best estimate (mean) for X uncertainty or error in the measurement (systematic, random or statistical) best

Similar documents
ENGI 4421 Propagation of Error Page 8-01

CHAPTER VI Statistical Analysis of Experimental Data

Functions of Random Variables

Lecture Notes Types of economic variables

Econometric Methods. Review of Estimation

Summary of the lecture in Biostatistics

hp calculators HP 30S Statistics Averages and Standard Deviations Average and Standard Deviation Practice Finding Averages and Standard Deviations

Introduction to local (nonparametric) density estimation. methods

UNIVERSITY OF OSLO DEPARTMENT OF ECONOMICS

Chapter 8. Inferences about More Than Two Population Central Values

Mean is only appropriate for interval or ratio scales, not ordinal or nominal.

Simulation Output Analysis

Lecture 3. Sampling, sampling distributions, and parameter estimation

Multiple Choice Test. Chapter Adequacy of Models for Regression

Point Estimation: definition of estimators

Evaluation of uncertainty in measurements

ESS Line Fitting

UNIVERSITY OF OSLO DEPARTMENT OF ECONOMICS

ENGI 3423 Simple Linear Regression Page 12-01

Lecture 3 Probability review (cont d)

Lecture 1 Review of Fundamental Statistical Concepts

A Study of the Reproducibility of Measurements with HUR Leg Extension/Curl Research Line

THE ROYAL STATISTICAL SOCIETY 2016 EXAMINATIONS SOLUTIONS HIGHER CERTIFICATE MODULE 5

Chapter 5 Properties of a Random Sample

Chapter 14 Logistic Regression Models

Special Instructions / Useful Data

Class 13,14 June 17, 19, 2015

Random Variables and Probability Distributions

12.2 Estimating Model parameters Assumptions: ox and y are related according to the simple linear regression model

Simple Linear Regression

ENGI 4421 Joint Probability Distributions Page Joint Probability Distributions [Navidi sections 2.5 and 2.6; Devore sections

Estimation of Stress- Strength Reliability model using finite mixture of exponential distributions

X ε ) = 0, or equivalently, lim

STA302/1001-Fall 2008 Midterm Test October 21, 2008

Ordinary Least Squares Regression. Simple Regression. Algebra and Assumptions.

Analysis of Variance with Weibull Data

STK4011 and STK9011 Autumn 2016

Parameter, Statistic and Random Samples

Example: Multiple linear regression. Least squares regression. Repetition: Simple linear regression. Tron Anders Moger

Module 7. Lecture 7: Statistical parameter estimation

Simple Linear Regression

STATISTICAL PROPERTIES OF LEAST SQUARES ESTIMATORS. x, where. = y - ˆ " 1

Continuous Distributions

2.28 The Wall Street Journal is probably referring to the average number of cubes used per glass measured for some population that they have chosen.

Lecture 2 - What are component and system reliability and how it can be improved?

Multivariate Transformation of Variables and Maximum Likelihood Estimation

Chapter -2 Simple Random Sampling

is the score of the 1 st student, x

Chapter -2 Simple Random Sampling

Chapter 8: Statistical Analysis of Simulated Data

THE ROYAL STATISTICAL SOCIETY HIGHER CERTIFICATE

Lecture 7. Confidence Intervals and Hypothesis Tests in the Simple CLR Model

f f... f 1 n n (ii) Median : It is the value of the middle-most observation(s).

Chapter Statistics Background of Regression Analysis

{ }{ ( )} (, ) = ( ) ( ) ( ) Chapter 14 Exercises in Sampling Theory. Exercise 1 (Simple random sampling): Solution:

Multiple Linear Regression Analysis

MEASURES OF DISPERSION

A Method for Damping Estimation Based On Least Square Fit

Wu-Hausman Test: But if X and ε are independent, βˆ. ECON 324 Page 1

The expected value of a sum of random variables,, is the sum of the expected values:

Bounds on the expected entropy and KL-divergence of sampled multinomial distributions. Brandon C. Roy

Quantitative analysis requires : sound knowledge of chemistry : possibility of interferences WHY do we need to use STATISTICS in Anal. Chem.?

THE ROYAL STATISTICAL SOCIETY GRADUATE DIPLOMA

LECTURE - 4 SIMPLE RANDOM SAMPLING DR. SHALABH DEPARTMENT OF MATHEMATICS AND STATISTICS INDIAN INSTITUTE OF TECHNOLOGY KANPUR

1 Mixed Quantum State. 2 Density Matrix. CS Density Matrices, von Neumann Entropy 3/7/07 Spring 2007 Lecture 13. ψ = α x x. ρ = p i ψ i ψ i.

LINEAR REGRESSION ANALYSIS

EVALUATION OF UNCERTAINITY IN MEASUREMENTS

Maximum Likelihood Estimation

L5 Polynomial / Spline Curves

ECONOMETRIC THEORY. MODULE VIII Lecture - 26 Heteroskedasticity

X X X E[ ] E X E X. is the ()m n where the ( i,)th. j element is the mean of the ( i,)th., then

Objectives of Multiple Regression

BIOREPS Problem Set #11 The Evolution of DNA Strands

Module 7: Probability and Statistics

Correlation and Simple Linear Regression

ρ < 1 be five real numbers. The

EXPERIMENTAL ERRORS. There are primarily two kinds of errors that affect experimental results:

Multiple Regression. More than 2 variables! Grade on Final. Multiple Regression 11/21/2012. Exam 2 Grades. Exam 2 Re-grades

The Mathematical Appendix

THE ROYAL STATISTICAL SOCIETY GRADUATE DIPLOMA

PROPERTIES OF GOOD ESTIMATORS

Chapter 13 Student Lecture Notes 13-1

Sampling Theory MODULE X LECTURE - 35 TWO STAGE SAMPLING (SUB SAMPLING)

Bayes Estimator for Exponential Distribution with Extension of Jeffery Prior Information

residual. (Note that usually in descriptions of regression analysis, upper-case

Chapter Business Statistics: A First Course Fifth Edition. Learning Objectives. Correlation vs. Regression. In this chapter, you learn:

Fitting models to data.

THE ROYAL STATISTICAL SOCIETY 2009 EXAMINATIONS SOLUTIONS GRADUATE DIPLOMA MODULAR FORMAT MODULE 2 STATISTICAL INFERENCE

Chapter 4 Multiple Random Variables

A New Family of Transformations for Lifetime Data

Discrete Mathematics and Probability Theory Fall 2016 Seshia and Walrand DIS 10b

TESTS BASED ON MAXIMUM LIKELIHOOD

The conformations of linear polymers

22 Nonparametric Methods.

STA 105-M BASIC STATISTICS (This is a multiple choice paper.)

Third handout: On the Gini Index

1. The weight of six Golden Retrievers is 66, 61, 70, 67, 92 and 66 pounds. The weight of six Labrador Retrievers is 54, 60, 72, 78, 84 and 67.

REVIEW OF SIMPLE LINEAR REGRESSION SIMPLE LINEAR REGRESSION

Unimodality Tests for Global Optimization of Single Variable Functions Using Statistical Methods

Analyzing Two-Dimensional Data. Analyzing Two-Dimensional Data

Transcription:

Error Aalyss Preamble Wheever a measuremet s made, the result followg from that measuremet s always subject to ucertaty The ucertaty ca be reduced by makg several measuremets of the same quatty or by mprovg the expermetal techque But for fudametal reasos ose sources such as thermal ose or quatum fluctuatos caot be elmated ad the ucertaty caot be reduced to zero Types of ucertates Radom ucertaty ad systematc error A batch of dgtal watches was purchased from a factory wth the expectato that they were all set to the correct tme before dspatch Oe mght mage that some weeks later a varety of tmes would be dsplayed by a selecto of the watches There could be a umber of reasos for the dffereces the tme dsplayed, such as a) the crystal oscllators are rug at dfferet rates; b) the watches were tally correctly set; c) the crystal oscllators mght be ustable; d) the tme stadard at the factory was correct Radom (statstcal) ucertates ca be observed whe repeated measuremets of the same quatty gve rse to dfferet values May radom factors cludg vbratos, varatos temperature ad electroc ose, usually combe to form the 'radom ucertaty' a expermet Systematc error refers to a costat factor whch flueces all readgs equally durg a expermet Such errors ormally arse from the effect of a fte umber of dsturbaces Sources of systematc error could be aythg from the cotact resstace of a wre a electrcal crcut, to a stopwatch rug too fast ad the use of a mproperly calbrated strumet I some expermets t mght be dffcult to see whether errors are due to radom or systematc effects Wheever possble, sources of systematc error should be elmated or corrected for Such sources of error are, however, ot easly detfed ad the expermeter may ever be aware of them I practce, the systematc ad radom ucertates are smply added up to obta the maxmum ucertaty estmate: u ( ) u ( ) u ( ) u( ) max sys ra Estmate of Ucertates Physcal Measuremets Measured value of a physcal quatty : x xbest u( ) x best u() best estmate (mea) for ucertaty or error the measuremet (systematc, radom or statstcal) Relatve ucertaty (error): u ( ) fract x best Expermetal ucertates should almost always be rouded to oe sgfcat fgure (dgt): measured g=(9800) ms -

The last sgfcat fgure ay stated aswer should usually be of the same order of magtude ( the same decmal posto) as the ucertaty Propagato of Ucertates If varous quattes,,, are measured wth small ucertates u( ) [u( )/ < 0] ad the measured values are used to calculate some quatty Y, the the ucertates the varous cause a ucertaty Y Ths ca be estmated by a Taylor expaso to frst order Example : Ucertates Sums ad Dffereces If Y s the sum or dfferece Y ( or), the u( Y) u( ) (maxmum error estmate) u (depedet radom errors) u( Y) ( ) Example : Ucertates Products ad Quotets If Y s the product ad quotet Y uy Y ( ) u ( ) m k k, the k, (maxmum error estmate) k, k, uy Y ( ) u ( ) k, k, k, (depedet radom errors) Example 3: Ucertaty a Power If Y s a power, Y= m, the u( Y) u( ) m Y If Y = Bx, where B s kow exactly, the u( Y) B u( ) Example 4: Ucertaty a Fucto of Oe Varable If Y s a fucto of a sgle varable (measured quatty), Y(), the dy u( Y) u( ) d

Geeral Case: Ucertaty a Fucto of Several Varables If Y s a fucto of several varables (measured quattes),,,, the Y u( Y) max u( ) (maxmum error estmate) Y u( Y) u( ) (depedet radom errors) Y Y u( ) u( ) (always) Remark: I some examples the dfferece u( ) betwee the results of expermetal measuremets ca also be used as a measure of ucertaty (error) Examples for the Maxmum Ucertaty Estmate from Physcs Determato of the desty of a cyldrcal metal rod measured quattes mass of the rod m legth of the rod l radus of the rod R m m Equato to calculate the desty V R l Relatve maxmum ucertaty estmate u ( ) ( ) ( ) ( ) max u m u l u R m l R Determato of the specfc heat capacty c f of a sold by a calorc expermet measured quattes temperature of the hot sold f tal temperature of the flud fl mxg temperature m mass of the flud m fl mass of the sold m f heat capacty of the calormeter C K specfc heat capacty of the flud c fl

( cfl mfl CK)( m fl) Equato to calculate c f cf m ( ) x f f m Maxmum ucertaty estmate ( m fl) ( cfl mfl CK) u( c f ) cfl u( mfl) u( CK) u( mf ) mf ( f m ) mf ( cfl mfl CK) ( m fl) ( f fl) u( fl) u( f ) u( m ) mf ( f m ) ( f m ) ( f m ) Statstcal Aalyss of Radom Ucertates Let us cosder data pots of a measured quatty wth measured values x k, k =,, 3, These values of wll be dstrbuted about ther mea value, whch s some termedate value ot ecessarly cocdet wth ay of the data values The mea value of the data set s wrtte as x ad s defed to be x () Subtractg the mea value from ay data value produces a resdual We ca equally well vew the data set as a collecto of resduals (devatos from the mea), x x x, () ad we see that the mea s chose so that the sum of the resduals s always zero x ( x x) 0 (3) Sce the trasformato s lear the shape of the dstrbuto of data s ot altered The resduals wll have a mea value of zero A secod pece of formato from the data set, besdes the mea, s related to ts spread Ths s gve by a quatty called the stadard devato (s x ) ad ts square, called the varace Data showg a large stadard devato wll have the mea value poorly determed Coversely, f the stadard devato s small, the data cluster closely, ad the mea s well-determed The stadard devato for ay data pot s ofte called ts error, the sese of 'ucertaty' The better the data are determed the more detal they reveal The stadard devato (ad the varace) deped o the scatter of the data about the mea ad from t we fd the ucertaty of the mea The varace of the data set s defed as the mea value of the squares of the resduals : var ( x x) (4) The stadard devato of the sample, s x, s the square root of the varace of the sample,

sx ( x x) (5) Ths root-mea-square result for the stadard devato s tutve; t descrbes the ucertaty of every sgle measuremet pot The mea value, however, s better defed wth a smaller stadard devato Ideed, the stadard devato of the mea value s gve by sx s x x (6) x ( ) ( ) I expermetal practce we adopt for the measuremet ucertaty approxmately twce the stadard devato of the mea: u( ) s x (7) Normal or Gaussa Dstrbuto I addto to the measuremet value ad ts ucertaty oe should specfy the probablty that the true value falls to the terval x u( ) I order to do so we assume that the data are dstrbuted accordg to a ormal dstrbuto (Gaussa fucto) ( x x) px ( ) exp u u (8) Ths fucto s show the fgure Its wdth s gve by the ucertaty u = u() Sce the Gaussa fucto s ormalzed, the probablty that the true value falls the rage x ku( ), where k s a umercal factor, s gve by the area uder the curve wth ths rage Ths leads to the cofdece level of the measuremet, where the cofdece level smply s the probablty for the true value to fall the rage x ku( ) As see from the fgure the cofdece level depeds o k ad s gve by 683%, 955% ad 997% for k =,,3