Discussion of presentations Issues 1. imbalanced sample size 2

Slides:



Advertisements
Similar presentations
Statistics Review – Part II Topics: – Hypothesis Testing – Paired Tests – Tests of variability 1.
Advertisements

Happiness comes not from material wealth but less desire. 1.
PTP 560 Research Methods Week 9 Thomas Ruediger, PT.
Sampling: Final and Initial Sample Size Determination
Chi Square Tests Chapter 17.
Session 2. Applied Regression -- Prof. Juran2 Outline for Session 2 More Simple Regression –Bottom Part of the Output Hypothesis Testing –Significance.
Copyright ©2011 Brooks/Cole, Cengage Learning Testing Hypotheses about Means Chapter 13.
Comparing Two Population Means The Two-Sample T-Test and T-Interval.
Testing means, part III The two-sample t-test. Sample Null hypothesis The population mean is equal to  o One-sample t-test Test statistic Null distribution.
ANOVA notes NR 245 Austin Troy
PSY 307 – Statistics for the Behavioral Sciences
Independent Sample T-test Formula
Statistics II: An Overview of Statistics. Outline for Statistics II Lecture: SPSS Syntax – Some examples. Normal Distribution Curve. Sampling Distribution.
Ch 15 - Chi-square Nonparametric Methods: Chi-Square Applications
PSYC512: Research Methods PSYC512: Research Methods Lecture 9 Brian P. Dyre University of Idaho.
1 Inference About a Population Variance Sometimes we are interested in making inference about the variability of processes. Examples: –Investors use variance.
Chapter 9 Hypothesis Testing.
Chapter 9: Introduction to the t statistic
Standard error of estimate & Confidence interval.
Qian H. Li, Lawrence Yu, Donald Schuirmann, Stella Machado, Yi Tsong
Choosing Statistical Procedures
AM Recitation 2/10/11.
1 STATISTICAL HYPOTHESES AND THEIR VERIFICATION Kazimieras Pukėnas.
Education 793 Class Notes T-tests 29 October 2003.
Two Sample Tests Nutan S. Mishra Department of Mathematics and Statistics University of South Alabama.
Education Research 250:205 Writing Chapter 3. Objectives Subjects Instrumentation Procedures Experimental Design Statistical Analysis  Displaying data.
t(ea) for Two: Test between the Means of Different Groups When you want to know if there is a ‘difference’ between the two groups in the mean Use “t-test”.
Chapter 13 – Difference Between Two Parameters Math 22 Introductory Statistics.
© Copyright McGraw-Hill 2000
Statistical Inference for the Mean Objectives: (Chapter 9, DeCoursey) -To understand the terms: Null Hypothesis, Rejection Region, and Type I and II errors.
Academic Research Academic Research Dr Kishor Bhanushali M
© Copyright McGraw-Hill 2004
Copyright ©2011 Brooks/Cole, Cengage Learning Testing Hypotheses about Difference Between Two Means.
Chapter 13 Understanding research results: statistical inference.
Chapter 9: Introduction to the t statistic. The t Statistic The t statistic allows researchers to use sample data to test hypotheses about an unknown.
Statistical Inference for the Mean Objectives: (Chapter 8&9, DeCoursey) -To understand the terms variance and standard error of a sample mean, Null Hypothesis,
 Confidence Intervals  Around a proportion  Significance Tests  Not Every Difference Counts  Difference in Proportions  Difference in Means.
Chapter 10: The t Test For Two Independent Samples.
Analytical Similarity Assessment: Practical Challenges and Statistical Perspectives Richard Montes, Ph.D. Hospira, a Pfizer company Biosimilars Pharmaceutical.
Hypothesis Testing.
JMP for Biosimilars: Tools for Analytical Similarity Sept 16, 2015 W
I. ANOVA revisited & reviewed
Richard K Burdick Elion Labs MBSW Meetings May 2016
Logic of Hypothesis Testing
Quantitative Methods Varsha Varde.
Statistical Quality Control, 7th Edition by Douglas C. Montgomery.
Professor Ke-Sheng Cheng
Chapter 4. Inference about Process Quality
CHAPTER 13 Design and Analysis of Single-Factor Experiments:
STAT 312 Chapter 7 - Statistical Intervals Based on a Single Sample
Math 4030 – 10b Inferences Concerning Variances: Hypothesis Testing
Multiple Regression Analysis: Inference
Confidence Intervals and Hypothesis Tests for Variances for One Sample
Psychology 202a Advanced Psychological Statistics
Statistical inference: distribution, hypothesis testing
Hypothesis Testing: Hypotheses
Towson University - J. Jung
Georgi Iskrov, MBA, MPH, PhD Department of Social Medicine
9 Tests of Hypotheses for a Single Sample CHAPTER OUTLINE
Chapter 9 Hypothesis Testing.
Chapter 25: Paired Samples and Blocks
Hypothesis Testing.
Quantitative Methods in HPELS HPELS 6210
I. Statistical Tests: Why do we use them? What do they involve?
Psych 231: Research Methods in Psychology
Psych 231: Research Methods in Psychology
Statistics II: An Overview of Statistics
What are their purposes? What kinds?
Hypothesis Testing: The Difference Between Two Population Means
Georgi Iskrov, MBA, MPH, PhD Department of Social Medicine
Presentation transcript:

Discussion of presentations Issues 1. imbalanced sample size 2 Discussion of presentations Issues 1. imbalanced sample size 2. estimate of reference variance for margin 3. adjusted for correlated lots 4. equivalence test of effect size instead 5. Comparability 6. tier 2 – What to propose for k? - it is more a statistical scale instead of a statistical test

Tsong and Dong’s presentation on Sample Size Imbalance Adjustment Method B: when the sample size ratio > 1.5 Use all the reference sample to estimate μR and σR; Use nR* = Min(1.5×nT, nR) to compute the CI. Satterthwaite approximation is recommended for CI computation. DIA/FDA Stat Forum 2016

III. Sample Size Imbalance Adjustment Statistical properties of Method B Decision Rule: conclude stat. Equivalence in means if T1 > t1-α, df* and T2 < -t1-α, df* . Power: p(u1,u2)~bivariate t(df*,df*, θ1, θ2,1) DIA/FDA Stat Forum 2016

III. Sample Size Imbalance Adjustment Method B: Type I error rate vs. SS Ratio the type I error rate is a monotone function of the sample size ratio for given variance. More specifically, the type I error rate is more conservative if the sample size ratio (imbalance) is large. DIA/FDA Stat Forum 2016

III. Sample Size Imbalance Adjustment Method B: Power Comparison(σT = σR) the type I error rate is a monotone function of the sample size ratio for given variance. More specifically, the type I error rate is more conservative if the sample size ratio (imbalance) is large. DIA/FDA Stat Forum 2016

  Use nR*=min(1.5nT,nR) Power Type I Error  

Very similar results But the issues are The need of information of biosimilar product There is no blindness or randomization of the study No control of change design through interim analysis Power rewarding based on product of extremely large sample size

II. Estimated vs. fixed equivalence margin Guidance will only focus on fixed margin Estimation margin but test as fixed will inflate type I error rate and reduce power Alternative approaches are in in the process of research and evaluation Generalized confidence interval of effect size approach has been proposed

III. Correlation between lots Tsong’s presentation on adjustment of lot correlation Parameters of a population may be estimated unbiasedly if the sample is taken randomly and representatively Which means every member of the population should have the same chance to be sampled Or randomly sample clusters before members but each clusters should represent the population The practical analytical population may consists of clusters do not represent the population And the random sample are taken from clusters only available during the study time In conclusion, study sample define a population which is a population ℙ generalized from the sample

Practical sampling design of analytical biosimilarity assessment

True batch population and sample defined population

Out of 30 data set we examined, 70% of data set has observed reference variability is larger than the biosimilar variability

Population, sample defined population ℙ, random sample and correlated samples Given the fact that there is no formula, example or knowledge of how the correlation is calculated, or how large it is based on the data available to the biosimilar sponsor, the adjustment proposed by researcher or sponsor becomes unrealistic Under the assumption that such knowledge is available, we conducted a research and found the conventional t-test used for equivalence assessment inflates type I error rate and lowers power (Shen, Wang (2016), to submit to JBS) 2016 MBSW

90% CI 4. Equiv. test of effect size Is [1-(confidence level)]/2 = type I error rate of two one-sided test? Is such a test monotone from -∞ to 0? Tsong’s presentation “hypothesis testing and confidence interval – where is the duality?

5. Comparability Comparability is to demonstrate the consistency of the same product before and after some change to its manufacturing process Equivalence test of means before and after change(s) Profile comparison – multivariate equivalence Normal vs. non-normal, parametric vs. nonparametric, equal vs. unequal variance – small sample problem Dissolution profile comparison Model dependent vs. model independent methods SK (saranadasa & krishnamoorthy (2015)) using mean difference at all time point IUT using non-constant margin at each time point Parallelism testing