Notes on Power and Sample Size D. Keith Williams PhD Department of Biostatistics.

Slides:



Advertisements
Similar presentations
“Students” t-test.
Advertisements

Psych 5500/6500 t Test for Two Independent Groups: Power Fall, 2008.
PTP 560 Research Methods Week 9 Thomas Ruediger, PT.
The two-sample t-test Expanding t to two groups. t-tests used for population mean diffs With 1-sample t, we have a single sample and a population value.
Business 205. Review Sampling Continuous Random Variables Central Limit Theorem Z-test.
Sample size computations Petter Mostad
Hypothesis : Statement about a parameter Hypothesis testing : decision making procedure about the hypothesis Null hypothesis : the main hypothesis H 0.
Statistics for the Social Sciences Psychology 340 Fall 2006 Hypothesis testing.
Understanding Statistics in Research
Statistics for the Social Sciences Psychology 340 Spring 2005 Hypothesis testing.
Intro to Statistics for the Behavioral Sciences PSYC 1900 Lecture 11: Power.
1 Confidence Interval for Population Mean The case when the population standard deviation is unknown (the more common case).
Sample size and study design
Hypothesis Testing Is It Significant?. Questions What is a statistical hypothesis? What is the null hypothesis? Why is it important for statistical tests?
Statistics for the Social Sciences
Tuesday, September 10, 2013 Introduction to hypothesis testing.
Comparing Means From Two Sets of Data
Section #4 October 30 th Old: Review the Midterm & old concepts 1.New: Case II t-Tests (Chapter 11)
Psy B07 Chapter 8Slide 1 POWER. Psy B07 Chapter 8Slide 2 Chapter 4 flashback  Type I error is the probability of rejecting the null hypothesis when it.
The paired sample experiment The paired t test. Frequently one is interested in comparing the effects of two treatments (drugs, etc…) on a response variable.
1 Clinical Investigation and Outcomes Research Statistical Issues in Designing Clinical Research Marcia A. Testa, MPH, PhD Department of Biostatistics.
Copyright © 2012 by Nelson Education Limited. Chapter 7 Hypothesis Testing I: The One-Sample Case 7-1.
Statistical Analysis Mean, Standard deviation, Standard deviation of the sample means, t-test.
Paired-Sample Hypotheses -Two sample t-test assumes samples are independent -Means that no datum in sample 1 in any way associated with any specific datum.
Determining the Sample Size. Doing research costs… Power of a hypothesis test generally is an increasing function of sample size. Margin of error is generally.
Step 3 of the Data Analysis Plan Confirm what the data reveal: Inferential statistics All this information is in Chapters 11 & 12 of text.
1 Lecture note 4 Hypothesis Testing Significant Difference ©
A Casual Tutorial on Sample Size Planning for Multiple Regression Models D. Keith Williams M.P.H. Ph.D. Department of Biostatistics.
Comparing Samples. Last Time I talked about what could go wrong in an experiment where you compared a sample mean against a population with a known population.
Statistical Power The power of a test is the probability of detecting a difference or relationship if such a difference or relationship really exists.
IS 4800 Empirical Research Methods for Information Science Class Notes March 13 and 15, 2012 Instructor: Prof. Carole Hafner, 446 WVH
Chapter 9 Introduction to the t Statistic. 9.1 Review Hypothesis Testing with z-Scores Sample mean (M) estimates (& approximates) population mean (μ)
General Linear Model 2 Intro to ANOVA.
Statistical Inference for the Mean Objectives: (Chapter 9, DeCoursey) -To understand the terms: Null Hypothesis, Rejection Region, and Type I and II errors.
I271B The t distribution and the independent sample t-test.
AP Statistics Section 11.4 B
Chapter 12 For Explaining Psychological Statistics, 4th ed. by B. Cohen 1 Chapter 12: One-Way Independent ANOVA What type of therapy is best for alleviating.
Reasoning in Psychology Using Statistics Psychology
Research Methods: 2 M.Sc. Physiotherapy/Podiatry/Pain Inferential Statistics.
Power and Sample Size Anquan Zhang presents For Measurement and Statistics Club.
Anthony Greene1 Two Sample t-test: Hypothesis of Differences Between Two Groups 1.Is Group “A” Different Than Group “B”? 2.Does an Experimental Manipulation.
AP Statistics.  If our data comes from a simple random sample (SRS) and the sample size is sufficiently large, then we know that the sampling distribution.
AP Statistics Section 11.4 B. A significance test makes a Type I error when ___________________________________ P(Type 1 error ) = ___ A significance.
MATB344 Applied Statistics I. Experimental Designs for Small Samples II. Statistical Tests of Significance III. Small Sample Test Statistics Chapter 10.
Oneway/Randomized Block Designs Q560: Experimental Methods in Cognitive Science Lecture 8.
Hypothesis test flow chart
Lec. 19 – Hypothesis Testing: The Null and Types of Error.
Statistical Inference for the Mean Objectives: (Chapter 8&9, DeCoursey) -To understand the terms variance and standard error of a sample mean, Null Hypothesis,
Statistical Inferences for Variance Objectives: Learn to compare variance of a sample with variance of a population Learn to compare variance of a sample.
Reasoning in Psychology Using Statistics Psychology
 List the characteristics of the F distribution.  Conduct a test of hypothesis to determine whether the variances of two populations are equal.  Discuss.
Slides to accompany Weathington, Cunningham & Pittenger (2010), Chapter 11: Between-Subjects Designs 1.
Hypothesis Testing Is It Significant?.
INF397C Introduction to Research in Information Studies Spring, Day 12
Hypothesis testing using contrasts
Hypothesis Testing Is It Significant?.
Elementary Statistics
Data Analysis and Interpretation
Central Limit Theorem, z-tests, & t-tests
Reasoning in Psychology Using Statistics
INTRODUCTORY STATISTICS FOR CRIMINAL JUSTICE Test Review: Ch. 7-9
Reasoning in Psychology Using Statistics
Introduction to ANOVA.
One sample problems Statistics 2126.
Reasoning in Psychology Using Statistics
Reasoning in Psychology Using Statistics
Reasoning in Psychology Using Statistics
Reasoning in Psychology Using Statistics
Type I and Type II Errors
Rest of lecture 4 (Chapter 5: pg ) Statistical Inferences
Presentation transcript:

Notes on Power and Sample Size D. Keith Williams PhD Department of Biostatistics

Area =

Area =

Area =

3.87 Area = 0.955

Goals n Remove the ‘mystery’ of power and sample size n Introduce the main ideas n See how a ‘statistician’ views the topic n Provide information on how to do it yourself, or at least get started (Its no big deal)

Buzzwords n Alpha (α) = P(Type I error) = P(Conclude experimental groups are different when they really are the same) n Beta (  ) = P(Type II error) = P(Conclude the experimental groups are the same when they really are different) n Power = 1 -  = P(Conclude experimental groups are different when they really are!)

Thoughts n When planning an experiment, one should determine a sample size that results in a statistical test powerful enough to declare significance for a reasonable difference in the means…if that difference truly exists in the population

Thoughts n Generally speaking...in order to calculate power/sample size, one needs a ‘guess’ about the pattern of the population means and an estimate of their variance n Otherwise the statistician feels that they have the role of dreaming up what the population means and variances are…YIKES!

Thoughts n α n 1-  n Variance n Population means n N: n Represent the five items involved in power and sample size. n One needs to recognize that that you must input 4 of these items to get the fifth.

One or the other…. n Input ⇒ (α, 1- , variance,population means) ⇒ gives N n Input ⇒ (α, variance,population means, N) ⇒ gives 1-  n One usually ends up iterating between the above to arrive at a sample size that has a desirable level of power.

How to Help Your Statistician Help You! n Usually a study has several questions to be answered…and a statistical test that goes with each. n Prioritize which of these are most important and arguably the ones power should be based on. n Organize your best bet on the population means and their variances…or some scenarios that are clinically important that you wish to detect (if they truly exist in the population).

How to Help Your Statistician Help You! n Determine what the resources of the study are…how many subjects can you afford. Communicate this up front. n Try to do some preliminary power calculations on your own.

The Non Centrality Parameter Two Group t-test

Scenario 1 n Alpha =0.05, sigma=2 n |mu1 – mu2| = 2, that is, a two unit diff in means for a population n Propose n1 = 10 and n2 = 10

Rejection region for two tailed t- test alpha=0.05, df = 18

Noncentrality value =2.236, Critical value = |2.101| Table B.5, Values between 2.0 and 3.0, alpha = 0.05, df = 18 Power between 0.47 and 0.81, SAS calculation

Now one has a couple of choices n Decide that a 2 unit difference in the means is reasonable and you can afford 30 subjects in each group

Rejection region for two tailed t- test alpha=0.05, df = 58

Noncentrality value =3.87, Critical value = |2.00| Table B.5, Values between 3.0 and 4.0, alpha = 0.05, df = 58 (60) Power between 0.84 and 0.98, SAS calculation

Now one has a couple of choices n Decide that a 3 unit difference in the means is reasonable and you can only afford 10 subjects in each group

Noncentrality value =3.35, Critical value = |2.101| Table B.5, Values between 3.0 and 4.0, alpha = 0.05, df = 18 Power between 0.81 and 0.97, SAS calculation

Now Lets Turn it Around Sample Sizes for Given Power Values n Δ = max(mu) – min(mu) n Determine k = Δ/s.d. (‘effect size’) n Use table B.12 n Different levels of alpha = 0.2, , and 0.01 n 1 – β = 0.7, 0.8, 0.9, and 0.95 n r : number of treatments, 2, 3, …., 10

From Our Earlier Anova Setting.. n mu1 = 20, mu2 = 15, mu3 = 15, mu4 = 12 n Δ = 20 – 12 = 8, sigma = 4 n K = Δ/sigma = 8/4 = 2, r = 4 Powern per groupTotal N