 : µD = 0.alternative=”greater”) gives P-value for HA : µD > 0.paired=TRUE) gives P-value for HA : µD = 0.test(sample1.sample2.2 Paired observations 8.paired=TRUE. 7 / 19 .Section 8. t.sample2. t.test(sample1. t.test(sample1.

4.15375 We estimate µD as 1.4.4.58.2 Paired observations 8.3.15 ml/min/g. 8 / 19 . df = 7.44. p-value = 0.3.4 Sign test R code for bloodﬂow data > baseline=c(6.68 ml/min/g greater in the control group.52.6796357 sample estimates: mean of the differences 1.11.70.96.paired=TRUE) Paired t-test data: baseline and caffeine t = 5.81.5.caffeine.3.5.test(baseline.3.70.5.00127 alternative hypothesis: true difference in means is not equal to 0 95 percent confidence interval: 0.22.69.1878.37.5.06.53) > caffeine=c(4.5.89. We reject H0 : µD = 0 at the 5% level because P-value = 0.0013 < 0.2.20) > t. Caﬀeine signiﬁcantly reduces bloodﬂow.05.63 and 1.6278643 1.Section 8.4.27. We are 95% conﬁdent that the true mean bloodﬂow is between 0.

The paired sample t-test and conﬁdence interval are valid if (a) The sample size is large enough. or (b) the diﬀerences are approximately normal.2 Paired observations 8. If the two samples are sample1 and sample2. Normality can be checked with a normal probability plot. n > 30. say.Section 8. 9 / 19 .4 Sign test Validity of paired t-test (p. 306) Let n be the number of paired observations. type qqnorm(sample1-sample2) in R.

2.2 Paired observations 8. or else was given the placebo for the ﬁrst two weeks and then mCPP for the second two weeks.4 Sign test Example 8.Section 8. As part of the study the subjects were asked to rate how hungry they were at the end of each two-week period. 10 / 19 .4 Hunger rating During a weight loss study each of n = 9 subjects was given either the active drug m-chlorophenylpiperazine (mCPP) for two weeks and then a placebo for another two weeks.

Section 8.4 Sign test Hunger rating data 11 / 19 .2 Paired observations 8.

2 Paired observations 8.4 Sign test Hunger rating dotplot & normal probability plot 12 / 19 .Section 8.

4 Sign test R code for hunger rating > > > > drug=c(79.25. We are 95% conﬁdent that the drug reduces hunger between 4 and 55 points.54.placebo.99.107.101.7014.61. We reject H0 : µD = 0 at the 5% level because P-value = 0.test(drug. The drug signiﬁcantly reduces hunger.55556 We estimate µD as −30.Section 8.15.94.paired=TRUE) Paired t-test data: drug and placebo t = -2.5) placebo=c(78.027 < 0. 13 / 19 .77.54.326402 sample estimates: mean of the differences -29.52.05. df = 8.142.64) qqnorm(drug-placebo) t.107.784709 -4.48.02701 alternative hypothesis: true difference in means is not equal to 0 95 percent confidence interval: -54. p-value = 0.2 Paired observations 8.

2 Paired observations 8.3 Tryglycerides measured before and after exercise. 8. Need to have the paired observations come from very similar experimental units.4 Sign test 8. Examples: Ex.3 Paired designs Paired analyses reduce variability and make it easier to reject H0 : µD = 0.3.2 Case-control data from people matched on gender. 8.3. 8.3.Section 8. 14 / 19 . Ex. age.1 Two plants grown in the same container. Ex.

4 Sign test Example 8.57 1.98 1.3.09 1.4 Triglycerides and exercise Triglycerides play a role in coronary artery disease.87 1.43 1. Subject 1 2 3 4 5 6 7 Before 0.14 2.60 After 0.47 1.20 1.2 Paired observations 8.Section 8.14 2.03 1.18 1. Researchers measured blood triglycerides in seven men before and after a 10-week exercise program.13 3.51 15 / 19 .

This test looks at the number of diﬀerences D = Y1 − Y2 that are positive N+ and the number that are negative N− .2 Paired observations 8.g. 0. A P-value is based on the binomial distribution.4 The sign test The paired t-test assumes that diﬀerences follow a normal distribution. These numbers should be similar if H0 : ηD = 0 is true. If the data aren’t normal and the sample size is small. n < 30. The sign test focuses on the median diﬀerence ηD rather than the mean µD .4 Sign test 8. then you can use the sign test.5). 16 / 19 . Under H0 : ηD = 0. e.Section 8. N+ ∼ bin(n.

HA : ηD = 0. 17 / 19 .2 Paired observations 8. Ignore all output except the P-value.alternative=”less”). For HA : ηD > 0 use binom.alternative=”greater”). For HA : ηD < 0 use binom. second number is sample size. binom.n. Need to count the number of +’s and put that as ﬁrst number.n.test(N+ .test(N+ .4 Sign test Sign test in R In R.test(N+ .n) tests H0 : ηD = 0 vs.Section 8.

> binom.03 1.14 2.4 Sign test Example 8.14 2.51 Sign + + + + + + + N+ = 7 and N− = 0. p-value = 0.87 1. P-value should be small.3.Section 8.01563 18 / 19 .98 1.09 1.47 1.43 1.13 3.2 Paired observations 8. number of trials = 7.test(7.57 1.18 1.20 1.4 Triglycerides and exercise Subject 1 2 3 4 5 6 7 Before 0.60 After 0.7) number of successes = 7.

2 Paired observations 8.test(2. The t-test has greater power to reject H0 when data are really normal. p-value = 0.Section 8.4 Sign test Two more examples Hunger rating > binom.007812 P-value from t-test is 0. number of trials = 9.9) number of successes = 2. p-value = 0. number of trials = 8.test(8.02701.00127. not close at all. Caﬀeine and blood ﬂow > binom.8) number of successes = 8. 19 / 19 . fairly similar but t-test has smaller P-value (more power if diﬀerences really are normal).1797 P-value from t-test is 0.

- Effect of the Changing Nature
- How to Write a Reflection Essay
- Operations Management.pdf
- Forecasting Summary Notes.docx
- Assembly Line Balancing
- 92699530-Chapter-8.docx
- McDonalds Corporation Strategic Management Analysis
- strategicmanagementfullnotes-110824114832-phpapp01 (1).doc
- 119925579-Outline-for-Chapter-7-of-Strategic-Management (1).docx
- Types of Networks
- Cyber Crime
- Tax Evasion
- TAX EVASION.pdf
- XII-CHAP-1
- Financial Accounting
- Paper
- Accounting
- Research
- Financial Accounting Volume Two
- Financial Accounting Volume One
- Financial Accounting Volume Three
- Great Accounting Topics for Your Research Paper
- Sample Student Results & Verification Letters
- Project Summary (Introduction), Management Study and Accounting System - Revised.docx
- Fallacies of Relevance.docx

Sign up to vote on this title

UsefulNot usefulStatistics

Statistics

- Two Ind Pops Summary V3
- BiostatisticsQUIZ Answers
- bbaldi_ips_chapter07
- Independent Sample T-test
- Ti Instructions for Statistics
- Lumley - Permutation Tests
- Statistics Review 5 Comparison of Means
- A to Z Statistical Significance GET IT NOW
- A13-Two Sample T [Key]
- 07Ttests
- Comandi Per Verifica Di Ipotesi Con R v2
- Setar Models
- The Effect of Using Pictures on Students’ Speaking
- P ValueStatement
- Thesis
- Statistics Handbk Act09
- SPSS Exact Tests
- Solution Test 2
- The ASA s Statement on p Values Context Process and Purpose
- JMET Vol 7 No 2 Paper 1
- 1-s2.0-S1063458412007789-main.pdf
- Ap13 Statistics q4
- Multiple Regression Analysis- Inference
- The magical P-value
- Copy of Wk5_checkpoints GM 533
- Unit II - Parametric & Non-parametric Tests
- P-Value.rtf
- Biostatistics (Awwab) II
- Significance of P-Value, Box-Whisker Plots in Statistical Testing 260811
- _newbold_ism_10.pdf
- Paired Observation.pdf

Are you sure?

This action might not be possible to undo. Are you sure you want to continue?

We've moved you to where you read on your other device.

Get the full title to continue

Get the full title to continue reading from where you left off, or restart the preview.

scribd