Comparing two sample means Dr David Field. Comparing two samples Researchers often begin with a hypothesis that two sample means will be different from.

Slides:



Advertisements
Similar presentations
Chapter 16 Inferential Statistics
Advertisements

1 COMM 301: Empirical Research in Communication Lecture 15 – Hypothesis Testing Kwan M Lee.
Lecture (11,12) Parameter Estimation of PDF and Fitting a Distribution Function.
Personal Response System (PRS). Revision session Dr David Field Do not turn your handset on yet!
Hypothesis: It is an assumption of population parameter ( mean, proportion, variance) There are two types of hypothesis : 1) Simple hypothesis :A statistical.
Hypothesis Testing A hypothesis is a claim or statement about a property of a population (in our case, about the mean or a proportion of the population)
Inference Sampling distributions Hypothesis testing.
Chapter 10 Section 2 Hypothesis Tests for a Population Mean
THE MEANING OF STATISTICAL SIGNIFICANCE: STANDARD ERRORS AND CONFIDENCE INTERVALS.
Chapter 7 Sampling and Sampling Distributions
Statistical Concepts (continued) Concepts to cover or review today: –Population parameter –Sample statistics –Mean –Standard deviation –Coefficient of.
4-1 Statistical Inference The field of statistical inference consists of those methods used to make decisions or draw conclusions about a population.
Inference about a Mean Part II
Independent Sample T-test Often used with experimental designs N subjects are randomly assigned to two groups (Control * Treatment). After treatment, the.
Chapter 9 Hypothesis Testing.
Independent Sample T-test Classical design used in psychology/medicine N subjects are randomly assigned to two groups (Control * Treatment). After treatment,
INFERENTIAL STATISTICS – Samples are only estimates of the population – Sample statistics will be slightly off from the true values of its population’s.
AM Recitation 2/10/11.
McGraw-Hill/IrwinCopyright © 2009 by The McGraw-Hill Companies, Inc. All Rights Reserved. Chapter 9 Hypothesis Testing.
PYPR1 lecture 2 : Populations and Samples Dr David Field.
Overview Definition Hypothesis
Review of Statistical Inference Prepared by Vera Tabakova, East Carolina University ECON 4550 Econometrics Memorial University of Newfoundland.
4-1 Statistical Inference The field of statistical inference consists of those methods used to make decisions or draw conclusions about a population.
Education 793 Class Notes T-tests 29 October 2003.
1 Level of Significance α is a predetermined value by convention usually 0.05 α = 0.05 corresponds to the 95% confidence level We are accepting the risk.
More About Significance Tests
Go to Index Analysis of Means Farrokh Alemi, Ph.D. Kashif Haqqi M.D.
Fundamentals of Data Analysis Lecture 4 Testing of statistical hypotheses.
Today’s lesson Confidence intervals for the expected value of a random variable. Determining the sample size needed to have a specified probability of.
1 CSI5388: Functional Elements of Statistics for Machine Learning Part I.
1 Statistical Inference Greg C Elvers. 2 Why Use Statistical Inference Whenever we collect data, we want our results to be true for the entire population.
Population All members of a set which have a given characteristic. Population Data Data associated with a certain population. Population Parameter A measure.
Review: Two Main Uses of Statistics 1)Descriptive : To describe or summarize a collection of data points The data set in hand = all the data points of.
Statistics 101 Chapter 10. Section 10-1 We want to infer from the sample data some conclusion about a wider population that the sample represents. Inferential.
Statistics and Quantitative Analysis U4320
Making decisions about distributions: Introduction to the Null Hypothesis 47:269: Research Methods I Dr. Leonard April 14, 2010.
Statistical Decision Making. Almost all problems in statistics can be formulated as a problem of making a decision. That is given some data observed from.
Learning Objectives In this chapter you will learn about the t-test and its distribution t-test for related samples t-test for independent samples hypothesis.
1 rules of engagement no computer or no power → no lesson no SPSS → no lesson no homework done → no lesson GE 5 Tutorial 5.
The Practice of Statistics Third Edition Chapter 10: Estimating with Confidence Copyright © 2008 by W. H. Freeman & Company Daniel S. Yates.
Chapter 7 Estimation Procedures. Basic Logic  In estimation procedures, statistics calculated from random samples are used to estimate the value of population.
1 Lecture note 4 Hypothesis Testing Significant Difference ©
Statistical Hypotheses & Hypothesis Testing. Statistical Hypotheses There are two types of statistical hypotheses. Null Hypothesis The null hypothesis,
10.1: Confidence Intervals Falls under the topic of “Inference.” Inference means we are attempting to answer the question, “How good is our answer?” Mathematically:
McGraw-Hill/Irwin Copyright © 2007 by The McGraw-Hill Companies, Inc. All rights reserved. Chapter 8 Hypothesis Testing.
Statistical Inference for the Mean Objectives: (Chapter 9, DeCoursey) -To understand the terms: Null Hypothesis, Rejection Region, and Type I and II errors.
Chapter 8 Parameter Estimates and Hypothesis Testing.
Lecture PowerPoint Slides Basic Practice of Statistics 7 th Edition.
Ex St 801 Statistical Methods Inference about a Single Population Mean.
Stats Lunch: Day 3 The Basis of Hypothesis Testing w/ Parametric Statistics.
Business Statistics for Managerial Decision Farideh Dehkordi-Vakil.
Statistical Inference Statistical inference is concerned with the use of sample data to make inferences about unknown population parameters. For example,
1 Probability and Statistics Confidence Intervals.
SAMPLING DISTRIBUTION OF MEANS & PROPORTIONS. SAMPLING AND SAMPLING VARIATION Sample Knowledge of students No. of red blood cells in a person Length of.
SAMPLING DISTRIBUTION OF MEANS & PROPORTIONS. SAMPLING AND SAMPLING VARIATION Sample Knowledge of students No. of red blood cells in a person Length of.
1 Testing Statistical Hypothesis The One Sample t-Test Heibatollah Baghi, and Mastee Badii.
Chapter 13 Understanding research results: statistical inference.
Fundamentals of Data Analysis Lecture 4 Testing of statistical hypotheses pt.1.
Statistical Inference for the Mean Objectives: (Chapter 8&9, DeCoursey) -To understand the terms variance and standard error of a sample mean, Null Hypothesis,
Review Statistical inference and test of significance.
Chapter 9 Introduction to the t Statistic
The Single-Sample t Test Chapter 9. t distributions >Sometimes, we do not have the population standard deviation, σ. Very common! >So what can we do?
Confidence Intervals.
Inference: Conclusion with Confidence
Chapter 9 Hypothesis Testing.
Chapter 6 Inferences Based on a Single Sample: Estimation with Confidence Intervals Slides for Optional Sections Section 7.5 Finite Population Correction.
Significance Test for the Difference of Two Proportions
Statistics in Applied Science and Technology
CHAPTER 14: Confidence Intervals The Basics
Chapter 9 Hypothesis Testing.
Presentation transcript:

Comparing two sample means Dr David Field

Comparing two samples Researchers often begin with a hypothesis that two sample means will be different from each other In practice, two sample means will almost always be slightly different from each other Therefore, statistics are used to decide whether the observed difference between two samples is meaningful or not To do this, we test the null hypothesis that the two samples were both drawn randomly from the same population

Test statistics To test the null hypothesis we need to quantify the strength of the evidence against it This is done using test statistics –when the test statistic is larger, there is more evidence against the null hypothesis What makes test statistics different from other statistics is that they have known probability distributions when the null hypothesis is true –we know the p of a test statistic of 1 or >1 occurring purely due to sampling variation from a null distribution –the p of a test statistic of 2 or > 2 will be lower than the p of a test statistic of >1 –if the p of the test statistic occurring purely due to sampling variation is < 0.05 (5%) the null hypothesis is rejected Test statistics with known probability distributions under the null hypothesis include z, t, r, and chi-square –Mean, Median, SD are not test statistics

Confidence intervals as a test Lecture 2 explained how to calculate a 95% confidence interval around a single sample mean –this was achieved using the SE of an inferred sampling distribution of the mean –collecting two samples and calculating two separate confidence intervals establishes that the two samples are from different populations if the confidence intervals do not overlap –but it does not allow a conclusion to be reached when the confidence intervals do overlap To calculate a test statistic to directly test the null hypothesis we need to consider a slightly different sampling distribution – the sampling distribution of the difference between two means

Sampling distribution of the difference between two means Normally, you are only able to measure 2 samples and calculate 2 means and the difference between them But test statistics are based on properties of an assumed underlying sampling distribution of the difference between two means The best way to understand test statistics is to consider unusual or artificial examples where full population data and sampling distributions are available Therefore….

The two populations Mean 4.5 SD 0.8 Mean 4.0 SD 0.4

Weights of British cats (Kg) Weights of Greek cats (Kg) Mean SD SE Small sample size (N = 5) Large sample size (N = 12)

Weights of British cats (Kg) Weights of Greek cats (Kg) Mean SD SE Small sample size (N = 5) Large sample size (N = 12)

Sampling distribution of the difference between two means Take a large number of samples of 5 cats from the UK population –Arrange the samples in pairs and for each pair calculate the difference between the two means –Half the differences will be negative and half of them will be positive –Therefore the mean of this sampling distribution will be zero. This differs from the sampling distribution of a single sample mean, which has a mean equal to the underlying population mean –The sampling distribution of the difference between two means will be normally distributed

ORIGINAL DISTRIBUTION is the population frequency distribution of weight differences between pairs of individual cats Black solid curves are sampling distributions of weight differences between 2 sample means, for samples of of 4, 16, and 64 cats 1 SE 1 SD of population

Standard error of the difference between two sample means σ (sigma) means the SD of the population of difference scores N1 and N2 are the two sample sizes –the formula allows the SE of the sampling distribution to be calculated when the two samples differ in size Like the SE of a single sample mean, this SE gets smaller as N increases and gets smaller as the SD gets smaller Smaller SE makes it easier to reject null hypothesis SE = 1 N1 σ + 1 N2

SE of the difference between mean Kg for two samples of 5 UK cats 1/5 (or 1/2, or 1/3, or 1/20) is a number less than 1 The square root makes the number larger, but never makes it greater than 1 So, the population SD gets multiplied by a number smaller than 1, which is why the SE is always smaller than the SD of the population Kg =

Weights of British cats (Kg) Mean SD SE Small sample size (N = 5) For the highlighted pair of samples the difference between the means is 0.5Kg What percentage of sample pairs have a difference of 0.5Kg or larger? If we expressed the difference of 0.5Kg in units of SE we could answer that question This is because the converted score is a Z score Remember that in this theoretical example we know that both samples are from the same population, and the purpose is to calculate the p of a difference this big or bigger occurring when that is the case

Converting the difference between 2 sample means to a Z score Z = Z = 0.99 The difference between the means The SE formula

16.1% of the total area under the normal curve corresponds to values of 0.99 or greater 16.1% of differences between means of sample size 5 will have Z scores greater than 0.99

From Z back to Kg So, 16.1% of differences between pairs of samples of N=5 drawn from the population of UK cats will be 0.5Kg or larger This is the same as saying the probability of a single comparison producing a difference of 0.5Kg or greater is 16.1%

SD1 2 What if the population SD (σ) is unknown? Usually, researchers only have two samples to compare, and the population parameters are unknown. In this situation the sample SD is used instead of the population SD, and the SE formula is modified SE = N1 + SD2 2 N2

Weights of British cats (Kg) Mean SD SE Small sample size (N = 5) For the highlighted pair of samples the mean difference is 0.5Kg The sample SD’s will be used in the modified formula instead of the unknown population SD

0.5 2 Converting the difference between 2 means to a Z score when σ is unknown Z = =

How much evidence is there against the null hypothesis? 9.8% of Z statistics are > 1.29, so we would not conclude that the two samples of cats are from different countries if we used the 5% cut off In this example, we know that the two samples were from the same population, so we can verify that this was the correct conclusion On the other hand, if two samples had a mean difference of 0.8Kg, then assuming the sample SD’s remain the same, the resulting Z statistic would be 2.07 Only 1.9% of Z statistics are greater than 2.07, and if we didn’t know that the two samples came from the same population we would reject the null hypothesis, and by doing so commit a Type I error

Weights of British cats (Kg) Weights of Greek cats (Kg) Mean SD SE Small sample size (N = 5) Large sample size (N = 12)

0.5 2 The Z score of the difference between samples of 5 UK and 5 Greek cats Z = 4.1 – =

How much evidence is there against the null hypothesis? 6.3% of Z statistics are > 1.53, so we would be unable to conclude that the two samples of cats are from different countries if we used the 5% cut off In this example we know that the two samples were from different populations, so we have committed a Type II error by failing to reject the null hypothesis Type II errors like this are common when the sample size is small

0.6 2 The Z score of the difference between samples of 12 UK and 12 Greek cats Z = 4.6 – =

How much evidence is there against the null hypothesis? 0.032% of Z statistics are > 2.73, so we would conclude that the two samples of cats are from different countries if we used the 5% cut off In this example we know that the two samples were from different populations, so we have correctly rejected the null hypothesis

Important caveat What I have described today is called a “Z test” But, the formula for estimating the SE of the difference between 2 means used in the Z test is only accurate when the individual sample sizes are 30 or more –This is because the estimate of the population SD is not accurate There is a different test that uses an accurate estimate of the SE when sample size is less than 30 –the “t test”, which is covered in the next lecture Because the t test produces the same results as the Z test when the sample size is >30 computer programs like SPSS generally only give the option of a t test Both tests work on the same principle, but the Z test is less complicated and easier to understand

General principle of test statistics test statistic = variation in the DV due to the IV other variation in the data (error) All test statistics have known probability distributions when variation in the DV due to the IV is zero (i.e. the null hyp is true) Z has the distribution of the standard normal distribution Other test statistics have different shaped distributions, and different calculation formulas, but the general principle for converting the test statistic to a p value is the same.

List of statistical terms for revision This lecture made use of terms introduced in previous lectures, and only introduced one new term –sampling distribution of the difference between two means