Chapter 5: Hypothesis testing

Similar documents
Hypothesis Testing (2) Barrow, Statistics for Economics, Accounting and Business Studies, 4 th edition Pearson Education Limited 2006

Common Large/Small Sample Tests 1/55

Notes on Hypothesis Testing, Type I and Type II Errors

Chapter 22. Comparing Two Proportions. Copyright 2010, 2007, 2004 Pearson Education, Inc.

Introduction to Econometrics (3 rd Updated Edition) Solutions to Odd- Numbered End- of- Chapter Exercises: Chapter 3

Chapter 22. Comparing Two Proportions. Copyright 2010 Pearson Education, Inc.

Recall the study where we estimated the difference between mean systolic blood pressure levels of users of oral contraceptives and non-users, x - y.

MOST PEOPLE WOULD RATHER LIVE WITH A PROBLEM THEY CAN'T SOLVE, THAN ACCEPT A SOLUTION THEY CAN'T UNDERSTAND.

1036: Probability & Statistics

Math 140 Introductory Statistics

Chapter 13: Tests of Hypothesis Section 13.1 Introduction

STA Learning Objectives. Population Proportions. Module 10 Comparing Two Proportions. Upon completing this module, you should be able to:

University of California, Los Angeles Department of Statistics. Hypothesis testing

Statistics. Chapter 10 Two-Sample Tests. Copyright 2013 Pearson Education, Inc. publishing as Prentice Hall. Chap 10-1

MA238 Assignment 4 Solutions (part a)

Data Analysis and Statistical Methods Statistics 651

LESSON 20: HYPOTHESIS TESTING

If, for instance, we were required to test whether the population mean μ could be equal to a certain value μ

This chapter focuses on two experimental designs that are crucial to comparative studies: (1) independent samples and (2) matched pair samples.

Chapter 4 Tests of Hypothesis

Topic 18: Composite Hypotheses

PSYCHOLOGICAL RESEARCH (PYC 304-C) Lecture 9

- E < p. ˆ p q ˆ E = q ˆ = 1 - p ˆ = sample proportion of x failures in a sample size of n. where. x n sample proportion. population proportion

Tests of Hypotheses Based on a Single Sample (Devore Chapter Eight)

Lecture 6 Simple alternatives and the Neyman-Pearson lemma

Overview. p 2. Chapter 9. Pooled Estimate of. q = 1 p. Notation for Two Proportions. Inferences about Two Proportions. Assumptions

Lecture 5: Parametric Hypothesis Testing: Comparing Means. GENOME 560, Spring 2016 Doug Fowler, GS

Math 152. Rumbos Fall Solutions to Review Problems for Exam #2. Number of Heads Frequency

HYPOTHESIS TESTS FOR ONE POPULATION MEAN WORKSHEET MTH 1210, FALL 2018

Lesson 2. Projects and Hand-ins. Hypothesis testing Chaptre 3. { } x=172.0 = 3.67

Final Examination Solutions 17/6/2010

Chapter 22: What is a Test of Significance?

Successful HE applicants. Information sheet A Number of applicants. Gender Applicants Accepts Applicants Accepts. Age. Domicile

Hypothesis Testing. Evaluation of Performance of Learned h. Issues. Trade-off Between Bias and Variance

A statistical method to determine sample size to estimate characteristic value of soil parameters

This is an introductory course in Analysis of Variance and Design of Experiments.

Stat 200 -Testing Summary Page 1

Properties and Hypothesis Testing

April 18, 2017 CONFIDENCE INTERVALS AND HYPOTHESIS TESTING, UNDERGRADUATE MATH 526 STYLE

ST 305: Exam 3 ( ) = P(A)P(B A) ( ) = P(A) + P(B) ( ) = 1 P( A) ( ) = P(A) P(B) σ X 2 = σ a+bx. σ ˆp. σ X +Y. σ X Y. σ X. σ Y. σ n.

2 1. The r.s., of size n2, from population 2 will be. 2 and 2. 2) The two populations are independent. This implies that all of the n1 n2

Power and Type II Error

Class 23. Daniel B. Rowe, Ph.D. Department of Mathematics, Statistics, and Computer Science. Marquette University MATH 1700

Interval Estimation (Confidence Interval = C.I.): An interval estimate of some population parameter is an interval of the form (, ),

Mathematical Notation Math Introduction to Applied Statistics

Statistical Inference About Means and Proportions With Two Populations

Composite Hypotheses

1 Inferential Methods for Correlation and Regression Analysis

Chapter 13, Part A Analysis of Variance and Experimental Design

1 Constructing and Interpreting a Confidence Interval

Homework 5 Solutions

Section 9.2. Tests About a Population Proportion 12/17/2014. Carrying Out a Significance Test H A N T. Parameters & Hypothesis

Lecture Notes 15 Hypothesis Testing (Chapter 10)

SOLUTIONS y n. n 1 = 605, y 1 = 351. y1. p y n. n 2 = 195, y 2 = 41. y p H 0 : p 1 = p 2 vs. H 1 : p 1 p 2.

Comparing your lab results with the others by one-way ANOVA

Comparing Two Populations. Topic 15 - Two Sample Inference I. Comparing Two Means. Comparing Two Pop Means. Background Reading

Statistical inference: example 1. Inferential Statistics

Last Lecture. Wald Test

BIOS 4110: Introduction to Biostatistics. Breheny. Lab #9

6 Sample Size Calculations

One-Sample Test for Proportion

Frequentist Inference

STATISTICAL INFERENCE

Chapter 23: Inferences About Means

Chapter two: Hypothesis testing

Chapter 6 Sampling Distributions

GG313 GEOLOGICAL DATA ANALYSIS

Expectation and Variance of a random variable

Sampling Distributions, Z-Tests, Power

Introductory statistics

Sample Size Determination (Two or More Samples)

ENGI 4421 Confidence Intervals (Two Samples) Page 12-01

A Confidence Interval for μ

MIT : Quantitative Reasoning and Statistical Methods for Planning I

Problem Set 4 Due Oct, 12

1 Constructing and Interpreting a Confidence Interval

Sampling Error. Chapter 6 Student Lecture Notes 6-1. Business Statistics: A Decision-Making Approach, 6e. Chapter Goals

Hypothesis Testing. H 0 : θ 1 1. H a : θ 1 1 (but > 0... required in distribution) Simple Hypothesis - only checks 1 value

STAT431 Review. X = n. n )

Worksheet 23 ( ) Introduction to Simple Linear Regression (continued)

The Hong Kong University of Science & Technology ISOM551 Introductory Statistics for Business Assignment 3 Suggested Solution

Chapter 8: Estimating with Confidence

FACULTY OF MATHEMATICAL STUDIES MATHEMATICS FOR PART I ENGINEERING. Lectures

STAT 155 Introductory Statistics Chapter 6: Introduction to Inference. Lecture 18: Estimation with Confidence

Topic 6 Sampling, hypothesis testing, and the central limit theorem

Working with Two Populations. Comparing Two Means

Econ 325 Notes on Point Estimator and Confidence Interval 1 By Hiro Kasahara

Chapter 20. Comparing Two Proportions. BPS - 5th Ed. Chapter 20 1

Biostatistics for Med Students. Lecture 2

Chapter 11: Asking and Answering Questions About the Difference of Two Proportions

Sampling, Sampling Distribution and Normality

Statistics 20: Final Exam Solutions Summer Session 2007

Topic 9: Sampling Distributions of Estimators

Exam II Review. CEE 3710 November 15, /16/2017. EXAM II Friday, November 17, in class. Open book and open notes.

Lecture 2: Monte Carlo Simulation

(7 One- and Two-Sample Estimation Problem )

Lecture 9: Independent Groups & Repeated Measures t-test

independence of the random sample measurements, we have U = Z i ~ χ 2 (n) with σ / n 1. Now let W = σ 2. We then have σ 2 (x i µ + µ x ) 2 i =1 ( )

Chapter 9, Part B Hypothesis Tests

A quick activity - Central Limit Theorem and Proportions. Lecture 21: Testing Proportions. Results from the GSS. Statistics and the General Population

Transcription:

Slide 5. Chapter 5: Hypothesis testig Hypothesis testig is about makig decisios Is a hypothesis true or false? Are wome paid less, o average, tha me? Barrow, Statistics for Ecoomics, Accoutig ad Busiess Studies, 5 th editio Pearso Educatio Limited 009

Slide 5. Priciples of hypothesis testig The ull hypothesis is iitially presumed to be true Evidece is gathered, to see if it is cosistet with the hypothesis If it is, the ull hypothesis cotiues to be cosidered true (later evidece might chage this) If ot, the ull is rejected i favour of the alterative hypothesis Barrow, Statistics for Ecoomics, Accoutig ad Busiess Studies, 5 th editio Pearso Educatio Limited 009

Slide 5.3 Two possible types of error Decisio makig is ever perfect ad mistakes ca be made Type I error: rejectig the ull whe true Type II error: acceptig the ull whe false Barrow, Statistics for Ecoomics, Accoutig ad Busiess Studies, 5 th editio Pearso Educatio Limited 009

Slide 5.4 Type I ad Type II errors True situatio Decisio H 0 true H 0 false Accept H 0 Reject H 0 Correct decisio Type I error Type II error Correct decisio Barrow, Statistics for Ecoomics, Accoutig ad Busiess Studies, 5 th editio Pearso Educatio Limited 009

Slide 5.5 Avoidig icorrect decisios We wish to avoid both Type I ad II errors We ca alter the decisio rule to do this Ufortuately, reducig the chace of makig a Type I error geerally meas icreasig the chace of a Type II error Hece a trade off Barrow, Statistics for Ecoomics, Accoutig ad Busiess Studies, 5 th editio Pearso Educatio Limited 009

Slide 5.6 Diagram of the decisio rule f x H H 0 Type I error Type II error x Rejectio regio x D No-rejectio regio Barrow, Statistics for Ecoomics, Accoutig ad Busiess Studies, 5 th editio Pearso Educatio Limited 009

Slide 5.7 How to make a decisio Where do we place the decisio lie? Set the Type I error probability to a particular value. By covetio, this is 5%. This is kow as the sigificace level of the test. It is complemetary to the cofidece level of estimatio. 5% sigificace level 95% cofidece level. Barrow, Statistics for Ecoomics, Accoutig ad Busiess Studies, 5 th editio Pearso Educatio Limited 009

Slide 5.8 Example: How log do LEDs last? A maufacturer of LEDs claims its product lasts at least 5,000 hours, o average. A sample of 50 LEDs is tested. The average time before failure is 4,900 hours, with stadard deviatio 500 hours. Should the maufacturer s claim be accepted or rejected? Barrow, Statistics for Ecoomics, Accoutig ad Busiess Studies, 5 th editio Pearso Educatio Limited 009

Slide 5.9 The hypotheses to be tested H 0 : m = 5,000 H : m < 5,000 This is a oe tailed test, sice the rejectio regio occupies oly oe side of the distributio Barrow, Statistics for Ecoomics, Accoutig ad Busiess Studies, 5 th editio Pearso Educatio Limited 009

Slide 5.0 Should the ull hypothesis be rejected? Is 4,900 far eough below 5,000? Is it more tha.64 stadard errors below 5,000? (.64 stadard errors below the mea cuts off the bottom 5% of the Normal distributio.) z x m s 4,900 5,000 500 80.79 Barrow, Statistics for Ecoomics, Accoutig ad Busiess Studies, 5 th editio Pearso Educatio Limited 009

Slide 5. Should the ull hypothesis be rejected? (cotiued) 4,900 is.79 stadard errors below 5,000, so falls ito the rejectio regio (bottom 5% of the distributio) Hece, we ca reject H 0 at the 5% sigificace level or, equivaletly, with 95% cofidece. If the true mea were 5,000, there is less tha a 5% chace of obtaiig sample evidece such as x 4,900 from a sample of = 80. Barrow, Statistics for Ecoomics, Accoutig ad Busiess Studies, 5 th editio Pearso Educatio Limited 009

Slide 5. Formal layout of a problem. H 0 : m = 5,000 H : m < 5,000. Choose sigificace level: 5% 3. Look up critical value: z* =.64 4. Calculate the test statistic: z = -.79 5. Decisio: reject H 0 sice -.79 < -.64 ad falls ito the rejectio regio Barrow, Statistics for Ecoomics, Accoutig ad Busiess Studies, 5 th editio Pearso Educatio Limited 009

Slide 5.3 Oe vs two tailed tests Should you use a oe tailed (H : m < 5,000) or two tailed (H : m 5,000) test? If you are oly cocered about fallig oe side of the hypothesised value (as here: we would ot worry if LEDs lasted loger tha 5,000 hours) use the oe tailed test. You would ot wat to reject H 0 if the sample mea were aywhere above 5,000. If for aother reaso, you kow oe side is impossible (e.g. demad curves caot slope upwards), use a oe tailed test. Barrow, Statistics for Ecoomics, Accoutig ad Busiess Studies, 5 th editio Pearso Educatio Limited 009

Slide 5.4 Oe vs two tailed tests (cotiued) Otherwise, use a two tailed test. If usure, choose a two tailed test. Never choose betwee a oe or two tailed test o the basis of the sample evidece (i.e. do ot choose a oe tailed test because you otice that 4,900 < 5,000). The hypothesis should be chose before lookig at the evidece! Barrow, Statistics for Ecoomics, Accoutig ad Busiess Studies, 5 th editio Pearso Educatio Limited 009

Slide 5.5 Two tailed test example It is claimed that a average child speds 5 hours per week watchig televisio. A survey of 00 childre fids a average of 4.5 hours per week, with stadard deviatio 8 hours. Is the claim justified? The claim would be wrog if childre sped either more or less tha 5 hours watchig TV. The rejectio regio is split across the two tails of the distributio. This is a two tailed test. Barrow, Statistics for Ecoomics, Accoutig ad Busiess Studies, 5 th editio Pearso Educatio Limited 009

Slide 5.6 A two tailed test diagram f x H H 0 H.5%.5% x Reject H 0 Reject H 0 Barrow, Statistics for Ecoomics, Accoutig ad Busiess Studies, 5 th editio Pearso Educatio Limited 009

Slide 5.7 Solutio to the problem. H 0 : m = 5 H : m 5. Choose sigificace level: 5% 3. Look up critical value: z* =.96 4. Calculate the test statistic: z x m 4.5 5 s 8 00 0.65 5. Decisio: we do ot reject H 0 sice 0.65 <.96 ad does ot fall ito the rejectio regio Barrow, Statistics for Ecoomics, Accoutig ad Busiess Studies, 5 th editio Pearso Educatio Limited 009

Slide 5.8 Why 5%? The choice of sigificace level Like its complemet, the 95% cofidece level, it is a covetio. A differet value ca be chose, but it does set a bechmark. If the cost of makig a Type I error is especially high, the set a lower sigificace level, e.g. %. The sigificace level is the probability of makig a Type I error. Barrow, Statistics for Ecoomics, Accoutig ad Busiess Studies, 5 th editio Pearso Educatio Limited 009

Slide 5.9 The prob-value approach A alterative way of makig the decisio Returig to the LED problem, the test statistic z = -.79 cuts off 3.67% i the lower tail of the distributio. 3.67% is the prob-value for this example Sice 3.67% < 5% the test statistic must fall ito the rejectio regio for the test Barrow, Statistics for Ecoomics, Accoutig ad Busiess Studies, 5 th editio Pearso Educatio Limited 009

Slide 5.0 Two ways to rejectio... Reject H 0 if either or z < -z* (-.79 < -.64) the prob-value < the sigificace level (3.67% < 5%) Barrow, Statistics for Ecoomics, Accoutig ad Busiess Studies, 5 th editio Pearso Educatio Limited 009

Slide 5. Testig a proportio Same priciples: reject H 0 if the test statistic falls ito the rejectio regio. To test H 0 : = 0.5 vs H : 0.5 (e.g. a coi is fair or ot) the test statistic is z p p 0.5 0.5 0.5 Barrow, Statistics for Ecoomics, Accoutig ad Busiess Studies, 5 th editio Pearso Educatio Limited 009

Slide 5. Testig a proportio (cotiued) If the sample evidece were 60 heads from 00 tosses (p = 0.6) we would have z 0.6 0.5 0.5 0.5 00 so we would (just) reject H 0 sice >.96. Barrow, Statistics for Ecoomics, Accoutig ad Busiess Studies, 5 th editio Pearso Educatio Limited 009

Slide 5.3 Testig the differece of two meas To test whether two samples are draw from populatios with the same mea H 0 : m = m or H 0 : m - m = 0 H : m m or H 0 : m - m 0 The test statistic is z x x m m s s Barrow, Statistics for Ecoomics, Accoutig ad Busiess Studies, 5 th editio Pearso Educatio Limited 009

Slide 5.4 Barrow, Statistics for Ecoomics, Accoutig ad Busiess Studies, 5 th editio Pearso Educatio Limited 009 Testig the differece of two proportios To test whether two sample proportios are equal H 0 : = or H 0 : - = 0 H : or H 0 : - 0 The test statistic is ˆ ˆ ˆ ˆ p p z ˆ p p

Slide 5.5 Two cosequeces: Small samples ( < 5) the t distributio is used istead of the stadard ormal for tests of the mea t x m s ~ t tests of proportios caot be doe by the stadard methods used i the book Barrow, Statistics for Ecoomics, Accoutig ad Busiess Studies, 5 th editio Pearso Educatio Limited 009

Slide 5.6 Testig a mea A sample of cars of a particular make average 35 mpg, with stadard deviatio 5. Test the maufacturer s claim of 40 mpg as the true average. H 0 : m = 40 H : m < 40 Barrow, Statistics for Ecoomics, Accoutig ad Busiess Studies, 5 th editio Pearso Educatio Limited 009

Slide 5.7 Testig a mea (cotiued) The test statistic is t 35 40 5.5 The critical value of the t distributio (df =, 5% sigificace level, oe tail) is t* =.796 Hece we caot reject the maufacturer s claim Barrow, Statistics for Ecoomics, Accoutig ad Busiess Studies, 5 th editio Pearso Educatio Limited 009

Slide 5.8 Barrow, Statistics for Ecoomics, Accoutig ad Busiess Studies, 5 th editio Pearso Educatio Limited 009 Testig the differece of two meas The test statistic is where S is the pooled variace S S x x t m m s s S

Slide 5.9 Summary The priciples are the same for all tests: calculate the test statistic ad see if it falls ito the rejectio regio The formula for the test statistic depeds upo the problem (mea, proportio, etc) The rejectio regio varies, depedig upo whether it is a oe or two tailed test Barrow, Statistics for Ecoomics, Accoutig ad Busiess Studies, 5 th editio Pearso Educatio Limited 009